U.S. patent application number 13/519843 was filed with the patent office on 2013-07-25 for erythropoietic stimulating agent (esa) dosage determination.
This patent application is currently assigned to MAYO FOUNDATION FOR MEDICAL EDUCATION AND RESEARCH. The applicant listed for this patent is David Dingli, Edward J. Gallaher, Craig L. Hocum, James T. McCarthy, James L. Rogers, David P. Steensma. Invention is credited to David Dingli, Edward J. Gallaher, Craig L. Hocum, James T. McCarthy, James L. Rogers, David P. Steensma.
Application Number | 20130191097 13/519843 |
Document ID | / |
Family ID | 43707935 |
Filed Date | 2013-07-25 |
United States Patent
Application |
20130191097 |
Kind Code |
A1 |
Hocum; Craig L. ; et
al. |
July 25, 2013 |
ERYTHROPOIETIC STIMULATING AGENT (ESA) DOSAGE DETERMINATION
Abstract
An Erythropoietic Stimulating Agent (ESA) dosing system/method
determines patient-specific ESA therapies for patients affected by
insufficient hemoglobin production that may benefit from ESA
treatment. The ESA dosing system includes a model that represents a
process by which red blood cells are produced in humans. The model
may include one or more parameters, the values of which are
patient-specific. The model takes into account patient-specific
historical hemoglobin (Hgb) data and corresponding historical ESA
dosage data to estimate the patient-specific values of the model
parameters, and determines a target therapeutic dose of the ESA
that may maintain the patient's Hgb within a target range.
Inventors: |
Hocum; Craig L.; (Winona,
MN) ; McCarthy; James T.; (Rochester, MN) ;
Steensma; David P.; (Wellesley, MA) ; Dingli;
David; (Rochester, MN) ; Rogers; James L.;
(Rochester, MN) ; Gallaher; Edward J.; (Hillsboro,
OR) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Hocum; Craig L.
McCarthy; James T.
Steensma; David P.
Dingli; David
Rogers; James L.
Gallaher; Edward J. |
Winona
Rochester
Wellesley
Rochester
Rochester
Hillsboro |
MN
MN
MA
MN
MN
OR |
US
US
US
US
US
US |
|
|
Assignee: |
MAYO FOUNDATION FOR MEDICAL
EDUCATION AND RESEARCH
Rochester
MN
|
Family ID: |
43707935 |
Appl. No.: |
13/519843 |
Filed: |
January 4, 2011 |
PCT Filed: |
January 4, 2011 |
PCT NO: |
PCT/US11/20120 |
371 Date: |
November 13, 2012 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
61292087 |
Jan 4, 2010 |
|
|
|
Current U.S.
Class: |
703/11 |
Current CPC
Class: |
A61P 7/06 20180101; G16B
5/00 20190201; G16H 50/50 20180101 |
Class at
Publication: |
703/11 |
International
Class: |
G06F 19/12 20060101
G06F019/12 |
Claims
1. A system, comprising: a biophysical simulation engine that
represents a process by which red blood cells are produced in
humans, wherein the biophysical simulation engine includes a
plurality of parameters the values of which are patient-specific;
and a processing unit that receives patient-specific historical
hemoglobin (Hgb) data and corresponding historical erythropoietic
stimulating agent (ESA) dosage data, applies the Hgb data and the
ESA dosage data to the biophysical simulation engine to estimate
the patient-specific values of the plurality of parameters, and
determines a therapeutic dose of the ESA based on the
patient-specific values of the plurality of parameters.
2. The system of claim 1, wherein the processing unit further
identifies, based on the therapeutic dose, an equivalent dosing
regimen titrated to available commercial doses.
3. The system of claim 1, wherein the parameters include one or
more of an Erythroblast Production Rate, a Blast Mortality
Fraction, a Reticulocyte Mortality Fraction, a Hepatic EPO, a RBC
Avg Lifespan, an EC50, and a setup EPO.
4. The system of claim 1, wherein the parameters include one or
more of a Blast Forming Unit Input, a Colony Forming Unit Survival,
a Recticulocyte Survival, an Erythropoietin Receptor Multiplier, a
Red Blood Cell Lifespan, and an Erythropoietin Setup Rate.
5. The system of claim 1, wherein the processing unit applies an
optimization algorithm to estimate the patient-specific values of
the plurality of parameters.
6. The system of claim 1, further comprising a user interface that
displays a graph of the patient-specific historical hemoglobin
(Hgb) data and the corresponding historical erythropoietic
stimulating agent (ESA) dosage data during a descriptive phase.
7. The system of claim 1, wherein the user interface further
displays a graph of patient-specific simulated hemoglobin data and
the therapeutic dose of the ESA during a prescriptive phase.
8. A method comprising: receiving patient-specific historical
hemoglobin (Hgb) and corresponding patient-specific historical
erythropoietic stimulating agent (ESA) dosage data; estimating
patient-specific values for each of a plurality of parameters of a
biophysical simulation model that represents a process by which red
blood cells are produced in humans based on the patient-specific
historical Hgb and corresponding patient-specific historical ESA
dosage data; and determining a therapeutic dose that results in a
predicted Hgb level stabilized within a target range based on the
patient-specific values for each of the plurality of
parameters.
9. The method of claim 8, wherein estimating patient-specific
values for each of the plurality of parameters comprises applying
Monte Carlo methods to estimate the patient specific values.
10. The method of claim 8, further comprising identifying one or
more dosing regimens that deliver the equivalent of the therapeutic
dose.
11. The method of claim 8, further comprising: receiving patient
response data concerning actual Hgb levels of the patient under the
identified dosing regimen; comparing the actual Hgb levels of the
patient under the identified dosing regimen to the predicted,
stabilized Hgb level; identifying variations from the predicted,
stabilized Hgb level; and re-estimating the patient-specific values
for each of the plurality of parameters of the biophysical
simulation model to obtain an updated therapeutic dose based on the
identified variations.
12. The method of claim 8, wherein determining the therapeutic dose
comprises: receiving a target Hgb level; and iteratively applying
the patient-specific values for each of the parameters to a series
of proposed therapeutic dosages until the target Hgb is
obtained.
13. A method comprising: receiving patient-specific historical
hemoglobin (Hgb) and corresponding patient-specific historical
erythropoietic stimulating agent (ESA) dosage data obtained during
a descriptive period; estimating patient-specific values for each
of a plurality of parameters of a model that represents a process
by which red blood cells are produced in humans based on the
patient-specific historical Hgb and corresponding patient-specific
historical ESA dosage data; simulating patient-specific Hgb values
for a prescriptive period based on the patient-specific values; and
identifying a therapeutic dose of the ESA that maintains the
patient-specific simulated Hgb values in a target range during the
prescriptive phase.
14. The system of claim 1, wherein the parameters include one or
more of a Blast Forming Unit Input, a Colony Forming Unit Survival,
a Recticulocyte Survival, an Erythropoietin Receptor Multiplier, a
Red Blood Cell Lifespan, and an Erythropoietin Setup Rate.
15. A method comprising: receiving patient-specific historical
hemoglobin (Hgb) and corresponding patient-specific historical
erythropoietic stimulating agent (ESA) dosage data; optimizing
patient-specific values for each of a plurality of parameters of a
model that represents a process by which red blood cells are
produced in humans to determine a best fit with the
patient-specific historical Hgb and corresponding patient-specific
historical ESA dosage data; and determining a therapeutic dose of
the ESA that maintains a patient-specific simulated Hgb values in a
target range.
Description
TECHNICAL FIELD
[0001] The disclosure relates to modeling of biophysical parameters
to determine pharmaceutical dosages.
BACKGROUND
[0002] Anemia causes an increased sense of fatigue, decreased
stamina and exercise tolerance, fatigue, shortness of breath,
decreased appetite, and decreased CNS functioning. Anemia can lead
to the need for red blood cell (RBC) transfusions, with associated
risks including bacterial and viral infections, volume overload,
iron overload, and a variety of transfusion reactions.
[0003] Chronic Kidney Disease (CKD) and End Stage Renal Disease
(ESRD) patients are at risk for anemia since RBC homeostasis
requires normal kidney function. The kidneys play a critical role
in erythropoiesis. Erythropoietic Stimulating Agents (ESAs) are
used among these patients as a pharmacological replacement for the
hormone erythropoietin (EPO), produced primarily by healthy
kidneys, and to a small extent by the liver (Hepatic EPO). Other
patient populations, including cancer patients, may also experience
reduced levels of hemoglobin and may benefit from ESA therapy.
[0004] Hemoglobin (Hgb) values are a primary indicator of anemia.
The Centers for Medicare & Medicaid Services (CMS) and National
Kidney Foundation (NKF) have established the target range for Hgb
values among ESRD patients to be between 10 g/dL and 12 g/dL. Hgb
values below the desired minimum lead to an increased sense of
fatigue and decreased stamina and are considered to be a risk
factor for increased cardiovascular morbidity and mortality in ESRD
patients. Patients with Hgb values under 10 g/dL suffer from the
effects of anemia, including fatigue and reduced stamina and
exercise tolerance, shortness of breath, decreased appetite and
decreased CNS functioning, and reduced compliance. Anemia can lead
to the need for red blood cell (RBC) transfusions, with associated
risks including bacterial and viral infections, volume overload,
iron overload, and a variety of transfusion reactions.
[0005] Hgb values above 12.0 g/dL are believed to create an
increased risk of cardiovascular events such as stroke and
myocardial infarction, cerbrovascular and cardiovascular mortality
and morbidity. Patients with Hgb values over 12 g/dL are at risk of
thrombosis, vascular access clotting (compromising effective
dialysis therapy), hypertension, and increased risk of acute
coronary syndromes or cerbrovascular accidents. These observations
have led to the development of regulatory and quality standards
which lead practitioners to try and maintain the hemoglobin values
of ESRD patients within the narrow range of 10.0-12.0 g/dL.
[0006] In ESRD patients, as well as in other patient populations
experiencing reduced hemoglobin levels, the biophysical system that
regulates erythropoietin production does not function properly.
ESAs are often prescribed to manage hemoglobin levels (anemia) in
ESRD patients and in other patient populations. An ESA prescription
may include, for example, intravenous injection of darbepoetin alfa
(Aranesp.RTM.) or Recombinant Human Erythropoietin (rHuEPO). The
current protocol for developing ESA prescriptions produces patterns
of hemoglobin (Hgb) oscillation that subject patients to a cycle of
overshoot and undershoot of target Hgb values. For example, when
the patient exhibits a low Hgb, the dosage may be dramatically
increased in an attempt to quickly raise Hgb levels. When the
patient exhibits a high Hgb, interruption of ESA therapy (by
greatly reducing the dose or withholding administration) may lead
to under-dosing of the ESA, which, in turn, leads to an undershoot
of Hgb values. The result is an undesirable fluctuation of Hgb
levels above and below the target range. The period of the
High-Low-High may take up to nine months for a complete cycle. Hgb
values are often measured monthly, rendering Hgb cycling
practically imperceptible.
[0007] In addition to the effects of low or high Hgb values upon
ESRD patients, there are considerable administrative and financial
impacts upon a dialysis facility if Hgb values are not maintained
within the desired range. For example, the current protocol
requires an ESA prescription to be developed one time per month per
patient. Due to the cyclic variation in both Hgb levels and ESA
dosage, considerable personnel time is used to review and adjust
ESA dosage.
[0008] The current protocol also cannot project future actual ESA
requirements, leading to difficulties with ESA inventory
management. As a result of this uncertainty, dialysis facilities
will often maintain large ESA inventories. However, because ESAs
are relatively expensive, maintenance of large unused ESA
inventories may not be financially optimal. In addition, Medicare
and/or other insurance providers may impose penalties when patient
Hgb levels exceed 12.0 g/dL for varying periods of time. These
denials may occur retrospectively; that is, after the ESA has
already been administered and the cost has been incurred by the
dialysis facility.
[0009] In addition, patient Hgb values are often monitored monthly
by regulatory agencies. A systematic pattern of high Hgb values can
cause sanctions to be applied, which may include the creation of
monitored compliance plans or even closure of a dialysis facility
until a plan to achieve compliance is approved.
SUMMARY
[0010] In general, the disclosure describes system(s) and/or
method(s) for determining Erythropoietic Stimulating Agent (ESA)
dosing.
[0011] In one example, the disclosure is directed to a system
comprising a biophysical simulation engine that represents a
process by which red blood cells are produced in humans, wherein
the biophysical simulation engine includes a plurality of
parameters the values of which are patient-specific and a
processing unit that receives patient-specific historical
hemoglobin (Hgb) data and corresponding erythropoietic stimulating
agent (ESA) dosage data, applies the Hgb data and the ESA dosage
data to the biophysical simulation engine to estimate the
patient-specific values of the plurality of parameters, and
determines a target therapeutic dose of the ESA based on the
patient-specific values of the plurality of parameters. The system
may further identify, based on the target therapeutic dose, an
equivalent dosing regimen titrated to available commercial doses.
The parameters include one or more of an Erythroblast Production
Rate, a Blast Mortality Fraction, a Reticulocyte Mortality
Fraction, a Hepatic EPO, a RBC Average Lifespan, an EC50, and a
setup EPO. Alternatively, the parameters include one or more of a
Blast Forming Unit Input, a Colony Forming Unit Survival, a
Recticulocyte Survival, an Erythropoietin Receptor Multiplier, a
Red Blood Cell Lifespan, and an Erythropoietin Setup Rate. The
processing unit may apply an optimization algorithm to estimate the
patient-specific values of the plurality of parameters.
[0012] In another example, the disclosure is directed to a method
comprising receiving patient-specific historical hemoglobin (Hgb)
and corresponding patient-specific historical erythropoietic
stimulating agent (ESA) dosage data, estimating patient-specific
values for each of a plurality of parameters of a biophysical
simulation model that represents a process by which red blood cells
are produced in humans based on the patient-specific historical Hgb
and corresponding patient-specific historical ESA dosage data, and
determining a therapeutic dose that results in a predicted Hgb
level stabilized within a target range based on the
patient-specific values for each of the plurality of
parameters.
[0013] In another example, the disclosure is directed to a method
comprising receiving patient-specific historical hemoglobin (Hgb)
and corresponding patient-specific historical erythropoietic
stimulating agent (ESA) dosage data obtained during a descriptive
period, estimating patient-specific values for each of a plurality
of parameters of a biophysical simulation model that represents a
process by which red blood cells are produced in humans based on
the patient-specific historical Hgb and corresponding
patient-specific historical ESA dosage data, determining
patient-specific simulated Hgb values for a prescriptive period
based on the patient-specific values, and identifying a target
therapeutic dose of the ESA that maintains the patient-specific
simulated Hgb values in a target range during the prescriptive
phase.
[0014] The details of one or more examples are set forth in the
accompanying drawings and the description below. Other features and
advantages will be apparent from the description and drawings, and
from the claims.
BRIEF DESCRIPTION OF DRAWINGS
[0015] FIG. 1 is a block diagram illustrating an example system
that determines a weekly therapeutic dose of an ESA that will
result in stabilization of a patient's Hgb to a target level.
[0016] FIG. 2 is a chart of historical Hgb levels and ESA dosage
over time for a patient under the existing ESA dosage protocol.
[0017] FIG. 3 is a chart of Hgb levels and ESA dosage over time for
the patient in FIG. 2 under the weekly therapeutic dosage.
[0018] FIG. 4 is a diagram illustrating four example building
blocks of a commercially available dynamic modeling
application.
[0019] FIG. 5 is a diagram illustrating the core model of the
biophysical simulation engine.
[0020] FIG. 6 is a diagram illustrating the configuration of five
of seven parameters used in the biophysical simulation engine.
[0021] FIG. 7 is a diagram illustrating the configuration of two of
seven parameters used in the biophysical simulation model: hepatic
EPO and setup EPO rate.
[0022] FIG. 8 is a diagram containing a screenshot of a portion of
the biophysical simulation model user interface that controls the
use of model parameters.
[0023] FIG. 9 is a screenshot of the portion of the biophysical
simulation model that controls a Partial Monte Carlo
simulation.
[0024] FIG. 10 is a screenshot that is used to specify patient
specific hemorrhages used optionally in the example biophysical
simulation model.
[0025] FIG. 11 is a graph illustrating the relationship of a
specific patient's weekly therapeutic dose and the resultant steady
state Hgb value.
[0026] FIG. 12 is a composite figure, consisting of a graph
illustrating the relationship of the weekly therapeutic dose to a
specific dosing regimen, together with a screenshot of the device
that is used to specify dosing regimens.
[0027] FIG. 13 is diagram illustrating an example process by which
the ESA dosing techniques described herein achieve and maintains
stable Hgb levels for ESRD patients receiving ESA therapy.
[0028] FIG. 14 is a screenshot of a behavior over time chart that
displays selected variables.
[0029] FIG. 15 is a graph illustrating an example curve fitting
result for the descriptive phase.
[0030] FIG. 16 is a graph illustrating an example weekly
therapeutic dose (WTD) calculation result.
[0031] FIGS. 17A and 17B are flowcharts illustrating example
processes by which a processor determines a weekly therapeutic dose
(WTD) that will result in stabilization of Hgb to a target level
and monitors the patient response.
[0032] FIG. 18 is a flowchart illustrating an example process by
which individual patient parameters may be determined.
[0033] FIG. 19 is a flowchart illustrating an example process by
which a weekly therapeutic dose (WTD) may be determined.
[0034] FIG. 20 is a block diagram of another example ESA dosing
system.
[0035] FIG. 21 illustrates an example diagram illustrating part of
the data acquisition/management component of the ESA dosing
system.
[0036] FIGS. 22 and 23 are diagrams illustrating a setup for an
example Monte Carlo simulation that determines the best fit
patient-specific parameter values for the patient's historical
hemoglobin data.
[0037] FIG. 24 is a diagram representing an example calculation of
a mean square error (MSE) for one run of the Monte Carlo
simulation.
[0038] FIG. 25 is a diagram representing an example amount of
Aranesp administered at a prescribed interval based on a
prescription regimen equivalent to a simulated therapeutic
dose.
[0039] FIG. 26 is a diagram representing one possible example set
of variables which may be used to define a recommended Aranesp
prescription regimen.
[0040] FIG. 27 is a diagram representing an example determination
of the circulating Aranesp concentration.
[0041] FIG. 28 is a diagram representing an example amount of
Epogen administered at a prescribed interval based on a recommended
prescription regimen equivalent to a simulated therapeutic
dose.
[0042] FIG. 29 is a diagram representing one possible set of
variables which may be used to define an Epogen prescription
regimen.
[0043] FIG. 30 is a diagram representing an example determination
of the circulating Epogen concentration.
[0044] FIG. 31 is a diagram illustrating an example EPOR
(erythropoietin receptor) binding for Epogen and Aranesp.
[0045] FIG. 32 is a diagram illustrating an example model of
reticulocyte production in bone marrow.
[0046] FIG. 33 is a diagram is an example model that simulates the
total number of red blood cells in circulation.
[0047] FIG. 34 is an example user interface through which a user
may interact with and/or control various aspects of the ESA dosing
system.
[0048] FIG. 35 is an example graph displaying historical Hgb
levels, historical ESA dosages, and simulated Hgb levels for the
pre-descriptive setup period, the descriptive period, and the
prescriptive period of a patient.
DETAILED DESCRIPTION
[0049] The disclosure generally relates to systems and/or methods
that design patient-specific Erythropoietic Stimulating Agent (ESA)
dosing regimens. The ESA dosing system and/or methods described
herein may result in determination of patient-specific ESA dosing
that achieves and sustains adequate Hgb values for patients
receiving ESA therapy.
[0050] The ESA dosing techniques described herein may be used to
determine patient-specific ESA dosing for any available ESA
therapy. These ESAs may include, but are not limited to,
Erythropoietin; Epoetin alpha (Procrit.RTM., Epogen.RTM.,
Eprex.RTM.); Epoetin beta; darbepoetin alpha (Aranesp.RTM.);
Methoxy polyethylene glycol-epoetin beta; Dynepo; Shanpoeitin;
Zyrop; Betapoietin; and others.
[0051] In addition, the ESA dosing techniques described herein may
also be applicable to a wide variety of patient populations,
including, for example, End Stage Renal Disease (ESRD) patients,
Chronic Kidney Disease (CKD) patients, cancer therapy patients, or
any other patient population having insufficient hemoglobin
production that may benefit from ESA treatment such as anemia
secondary to HIV infection. In addition, the ESA dosing techniques
described herein may also be applicable to multiple modes of ESA
therapy delivery, including intravenous (IV) delivery, subcutaneous
delivery, oral delivery, biopump, implantable drug delivery
devices, etc.
[0052] In recent years there has been much controversy regarding
optimal Hgb target values in ESRD patients, as well as controversy
over the impact of swings in Hgb values. Based on currently
available clinical data, there is widespread agreement that the
target value for Hgb should be somewhere between 10 and 12
grams/deciliter (g/dL), and that the probable desired optimal range
is 11.0-12.0 g/dL. There is also growing agreement that stable Hgb
values are more conducive to patient well-being than are wide
oscillations in Hgb values. The system described herein enables
care providers to identify dosing regimens that will establish and
maintain Hgb values in the target range for the majority of their
patients.
[0053] The system includes a patient-specific biophysical
simulation model that, based on a patient's historical response to
ESA therapies, determine a target dosing level which can be
translated to a dosing regimen titrated to available commercial
doses. The dosing regimen thus obtained can be configured to
simultaneously achieve and sustain adequate and stable Hgb values
for extended periods of time as well as minimize or eliminate Hgb
oscillations (commonly known as Hgb cycling). The total amount (and
cost) of ESA administered may also be reduced or minimized. If the
patient's overall medical condition remains stable, Hgb values have
been shown, using the techniques described herein, to remain stable
at a given target level. If the patient's underlying medical
condition changes, the system includes a diagnostic system'which
can be used to establish a new target dosing level that may restore
Hgb values to a desired target level in a minimum of time.
[0054] In some examples, the system/method creates a recommended
intravenous (IV) ESA dosing regimen including a dose level and dose
administration frequency. Care providers can tailor the frequency
of ESA administration enabling effective and efficient use of
supporting staff time.
[0055] FIG. 1 is a block diagram illustrating an example system 10
that determines a weekly therapeutic dose of an ESA that will
result in stabilization of a patient's Hgb to a target level.
System 10 includes a processing unit 20 and an assortment of data
processing and management tools. For example, system 10 includes a
biophysical simulation engine 24 that predicts red blood cell (RBC)
production (Hgb is contained within RBCs, so RBC production and
hemoglobin production are used interchangeably in this document),
ESA prescriptive tools 26, patient data management tools 28,
outcome tracking tools 30, reporting tools 32, and change
management tools 34 to maintain adequate and stable Hgb values
through adjustments to the indicated therapy. A user interface 22
permits one or more users to input patient historical data (either
manually or electronically), run the tools and view and manipulate
the results.
[0056] The purpose of system 10 is to help care providers develop
ESA dosing strategies that avoid creating the oscillations in Hgb
values for patients that are characteristically created by existing
protocols, and that provide stabilized Hgb levels within a target
Hgb range.
[0057] Patients with ESRD have a deficiency of the hormone
erythropoietin (the endogenous ESA), and, as a result, they are
severely anemic. Anemia (hemoglobin<10.0 g/dL) is a risk factor
for mortality in ESRD patients, and patients with anemia have
poorer quality of life than non-anemic patients. Patients receiving
an ESA also have an increased risk of cardiovascular events
(stroke, myocardial infarction) if their hemoglobin rises above
12.0 g/dL. These observations have led to the development of
regulatory and quality standards which lead practitioners to try
and maintain the hemoglobin values of ESRD patients within the
range of 10.0-12.0 g/dL. In addition, other patient populations may
also receive ESA therapy, including CKD patients, cancer therapy
patients, and other patients who would benefit from ESA therapy,
and it shall be understood that ESA dosing system 10 may also be
applicable to these and other patient populations. Thus, although
some portions of this description may refer specifically to ESRD or
CKD patients, it shall be understood that ESA dosing system 10 and
the techniques implemented therein may also be applicable to other
patient populations.
[0058] Patient-specific responses to ESA therapy are dependent upon
a variety of factors, including total body iron storage status,
extracellular volume fluid status, inflammation, residual kidney
function, hemorrhage, and variations in the dose effectiveness of
ESA among ESRD patients.
[0059] The majority of ESRD patients who need ESA therapy are
currently receiving one of two formulations: rHuEPO, FDA approved
for the treatment of anemia in patients with chronic renal failure
in 1985, or darbepoetin alfa (Aranesp.RTM.), FDA approved in 2001.
rHuEPO has an average half life of five to seven hours, requiring
frequent administration. darbepoetin alfa was designed to have a
longer half-life of 25-27 hours. The example simulation engine was
designed to monitor patient response to darbepoetin alfa. However,
adjustments to the simulation engine may allow for similar
simulations to be conducted for patients receiving o rHuEPO, in
transition from rHuEPO to darbepoetin alfa, or other ESA
therapies.
[0060] Due its longer half life, darbepoetin alfa requires
approximately five days for complete elimination. This allows
providers to administer the drug less frequently. But the extended
half-life of darbepoetin alfa, in combination with red blood cell
dynamics, contributes to a confounding physiological consequence.
After an administration of darbepoetin alfa, RBC production is
enhanced for up to 26 days. This delay, if not factored into the
design of the prescription, sets up Hgb oscillation. It is not
uncommon for patients to experience 12-18 months of Hgb "overshoot"
and "undershoot" as providers try to reestablish an adequate and
stable Hgb level following existing protocols.
[0061] FIG. 2 is a graph illustrating an example of an actual
oscillating Hgb pattern for an ESRD patient that was generated by
following existing ESA protocol for the period May 2007 through
December 2008. FIG. 2 illustrates a delayed response between ESA
dosage (represented by curve 52) and the measured Hgb value
(represented by curve 50). The delayed response makes it difficult
to identify the appropriate dose when the provider considers only
the most recent Hgb values. In addition, as mentioned above,
response to ESA therapy is highly patient specific and cannot be
generalized to a larger population. This case is a typical pattern
observed among ESRD patients on dialysis receiving darbepoetin
alfa.
[0062] The system uses an operational approach that includes all
the factors that generate patient Hgb values. The system and
associated processes and models described herein may help providers
design ESA therapies that eliminate Hgb oscillations and achieve
adequate and stable Hgb values within target levels.
[0063] FIG. 3 is a chart of Hgb levels and ESA dosage over time for
the ESRD patient of FIG. 2 both during a "descriptive period" and a
"prescriptive period." In FIG. 3, the period between Mar. 1, 2008
and Jan. 11, 2009 is defined as the "descriptive period." The
descriptive period includes historical data for a specific ESRD
patient that includes monitored actual Hgb levels (curve 50) and
ESA dosage over time (curve 52). The data collected during the
descriptive period is used in a biophysical simulation that
calculates the values for seven (in this example) patient-specific
parameters that define an ESRD patient's response to ESA therapy.
The biophysical simulation calculates these patient-specific
parameters such that simulated descriptive Hgb values (represented
by curve 54) match historical actual Hgb values during the
descriptive period with a specified error. The patient-specific
parameter values are then used as the basis for an ESA prescription
for a "prescriptive period." The prescriptive period is the chosen
time period for a simulation during which a recommended
prescription (the therapeutic dose, which may be a per session
therapeutic dose or a weekly therapeutic dose (WTD) depending,
among other things, upon the particular ESA to be prescribed) will
be designed, with the intent of determining a dosing regimen that
will stabilize Hgb values within a target range. In this example,
the prescriptive period is January 2009 through June 2009. FIG. 3
illustrates simulated prescriptive Hgb values (represented by curve
56) obtained using an optimized ESA dosage for the patient of FIG.
2. FIG. 3 illustrates that the optimized ESA dosage created an
adequate and stable Hgb level for the prescriptive period within
the target range.
[0064] Due to its longer half life, darbepoetin alfa requires
approximately five days for complete elimination from the serum,
and has a prolonged period of pharmacological activity. This allows
providers to administer the drug less frequently. But the extended
half-life of darbepoetin alfa, in combination with red blood cell
dynamics, creates a physiological consequence. After an
administration of darbepoetin alfa, RBC production is enhanced for
up to 26 days. This delay, if not factored into the design of the
prescription, sets up Hgb cycling. It is not uncommon for patients
to experience 12-18 months of Hgb "overshoot" and "undershoot" as
providers try to establish an adequate and stable Hgb level
following existing protocols. The system accounts for feedback and
delay in the erythropoietic process by establishing a target dosing
level, assisting with the design of a dosing regimen, and
monitoring results over time.
[0065] In current practice it is often desirable to reduce the
frequency of administration in order to capture reduced
administrative costs. When a patient is switched from bi-weekly to
monthly dosing, for example, current practice is to double the dose
and then seek the optimal regimen using current protocols. This
introduces a potential round of Hgb cycling. However, it has been
found using the presently described biosimulation techniques that
reducing the frequency of administration by a factor of two
requires far more than twice the previous dose. If such a decrease
in frequency is otherwise desirable, the system permits the
identification of the required dose to sustain adequate Hgb values.
In other examples, the system may determine that increasing the
frequency of administration with optimal doses may, in spite of
increased administrative costs, reduce total cost due to
significantly reduced amount of the drug required.
[0066] The presently described biosimulation techniques utilize
dynamic modeling. Dynamic modeling is a framework, consisting of a
language and a set of concepts. These are embedded in a process for
representing, understanding, explaining and improving how dynamic
systems (erythropoiesis, for example) work, how they perform over
time, and how they respond to inputs (such as ESA
administration).
[0067] There are several commercial packages available to build and
simulate dynamic models, including iThink.RTM., available from Isee
Systems, Inc.; Stella.RTM., available from Isee Systems, Inc.;
Vensim.RTM., available from Ventana Systems, Inc.; Powersim Studio
8, available from Powersim Software AS; Berkeley Madonna.TM.,
developed by Robert Macey and George Oster of the University of
California at Berkeley; and other commercially available software
packages. The example described herein was implemented using iThink
version 9.3 and the examples provided herein are described using
iThink syntax and conventions. It shall be understood, however,
that the specific implementation described herein is one example of
how the biosimulation model may be implemented, and that equivalent
models may be constructed in each of the aforementioned commercial
packages, in other commercially available packages, in customized
software packages, or in application specific software programs
and/or systems, and that the disclosure is not limited in this
respect.
[0068] The model manages the dynamic linkage that exists between
the pharmacokinetics and pharmacodynamics of the ESA in question
with the dynamics of the RBC chain. Further, the model may be
embedded in a data processing system that enables effective ESRD
anemia management both at the individual and group level.
[0069] FIG. 4 shows the four elements of the syntax used in the
selected commercial package (iThink, in this example). A Stock
(301) represents an accumulation at a point in time, such as total
RBC count. A Flow (302) represents rates of flow over time. FIG. 4
contains two flows, an inflow and an outflow, such as RBC's created
per day and RBC's destroyed per day, respectively. The values of
stocks and flows are evaluated at each point in time in a
simulation using user-supplied mathematical expressions. A
Converter (303) represents and contains a mathematical expression
that may be as simple as a constant value or as complex as an
aggregate of a generalized subsystem. Connectors (304) indicate
relationships between variables in the model, both graphically and
mathematically. Cloud icons (305) represent boundaries of the
model. When all the required mathematical relationships with a
model's design are described, the behavior of the modeled system
may be simulated for a period of time. This simulation is performed
by calculating the current state of the system from the beginning
of the simulation time period to the end, stepwise and
incrementally, using a selected time increment, delta t, referred
to as DT. By observing the dynamic behavior of various variables
(RBC counts and Hgb values, for example) model users are able to
confirm or refine their understanding of how erythropoiesis
operates and create successive model improvements until the
simulated behavior effectively matches known data.
[0070] FIG. 5 is a diagram illustrating an example of a biophysical
simulation model. The model simulates the relationship between the
concentration of darbepoetin alfa and Hgb values for individual
patients over time. In other examples, the model may simulate the
relationship between the concentration of other ESAs and Hgb
values.
[0071] FIG. 6 is a diagram illustrating the configuration of five
of seven parameters used in the biophysical simulation model:
erythroblast production in ten millions (609), baseline blast
mortality fraction (610), baseline reticulocyte mortality fraction
(611), EC50 (607), and avg lifetime (606). Note that FIG. 6,
concerning the baseline blast mortality fraction (610), provides a
more detailed description than FIG. 5 provides in the description
of blast mortality fraction (404). On the other hand, in FIG. 6,
the icon Aranesp Concentration (612), represents all of the detail
shown in FIG. 5, Aranesp Dosing and Pharmacokinetics (412). The
purpose of FIG. 6 is to illustrate the configurations of the five
parameters listed above in this example of the biophysical
simulation model.
[0072] The process of constructing the example model was to consult
with subject matter experts to learn how selected variables are
related and then to translate those relationships to a specific
model in the chosen syntax. Various decisions are made in the model
building process concerning levels of aggregation/disaggregation
required to achieve the model's purpose. In other examples, levels
of aggregation/disaggregation may be modified to achieve the same
purpose, while exposing differing biophysical behaviors over
time.
[0073] The model building process includes decisions about
boundaries of the model that are consistent with the model purpose.
This is referred to as establishing the extensity of the model.
Other examples of the model may include revisions to the model's
extensity as described below.
[0074] The specific example of the model described with respect to
FIG. 5 includes the following boundaries: dynamics of progenitor
cells in the marrow are excluded; the impact of eliminated RBC's is
excluded; tissue oxygenation is excluded; iron metabolism is
excluded; the model is aimed at simulating Hgb response profiles of
iron replete patients; and plasma fluids are excluded. However, it
shall be understood that in other examples of the model, such as
the model shown and described with respect to FIGS. 20-35, one or
more of these factors could be taken into account. For example,
other examples of the model could include plasma fluids and
simulation of Hematocrit values, a measure often used instead of
Hgb values. Any or all of these boundaries, or related boundaries,
may be included while maintaining the fundamental purpose of
simulating an individual patient's response to ESA therapy. In
addition, the model may also include simulation of Hgb response
profiles in patients that are not iron replete.
[0075] Typically in scientific studies of the factors that relate
to a so-called dependent variable, one performs various studies of
correlation, analysis of variance, principal components, etc.
Parameters of a Dynamic Model, however, are identified and used
differently than in statistical studies. Once the boundary of a
dynamic model is defined, the parameters describe exogenous inputs
to the model. The parameters of a dynamic model describe
operational variables (as, in general, do endogenous model
variables as well) in that they describe causal factors of the
behavior being simulated. As an example, the parameter Erythroblast
Production Rate (EPR) is one parameter to the biosimulation model.
This parameter describes the rate at which erythroblasts are
created per day. The value of this parameter is not merely
correlated to the RBC count, but, all other parameters equal, a
given value for EPR will cause a certain number of RBC's to exist.
Note that parameters may be constants or complex mathematical
expressions, representing aggregates of external subsystems. Note
that parameters selected and defined for a dynamic model depend
upon the definition of the model's boundary.
[0076] The system develops a targeted dosing plan, and then
anticipates changes in the patient's response to the ESA therapy,
which allows for an equally targeted response that reduces or
avoids Hgb oscillations. The result may be that more ESA therapy
patients have hemoglobin values maintained within the desired range
of 10.0-12.0 g/dL, or within 11.0-12.0 g/dL.
[0077] The modeling process takes into account not only endogenous
physiological factors that regulate red blood cell (hemoglobin)
values, but also the patient in order to achieve and maintain an
adequate, stable Hgb level. This model is unique for each
individual patient, and this allows for inclusion of
patient-specific components of anemia management.
[0078] Table 1 lists the seven patient-specific parameters utilized
in this example, provides a definition of each, and sets forth
example default minimums and maximums used in this example of the
model. Alternative examples may use a different parameter set, yet
still describe the dynamics of Hgb response to various dosing
regimens.
TABLE-US-00001 TABLE 1 Default Minimum Default Maximum Name
Description Search Value Search Value Erythroblast Production The
rate at which patient produces 60 90 Rate erythroblasts per day
(.times.10.sup.6) Blast Mortality Daily mortality of erythroblasts
60% 90% Fraction Reticulocyte Mortality Daily mortality of
reticulocytes 40% 60% Fraction Hepatic EPO Endogenous
erythropoietin created 0 0 by the liver. Assumed to be zero for
most ESRD patients, but may be a factor for some. May also be used
to model Endogenous EPO produced through residual kidney function.
RBC Avg Lifespan Average number of days for patient's 50 100 RBC's.
EC50 Represents a patient's resistance to 15 25 ESA therapy A
measure of patient sensitivity to EPO therapy. High EPO resistance
indicates low sensitivity to therapy. Setup EPO A mathematical
construct used to 1 5 initialize the simulation model for "day 1"
of the simulation. A mathematical EPO dose applied to a simulation
model during model initialization for the purpose of stabilizing
simulated Hgb to the value of the first historical Hgb in the
patient's descriptive period.
[0079] The Erythroblast Production Rate (EPR) represents the number
of erythroblasts created per day within the bone marrow, but
outside the boundary of this example of the model. FIG. 5 shows
that EPR (403) is an inflow to the conveyor that accumulates
maturing erythroblasts (401). In this example, the value of EPR is
held fixed. Alternative examples may include time varying values
for EPR. Still other examples may subsume the parameter, rendering
it an endogenous model variable that may be a constant, a complex
mathematical expression or a variable that is dependent on other
model variables or parameters. These comments concerning
alternative examples of the model apply equally well to each of the
parameters described below. Typical ranges for EPR values were
obtained, for example, from subject matter experts in
hematology.
[0080] The leakage flow named blast mortality (602 in FIG. 6)
represents the fraction of maturing erythroblasts that go through
programmed cell death each day. Typical ranges for this fraction,
as well as for the other parameters described below, were obtained
from subject matter experts in hematology. A common misconception
in the art is that ESAs enhance the creation of erythroblasts in
the marrow, or as in FIG. 6, erythroblast production in ten
millions (609). Although ESAs do stimulate erythroblast production,
the Applicants have identified that a relatively more significant
effect of ESAs is to inhibit blast mortality (602) (as well as
other factors described below) which allows a larger fraction of
maturing erythroblasts to survive, thus creating more RBCs, all
other factors being equal. In this example of the model,
darbepoetin alfa stimulates RBC production by increasing the
survival rate of precursor cells. FIG. 6 shows specifically how
this example of the model operates in this regard. The exposed
detail FIG. 6 provides (relative to FIG. 5) the baseline blast
mortality fraction (610) as the actual model parameter. Blast
Mortality Fraction (601) is a value that is determined by the value
of baseline blast mortality fraction (610) as moderated by the
variable Aranesp fractional effect (605). In an alternative example
of the model, Aranesp fractional effect (605) may be named (and
appropriately mathematically revised) ESA fractional effect, thus
representing different types of ESAs that operate in the same
manner.
[0081] The Reticulocyte Mortality Fraction (RMF) is similar in
effect to the BMF. In this example of the model, darbepoetin alfa
stimulates RBC production by increasing the survival rate of
reticulocytes in the marrow. FIG. 6 shows specifically how this
aspect of the example model operates. The parameter is more
correctly named the Baseline Reticulocyte Mortality Fraction (603).
The RMF (604) is determined by the combination of the Baseline
Reticulocyte Mortality Fraction (603) as moderated by the variable
named Aranesp Fractional Effect (605), which in alternative
examples of the model may be named ESA Fractional Effect, thus
applying to different types of ESAs that operate in the same
manner. Note that in this example of the model, the variable
Aranesp Fractional Effect (605) is the same for both BMF and RMF.
In other examples, different values for Aranesp Fractional Effect
relative the BMF and RMF may be used.
[0082] Hepatic EPO is shown in FIG. 7 (701). Hepatic EPO is a form
of endogenous epoetin produced by the liver. As explained below,
the parameter Hepatic EPO is one of several inputs to the serum
concentration of the ESAs. Originally envisioned as an exogenous
model parameter, experience with this example of the model has
confirmed what is clear from medical literature: the impact of
Hepatic EPO is insignificant relative to the impact of administered
darbepoetin alfa. In this example, therefore, the Hepatic EPO
parameter is uniformly fixed to zero. Other examples of the model
may include non-zero values for Hepatic EPO.
[0083] RBC Average Lifespan is included as a parameter for this
example of the model. It is known that, while in healthy
individuals, the RBC Lifespan is about 120 days, for ESRD patients
on dialysis, the RBC Lifespan is shorter in duration. RBC Average
Lifespan is named Average Lifetime (606) in FIG. 6. Alternative
examples of the model may represent the lifespan of RBC's
differently, allowing various RBC mortality rates for RBC's of
different vintages.
[0084] In practice, the EC50 of an agent is the concentration that
produces a response half way between the baseline and maximum
response for a given time period. Usually a measure of potency,
this parameter is used differently in this example of the model as
a measure of what is known in the field as EPO resistance. Various
medical conditions such as inflammation, infection, and the
presence of ESA antibodies can decrease an individual patient's
response the ESA therapy. This example of the model uses a single
measure of EPO resistance. Alternative examples may use separate
values for each cause of EPO resistance which could potentially
produce an improved simulation. FIG. 6 (bone marrow) shows how EC50
(607) is configured in the model. The mathematical expression used
to evaluate the Aranesp Fractional Effect (605) includes factors
related to EC50 (607).
[0085] As explained above, the RBC chain is represented in this
example of the model as an array of 12 so-called bins of RBC cells.
In the initial phase of a simulation, named the Setup Phase, a
steady state RBC count is established in each of the 12 bins,
corresponding to the initial actual Hgb value for an individual
patient in the second, Descriptive Phase of the simulation. (The
Descriptive Phase is described below.) To establish a steady state
RBC count, this example of the model is provided with what is
mathematically equivalent to an externally administered dose. FIG.
6 shows how the parameter setup EPO rate (702) is configured in
this example of the model to achieve this result. As described
below, setup EPO rate (702) is in effect during the Setup Phase of
a simulation; its value diminishes to zero during the subsequent
Descriptive and Prescriptive Phases of a simulation.
[0086] The user of the model, or the software system into which the
model is integrated, develops an initial estimate of
patient-specific parameter values, as described below. Initial
estimates of parameter values may be manually adjusted using the
interface to the biophysical simulation engine. FIG. 8 depicts that
part of the user interface which enables manual manipulation of
parameter values in this example. The interface is constructed so
that a given simulation may use either initial estimates or
manually revised estimates of parameter values. In FIG. 8, the
tilde (310) is in the off position, indicating that the value 73.8
(311) is to be applied in the simulation as a substitute for the
initial parameter value estimate. The expression "eqn on" (312),
indicates that the initial estimate is to be used for the
corresponding parameter value.
[0087] A dynamic model, as expressed using the selected
commercially available simulation package or any of the others
listed above, is defined by its degree of aggregation among
selected variables, model boundaries, exogenous parameter values,
time period to be simulated, and time increment (DT) to use for the
simulation. A model so expressed simulates proposed causal
relationships among its elements, as distinct from correlated
relationships. As such, a model so defined represents a theory of
dynamic behavior of a system that can be tested in a laboratory,
confirmed, and refined. The dynamic modeling process often includes
simulation and testing of a proposed dynamic hypothesis using a
specific model, testing and validation of the dynamic hypothesis,
followed by revisions to any of the model elements to improve model
performance. Dynamic modeling is an iterative process in which the
dynamics of the system are represented, understood, and explained
in order to improve the simulation of the dynamic system under
study, in this case erythropoiesis for iron replete ESRD patients
on dialysis receiving darbepoetin alfa. The scope of the claims
presented below shall include all the iterates of the example
models described herein and those which may evolve in future
examples.
[0088] Referring again to the example core model shown in FIG. 5,
elements of the model as a whole that are not part of the core
model are generally elements that inform, control, and report on
values of elements that are within the core model. The core model,
when supplied with appropriate patient specific parameter values
and a dosing regimen (409) simulates an individual patient's
response in terms of Total RBC Count (410). Total RBC count is then
converted to a Hgb value. In an alternative example of the model,
the contribution of peripheral reticulocytes (411) may be included
in the calculation of Hgb values. Extension of the model boundary
to include plasma fluids may enable reporting on hematocrit values
in addition to Hgb.
[0089] FIG. 5 contains two syntax items not previously described,
namely Conveyors (401, for example) and a Stock Array (402).
Conveyors are specialized stocks that have an inflow and up to two
outflows. Conveyors follow a First in First Out Rule in which
quantities that flow in to the conveyor exit the conveyor in the
same order as they entered, after a specified conveyor transit
time. The outflow and contents of a conveyor can also be modified
by a second optional outflow, named a Leakage flow. The rate of
flow through a leakage flow is specified as a fraction of the
inflow at each time increment of DT. A Stock Array (402), as
implemented in this example, is a sequence of 12 stocks in which
the outflow of the first stock in the sequence is the inflow to the
second, and so on. The Core Model in FIG. 5 represents RBC's as
twelve sequential stocks. The first stock in the array represents
RBC's that are one day old to a value equal to the Time Constant
(408) divided by 12. Successive stocks in the array represent RBC's
at correspondingly older vintages, as determined by the time
constant (408).
[0090] Table 2 lists the correspondence between the seven
patient-specific parameters listed in Table 1 and the variables
shown in FIG. 5.
TABLE-US-00002 TABLE 2 Generalized Variable Name Name Description
in Core Model Erythroblast The rate at which patient produces
Erythroblast Production erythroblasts per day (.times.10.sup.6)
Production Rate Blast Daily mortality of erythroblasts Blast
Mortality Mortality Fraction Fraction Reticulocyte Daily mortality
of reticulocytes Retic Mortality Mortality Fraction Fraction
Hepatic EPO Endogenous erythropoietin cre- Release of Hepatic ated
by the liver. Assumed to be EPO zero for most ESRD patients, but
may be a factor for some. May also be used to model Endoge- nous
EPO produced through residual kidney function. RBC Avg Average
lifespan of a patient's Time Constant Lifespan RBC's (number of
days). EC50 Represents a patient's resistance Sensitivity to to ESA
therapy. A measure of Aranesp patient sensitivity to EPO therapy.
High EPO resistance indicates low sensitivity to therapy. Setup EPO
A mathematical construct used to Setup EPO rate initialize the
simulation model for "day 1" of the simulation. A mathematical EPO
dose applied to a simulation model during model initialization for
the purpose of stabilizing simulated Hgb to the value of the first
historical Hgb in the patient's descriptive period.
[0091] The following are illustrative equations for the example
model shown in FIGS. 5-7, as expressed in the syntax of a
commercially available modeling application (iThink.RTM., available
from Isee Systems, Inc., in this example). The equations describe
the relationships between model variables for a specific patient.
Also shown are definitions for core model variables, some of which
are not shown in FIG. 5. Although an example implementation using
iThink.RTM. is shown, it shall be understood that the ESA dosing
techniques described herein may also be implemented using other
commercially available or customized software applications.
TABLE-US-00003 Aranesp_in_circulation(t) = Aranesp_in_circulation(t
- dt) + (EPO_input - Aranesp_degradation) + dt INIT
Aranesp_in_circulation = 1 DOCUMENT: Aranesp concentration at time
t. INFLOWS: EPO_input = setup_EPO_input + hepatic_EPO + ( IF time +
DOSE_A_START THEN CLINICAL_ARANESP_DOSE ELSE Aranesp_Pulse )
DOCUMENT: The IV dose to be administered, either in the descriptive
period when parameters are being sought, or in the prescriptive
period when either the weekly therapeutic dose or a proposed dose
is being proposed. OUTFLOWS: Aranesp_degradation =
Aranesp_in_circulation*Aranesp_halflife DOCUMENT: Elimination of
the drug at time t. marrow_retics(t) = marrow_retics(t - dt) +
(nuclear_exclusion - retic_mortality - retic_release) + dt INIT
marrow_retics = 7e8 TRANSIT TIME = 2 INFLOW LIMIT = INF CAPACITY =
INF DOCUMENT: Count of marrow reticulocytes at time t. INFLOWS:
nuclear_exclusion = CONVEYOR OUTFLOW DOCUMENT: Rate of surviving
erythroblasts at time t. OUTFLOWS: retic_mortality = LEAKAGE
OUTFLOW LEAKAGE FRACTION = retic_mortality_fraction NO-LEAK ZONE =
0 DOCUMENT: Rate of reticulocyte mortality at time t. retic_release
= CONVEYOR OUTFLOW DOCUMENT: Rate of surviving marrow reticulocytes
at time t. maturing_erythro_blosts(t) = maturing_erythro_blasts(t -
dt) + (erythroblast_production - nuclear_exclusion -
blast_mortality) * dt INIT maturing_erythro_blasts = 5e9 TRANSIT
TIME = 15 INFLOW LIMIT = INF CAPACITY = INF DOCUMENT: Count of
erythroblast cells at time t. INFLOWS: erythroblast_production =
erythroblast_production_in_ten_millions * 1e7 DOCUMENT: A main
model parameter, estimating overall erythroblast production rate
for the duration of the simulation. This value is found using the
partial monte carlo simulation. OUTFLOWS: nuclear_exclusion =
CONVEYOR OUTFLOW DOCUMENT: Rate of surviving erythroblasts at time
t. blast_mortality = LEAKAGE OUTFLOW LEAKAGE FRACTION =
blast_mortality_fraction NO-LEAK ZONE = 0 DOCUMENT: Rate of
erythroblast mortality at time t. peripheral_retics(t) =
peripheral_retics(t - dt) + (retic_release - retic_maturation -
periph_retic_mortality) * dt INIT peripheral_retics = 6e8 TRANSIT
TIME = 2 INFLOW LIMIT = INF CAPACITY = INF DOCUMENT: Count of
peripheral reticulocytes at time t. INFLOWS: retic_release =
CONVEYOR OUTFLOW DOCUMENT: Rate of surviving marrow reticulocytes
at time t. OUTFLOWS: retic_maturation = CONVEYOR OUTFLOW DOCUMENT:
Rate of maturing reticulocytes at time t. periph_retic_mortality =
LEAKAGE OUTFLOW LEAKAGE FRACTION = If HEM >0 Then HEM Else 0
NO-LEAK ZONE = 0 DOCUMENT: Rate of reticulocyte mortality at time t
given by the sum of all user supplied hemorrhages.
RBC_population[Delay_Chain_D12](t) =
RBC_population[Delay_Chain_D12](t - dt) +
(mature_RBC_input[Delay_Chain_D12] - RBC_loss[Delay_Chain_D12]) *
dt INIT RBC_population[Delay_Chain_D12] = initial_Hgb*1e9/12
DOCUMENT: An aging chain of RBC's, implemented in this package as
an array of 12 bins, each bin being 10 days in duration. The
maximum assumed lifetime of on RBC is assumed to be 120 days.
INFLOWS: mature_RBC_input[1] = retic_maturation +
0*RBC_population[1]/time_constant DOCUMENT: Rate of maturing
reticulocytes flowing in to bin 1 (1 to 10 days old) at time t.
mature_RBC_input[2] = 0*retic_maturation + RBC_population[1] /
time_constant DOCUMENT: Rate of RBC's flowing from bin 1 to bin 2
(RBC's that are 11 to 20 days old) at time t. mature_RBC_input[3] =
0*retic_maturation + RBC_population[2] / time_constant DOCUMENT:
Rate of RBC's flowing from bin 2 to bin 3 (RBC's that are 21 to 30
days old) at time t. mature_RBC_input[4] = 0*retic_maturation +
RBC_population[3] / time_constant DOCUMENT: Rate of RBC's flowing
from bin 3 to bin 4 (RBC's that are 31 to 40 days old) at time t.
mature_RBC_input[5] = 0*retic_maturation + RBC_population[4] /
time_constant DOCUMENT: Rate of RBC's flowing from bin 4 to bin 5
(RBC's that are 41 to 50 days old) at time t. mature_RBC_input[6] =
0*retic_maturation + RBC_population[5] / time_constant DOCUMENT:
Rate of RBC's flowing from bin 5 to bin 6 (RBC's that are 51 to 60
days old) at time t. mature_RBC_input[7] = 0*retic_maturation +
RBC_population[6] / time_constant DOCUMENT: Rate of RBC's flowing
from bin 6 to bin 7 (RBC's that are 61 to 70 days old) at time t.
mature_RBC_input[8] = 0*retic_maturation + RBC_population[7] /
time_constant DOCUMENT: Rate of RBC's flowing from bin 7 to bin 8
(RBC's that are 71 to 80 days old) at time t. mature_RBC_input[9] =
0*retic_maturation + RBC_population[8] / time_constant DOCUMENT:
Rate of RBC's flowing from bin 8 to bin 9 (RBC's that are 81 to 90
days old) at time t. mature_RBC_input[10] = 0*retic_maturation +
RBC_population[9] / time_constant DOCUMENT: Rate of RBC's flowing
from bin 9 to bin 10 (RBC's that are 91 to 100 days old) at time t.
mature_RBC_input[11] = 0*retic_maturation + RBC_population[10] /
time_constant DOCUMENT: Rate of RBC's flowing from bin 10 to bin 11
(RBC's that are 101 to 110 days old) at time t.
mature_RBC_input[12] = 0*retic_maturation + RBC_population[11] /
time_constant DOCUMENT: Rate of RBC's flowing from bin 11 to bin 12
(RBC's that are 111 to 120 days old) at time t. OUTFLOWS:
RBC_loss[Delay_Chain_D12] = ( RBC_population[Delay_Chain_D12] /
time_constant ) + RBC_population[Delay_Chain_D12] * HEM DOCUMENT:
Rate of RBC death from each bin in the RBC chain plus losses due to
user supplied hemorrhages at time t. .largecircle. Aranesp_halflife
= .9 DOCUMENT: Drug elimination half life, measured in days.
.largecircle. ARANESP_Volume_of_Distribution = 1 DOCUMENT:
Effective volume into which drug is distributed, thus providing its
observed concentration. .largecircle. Aranesp_effect_on_mortality =
Aranesp_in_circulation/(Aranesp_in_circulation+EC50) DOCUMENT:
Moderates erythroblast and reticulocyte mortality rates, based upon
drug concentration and epo resistance at time t. .largecircle.
Aranesp_concentration_in_circulation =
Aranesp_in_circulation/ARANESP_Volume_of_Distribution .largecircle.
blast_mortality_fraction = IF (Aranesp_effect_on_mortality>
.001) THEN
(baseline_blast_mortality_fraction*(1-Aranesp_effect_on_mortality))
ELSE (baseline_blast_mortality_fraction) DOCUMENT: Value of
baseline erythroblast mortality fraction as moderated by drug
fractional effect. .largecircle. CLINICAL_ARANESP_DOSE = IF
((mod(time.1) = .5) AND (CLINICAL_ARANESP_DATA > 1)) THEN (Pulse
(CLINICAL_ARANESP_DATA,time,99999)) ELSE 0 DOCUMENT: Historical
drug dose administered at time t during the descriptive period of a
simulation. .largecircle. Release_of_hepatic_EPO = 1 .largecircle.
retic_mortality_fraction = IF (Aranesp_effect_on_mortality>.01)
THEN
(baseline_reticulocyte_mortality_fraction*(1-Aranesp_effect_on_mortality)-
) ELSE (baseline_reticulocyte_mortality_fraction) DOCUMENT: Value
of baseline reticulocyte mortality fraction as moderated by drug
fractional effect. .largecircle. sensitivity_to_Aranesp = { Place
right hand side of equation here... } .largecircle. time_constant =
Avg_Lifetime/12 DOCUMENT: Evenly allocates Average Lifetime to the
respective durations in each successive RBC bin in the RBC aging
chain. .largecircle. Total_RBC_Count = ARRAYSUM(RBC_population[*])
DOCUMENT: Cumulative number of RBC cells in the 12 bins of the RBC
chain of the model.
[0092] The simulation performed using the equations described above
results in a numerical approximation of the solution to a set of
differential equations that describe accumulations in the chosen
stocks (which represent integrals) as determined by their
respective inflows and outflows (which represent derivatives).
Specifically, the user of this model supplies, for an individual
patient, historical Hgb values, historical darbepoetin alfa doses,
the time period to be simulated, and the time increment, DT. The
simulation, embedded in a simplified optimization routine
(described below) then enables the user to determine a target
dosing level and an associated dosing regimen that will obtain the
desired Hgb values as long as the patient's current medical
condition remains relatively unchanged.
[0093] In one example, the biophysical simulation model may employ
an adaptation of the Monte Carlo method to estimate parameter
values. However, it shall be understood that other non-linear
optimization routines may also be used, and that the disclosure is
not limited in this respect. FIG. 9 presents the structure (501)
that generates a collection of parameter values associated with
simulation runs from which a best fit in the collection may be
chosen by external (to the model) processing. The user, or the
software system into which the model is integrated, specifies the
number of simulation runs by providing a value for the converter
named Simulation Number (502). The converter named Partial Monte
Carlo Switch (503) is an on-off switch that controls the mode of
the simulator: single simulation or multiple simulation. The Monte
Carlo Switch is replicated (508) for each of the model parameters,
informing the respective control converters (506) which parameter
values are to be used in a given simulation: either the values in
the CALC converters (507) in the case of a single simulation, or
values in the in the respective stocks (505), which is the case
when the Partial Monte Carlo Switch (503) is in the on position.
The collection of parameter values that is generated by a Partial
Monte Carlo simulation may be exported to a commercially available
spreadsheet software which may be used to select the set of
parameter values which produces the best fit between simulated Hgb
values and the patient's historical values for the descriptive
phase of the simulation, described below. Selected parameter values
may then be reimported to the model for further processing and use
in the Prescriptive Phase of the simulation, described below.
Alternatively, a fully automated software system may perform these
tasks. In that event, the data need not be exported to an external
software application. Alternative examples of the Monte Carlo
structure within the model (501), the method of best fit selection,
and the movement of data exported from and imported into the model
may be performed by a variety of methods, including a fully
automated software system, all achieving the same purpose: to
identify values for these, or other, parameter values and then make
use of the parameters to find the weekly therapeutic dose
(described below) which leads to the desired dosing regimen.
[0094] In this example, 100 or fewer simulations of the Monte Carlo
method may be to choose an optimum value. However, in alternative
examples of the model, thousands or tens of thousands of
simulations might be run in a reasonable amount of time, which may
allow for a more complete assessment of the distributions of each
of the seven parameters. Further, this example of the model may not
provide a unique solution. The same proposed dosing regimen might
be developed for one patient with a low EPR, BMF, and RMF as for a
patient with a high EPR, BMF, and RMF. Alternative examples of the
model may allow potential classification of patients of the first
or second type. In practice, however, proposed dosing regimens,
though non-unique, may be quite adequate, resulting in 60% to 90%
or more of the patients at a DCF achieving and sustaining Hgb
values within the target range.
[0095] Each simulation is executed in three phases: Setup,
Descriptive, and Prescriptive. Each phase is defined over a
specific number of days. As described below, the Setup Phase
extends from Day -200 to Day 1, the Descriptive Phase extends from
the day number associated with a patient's first historical Hgb
value (chosen by an analyst or chosen automatically by an automated
software system) to the most recently available Hgb value or
administered darbepoetin alfa dose. The Prescriptive Phase extends
from the simulation day number of the first potential dosing date
(generally one week after the end of the Descriptive Phase) to a
simulation day number at which a proposed dosing regimen, given the
selected parameter values, produces stabilized simulated Hgb values
at the desired target value. The extent of each phase is identical
for the two modes of simulation: the Monte Carlo Mode that
generates a collection of random parameter values and associated
simulated Hgb values, and the single simulation mode, in which
proposed dosing regimens are developed or revised.
[0096] For the individual patient, the Descriptive Phase is used to
select a set of parameter values that produces simulated Hgb values
that match the historically observed Hgb values in response to
historical drug doses over the duration of the Descriptive Phase.
There will be a combined set of parameter values (e.g. the baseline
progenitor input, baseline progenitor mortality in the absence of
darbepoetin alfa, the EC50 of darbepoetin alfa, the level of
protection provided, and the lifespan of circulating RBCs). This
phase serves to define the `pathophysiological state` of the
patient, and the patient's sensitivity to the drug.
[0097] The setup phase begins at `-200 days`, i.e. prior to `Day
1`, the day at which historical data is available (the beginning of
the Descriptive Phase). The purpose of the setup phase is to
identify a set of parameters that bring the system into a steady
state (flat-line Hgb level equal to the patient's first Hgb value)
prior to Day 1, and then simultaneously enable the system to follow
the patient's response to darbepoetin alfa during the Descriptive
Phase. The parameter Setup EPO (see Table 1) is used primarily in
the Setup Phase to represent a mathematical dose of darbepoetin
alfa, which, together with other parameter values in play at during
the Setup Phase, achieves the results described immediately above.
The parameter Setup EPO has no effect in subsequent phases but is
rather replaced by either historical doses (in the Descriptive
Phase) or proposed doses (in the Prescriptive Phase).
[0098] Given an appropriate set of parameter estimates, simulated
Hgb values will respond to historical doses during the Descriptive
Phase by generating simulated Hgb values that approximate the
waxing and waning of historical Hgb values. The Partial Monte Carlo
method, complemented by additional manual adjustments and/or
automated adjustments, if required, is used to identify the best
fit described above. Further, the best fit is defined as the
simulated Hgb values within the extent of the Descriptive Phase,
selected from a collection of simulations, which have a mean square
error with respect to actual Hgb values in the Descriptive Phase of
approximately 0.25 g/dL.
[0099] During the Prescriptive Phase, one or more of the following
steps may be performed with fixed parameter values. A therapeutic
dose (which may be a per session therapeutic dose or a weekly
therapeutic dose (WTD) depending, among other things, upon the
particular ESA to be prescribed) may be determined. In addition, in
the event of a WTD, an analyst or automated system may introduce
`sample` dosage regimens, searching for a dosing regimen (dose and
frequency) that delivers the equivalent of the WTD at a minimum
cost. Frequently dose "pulses" may be required to quickly elevate
Hgb values or avoid a projected undershoot. The dosage regimen is
refined to bring the patient quickly and smoothly within the target
Hgb range and to sustain that value. The selected dosing regimen
may be extended several months into the future and remains
effective as long as the patient's underlying medical condition
remains relatively stable. In alternative examples, the search for
a WTD and the selected dosing regimen may be implemented using
automated software algorithms.
[0100] As described above, in this example of the biophysical
simulation model, parameter values are optimized across the Setup
and Descriptive phases using non-linear optimization methods (such
as Monte Carlo techniques). However, it shall be understood that
the present disclosure is not limited in this respect. Alternative
examples may include, for example, other optimization strategies
such as simplex algorithms and maximum likelihood estimators or
other non-linear computational algorithms known to those of skill
in the art.
[0101] The pharmacokinetic (PK) section of the model (FIG. 7,
"Aranesp Dosing and Pharmokinetics") simulates the dynamics of
circulating drug concentrations over time in response to various
types (mathematical, historical, proposed) of dosing regimens and a
simulated elimination rate. Alternative examples of the model may
contain more extensive or refined PK representations. The
pharmacodynamic (PD) section simulates the concentration-response
influence of darbepoetin alfa on the time course and magnitude of
the RBC count and release into the circulation. Once RBCs enter the
circulation the clinician has no influence over their lifespan.
Effective therapy is dependent upon an awareness of two critical
delays within the process of erythropoiesis. First, the immediate
effect of darbepoetin alfa is to increase (predictably delayed) Hgb
values by replication and maturation processes within the bone
marrow. Second, the decrease of Hgb levels is (predictably) delayed
as a result of the persistent lifespan of circulating RBCs.
[0102] It is recognized that additional pathophysiological
subsystems (and comorbid conditions) may influence Hgb levels over
time. Alternative examples of the model may include, for example,
iron availability and treatment regimens, bleeding, and the
influence of inflammation on EPO resistance.
[0103] Other examples of the model may include additional parameter
refinements, supporting disaggregation of biophysical subsystems,
when these variations improve fulfillment of the purpose of the
model.
[0104] ESRD patients on dialysis frequently suffer blood loss for
various reasons, such as bleeding from the access point to the
patient's bloodstream or gastrointestinal bleeding. Other patient
populations may also experience hemorrhages for various reasons.
This example of the model includes the ability to specify up to
four periods during which the patient experiences a hemorrhage.
FIG. 6 shows how the variable HEM (608) is configured in the model.
Note from this figure that hemorrhages are applied to RBC counts
and peripheral reticulocytes in circulation. A portion of the user
interface shown in FIG. 10 shows the control device used to specify
hemorrhages. Alternative tools will have different but equivalent
representations. In the illustrative example shown in FIG. 10,
there are two active hemorrhages, A and B, specified by (801) and
(802), having values set to the value 1, i.e., (803) and (807). The
magnitudes of these two hemorrhages are specified as 2% per day and
4% per day, indicated by the values aligned with Hem A Magnitude
and Hem B Magnitude (804) and (807), respectively. Items (805),
(806), (809) and (810) specify the start and stop days for the two
hemorrhages. Note that hemorrhages A and B overlap between days 170
and 192, in which case, the cumulative effect is used by the model.
A negative magnitude may also be specified to simulate the effect
of blood transfusions which are frequently administered to ESRD
patients on dialysis, for example.
[0105] Although it is possible to identify individual factors that
influence Hgb values over time, it is the interaction between these
factors that influence the time course and magnitude of Hgb values
in response to darbepoetin alfa and other ESAs. The present ESA
dosing system includes a mathematical model that provides
definitions of individual components, and then allows the
controller to conduct simulations that reliably predict the
behavior of the system as a result of perturbations that may be
systematically introduced by an analyst or automatically by an
automated software system.
[0106] The `Weekly Therapeutic Dose` (WTD) is a theoretical value
that defines the weekly dose that will ultimately maintain the
patient's Hgb values at the midpoint of the target range. WTD is
determined by systematically varying a fixed weekly dose and
observing Hgb concentrations during the prescriptive phase of the
simulation. The therapeutic dose determined by the ESA dosing
system may be a per session therapeutic dose (PSTD) or a weekly
therapeutic dose (WTD) depending, among other things, upon the
particular ESA to be prescribed. In the case of Aranesp.RTM., the
system determines a WTD and then determines a dosing regimen that
will deliver the equivalent of the WTD. In the case of other ESAs,
the therapeutic dose determined by the system may be equivalent to
the actual dosing regimen recommended for the patient.
[0107] FIG. 11 depicts three responses in simulated Hgb values,
using parameter values obtained previously during the Prescriptive
Phase, derived from historical darbepoetin alfa doses and actual
Hgb values, to applied WTD's of 8, 12, and 16 mcg of darbepoetin
alfa respectively. (901), (902), and (903) are the Setup,
Descriptive, and Prescriptive Phases of the simulation,
respectively. The WTD is applied in the model during the
Prescriptive Phase only. The portion of the curves (904), (905),
and (906) that are within the Prescriptive Phase of the simulation
are simulated Hgb values, stabilized at 11.5, 12.3, and 13.0 g/dL
respectively in response to the three WTD's described. This example
of the model provides the user with projections of future Hgb
values in response to various WTD's and associated dosing
regimens.
[0108] The WTD derived in the Prescriptive Phase of the simulation
is the clinician's guide to developing a dosing regimen. A dosing
level and the frequency at which to administer the chosen doses may
then be chosen from commercially available doses. For some ESAs,
the WTD may be unequal to commercially available doses (in the case
of Aranesp.RTM., for example), and this may require a mix of doses
be applied at various frequencies that together will deliver a dose
equivalent to the WTD. Further, on the date the dosing regimen is
to be started, the patient may currently be either above or below
the target range, with either an upward or downward trend in Hgb
values. In such cases "pulse" doses must be found that will quickly
and smoothly achieve Hgb values within the target range. FIG. 12
displays a complete scenario:
[0109] (1001) denotes the Setup Phase
[0110] (1002) denotes the Descriptive Phase
[0111] (1003) denotes the Prescriptive Phase
[0112] (1004) portrays simulated Hgb values in each phase
[0113] (1005) reports historical ESA doses administered in the
Descriptive Phase
[0114] (1006) reports historical Hgb values measured during the
Descriptive Phase
[0115] (1007) reports three pulse doses to initiate the
Prescriptive Phase, designed to arrest the concurrent downward
trend in Hgb values
[0116] (1008) reports the proposed dosing regimen that will sustain
Hgb values at 11.5 g/dL
[0117] (1009) is the simulation day number of the first dose in the
three dose pulse (1007)
[0118] (1010) is the amount of the dose to be administered as Dose
A
[0119] (1011) is the interval in days for Dose A
[0120] (1012) is the end date for Dose A
[0121] (1013)-(1016) are the analogues of Dose A specifications, to
be applied as Dose B
[0122] (1017)-(1020) are the analogues of Dose A specifications, to
be applied as Dose C
[0123] Thus, in this case, the steady state dosing regimen is found
to be: "Starting on day 125, give three weekly doses of 25 mcg,
followed by alternating doses of 25 mcg and 40 mcg every 21
days".
[0124] One concept of dynamic modeling in the example system
described herein is that recommended strategies (i.e., dosing
regimens) are hypotheses as opposed to "black box answers". These
hypotheses are to be tested by follow up measurements of actual
future Hgb values of the patient and either confirmed or rejected.
Both confirmation and rejection of a hypothesis provide insight and
understanding of how the process under study actually operates,
which is a major goal to be achieved from a dynamic modeling
perspective, generally and particularly. To that end, the system
may also include, for example, components and tools for follow up,
analysis, learning, revision, improved anemia management skills,
and ultimately the well being of the patient.
[0125] The example system described is a clinically applicable set
of tools designed to address and resolve Hgb cycling. The example
system includes of a collection of components that have been
loosely coupled by means of various software components.
Alternative examples of the system may include tightly integrated
modules of functionality, providing an information system that
supports anemia management for individual patients receiving
therapy at one or more Dialysis Care Facilities (DCF).
[0126] The purpose of the system is to capture, cleanse, maintain,
transform, analyze, and create data and information required by
clinicians to effectively manage anemia concerns for a population
of individual patients.
[0127] FIG. 13 is a diagram illustrating an example of the overall
process by which the system achieves and maintains stable Hgb
levels for patients receiving ESA therapy. The first aspect of the
process includes gathering and maintenance of historical hemoglobin
(Hgb) data and ESA dosing data for a particular patient. Historical
measured actual Hgb levels and corresponding ESA dose history,
along with identifying information and other relevant information
(such as hospitalizations, iron studies, transfusions, infections,
and other factors that may affect Hgb levels) for all ESRD patients
at, for example, a DCF or group of DCFs, may be obtained and stored
in a database or other medium for storage and retrieval of the
information. Behavior over time (BOT) charts (such as those shown
in FIGS. 2 and 3) are generated that assist the analyst in each of
the three phases of a simulation.
[0128] The first step in treating a population of patients is to
select applicable patients. In one example, the system is designed
to treat iron replete patients who have a minimum of 6 recorded Hgb
values. Patient data described above is assembled and organized for
processing and maintenance. In other examples, the system may
include iron metabolism components, potentially enabling the
inclusion of patients who were not iron replete during the period
in which Hgb values were obtained.
[0129] Obtaining and maintaining individualized simulation
parameters, whether those described above, or refinements and
parameter improvements, are stored in a database. Alternative
examples may include effective classifications of Hgb response
profiles, potentially creating improved methods of treatment.
[0130] Individualized recommended prescriptions are stored in a
database which permits overall analysis of ESA consumption and
improved management of associated costs. In this example of the
system, the database is implemented in a commercially available
spreadsheet program. Alternative examples may include
implementation of customized database application using
commercially available database engines, data transformation tools,
analysis and reporting.
[0131] Recommended prescriptions are reviewed and approved by
authorized providers. The example system provides recommend dosing
regimens. In addition, clinicians' anemia management skills may be
improved through use of the system.
[0132] Approved prescriptions may be entered into a provider's
order management system. The ESA dosing system may be loosely
coupled with a medical order management system. Alternative
examples may include software components that tightly integrate
each step in the process from a data and information perspective.
Alternative examples may also provide improved ESA consumption
management which may significantly reduce a variety of operational
costs, such as reduced drug consumption, reduced carrying
inventories, reduced spoilage, reduced administrative costs, and
reduced administrative costs associated with preventable emergent
medical issues.
[0133] The ESA dosing system may also include data collection tools
designed to monitor the compliance of drug administration with
medical orders. Hgb cycling often arises from dose
misadministration. Using these tools, the ESA dosing system may
detect dose misadministration and prompt as well as design
corrective interventions.
[0134] As described above, the recommended dosing regimen, once
approved and ordered, is a hypothesis awaiting confirmation or
rejection, each of which improves insight. The ESA dosing system
may include, for example, weekly monitoring of Hgb values for a
minimum of 12 weeks, allowing the clinician to detect and diagnose
the causes of observed deviations of actual Hgb values from those
that were predicted by the simulation.
[0135] The BOT chart is a tool used by clinicians in the
implementation of the methodology supported by the system. FIG. 14
shows an example of an individual's BOT chart that tells the story
of anemia management effectiveness. FIG. 14 contains the following
information for a 10 month period: [0136] 1101 administered iron
(Venofer.RTM.) [0137] 1102 Mean Corpuscular Volume (MCV) [0138]
1103 simulated Hgb values for a portion of the Descriptive period
[0139] 1104 actual Hgb values [0140] 1105 projected Hgb values in
response to planned therapy [0141] 1106 hospitalized 7 days for
pneumonia [0142] 1107 transferrin saturation and iron values
(resting upon one another, but with different scales) [0143] 1108
boxes represent recommended and approved dosing regimen [0144] 1109
dots within boxes represent actual aranesp doses administered
[0145] 1110 open box represents recommended and approved dose
either missed or not yet administered [0146] 1111 upper bound of
target Hgb range (13.0 g/dL) [0147] 1112 upper bound of optimal
target Hgb range (12.0 g/dL) [0148] 1113 lower bound of optimal
target Hgb range (11.0 g/dL) [0149] 1114 lower bound of target Hgb
range (10.0 g/dL)
[0150] A given patient's BOT, containing these and other data (such
as the time course of vital signs) enables the clinician to develop
a comprehensive picture of the patient's overall condition. Axis
labels and scales are not shown in FIG. 14 for brevity. The example
system includes an underlying database of patient results and a web
based report (the BOT), along with various filters that quickly
isolate patients with Hgb values deviating from expected values.
The example system enabled s one physician assistant to monitor the
status of 370 patients and recommend interventions in a four hour
period.
[0151] The process may also include a structured change control
process. The ESA dosing system may be designed to anticipate
changes or replacements to any or all of the system components. For
example, the ESA dosing system may anticipate changes in an
individual patient's underlying medical condition. These changes
may require resimulation of a new recommended prescription, or
searching for a new set of parameter values and then developing a
new recommended prescription. The ESA dosing system may include
tools whereby an analyst can retrieve previously modeled patients
and begin anew. Alternative examples may include information system
components that maintain histories of identified parameter values,
prescriptions, and changes over time of the patient's medical
condition. This information may improve insights and an operational
understanding of the relationship between the progression of CKD
for ESRD patients on dialysis and Hgb response profiles.
[0152] FIG. 15 is an example screenshot of the simulation engine
control panel for the pre-descriptive setup period and the
descriptive period. This is an example screenshot that could be
displayed on user interface 22 (FIG. 1). A control panel 152 allows
the user to set minimum and maximum search values for each of the
patient-specific parameters. A graphing area 154 graphically
displays historical Hgb levels, historical ESA dosages and
simulated Hgb levels for the pre-descriptive setup and descriptive
periods. A patient ID 156, Count Clinical Hgb 157 and the mean
standard error (MSE) for the currently displayed simulation are
also displayed. A series of function buttons 160 permit the user to
run, pause, resume, stop, restore graphs and tables, and/or perform
other relevant functions related to the biophysical simulation.
Although specific data, graphs, and functional interfaces are shown
in FIG. 6, it shall be understood that the disclosure is not
limited in this respect, and that other relevant data, graphs,
tables, charts or other ways of displaying data may also be
displayed, and that other types of functional interfaces, such as
touch screen, mouse, stylus, keyboard, multi-touch, mobile devices,
or other method of interacting with the program may be used without
departing from the scope of the present disclosure.
[0153] The graph 154 of FIG. 15 illustrates an example curve
fitting result for the descriptive phase. In this example, the
descriptive period for this patient was 371 days in duration.
During that period, 18 actual Hgb values were measured, and those
values display the typical oscillation. 37 doses of darbepoetin
alfa were administered in the descriptive phase. When Hgb values
were too high, darbepoetin alfa was withheld. When Hgb values were
too low, darbepoetin alfa doses were increased.
[0154] The model uses a so called pre-descriptive period to
establish an erythropoietic equilibrium with an RBC count, which
reflects the Hgb level that is near the first observed Hgb result
in the descriptive period. In this example, the pre-descriptive
period in the model is 201 days in duration, running from day -200
to day 0. This is the period of time the body requires to establish
equilibrium in the presence of a theoretical (mathematically
applied) daily ESA dose. The model uses the parameter values
displayed on the left of the FIG. 15 to simulate an Hgb value from
Day -200 to Day 371.
[0155] The search for parameter values stops when the Mean Square
Error (MSE) between the simulated Hgb values and observed Hgb
values in the descriptive period is sufficiently small. In FIG. 15,
the MSE is reported as 0.22, meaning that on average, the simulated
Hgb values are within +/-0.47 g/dL of the actual Hgb values.
[0156] The simulation based approach solves the problem of
overshoot and undershoot by associating the post administration
exponential decay in ESA concentration levels with the delays
involved in red blood cell production. The model accounts for the
production and mortality of RBC precursor cells. By providing
estimates of RBCs "in the pipeline", the provider can extract
advice from the biophysical simulation engine that will create a
dosing plan that will achieve an adequate and stable Hgb value
within a target range.
[0157] As discussed above, an output of the process is the
calculation of the weekly therapeutic dose (WTD) required for a
stable Hgb at the target value or within a target range. The WTD is
the theoretical weekly dose required to achieve the intended
therapeutic response. That is, a dose level which, if administered
weekly, would stabilize a patient's Hgb at the target level.
[0158] FIG. 16 is a graph illustrating an example WTD calculation
result for the same patient shown in FIG. 15. This graph could be
displayed in the graphical portion 154 of the control panel 150 of
FIG. 15. In this example, the descriptive period was 370 days in
duration (day 1 to day 371). The prescriptive period is extended to
700 days from day 0 (or a total of 330 days). Typically, this
should be ample time for the RBC production chain to stabilize in
response to a proposed constant WTD. 18 actual Hgb values were
collected in the descriptive period. 37 doses of darbepoetin alfa
were administered in the descriptive phase. When Hgb values were
too high, darbepoetin alfa was withheld. When Hgb values were too
low, darbepoetin alfa doses were increased. These results display a
typical oscillation, with values well below and well above the
target range of 10-12 g/dL.
[0159] The simulation of the prescriptive phase of the patient
included a weekly dose titrated to deliver the equivalent of 25 mcg
of darbepoetin alfa per week. As shown by the simulated Hgb levels
for the prescriptive period, Hgb would stabilize at 11.5 g/dL after
a little over 60 days.
[0160] Once the system determines the optimized WTD, the system may
assist providers in finding the most effective combinations of
available dosing levels at the optimal frequency of administration
that will deliver the required WTD, and as a result, achieve and
maintain the desired Hgb value. This may include, for example,
titrating available dosing levels that will deliver the equivalent
of the WTD. The example illustrated in FIG. 16 indicates a WTD of
25 mcg. Since 25 mcg is a standard available unit doses, this is
the prescription the system may recommend.
[0161] For other patient simulations, WTD values that are not equal
to available unit doses require experimentation to determine the
optimal dosing strategy. For these patients, the process may guide
the provider to a titration scheme that achieves the intended
result.
[0162] After a proposed prescription is approved and administration
has begun, the patient's situation will invariably change. A
hospitalization, an infection that increases resistance to ESA
therapy, or a hemorrhage may occur that changes the patient's
response to ESA therapy.
[0163] In accordance with another aspect of the example ESA dosing
system, it has been determined that Hgb measurements taken in the
prescriptive period that differ from the projected (simulated)
response are reliable indicators that the value of at least one of
the patient-specific parameters has changed. Identification of
changes in observed Hgb levels from the projected response may be
used to prompt focused assessments about changes in the patient's
condition that may lead to effective corrections.
[0164] For example, variance in Hgb levels from the projected
response may be related to a condition or situation that was not
present during the descriptive phase. Re-modeling may be used at
this point to seek an alternative set of patient-specific
parameters values. The newly updated parameters may then be used to
yield an effective corrective action, that is, an updated WTD, to
restore an adequate and stable Hgb value.
[0165] Monitoring of Hgb levels during the prescriptive period may
therefore be a part of the systemic solution. This part of the
process is a probe that scans for changes in the patient's
condition, develops corrective actions, and communicates the
required changes in a timely and effective manner.
[0166] FIG. 17A is a flowchart illustrating an example process 200
by which system 10 (FIG. 1) or system 1240 (FIG. 20) may determine
a therapeutic dose that will result in stabilization of Hgb to a
target level or keep it within a target range. Historical Hgb and
corresponding ESA dosage data is received (202). Patient-specific
parameters are estimated (204). In one example, Monte Carlo methods
such as those described herein, or other non-linear optimization
routines, may be used to arrive at an approximation of model
parameters for individual patients. The parameters may be manually
or automatically adjusted to improve fit between historical and
simulated Hgb values for the descriptive period chosen.
[0167] The system determines a therapeutic dose that results in
stabilized Hgb within the target range (206). In the event that the
therapeutic dose is a WTD, the system may identify one or more
dosing regimens that deliver the equivalent of the WTD (208). The
one or more equivalent dosing regimens may be developed for a 30,
60, 90 or other day regimen, up to six months, for example. Other
variables for the equivalent dosing regimens may include the dosage
given per dialysis session and/or the number and frequency of
doses. Multiple dosing regimens from among available titrations of
ESA therapy may be identified, and a dosing regimen that minimizes
one or more variables such as the dosage given per dialysis
session, the number or frequency of doses, cost, or other
appropriate factors.
[0168] FIG. 17B is a flowchart illustrating an example process by
which system 10 monitors the patient response to the identified
dosing regimen and makes changes if necessary (210). The patient
response to the therapeutic dose is monitored (212). The measured
Hgb levels during the prescriptive period are compared to the
predicted Hgb level. Variations from the predicted response are
identified (214). The causes of the variation are assessed. The
model may be re-simulated to obtain an updated therapeutic dose and
equivalent dosing regimen, if necessary (216).
[0169] In addition, corrective therapies may be identified,
diagnostics may be ordered, statistics summarized and group
performance reports developed.
[0170] FIG. 18 is a flowchart illustrating another example process
240 by which the ESA dosing system may determine the
patient-specific values of the model parameters and the therapeutic
dose that may maintain the patient's Hgb within a target range. The
patient-specific historical Hgb levels and corresponding ESA dosing
data are received (232). The system optimizes the patient-specific
parameter values to determine a best fit with the patient's
historical Hgb data (234). For example, Monte Carlo or other
optimization methods may be used to determine the optimized
patient-specific parameter values. The parameters may be optimized
to result in a minimum Mean Squared Error (MSE). For purposes of
this description, the MSE refers to the sum of the squared
deviations of simulated Hgb values from the actual value obtained
divided by the'number of observations in a given time series of Hgb
values. In some examples, the parameters may be manually or
automatically adjusted to improve fit between historical and
simulated Hgb values for the descriptive period chosen.
[0171] If applicable, the patient-specific parameter values may be
manually or automatically adjusted to account for known hemorrhages
(236) or transfusions (238). The system then determines the
therapeutic dose based on the optimized patient-specific parameter
values that may maintain the patient's Hgb within a target range
(240).
[0172] FIG. 19 is a flowchart illustrating an example process 250
by which processing unit 20 (FIG. 1) or processing unit 1260 (FIG.
20) may determine a therapeutic dose that may maintain the
patient's Hgb within a target range. The target Hgb range is
received (252). The target Hgb may be input by the user via user
interface 22 (FIG. 1), or may be automatically determined. The
patient-specific parameter values are also received (254). The
system may iteratively simulate, based on the patient-specific
parameter values, the patient's Hgb response to a series of
proposed therapeutic doses until the simulated Hgb is maintained
within the target range (256). The proposed therapeutic dose that
results in stabilization of Hgb levels within the target range is
identified as the therapeutic dose (258).
[0173] In one example, "optimal" patient-specific parameter values
are identified through a type of Monte Carlo simulation that
minimizes the mean square error between simulated Hgb values and
the patient's actual Hgb lab history. A Monte Carlo simulation is a
method of randomly selecting model parameter values to be used in a
simulation in order to seek optimal values selected from the
results of a large number of simulations.
[0174] Simplified (e.g., less processing intensive) versions of the
Monte Carlo simulation may run only 100 or so simulations, whereas
more robust versions may allow thousands or tens of thousands of
simulations. Expanding the Monte Carlo sample space by orders of
magnitude may improve the reliability of the proposed prescriptions
and/or reduce the need for expert judgment. In addition, as
described herein, other non-linear optimization routines may also
be used to obtain the patient-specific parameter values, and the
disclosure is not limited in this respect.
[0175] The example biophysical simulation engine described above
was limited to those ESRD patients that were iron replete
throughout their descriptive periods. This permits exclusion of
iron metabolism components from the simulation engine. However, it
shall be understood that iron metabolism components may be added to
the example system to accommodate patients who experience periods
of iron deficiency, and possible reduced responsiveness to ESA
therapy.
[0176] As described above, the biophysical simulation engine
estimates Hgb values based on the red blood cell count. An
alternative would be to add model components that include whole
blood hematocrit calculations. For example, blood plasma and fluid
dynamics model components could be added. These alternatives may
account for any hemodilution and hemo-concentration effects. Once
the hematocrit is known, an estimate of the Hgb level could be
derived that would be more accurate than the current estimate.
[0177] As discussed above, variances from projected Hgb levels may
be used as a diagnostic, enabling users to anticipate and
potentially prevent undesirable outcomes. In each case where
significant variances were observed, a model based upon the
patient's new reality may be reconstructed, allowing the system to
create a revised prescription and to continue pursuing an adequate
and stable Hgb level with a revised therapy.
[0178] In addition to improvements in the management of anemia,
with adequate and stable Hgb values achieved, the patient will have
more stamina to comply with the rigors of life on dialysis. As a
result, hospitalizations may decrease, missed sessions may
decrease, mortality may be reduced, and dietary restrictions may be
more valued and observed. The patient may obtain the presence of
mind to effectively engage with managing the details of creating
their own health outcomes. Because missed ESA doses perturb Hgb
values, the information provided by this process greatly reduces
the risk of missed doses going uncorrected.
[0179] Elimination of Hgb cycling for an individual patient
eliminates a number of patient health risks. Hgb cycling is an
indicator that all the systems of the body are experiencing
alternating periods of excessive and diminished oxygen supply. It
is believed that this variation in oxygenation leads to increased
hospitalization rates and mortality. Stable Hgb values improve
patient quality of life and reduce overall health risks. In
addition, it may make it easier for health care providers to
recognize the onset of new comorbidities in their patients who have
a stable Hgb while on therapy since in general new medical problems
may lead to a fall in Hgb.
[0180] In one example, the ESA dosing system is applied to patients
receiving dialysis and darbepoetin alfa (or other ESA) therapy. CKD
patients who are not on dialysis, however, also may require ESA
therapy. CKD is a progressive disease and generally leads to the
initiation of renal replacement therapy, most frequently, dialysis.
Hgb cycling among CKD patients not on dialysis has also been
observed. Thus, in other examples, the system may also be applied
to those CKD patients currently not on dialysis, improving their
overall health and stamina, and designing therapies to postpone the
initiation of dialysis and the associated rigors and costs.
[0181] In other examples, the ESA dosing system may be applied to
any patient requiring ESA therapy.
[0182] In general, ESRD, CDK, and other patients with adequate and
stable Hgb values are easier for care providers to manage. In
addition, the ESA dosing system may create a new perspective for
care providers concerning RBC homeostasis. Insights gained by the
successful management of Hgb values using the system and methods
described herein may be, at least partially, transferable to
patients who have not been assessed using the ESA dosing system
described herein. For example, such insights may permit more
accurate and optimal dosing regimens to be designed. Thus the ESA
dosing system may reduce the complexity, time, and cost of caring
for patients and improve effectiveness at the same time.
[0183] The ESA dosing system and methods described herein may also
improve management of ESA inventories at Dialysis Care Facilities.
Using this system, accurate projections of ESA requirements may be
made, reducing excess inventory. Because ESA are relatively
expensive, this reduction in inventory may result in great savings
per year in inventory costs. Projected, precise dosing regimens for
each patient receiving dialysis at a Dialysis Care Facility (DCF)
for the future (90 days, for example) equips the DCF with a more
accurate estimation of the required ESA inventory levels. This can
reduce the cost of waste and other costs associated with carrying
excessive inventories.
[0184] The system and method described herein allows creation of
dosing regimens that achieve adequate and stable Hgb values that
also consume a minimum amount of ESA drugs. Retrospective
assessments of the data has produced an estimate that ESA costs may
be reduced by as much as 46% or more.
[0185] The system may include an analysis and reporting subsystem
that provides, for example, "at a glance" overviews of patients
with below target Hgb values, in range, or above target Hgb values.
Maintaining adequate and stable Hgb values for a higher percentage
of patients may enable providers to spend less time per patient,
and allocate more time to the care of patients with emergent
medical issues.
[0186] The ESA dosing system may be used to increase the efficacy
and efficiency of administered ESAs while achieving adequate and
stable Hgb values. This is a primary concern of CMS and may
represent significant cost savings.
[0187] The ESA dosing system may provide a proven evidence-based
assessment of the effectiveness of (or inadequacy of) various
dosing regimens or protocols. By means of the ESA dosing system
described herein, providers are equipped with a target dosing level
heretofore unknown. Other examples of the ESA dosing system may
include dosing regimen quality metrics that give providers the data
they need to continuously improve their anemia management
practices.
[0188] National and private insurers are moving for pay for
performance reimbursement policies. Other examples of the ESA
dosing system may include tools to assist with an objective
performance measurement system for the management of anemia.
[0189] By resulting in more stable Hgb levels, the ESA dosing
system described herein may also decrease the amount of
un-reimbursed ESA that has been administered. In sum, the derived
therapies may continuously improve patient outcomes, financial
performance for providers, and multidisciplinary care team
effectiveness.
[0190] Experience has shown that providers using rHuEpo are more
able to achieve target Hgb values than are providers using
darbepoetin alfa. However, rHuEpo alfa may be administered up to
three times per week, whereas darbepoetin alfa may be administered
weekly, bi-weekly, or even monthly. Providers have attempted to
switch to the use of darbepoetin alfa in order to reduce operating
costs only to decide at a later time to revert back to darbepoetin
alfa due to uncontrolled Hgb cycling. Other examples of the ESA
dosing system may include tools to assist providers in
transitioning from epoetin alfa to darbepoetin alfa and
simultaneously maintaining adequate and stable Hgb values.
[0191] Recombinant human erythropoietin (rHuEPO, Epogen, EPO) has a
shorter half-life than darbepoetin alfa and is therefore easier to
administer. However, dialysis providers utilizing rHuEPO as an ESA
must manage and administer rHuEPO at each dialysis session,
resulting in increased operational costs, increased risk of
infection, and higher turnover on their ESA inventories. Although
the techniques are described herein with respect to darbepoetin
alfa, it shall be understood that the techniques could be adapted
to any form of ESA. The ESA dosing system may be used to assist
dialysis providers worldwide in making a successful transition from
rHuEPO to darbepoetin alfa and secure the benefits of achieving
adequate and stable Hgb values along with the reduced operational
costs of less frequent ESA administration.
[0192] The successful utilization of the modeling described herein
has repercussions that may extend beyond the use of ESAs in
patients. For example, the systems, methods, and techniques
described herein may be extended to the administration of other
drugs with prolonged half-life or extended release formulation. The
pharmokinetic studies that are required by the FDA do not provide
renal or hepatic function. Likewise, genetically determined
differences in drug metabolism are not evident to clinicians until
an adverse effect of under- or over-dosing of the drug is noted.
The application of this methodology to drug administration may
allow faster determination and use of optimal drug dosages, and
highlight individual patient differences in the clearance and
metabolism of drugs. The current trial and error method of drug
administration requires improvement if we are to more safely
administer drugs in a patient population with increasing incidence
of kidney and liver disease, and increasing utilization of drugs
with longer half-lives.
[0193] The example ESA dosing model described with respect to FIG.
5 is directed for purposes of illustration to determining dosing of
the ESA darbepoetin alfa (Aranesp.RTM.). However, as mentioned
above, the ESA dosing techniques described herein may also be used
to determine patient-specific ESA dosing for any available ESA
therapy. These ESAs may include, but are not limited to,
Erythropoietin; Epoetin alpha (Procrit.RTM., Epogen.RTM.,
Eprex.RTM.); Epoetin beta; darbepoetin alpha (Aranesp.RTM.);
Methoxy polyethylene glycol-epoetin beta; Dynepo; Shanpoeitin;
Zyrop; Betapoietin; and others.
[0194] In addition, the ESA dosing techniques described herein may
also be applicable to a wide variety of patient populations,
including, for example, ESRD patients, CDK patients, cancer therapy
patients, HIV patients, or any other patient population having
insufficient hemoglobin production that may benefit from ESA
treatment. In addition, the ESA dosing techniques described herein
may also be applicable to multiple modes of ESA therapy delivery,
including intravenous (IV) delivery, subcutaneous delivery, oral
delivery, biopump, implantable drug delivery devices, etc.
[0195] FIGS. 20-35 illustrate another example ESA dosing system
1240 and the techniques implemented therein which may be used to
determine dosing of ESA therapies. In the examples, the ESA dosing
system 1240 is described with respect to darbepoetin alfa
(Aranesp.RTM.,) or epoetin alfa (Epogen.RTM.). However, it shall be
understood that the example ESA dosing system may also be used to
determine dosing of other ESA therapies.
[0196] FIG. 20 is a block diagram of an example ESA dosing system
1240. ESA dosing system 1240 is similar in many respects to ESA
dosing system 10 shown in FIG. 1. System 1240 includes a processing
unit 1260 and an assortment of data processing and management
software modules. For example, ESA dosing system 1240 includes
several component software modules: a data management module 1242,
an optimization module 1244, a pharmacokinetics (PK)
simulation/modeling module 1246, a pharmacodynamics (PD)
simulation/modeling module 1248, and a reporting module 1250. Data
management module 1242 is concerned with getting data in and out of
the system. Optimization module 1244 is concerned with
determination of the patient-specific parameters which cause the
model to simulate patient-specific erythropoietic responses to ESA
therapy. The PK simulation module 1246 models and simulates the
effect of the body on the drug, e.g., absorption, metabolism, and
elimination. The PD simulation module 1248 models and simulates the
effect the drug on the body, e.g., apoptosis sparing. Reporting
module 1250 is concerned with presenting the results of the
simulation in the form of reports, graphs, and/or other output in a
way that is meaningful for an analyst or provider.
[0197] FIG. 21 illustrates an example diagram 1200 that is part of
the data acquisition/management component of the ESA dosing system.
Diagram 1200 illustrates importation of the historical individual
patient data. Pt ID 1202 represents the patient identification
number. The top left portion of the diagram 1200 receives the
calendar dates concerning the historical data and maps the
simulation day numbers to the actual calendar days of the
descriptive and prescriptive periods (e.g., simulation day 476 may
be equivalent to Dec. 4, 2009, simulation day 477 would then be
Dec. 5, 2009, etc.). Last descriptive day number 1204 represents
the final calendar day of the descriptive period and First
prescriptive day number 1206 represents the first calendar day on
which an ESA dose may be administered. (The descriptive period is
historical and is used to determine the most likely patient
response to future ESA therapy. The prescriptive period is a
projection for the future which is developed based upon analysis of
the patient's historical response to ESA therapy.) SimDays in
descriptive period 1207 is the total number of days in the
descriptive period; that is, the total number of days for which
historical data will be entered into the model. Sim Start Day
Number 1208 is the day number that the descriptive period is to
start. In the examples given herein, this day has been designated
Day 0. First Prescriptive Sim Day number 1212 is the day number
corresponding to the calendar day of the First prescriptive day
number 1206 and is the first day on which a patient may receive an
ESA dose. Current day number 1210 represents the current day number
of the overall simulation as it progresses from day -200 to the
last day of the simulation. Weekly Therapeutic Dose 1216 may allow
the WTD determined from the model described above with respect to
FIG. 5 to be compared with the results of the model described below
with respect to FIGS. 32 and 33, if desired.
[0198] Model inputs 1222 represent the patient historical
hemoglobin data 1226 and the patient historical ESA dosage data to
be entered into the model. In this example, the historical ESA data
may include either historical Aranesp data 1228 or historical
Epogen data 1230, depending upon the ESA therapy used by the
particular patient. Other ESA dosage data may also be entered, and
the disclosure is not limited in this respect.
[0199] Fe (iron) status indicators 1224 represent the patient
historical iron data 1232 and/or the patient historical transferrin
saturation data, if any, to be entered into the model. This allows
the user to take the patient's iron levels into account when
running the ESA dosage simulation. Patients who are iron deficient,
indicated in part be a transferrin saturation value below 20
percent, may not have enough iron in the blood to combine with
mature reticulocytes produced by the bone marrow to create a
sufficient number of mature red blood cells containing hemoglobin.
If the simulation is not able to fit the historical data, the Fe
status indicators 1224 may help the user to better interpret the
patient's clinical status and define corrective therapies.
[0200] Hgb low 1218 and Hgb high 1220 represent the low and high
values of the desired hemoglobin range. For example, the Centers
for Medicare & Medicaid Services (CMS) and National Kidney
Foundation (NKF) have established the target range for Hgb values
among ESRD patients to be between 10 g/dL and 12 g/dL. These or
other hemoglobin values appropriate for the patient or the
patient's condition may be entered as the low and high hemoglobin
values, respectively.
[0201] As described above, the biophysical simulation may employ an
adaptation of the Monte Carlo method to estimate patient-specific
parameter values. It shall be understood that other optimization
routines may be employed, and the disclosure is not limited in this
respect. FIGS. 22 and 23 are diagrams illustrating a setup for an
example Monte Carlo simulation that determines the best fit
patient-specific parameter values for the patient's historical
hemoglobin data. In this example ESA dosing model, one or more
parameters representing various parameters of the patient's red
blood cell production chain may be used. These may include, for
example, one or more of the following six patient-specific
parameters:
[0202] Blast Forming Unit Input (BFU INPUT): the number of
erythroid burst forming units entering the erythropoietic process
each day.
[0203] Colony Forming Unit Survival (CFU SURV): the fraction of
colony forming units that survive apoptosis in the absence of an
ESA.
[0204] Reticulocyte Survival (RETIC SURV): the fraction of
reticulocytes that survive reticulocyte atrophy, which may be
caused by a deficiency in hemoglobin building blocks such as iron,
folate, or vitamin B12, among others.
[0205] Erythropoietin Receptor (EpoR) Multiplier (EPOR MULT): the
value by which Kd is amplified to generate a response within the
developing RBC cell a strong enough reaction to prevent
apoptosis.
[0206] Red Blood Cell Lifespan (RBC LIFESPAN): average lifespan (in
days) of a red blood cell.
[0207] Erythropoietin Setup Rate (EPO SETUP RATE): a mathematical
value applied during the setup period that raises the simulated
hemoglobin to a level equal to the observed hemoglobin on the first
day of the descriptive period.
[0208] For each of the six parameters shown in FIGS. 22 and 23, a
Monte Carlo switch 1282, 1292, 1302, 1312, 1322 and 1332 determines
whether a user selected value, a previously determined parameter
value obtained from a previous Monte Carlo run, or a randomly
generated value will be used for each simulation. If the Monte
Carlo switch for a particular parameter is turned off, the
simulation will take the value of that parameter from a
corresponding user selected value or from a previously determined
parameter value obtained from previous Monte Carlo run. These user
selected values may be input via any suitable user interface, for
example via sliders 1602, 1604, 1606, 1608, 1610, and 1612 shown in
FIG. 34. This may permit the user to manually control one or more
of the parameter values in order to obtain a better fit to the
historical hemoglobin data.
[0209] If the Monte Carlo switch for a particular parameter is
turned on, the simulation will obtain the value of that parameter
for each run of the Monte Carlo simulation via a random number
generator as described below.
[0210] For each of the six parameters shown in FIGS. 22 and 23, MIN
1281, 1291, 1301, 1311, 1321, and 1331 and MAX 1283, 1293, 1303
1313, 1323, and 1333, represent the ranges from which the randomly
generated parameter values are to be drawn for the individual
simulations of a Monte Carlo simulation. For example, BFU INPUT MIN
1281 and BFU INPUT MAX 1283 are the minimum and maximum values,
respectively, from which the random numbers to be used for each run
of the Monte Carlo simulation for the parameter BFU INPUT are to be
drawn.
[0211] For each parameter, MC 1285, 1295, 1305, 1315, 1325, and
1335, represents a function that generates a random number between
the values of MIN and MAX for each run of the Monte Carlo
simulation. For example, BFU INPUT MC 1285 represents a function
that generates a random number between the values of BFU INPUT MIN
1281 and BFU INPUT MAX 1283 for each run of the Monte Carlo
simulation.
[0212] BFU INPUT 1280, CFU SURV 1290, RETIC SURV 1300, EPOR MULT
1310, RBC LIFESPAN 1320 and EPO SETUP RATE 1330 are either the
previously determined parameter value obtained from a Monte Carlo
run or user selected parameter values (input, for example, via
sliders shown in FIG. 34 as described above when those sliders are
set to a mode to override previously obtained parameter values.)
used when the Monte Carlo switch is turned off for the
corresponding parameter.
[0213] BFU INPUT CALC 1284, CFU SURV CALC 1294, RETIC SURV CALC
1304, EPOR MULT CALC 1314, RBC LIFESPAN CALC 1324, EPO SETUP RATE
CALC 1334 are the values obtained from the best fit run of a Monte
Carlo simulation. Once the best fit run is determined, the
parameter values determined from that best fit run may be used to
determine a therapeutic dose that may be administered in the
prescriptive period. A therapeutic dose of an ESA is that dose
which causes a patient's hemoglobin values to achieve and sustain
the target hemoglobin value as long as the patient's clinical
condition remains stable. However, the user may want to attempt to
improve upon the initially obtained best fit results by manually
selecting a value for one or more of the parameters (such as via
the sliders shown in FIG. 34) and running additional simulations.
Alternatively or in addition, the user may attempt to improve upon
the initial or intermediate best fit results by narrowing the range
from which the random numbers are generated by adjusting the MIN
and MAX values (such as BFU INPUT MIN 1281 and/or BFU INPUT MAX
1283, CFU SURV MIN 1291 and/or CFU SURV MAX 1293, etc.) for one or
more parameters. Suggested values for narrowed parameter ranges may
be supplied by 1286, 1288, 1296, 1298, 1306, 1308, 1316, 1318,
1326, 1328, 1336, an 1338 on FIGS. 22-23.
[0214] Suggested BFU minimum (Sugg BFU min) 1286 and Suggested BFU
maximum (Sugg BFR max) 1288, Sugg CFU surv min 1296 and Sugg CFU
surv max 1298, sugg retic surv min 1306 and sugg retic surv max
1308, sugg EPOR mult min 1316 and sugg EPOR mult max 1318, sugg RBC
LIFE 1326 and sugg RBC LIFE 1328, and sugg EPO setup min 1336 and
sugg EPO setup min and sugg EPO setup max 1340 may be the results
obtained from a Monte Carlo run. Should the user choose to perform
a subsequent Monte Carlo run, these values may be the suggested
values to use for the respective minimum and maximum values to use
as lower and upper bounds from which random values will be drawn in
the subsequent Monte Carlo run.
[0215] FIG. 24 is a diagram 1340 representing an example
calculation of a mean square error (MSE) for one run of the Monte
Carlo simulation. In this example, the MSE of each run of the Monte
Carlo simulation is used to determine goodness of fit of the
simulated hemoglobin and the observed hemoglobin values of the
descriptive period. The run with the lowest MSE, drawn from, for
example, 100 individual simulations may be determined to be the
best fit run. However, other methods of minimizing MSE may also be
used, and the disclosure is not limited in this respect.
[0216] To determine the MSE for each run, the total number of
hemoglobin values in the descriptive period is counted (1341,
1344). The squared difference of the simulated hemoglobin 1348 and
the patient specific historical Hgb data 1226 is determined (1350).
The squared differences are summed over all days of the descriptive
period (1352, 1354). The MSE 1356 is the sum of the squares divided
by the total number of hemoglobin values in the descriptive period
1344 (displayed in this example in box 1342). The MSE for each run
is determined, and the run with the lowest MSE is determined to be
the "best fit run."
[0217] FIG. 25 is a diagram 1360 representing the amount of Aranesp
administered at a prescribed interval based on a prescription
regimen equivalent to a simulated therapeutic dose. For example, a
therapeutic Aranesp dose for a patient might be determined to be 32
mcg per week. The equivalent Aranesp dosing regimen, a titration of
available dose amounts might be 25 mcg in week 1 followed by 40 mcg
in week 2, the 25 mcg 1 week later and so on. Although this example
is described with respect to Aranesp, it shall be understood that
this model may also be applicable to other ESA therapies that
require similar methods of titration based upon their respective PK
parameters, Kd, and/or half life.
[0218] EPO SETUP RATE 1330 is one of the patient-specific
parameters to be determined during the simulation. CLINICAL ARANESP
DATA 1228 is the patient-specific historical Aranesp dosage data
during the descriptive period. In this example, a recommended
dosing regimen for Aranesp is divided into three separate doses to
be administered on given days, dose A 1370, dose B 1371, and dose C
1372, to deliver the equivalent of the weekly therapeutic dose
(WTD) determined by the simulation. Dose A START 1368 is the
earliest day on which dose A may be given, as well as the day on
which Dose A is to commence. ARANESP INPUT (micrograms) 1376 is a
the total dose on a given day of the simulation from all sources:
EPO setup rate 1330, Clinical Aranesp Data 1364, and Rx Protocol
1374. Rx Protocol 1374 can be of two types, it is either the sum of
Aranesp dose A 1370, Aranesp dose B 1371, and Aranesp dose C 1372
or, it is Aranesp WTD 1373, depending on the value of WTD Switch
1371. Aranesp WTD 1373 is equal to the value of Example 2 Weekly
Therapeutic Aranesp Dose Amount 1377, which is a user-entered dose,
to be applied in the simulation beginning on First Prescriptive Sim
Day Number 1212 (see FIG. 21) and on every subsequent seventh day
of the simulation. Once patient parameters have been found,
simulation experiments with different values of Example 2 Weekly
Therapeutic Dose Amount 1377 may enable the user to identify the
value of the weekly therapeutic Aranesp dose that will achieve and
sustain the desired hemoglobin level for this patient. Aranesp
input (picomoles) 1378 is a numerical conversion that converts the
dosage in micrograms per dose to picomoles per dose, regardless of
the source of the dose: setup, historical, WTD, or recommended
dosing regimen. Alternatively, rather than permitting or requiring
user input, these and other components of the model may be
implemented via an automated software system.
[0219] FIG. 26 is a diagram 1380 representing one possible set of
variables which may be used to define a recommended Aranesp
prescription regimen. Although this diagram is described with
respect to Aranesp, it may also be applicable to other ESA
therapies. Diagram 1382 represents the factors that determine
Aranesp dose A 1370, diagram 1384 represents the factors that
determine Aranesp dose B 1371, and diagram 1386 represents the
factors that determine Aranesp dose C 1372. For example, dose A,
dose B, and dose C each include a dose amt 1391A-1391C, a does
start 1392A-1392C, a dose interval 1393A-1393C, and a dose end
1394A-1394C, respectively. This permits the appropriate titration
to be determined based on the Aranesp WTD determined by the ESA
dosing system. The time periods for dose A, dose B, and/or dose C
may or may not overlap, depending upon what is required to obtain
the Aranesp WTD determined by the ESA dosing system, as well as any
initial corrective doses that might be required by patients
entering the prescriptive period with low hemoglobin values.
[0220] FIG. 27 is a diagram 1400 representing determination of the
circulating Aranesp concentration. Again, although this diagram is
described with respect to Aranesp, the model is equally applicable
to other ESA therapies. An idealized body weight 1402 (e.g., 70 kg)
is used to determine an idealized volume of distribution 1404 (the
distribution of a medication between plasma and the rest of the
body). The patient's actual body weight may be used, if desired.
Aranesp input (picomoles) 1378 is the value obtained as described
above with respect to FIG. 26 and is input at ESA input moles 1412.
The ARANESP AMOUNT 1414 is eliminated from the body as determined
by the Aranesp half life 1418. Aranesp elimination 1416 represents
a mathematical reduction per day of the ARANESP AMOUNT 1414 based
on the Aranesp half life 1418. The reduced Aranesp amount and the
volume of distribution 1404 determine the resulting Aranesp
concentration (picomoles) 1406.
[0221] FIG. 28 is a diagram 1420 representing the amount of Epogen
administered at a prescribed interval based on a recommended
prescription regimen equivalent to a simulated therapeutic dose.
Although this example is described with respect to Epogen, it shall
be understood that this model may also be applicable to other ESA
therapies. Diagram 1420 has the same structure as diagram 1360 of
FIG. 25. A similar diagram may thus apply to other ESA therapies.
EPO SETUP RATE 1330 is one of the patient-specific parameters to be
determined during the simulation. This parameter is input at setup
EPO input 1424. CLINICAL EPOGEN DATA 1230 is the patient-specific
historical Epogen dosage data during the descriptive period. This
data is input at historical Epogen doses 1428. In this example, a
recommended dosing regimen for Epogen is divided into three
separate doses to be administered on given days, dose A 1432, dose
B 1433, and dose C 1434, to arrive at the desired per session
therapeutic dose (PSTD) determined by the simulation. Dose A START
1430 is the earliest day on which dose A may be given, as well as
the day on which Dose A is to commence. EPOGEN INPUT (units) 1440
is a the total dose on a given day of the simulation from all
sources: EPO setup rate 1330, Clinical EPOGEN Data 1426, and Epogen
Rx Pulses 1436. Epogen Rx Pulses 1436 can be of two types, it is
either the sum of Epg Rx regimen A 1432, Epg Rx regimen B 1433 and
Epg Rx regimen C 1434 or, it is Epogen PSTD 1446, depending on the
value of WTD Switch 1371. Epogen PSTD 1446 is equal to the value of
Example 2 Per Session Epogen Therapeutic Dose 1444, which is a
user-entered dose, to be applied in the simulation beginning on
First Prescriptive Sim Day Number 1448 and on the day of every
subsequent dialysis session of the simulation. (Epogen is typically
administered three times per week at each dialysis session; Aranesp
is administered at least weekly, hence the difference in
therapeutic dosing conventions. Other ESA's may have differing
therapeutic dosing conventions based fundamentally on their
respective Kd and half lives.) Once patient parameters have been
found, simulated experiments with different values of Example 2 Per
Session Epogen Therapeutic Dose 1444 enables the user to identify
the value of the PSTD that will achieve and sustain the desired
hemoglobin level for this patient. Epogen input (picomoles) 1438 is
a numerical conversion that converts the dosage in units per dose
to picomoles per dose regardless of the source of the dose: setup,
historical, PSTD, or recommended dosing regimen. Alternatively,
rather than permitting or requiring user input, these and other
components of the model may be implemented via an automated
software system.
[0222] FIG. 29 is a diagram 1450 representing one possible set of
variables which may be used to define an Epogen prescription
regimen. Although this diagram is described with respect to Epogen,
it may also be applicable to other ESA therapies. In this example,
Epogen may be given in up to three different dosages, dose A, dose
B and dose C. Diagram 1452 illustrates the factors that determine
Epogen dose A 1432, diagram 1454 illustrates the factors that
determine Epogen dose B 1433, and diagram 1456 represents the
factors that determine Epogen dose C 1434. In this example, doses
A, B, and C are scheduled to dose on Monday, Wednesday, and Friday,
respectively. For example, close A, dose B, and dose C each include
a dose amt 1461A-1461C, a dose start 1462A-1462C, and a dose end
1463A-1463C, respectively. Epogen dose A1 1464, Epogen dose A2
1465, and Epogen dose A3 1466 permit the user flexibility in
setting up customized dosing regimens. However, the same dose could
be given each time rather than different doses on different days of
the week.
[0223] FIG. 30 is a diagram 1470 representing determination of the
circulating Epogen concentration. This diagram has the same
structure as diagram 1400 of FIG. 27. Again, although this diagram
is described with respect to Epogen, it may also be applicable to
other ESA therapies. An idealized volume of distribution 1472 is
entered into the system. Epogen input (picomoles) 1438 is the value
obtained as described above with respect to FIG. 28 and is input at
Epogen input 1480. The EPOGEN AMOUNT pM 1482 is eliminated from the
body as determined by the Epogen half life 1486. Epogen elimination
1484 represents a mathematical reduction per day of the EPOGEN
AMOUNT pM 1482 based on the Epogen half life 1486. The reduced
Epogen amount and the volume of distribution 1472 determine the
resulting Epogen concentration (picomoles) 1476.
[0224] FIG. 31 is a diagram 1490 illustrating EPOR (erythropoietin
receptor) binding for Epogen and Aranesp. However, it shall be
understood that this diagram may also be applicable to other ESA
therapies. If other ESA therapies are to be used, the ESA dosing
system may include similar diagrams and functionality corresponding
to those other ESA therapies. Diagram 1500 includes an ARANESP
switch 1504 and an EPOGEN switch 1508. Switches 1504, 1508 permit a
user to select which drug was used by the patient during the
descriptive period, and for which a proposed prescription during
the prescriptive period should be determined.
[0225] A published parameter referred to as the "ESA Kd" is stored
by the ESA dosing system and applied in the model. The ESA Kd is a
known value for each ESA that may be entered by a user and stored
by the system. Kd refers to the dissociation constant of the ESA
being used and the erythropoietin receptor (EPOR).
[0226] In this example, the ESA Kd values shown in FIG. 31 are the
Aranesp Kd 1494 and the Epogen Kd 1516. When the drug at issue is
Aranesp, for example, the ESA dosing system combines the Aranesp Kd
1494 and the Aranesp concentration 1408 (determined as shown in
FIG. 27) and determines a calculated value referred to as the "EPOR
fraction bound" (erythropoietin receptor fraction bound) 1510.
Similarly, when the drug at issue is Epogen, the ESA dosing system
combines the Epogen Kd 1516 and the Epogen concentration 1476
(determined as shown in FIG. 30) and determines the EPOR fraction
bound 1510. A similar calculation may be made when other ESAs are
being simulated.
[0227] The calculated value "EPOR fraction bound" refers to the
percentage of eporeceptors on the surface of BFU-E cells that have
bound the ESA being used for therapy. Once a minimum percentage is
reached, the rate of programmed cell death (apoptosis) is decreased
by means of reactions within the cell in response to bound
eporeceptors on the surface of the cell.
[0228] For example, if the EPOR fraction bound 1510 is greater than
a minimum percentage (such as 10%) then the apoptosis rate
decreases. Alternatively, if the EPOR fraction bound 1510 is less
than the minimum percentage, the apoptosis rate increases. The
effect of the EPOR fraction bound on the apoptosis rate is
described in more detail below with respect to FIG. 32.
[0229] FIG. 32 is a diagram 1530 illustrating an example model of
reticulocyte production in bone marrow. Diagram 1530 generally
represents the pharmacodynamics (PD) component of the model.
Diagram 1400 of FIG. 27 and diagram 1470 of FIG. 30 generally
represent the pharmacokinetics (PK) component of the model. The
patient-specific parameters that enter into Erythropoiesis in
Marrow 1530 of the ESA dosing model may include, for example, BFU
input 1280, CFU Survival 1290, EpoR multiplier 1310, and
Reticulocyte Survival 1300.
[0230] CFU/E chain 1536 represents cell replication that occurs
through a number of generations. The number of generations
considered by the model may vary in a specified range; for example,
from 12 to 31 generations. BFU INPUT 1280 is the number of blast
forming units committed to forming red blood cells which is input
into model at 1534. Due to cellular division, the input of each
successive generation in the CFU/E chain 1536 is twice the output
of the previous generation. A replication interval 1538 describes
the length of time required for each generation to replicate. For
example, the replication interval may be in the range of 0.75 to
1.25 days, or other appropriate interval. CFU/E cells are the cells
within the erythrocyte lineage that are subject to apoptosis and
responsive to apoptosis-sparing ESA therapy. The total number of
cells within CFU/E chain 1536 is reduced by apoptosis 1543, as
moderated by the ESA concentration. CFU Survival 1290 is a
patient-specific parameter defined as the fraction of colony
forming units that survive apoptosis in the absence of an ESA. EPOR
fraction bound 1510 and EPOR multiplier 1310 determine the
fractional amount of apoptosis sparing 1541 in the presence of an
ESA. This lowers the rate of apoptosis 1543, increasing the number
of dividing cells 1540 that survive apoptosis and go on to become
erythroblasts 1542, and eventually, erythrocytes.
[0231] Erythroblast/Reticulocyte development chain 1544 represents
maturation of the number of erythroblasts that survived apoptosis
1542, 1544. A maturation time 1546 describes the length of time
required for each generation of erythroblasts to mature. For
example, maturation time may be in the range of 3 to 5 days, or
other appropriate maturation time. The total number of
reticulocytes leaving development chain 1544 is reduced by
reticulocyte atrophy 1548. Reticulocyte atrophy 1548 is determined
by the patient-specific parameter Reticulocyte Survival 1300,
defined as the fraction of reticulocytes successfully mature in the
presence of required complementary materials, such as iron, folate,
and B12, for example. Reticulocyte survival is also influenced by
infection and inflammation. In the case of infection, bacteria
compete with the maturing cells for iron, reducing the iron
available to form hemoglobin in the maturing cell. In the case of
inflammation, available iron is sequestered, effectively reducing
iron availability to the maturing cell. In the current embodiment
of ESA Dosing System 1240, assessments on the status of
complementary materials, inflammation, and infection are made by an
expert model user. The ESA dosing model may include software
algorithms to simulate the status of complementary materials,
inflammation, and infection in regard to reticulocyte survival, and
the disclosure is not limited in this respect. The number of
maturing cells that survive reticulocyte atrophy is the number of
reticulocytes 1550 produced by the bone marrow.
[0232] FIG. 33 is a diagram 1551 is an example model to simulate
the total number of red blood cells in circulation. Reticulocyte
production 1550 (determined as described above with respect to FIG.
32) enters into reticulocytes in circulation chain 1554. HEM 1562
is information concerning hemorrhages (blood loss for any reason)
experienced by the patient. A hemorrhage reduces the number of
reticulocytes in circulation and the number of red blood cells
(RBCs) in circulation, and therefore hemoglobin. These hemorrhage
reduction effects are represented by hemorrhage reticulocyte
reduction from circulation 1560 (expressed, for example, as a
fraction) and hemorrhage RBC reduction from circulation 1564 (also
expressed, for example, as a fraction). The number of reticulocytes
maturing 1556 enters into the RBC chain 1558. The number of RBCs
leaving the RBC chain 1558 is reduced by any hemorrhage effects
(1564). The total number of mature RBCs is influenced by the
patient-specific parameter RBC Lifespan 1320, defined as the
average lifespan (in days) of a red blood cell. The total number of
RBCs in circulation is represented by total cells in circulation
1566 which is the sum of reticulocytes in circulation 1554 and
RBC's 1558.
[0233] The total cells in circulation 1566 (reticulocytes in
circulation+RBCs in circulation) together with the volume of
distribution 1472 gives the simulated hematocrit 1568. The
hematocrit multiplied by a constant gives the simulated hemoglobin
1570.
[0234] FIG. 34 is an example user interface 1600 through which a
user may control various aspects of the ESA dosing system, enter
various parameter values, run and control simulations, etc. These
user selected patient-specific parameter values may be input via
sliders 1602, 1604, 1606, 1608, 1610, and 1612 for one or more of
the patient-specific parameter values. Button 1614 takes the user
to a Monte Carlo setup screen, button 1616 takes the user to an ESA
dosing screen, button 1618 takes the user to a home screen, buttons
1620 allow the user to run a simulation, and button 1621 allows the
user to stop a simulation. It shall be understood that this
disclosure is not limited to the specific example methods of
navigation among the various elements of the simulation model
described herein, as other methods may be used alternatively or in
addition to the methods described herein.
[0235] FIG. 35 is an example graph 1620 displaying historical Hgb
levels 1622, historical ESA dosages 1626, and simulated Hgb levels
1624 for the pre-descriptive setup period 1630, the descriptive
period 1632, and the prescriptive period 1634. Graph 1620 also
displays recommended ESA dosing 1628 for the prescriptive period,
determined by simulated experiments described above. Graph 1620
could be displayed on, for example, user interface 1252 (FIG. 20)
or other suitable user interface. Although specific data and graph
are shown in FIG. 35, it shall be understood that the disclosure is
not limited in this respect, and that other relevant data, graphs,
tables, charts or other ways of displaying data may also be
displayed, and that other types of functional interfaces, such as
touch screen, mouse, stylus, keyboard, multi-touch, mobile devices,
or other method of interacting with the program may be used without
departing from the scope of the present disclosure.
[0236] The graph 1620 of FIG. 35 illustrates an example curve
fitting result for the descriptive period 1632. In this example,
the descriptive period for this patient was approximately 690 days
in duration. During that period, 30 actual Hgb values were
measured, and those values display a typical Hgb oscillation. 92
doses of Aranesp were administered in the descriptive phase. The
model uses a so called pre-descriptive period 1630 to establish an
erythropoietic equilibrium which simulates the Hgb value of the
first observed Hgb result in the descriptive period, in this
example equal to 12.8 g/dL. In this example, the pre-descriptive
period 1630 in the model is 201 days in duration, running from day
-200 to day 0. This is the period of time the simulation model
requires to establish equilibrium in the presence of a theoretical
(mathematically applied) daily ESA dose. Once the system determines
the optimized therapeutic dose for a specific ESA, the system may
further assist providers in finding the most effective combinations
of available dosing levels at the optimal frequency of
administration that will deliver the required therapeutic dose, and
as a result, achieve and maintain the desired Hgb value. The
recommended dosing regimen for the prescriptive period 1634 is
indicated by 1628.
[0237] As mentioned above, although the examples were presented
herein with respect to Aranesp and Epogen, it shall be understood
that the ESA dosing techniques described herein may also be
applicable to other types of ESA therapies, other patient
populations and alternative routes of administration. In general,
to apply the ESA dosing model to other ESA therapies, the
ESA-specific constants ESA half life and ESA Kd would be entered
into the ESA dosing model. For example, the ESA half life for the
particular ESA would be taken into account in the model at the same
point as the Aranesp half life 1418 in FIG. 27 or the Epogen half
life 1486 in FIG. 30. Similarly, the ESA Kd for the particular ESA
would be taken into account in the model at the same point as
either the Aranesp Kd 1494 or the Epogen Kd 1516 as shown in FIG.
31. Those of skill in the art will appreciate that the ESA dosing
model described herein may be applicable to a wide variety of ESA
therapies, as well as to a wide variety of patient populations
(ESRD patients, CKD patients, cancer therapy patients, HIV
patients, or any other patient population having insufficient
hemoglobin production and benefiting from ESA treatment). In
addition, the ESA dosing model may also be applicable to multiple
modes of delivery, including intravenous (IV) delivery,
subcutaneous delivery, oral delivery, biopump, implantable drug
delivery devices, etc.
[0238] The following are illustrative equations for the example
model shown in FIGS. 21-33, as expressed in the syntax of a
commercially available modeling application (iThink.RTM., available
from Isee Systems, Inc., in this example). Although an example
implementation using iThink.RTM. is shown, it shall be understood
that the ESA dosing techniques described herein may also be
implemented using other commercially available or customized
software applications.
TABLE-US-00004 ARENESP_AMOUNT(t) = ARENESP_AMOUNT(t - dt) +
(ESA_input_picomoles - Arenesp_elimination) * dt INIT
ARENESP_AMOUNT = 0 INFLOWS: ESA_input_picomoles =
Aranesp_input_picomoles OUTFLOWS: Arenesp_elimination =
ARENESP_AMOUNT * ( .693 / Aranesp_halftime ) BFU_INPUT_MC(t) =
BFU_INPUT_MC(t - dt) INIT BFU_INPUT_MC =
RANDOM(BFU_INPUT_MIN,BFU_INPUT_MAX) CFU\E[generation](t) =
CFU\E[generation](t - dt) + (input[generation] -
dividing_cells[generation] - apoptosis[generation]) * dt INIT
CFU\E[generation] = 0 INFLOWS: input[generation] = IF ARRAYIDX ( )
= 1 THEN BFU_INPUT ELSE 2 * dividing_cells[generation-1] { COMMENT
OUT: IF ARRAYIDX ( ) = 1 THEN Noname_1 ELSE 2 *
dividing_cells[generation-1] } OUTFLOWS: dividing_cells[generation]
= CONVEYOR OUTFLOW TRANSIT TIME = replication_interval apoptosis[1]
= LEAKAGE OUTFLOW LEAKAGE FRACTION = 0 NO-LEAK ZONE = 0%
apoptosis[2] = LEAKAGE OUTFLOW LEAKAGE FRACTION = 0 NO-LEAK ZONE =
0% apoptosis[3] = LEAKAGE OUTFLOW LEAKAGE FRACTION = 0 NO-LEAK ZONE
= 0% apoptosis[4] = LEAKAGE OUTFLOW LEAKAGE FRACTION = 0 NO-LEAK
ZONE = 0% apoptosis[5] = LEAKAGE OUTFLOW LEAKAGE FRACTION = 0
NO-LEAK ZONE = 0% apoptosis[6] = LEAKAGE OUTFLOW LEAKAGE FRACTION =
0 NO-LEAK ZONE = 0% apoptosis[7] = LEAKAGE OUTFLOW LEAKAGE FRACTION
= 0 NO-LEAK ZONE = 0% apoptosis[8] = LEAKAGE OUTFLOW LEAKAGE
FRACTION = 0 NO-LEAK ZONE = 0% apoptosis[9] = LEAKAGE OUTFLOW
LEAKAGE FRACTION = 0 NO-LEAK ZONE = 0% apoptosis[10] = LEAKAGE
OUTFLOW LEAKAGE FRACTION = 0 NO-LEAK ZONE = 0% apoptosis[11] =
LEAKAGE OUTFLOW LEAKAGE FRACTION = ( 1 - fractional_apopt_sparing )
* (1 - CFU_SURV ) NO-LEAK ZONE = 0% apoptosis[12] = LEAKAGE OUTFLOW
LEAKAGE FRACTION = 0 NO-LEAK ZONE = 0% CFU_SURV_MC(t) =
CFU_SURV_MC(t - dt) INIT CFU_SURV_MC = RANDOM(CFU_SURV_MIN ,
CFU_SURV_MAX) Count_of_Hgb_Values(t) = Count_of_Hgb_Value(t - dt) +
(Counting) * dt INIT Count_of_Hgb_Values = 0 INFLOWS: Counting = If
CLINICAL_Hgb_Data > 0 THEN 1 ELSE 0 EPOGEN_AMT_pM(t) =
EPOGEN_AMT_pM(t - dt) + (Epogen_input - Epogen_elim) * dt INIT
EPOGEN_AMT_pM = 0 INFLOWS: Epogen_input = Epogen_input_picomoles
OUTFLOWS: Epogen_elim = EPOGEN_AMT_pM * ( .693 / Epogen_halftime )
EPOR_MULT_MC(t) = EPOR_MULT_MC(t - dt) INIT EPOR_MULT_MC = RANDOM (
EPOR_MULT_MIN , EPOR_MULT_MAX ) EPO_SETUP_RATE_MC(t) =
EPO_SETUP_RATE_MC(t - dt) INIT EPO_SETUP_RATE_MC = RANDOM (
EPO_SETUP_RATE_MIN , EPO_SETUP_RATE_MAX )
Eryth\Retic_development(t) = Eryth\Retic_development(t - dt) +
(eryth_input - reticulocyte_prod - retic_atrophy) * dt INIT
Eryth\Retic_development = 0 TRANSIT TIME = varies INFLOW LIMIT =
INF CAPACITY = INF INFLOWS: eryth_input = erythroblast_production
OUTFLOWS: reticulocyte_prod = CONVEYOR OUTFLOW TRANSIT TIME =
maturation_time retic_atrophy = LEAKAGE OUTFLOW LEAKAGE FRACTION =
( 1 - RETIC_SURV ) NO-LEAK ZONE = 0% P7_MC(t) = P7_MC(t - dt) INIT
P7_MC = RANDOM ( P7_MIN , P7_MAX ) RBCs(t) = RBCs(t - dt) +
(reticulocytes_maturing - RBCs_lysing - hem_RBC_from_circul) * dt
INIT RBCs = 0 TRANSIT TIME = varies INFLOW LIMIT = INF CAPACITY =
INF INFLOWS: reticulocytes_maturing = CONVEYOR OUTFLOW OUTFLOWS:
RBCs_lysing = CONVEYOR OUTFLOW TRANSIT TIME = RBC_LIFESPAN
hem_RBC_from_circul = LEAKAGE OUTFLOW LEAKAGE FRACTION = If HEM
>0 Then HEM Else 0 NO-LEAK ZONE = 0% RBC_LIFESPAN_MC(t) =
RBC_LIFESPAN_MC(t - dt) INIT RBC_LIFESPAN_MC = RANDOM (
RBC_LIFESPAN_MIN , RBC_LIFESPAN_MAX )
reticulocytes_in_circulation(t) = reticulocytes_in_circulation(t -
dt) + (reticulocyte_release_from_marrow - reticulocytes_maturing -
hem_reticul_from_circul) * dt INIT reticulocytes_in_circulation = 0
TRANSIT TIME = 2 INFLOW LIMIT = INF CAPACITY = INF INFLOWS:
reticulocyte_release_from_marrow = reticulocyte_prod OUTFLOWS:
reticulocytes_maturing = CONVEYOR OUTFLOW hem_reticul_from_circul =
LEAKAGE OUTFLOW LEAKAGE FRACTION = If HEM >0 Then HEM Else 0
NO-LEAK ZONE = 0 RETIC_SURV_MC(t) = RETIC_SURV_MC(t - dt) INIT
RETIC_SURV_MC = RANDOM ( RETIC_SURV_MIN , RETIC_SURV_MAX )
Summed_Squared_Difference(t) = Summed_Squared_Difference(t - dt) +
(Summing) * dt INIT Summed_Squared_Difference = 0 INFLOWS: Summing
= Squared_Difference Aranesp_Conc_pM = ARENESP_AMOUNT / Vol_of_Dist
Aranesp_halftime = 25/24 Aranesp_input_picomoles = (
ARANESP_INPUT_ug ) * 1e6 / 37100 ARANESP_INPUT_ug = setup_EPO_input
+ ( IF time < Ar_DOSE_A_START THEN Historical_Aranesp_doses ELSE
Rx_protocol ) Aranesp_Kd = 400E-12 ARANESP_switch = 1 Aranesp_WTD =
IF Time < First_Prescriptive_Sim_Day_Number THEN 0 ELSE
PULSE(Example_2_Weekly_Therapeutic_Aranesp_Dose_Amount,First_Prescriptive_-
Sim_Day_Number,7) Ar_dose_A = IF ( TIME < Ar_dose_A_end ) THEN
PULSE ( Ar_dose_A_amt , Ar_DOSE_A_START , Ar_dose_A_interval ) ELSE
0 Ar_dose_A_interval = 7 Ar_DOSE_A_START = 700 Ar_dose_A_amt = 0
Ar_dose_A_end = 900 Ar_dose_B = IF ( TIME < Ar_dose_end ) THEN
PULSE ( Ar_dose_B_amt , Ar_DOSE_B_START , Ar_dose_B_interval ) ELSE
0 Ar_dose_B_amt = 0 Ar_dose_B_interval = 0 Ar_DOSE_B_START = 700
AR_dose_C = IF ( TIME < Ar_dose_C_end ) THEN PULSE (
AR_dose_C_amt , Ar_DOSE_C_START , Ar_dose_C_interval ) ELSE 0
Ar_dose_C_end = 0 Ar_DOSE_C_START = 700 AR_dose_C_amt = 0
Ar_dose_C_interval = 0 Ar_dose_end = 0 avg_lifetime_CALC =
86.0699996948242 baseline_blast_mortality_fraction_CALC =
0.680000007152557 baseline_reticulocyte_mortality_fraction_CALC =
0.569999992847443 BFU_INPUT = IF Monte_Carlo_switch = 1 THEN
BFU_INPUT_MC ELSE BFU_INPUT_CALC BFU_INPUT_CALC = 90.62
BFU_INPUT_MAX = 1e9 BFU_INPUT_MIN = 5e7 Body_Wt = 70 CFU_SURV = IF
Monte_Carlo_switch = 1 THEN CFU_SURV_MC ELSE CFU_SURV_CALC
CFU_SURV_CALC = 1.24 CFU_SURV_MAX = .35 CFU_SURV_MIN = .01
Current_Excel_Day_Number = Sim_Start_Excel_Day_Number+Time-2
EC50_CALC = 23.1800003051758 Epg_dose_amt_A = 0 Epg_dose_amt_B = 0
Epg_dose_amt_C = 0 Epg_dose_A_1 = IF ( TIME < Epg_dose_end_A - 3
) THEN PULSE ( Epg_dose_amt_A , Epg_DOSE_START_A , 7 ) ELSE 0
Epg_dose_A_2 = IF ( TIME < Epg_dose_end_A - 1 ) THEN PULSE (
Epg_dose_amt_A , Epg_DOSE_START_A + 2 , 7 ) ELSE 0 Epg_dose_A_3 =
IF ( TIME < Epg_dose_end_A + 1 ) THEN PULSE ( Epg_dose_amt_A ,
Epg_DOSE_START_A +4 , 7 ) ELSE 0 Epg_dose_B_1 = IF ( TIME <
Epg_dose_end_B - 3 ) THEN PULSE ( Epg_dose_amt_B , Epg_DOSE_START_B
, 7 ) ELSE 0 Epg_dose_B_2 = IF ( TIME < Epg_dose_end_B - 1 )
THEN PULSE ( Epg_dose_amt_B , Epg_DOSE_START_B + 2 , 7 ) ELSE 0
Epg_dose_B_3 = IF ( TIME < Epg_dose_end_B + 1 ) THEN PULSE (
Epg_dose_amt_B , Epg_DOSE_START_B +4 , 7 ) ELSE 0 Epg_dose_C_1 = IF
( TIME < Epg_dose_end_C - 3 ) THEN PULSE ( Epg_dose_amt_C ,
Epg_DOSE_START_C , 7 ) ELSE 0 Epg_dose_C_2 = IF ( TIME <
Epg_dose_end_C - 1 ) THEN PULSE ( Epg_dose_amt_C , Epg_DOSE_START_C
+ 2 , 7 ) ELSE 0 Epg_dose_C_3 = IF ( TIME < Epg_dose_end_C + 1 )
THEN PULSE ( Epg_dose_amt_C , Epg_DOSE_START_C +4 , 7 ) ELSE 0
Epg_dose_end_A = 1000 Epg_dose_end_B = 1000 Epg_dose_end_C = 1000
Epg_DOSE_START_A = 1000 Epg_DOSE_START_B = 1000 Epg_DOSE_START_C =
1000 Epg_Rx_regimen_A = Epg_dose_A_1 + Epg_dose_A_2 + Epg_dose_A_3
Epg_Rx_regimen_B = Epg_dose_B_1 + Epg_dose_B_2 + Epg_dose_B_3
Epg_Rx_regimen_C = Epg_dose_C_1 + Epg_dose_C_2 + Epg_dose_C_3
Epogen_Conc_pM = EPOGEN_AMT_pM / Vol_of_Dist Epogen_halftime = 1.2
Epogen_input_picomoles = ( EPOGEN_INPUT_U ) * 1e6 / 37100
EPOGEN_INPUT_U = setup_Epogen_input_U + ( IF time <
Epg_DOSE_START_A THEN Historical_Epogen_doses_U ELSE
Epogen_Rx_Pulses ) Epogen_Kd = 50e-12 Epogen_PSTD = IF Time <
First_Prescriptive_Sim_Day_Number THEN 0 ELSE
PULSE(Example_2_Per_Session_Epogen_Therapeutic_Dose,First_Prescriptiv-
e_Sim_Day_Number,7) Epogen_Rx_Pulses = (WTD_Switch - 1) *
(Epg_Rx_regimen_A + Epg_Rx_regimen_B + Epg_Rx_regimen_C) +
WTD_Switch*Epogen_PSTD EPOGEN_switch = 0 EPOR_fraction_bound =
ARANESP_switch * (Aranesp_Conc_pM * 1e-12) / ( Aranesp_Kd +
(Aranesp_Conc_pM * 1e-12 ) ) + EPOGEN_switch * ( Epogen_Conc_pM *
1e-12) / ( Epogen_Kd + (Epogen_Conc_pM * 1e-12 ) ) EPOR_MULT = IF
Monte_Carlo_switch = 1 THEN EPOR_MULT_MC ELSE EPOR_MULT_CALC
EPOR_MULT_CALC = 9.1 EPOR_MULT_MAX = 10 EPOR_MULT_MIN = 1
EPO_SETUP_RATE_CALC = 90.95 EPO_SETUP_RATE_MAX = 40
EPO_SETUP_RATE_MIN = 1 EPO_SETUP_RATE = IF Monte_Carlo_switch = 1
THEN EPO_SETUP_RATE_MC ELSE EPO_SETUP_RATE_CALC
erythroblast_production = dividing_cells[12]
erythroblast_production_CALC = 69.5800018310547
Example_2_Per_Session_Epogen_Therapeutic_Dose = 26
Example_2_Weekly_Therapeutic_Aranesp_Dose_Amount = 26
Example_Weekly_Therapeutic_Dose_Amount = 0
First_Prescriptive_Excel_Day_Number = 40482
First_Prescriptive_Sim_Day_Number =
First_Prescriptive_Excel_Day_Number- Sim_Start_Excel_Day_Number+2
fractional_apopt_sparing = MIN ( EPOR_fraction_bound * EPOR_MULT ,
1) HEM = 0 HEMATOCRIT = 42 * ( total_cells_in_circulation /
Vol_of_Dist ) / 5e12 HEMOGLOBIN = HEMATOCRIT * .34 hepatic_EPO_CALC
= 0 hgb_high = 12 hgb_low = 10 Historical_Aranesp_doses = IF
((mod(time,1) = .5) AND (CLINICAL_ARANESP_DATA > 1)) THEN (Pulse
( CLINICAL_ARANESP_DATA , time , 99999)) ELSE 0
Historical_Epogen_doses_U = IF ((mod(time,1) = .5) AND
(CLINICAL_EPOGEN_DATA > 1)) THEN (Pulse ( CLINICAL_EPOGEN_DATA ,
time , 99999)) ELSE 0 Last_Descriptive_Excel_Day_Number = 40480
maturation_time = 6 Monte_Carlo_switch = 0 MSE = if
Count_of_Hgb_Values > 0 THEN (Summed_Squared_Difference /
Count_of_Hgb_Values) ELSE 0 P7 = IF Monte_Carlo_switch = 1 THEN
P7_MC ELSE P7_CALC P7_CALC = 0 P7_MAX = 100 P7_MIN = 50
Plotted_Aranesp_Rx_Doses = 0 {Aranesp_Pulse / 8} plot_AR_dose_A =
Ar_dose_A/8 plot_historical_Ar_dose = Historical_Aranesp_doses/8
plot_setup_EPO_input = setup_EPO_input/8 Pt_ID = 6198 RBC_LIFESPAN
= IF Monte_Carlo_switch = 1 THEN RBC_LIFESPAN_MC ELSE
RBC_LIFESPAN_CALC RBC_LIFESPAN_CALC = 61.06 RBC_LIFESPAN_MAX = 120
RBC_LIFESPAN_MIN = 40 replication_interval = 1 RETIC_SURV = IF
Monte_Carlo_switch = 1 THEN RETIC_SURV_MC ELSE RETIC_SURV_CALC
RETIC_SURV_CALC = 96.03 RETIC_SURV_MAX = .8 RETIC_SURV_MIN = .2
Rx_protocol = (1 - Aranesp_WTD) * (Ar_dose_A + Ar_dose_B +
AR_dose_C) + WTD_Switch*Aranesp_WTD Scenario = 1
setup_Epogen_input_U = IF time < 8 THEN PULSE ( EPO_SETUP_RATE ,
-200, 2.33 ) ELSE 0 setup_EPO_input = IF time < 8 THEN PULSE (
EPO_SETUP_RATE , -200, 7 ) ELSE 0 setup_EPO_rate_CALC =
2.95000004768372 SimDays_in_Descriptive_Period =
Last_Descriptive_Excel_Day_Number-Sim_Start_Excel_Day_Number+1
Sim_Day_Number = Time Sim_Start_Excel_Day_Number = 40030
Squared_Difference = If CLINICAL_Hgb_Data > 0 Then (HEMOGLOBIN -
CLINICAL_Hgb_Data){circumflex over ( )}2 Else 0 sugg_BFU_max =
BFU_INPUT_CALC * 2 sugg_BFU_min = BFU_INPUT_CALC / 2
sugg_CFU_surv_max = MIN ( CFU_SURV_CALC + .2 , .80 )
sugg_CFU_surv_min = MAX ( CFU_SURV_CALC - .2 , .20 )
sugg_EPOR_mult_max = MIN ( EPOR_MULT_CALC * 2 , 10 )
sugg_EPOR_mult_min = MAX ( EPOR_MULT_CALC / 2 , 1 )
sugg_EPO_setup_max = MIN ( EPO_SETUP_RATE_CALC * 2 , 30 )
sugg_EPO_setup_min = MAX ( EPO_SETUP_RATE_CALC / 2 , 1 )
sugg_RBC_LIFE_max = RBC_LIFESPAN_CALC + 20 sugg_RBC_LIFE_min =
RBC_LIFESPAN_CALC - 20 sugg_retic_surv_max = MIN ( RETIC_SURV_CALC
+ 0.2 , .80 ) sugg_retic_surv_min = MAX ( RETIC_SURV_CALC - 0.2 ,
.20 ) total_cells_in_circulation = reticulocytes_in_circulation +
RBCs Vol_of_Dist = Body_Wt * .07 WTD_Switch = 0
[0239] The techniques described in this disclosure, including
functions performed by a processor, controller, control unit, or
control system, may be implemented within one or more of a general
purpose microprocessor, digital signal processor (DSP), application
specific integrated circuit (ASIC), field programmable gate array
(FPGA), programmable logic devices (PLDs), or other equivalent
logic devices. Accordingly, the terms "processor" "processing unit"
or "controller," as used herein, may refer to any one or more of
the foregoing structures or any other structure suitable for
implementation of the techniques described herein.
[0240] The various components illustrated herein may be realized by
any suitable combination of hardware, software, or firmware. In the
figures, various components are depicted as separate units or
modules. However, all or several of the various components
described with reference to these figures may be integrated into
combined units or modules within common hardware, firmware, and/or
software. Accordingly, the representation of features as
components, units, or modules is intended to highlight particular
functional features for ease of illustration, and does not
necessarily require realization of such features by separate
hardware, firmware, or software components. In some cases, various
units may be implemented as programmable processes performed by one
or more processors or controllers.
[0241] Any features described herein as modules, devices, or
components may be implemented together in an integrated logic
device or separately as discrete but interoperable logic devices.
In various aspects, such components may be formed at least in part
as one or more integrated circuit devices, which may be referred to
collectively as an integrated circuit device, such as an integrated
circuit chip or chipset. Such circuitry may be provided in a single
integrated circuit chip device or in multiple, interoperable
integrated circuit chip devices, and may be used in any of a
variety of pharmaceutical applications and devices.
[0242] If implemented in part by software, the techniques may be
realized at least in part by a computer-readable data storage
medium comprising code with instructions that, when executed by one
or more processors or controllers, performs one or more of the
methods described in this disclosure. The computer-readable storage
medium may form part of a computer program product, which may
include packaging materials. The computer-readable medium may
comprise random access memory (RAM) such as synchronous dynamic
random access memory (SDRAM), read-only memory (ROM), non-volatile
random access memory (NVRAM), electrically erasable programmable
read-only memory (EEPROM), embedded dynamic random access memory
(eDRAM), static random access memory (SRAM), flash memory, magnetic
or optical data storage media. Any software that is utilized may be
executed by one or more processors, such as one or more DSP's,
general purpose microprocessors, ASIC's, FPGA's, or other
equivalent integrated or discrete logic circuitry.
[0243] Various examples have been described. These and other
examples are within the scope of the following claims.
* * * * *