Today, nearly all dialysis patients receive EPO therapy
We used monthly hematocrit, hospitalization, prior EPO dose, iron use, blood transfusions, dialysis sessions, urea reduction ratio; and baseline age at ESRD onset, race, gender, initial EPO dose per administration, underlying cause of ESRD, pre-dialysis hematocrit, presence of cardiovascular and noncardiovascular co-morbidities, geographic region, dialysis chain membership, hypertension, body mass index, GFR, and serum creatinine
Thank you, Dr. Herann. I will now present our research findings.
In this study, we examined the relationship between Epo dose and population HCT response. This research is supported by R01 NIDDK grant and has been accepted by Kidney International.
This figure shows mean haemoglobin concentrations and weekly EPO doses in prevalent dialysis patients. The red line is for EPO dose and the blue is for Haemoglobin. Between 1991 and 2005, the mean administered dose of epoetin increased more than 300% in the US. However, the effect of epo dose and average HCT response in the ESRD population has not been evaluated.
The only exsiting clinical study that examined dose and HCT relationship is the Phase II trials. However, phase II did not fully capture the population dose response because of its strict patient selection criteria and that non-responsive patients were not included. Subsequent studies examining the relationship between EPO dose and hematocrit (or hemoglobin) have primarily used administrative databases. However, these studies often showed an inverse association between dose and HCT due to bias by indication. The true relationship between EPO dose and HCT should not be negative.
So our aim was to estimate the relationship between EPO dose and hematocrit by using causal inference techniques to control for treatemnt-by-indication bias. In doing so, we tried to mimic an RCT in which patients are randomized on EPO dose and then we compared the achieved hematocrit in each arm.
We used the 2003 and 2004 USRDS data, which are the most recent available data for researchers. USRD is an administrative database that contains EPO dose and HCT info based on monthly claims.
We restricted our analysis to patients who were ≥65 years of age (because EPO use before the initiation of dialysis can be reliably determined in this group), started hemodialysis and EPO treatment within 90 days after their first ESRD service date to avoid left censoring. Patients who used predialysis epo were excluded to ensure we have complete EPO therapy information, did not have a kidney transplant, HIV or cancer before starting dialysis (because these patients might respond differently to EPO therapy).
Patients were censored on these censoring events, whichever occurred first. We estimated the effect of average EPO dose in the first three months on dialysis on the achieved hematocrit at month four among incident dialysis patients who were EPO naïve. By choosing EPO naïve patients and limiting the observation period to the first three months of treatment, we selected the period in which change in hematocrit would be greatest . The outcome was chosen at the end of the fourth month period because a 2 – 4 week lag period has been reported for EPO to affect hematocrit.
We used margianl structural models to adjust for confounding by indication and we contructed a dose-response curve. Since Dr. Hernan has just presented an overview of how MSM works, I will focus on the results.
This is the distribution of patients by initialntial EPO doses.30% patients used initial doses recommend by FDA, while more than 60% received dose higher than recommended starting doses.
This is the distribution of patients by hematocrit at the end of first 3 months. More than 30% of patient achieved HCT greater than 39% and received highest epo doses in the first 3 months..
This is the estimated dose-response curve. x-axis is--. Y axis is --. The dotted lines indicate 95% percent confidence intervals. A starting dose of 13,500 units/week would result in an average hematocrit of 36%, t he greatest increases in the average hematocrit of the population take place with EPO doses between 9,000 and 22,500 units/week. At higher doses of EPO, the average population hematocrit plateaus at 38.5%. Red dots indicate the fda-recommended starting doses. which are located on the linear portion of the curve.
the curve based on standard linear regression analysis is flatter, plateaus lower and is is less biologically plausible than our estimated curve.
This study has several limitations . First, there might be residule confounding introducced by unmeasured clininal factors. I t's ideal for us to have treatment-to treatment data/from one epo administration to another. However, what we have in the claims data are HCT taken at the end of the month and the total monthly EPO dose. So we may not necessarily have the HCT that physicians are actually based upon when making dosing decisions. To the extent therefore that the r esidual confounding by unmeasured factors exists, it might underestimate the average hematocrit for large EPO doses, thus shifting the curve downwards . Inaddition, this administrative data source does not contain laboratory results such as nutritional status, blood pressure or inflammatory state, although in practice, decisions on EPO dosing are largely based on the archived hematocrit level. Other caveats include that the research question may not reflect current anemia management strategies and the study conclusions may not be generilizeble to different patient population.
Conclusions Dose-response is S-shaped. HCT plateaus at 38.5%. Normal HCT target might not be achievable for dialysis population, and starting doses recommended by FDA are appropriate.
MTPPI EPO Outcomes Research Presented to FDA /CDER Joint Meeting of the Cardiovascular and Renal Drugs & Drug Safety and Risk Management Advisory Committees September 11, 2007 Hilton Washington Gaithersburg, MD
Background and Context Dennis Cotter President of Medical Technology and Practice Patterns Institute (MTPPI) 4733 Bethesda Avenue #510 Bethesda, MD 20814
Decade-long study of EPO <ul><li>Identified Medicare and non-Medicare use of EPO </li></ul><ul><li>Quantified total EPO use among dialysis patients </li></ul><ul><li>Currently, PI on R01 grant focusing on the role of EPO dosing and patient outcomes </li></ul>
Hemoglobin values have increased steadily after EPO introduced Source: USRDS 2006 Annual Data Report
Widespread EPO use based on 2000 DOQI findings including: <ul><ul><li>Survival benefits </li></ul></ul><ul><ul><li>Decreased incidence of hospitalization </li></ul></ul><ul><ul><li>Partial regression of left ventricular hypertrophy (LVH) </li></ul></ul><ul><ul><li>Improved quality of life </li></ul></ul><ul><ul><li>Increased exercise capacity </li></ul></ul>
However, survival findings might have been confounded by EPO treatment itself
Application of causal modeling techniques Received R01 grant (5R01DK066011-02 Epoetin Therapy and Survival of Hemodialysis Patients) to examine the role of EPO treatment in patient outcomes
Introduction to Causal Modeling Miguel Hernán Associate Professor of Epidemiology Department of Epidemiology Harvard School of Public Health 677 Huntington Avenue Boston, MA 02115
Goal To estimate the effect of EPO on hematocrit and survival among renal failure patients with anemia A RCT would be ideal Next best thing is an observational study that mimics an RCT
Problem with observational studies Patients with worse prognosis tend to receive higher EPO doses (confounding by indication) Not a problem in ITT analyses of RCTs
Actually, there are 2 problems 1. Confounding may be unmeasured 2. Confounding may be measured but inappropriately adjusted for
Problem 1 Unmeasured confounding THE fundamental problem Need measurements of all important prognosis factors that are also indications for treatment but can never prove you have all confounders
Problem 2 Inappropriately adjusting for confounding Conventional statistical methods cannot appropriately adjust for confounding When the prognosis factors (e.g., hematocrit) that affect treatment decisions (e.g., EPO dose) are themselves affected by prior treatment decisions A solvable problem: just use inverse probability weighting (IPW)
IPW: Utility Can be used to mimic an RCT using observational data Under the assumption of no unmeasured confounding Even in the presence of time-varying confounders affected by prior treatment
IPW: Technical details Each subject is weighted by the inverse of the estimated probability of receiving the EPO dose that he actually received Essentially equivalent to standardization The corresponding weighted models estimate the parameters of marginal structural models
IPW: Examples of application IPW extensively used in HIV/AIDS research In fact, NIH required expertise on IPW when requesting applications for estimating the effects of antiretrovirals from observational data IPW replicated estimates from RCTs in the HIV/AIDS field
IPW: Our application We used IPW to estimate the survival and mean hematocrit of subjects randomly assigned to different EPO doses We needed IPW because hematocrit is a time-dependent confounder (predicts both EPO dose and outcome) and is affected by prior EPO dose
Research Findings <ul><ul><li>Yi Zhang </li></ul></ul><ul><ul><li>Senior Analyst </li></ul></ul><ul><ul><li>MTPPI </li></ul></ul>
The effect of EPO dose on hematocrit response among elderly hemodialysis patients in the U.S. Cotter D, Zhang Y, Thamer M, Kaufman J, Hernán MA. Kidney International 2007 [in press]
Mean monthly hemoglobin and mean EPO dose per week
Prior research <ul><li>Dose response relationship has not been examined since Phase II trials </li></ul><ul><ul><li>Stringent patient eligibility criteria </li></ul></ul><ul><ul><li>Limited dose </li></ul></ul><ul><li>Observational studies have shown an inverse relationship between EPO dose and hematocrit </li></ul><ul><ul><li>Confounding by indication </li></ul></ul>
Research goals To mimic an RCT in which subjects are randomly assigned to different arms, each receiving a different EPO dose To compare the achieved hematocrit in each arm
Data source <ul><li>United States Renal Data System (USRDS) </li></ul><ul><ul><li>administrative database on ESRD patients whose care is covered by Medicare </li></ul></ul><ul><ul><li>include extensive baseline and follow-up demographic and clinical data </li></ul></ul><ul><ul><li>outpatient EPO claims include monthly total EPO dose and hematocrit values </li></ul></ul><ul><ul><li>most recent USRDS data available for researchers </li></ul></ul>
Patient population <ul><li>Retrospective cohort study. </li></ul><ul><li>14,001 patients who started EPO and dialysis in 2003. </li></ul><ul><ul><li>>=65 years of age </li></ul></ul><ul><ul><li>had first claim with 90 days of their first ESRD service date </li></ul></ul><ul><ul><li>had not used EPO before </li></ul></ul><ul><ul><li>did not have a kidney transplant, HIV or cancer before starting dialysis. </li></ul></ul><ul><ul><li>were not censored during the first complete dialysis month </li></ul></ul>
Study variables <ul><li>Censoring events </li></ul><ul><ul><li>change of dialysis modality, transplantation, 30 days after change of dialysis provider, gap in outpatient dialysis services, or death </li></ul></ul><ul><li>Exposure: Average EPO dose in the first 3 months of dialysis </li></ul><ul><li>Outcome: HCT at month 4 </li></ul>
Statistical methods <ul><li>Estimated inverse probability weights to adjust for measured confounders, and then fit a weighted regression model </li></ul><ul><li>Constructed a dose-response curve </li></ul><ul><li>Each hematocrit-EPO dose point in the curve shows the estimated average hematocrit if subjects had been randomly assigned to that EPO dose </li></ul><ul><li>95% CI were based on bootstrap techniques </li></ul>
Distribution of patients by hematocrit group 23,400 21,000 21,100 21,500 26,000 Average EPO dose (U/week)
4 Dose response curve and 95% confidence intervals based on MSM
Dose response curve based on standard adjustment
Study limitations <ul><li>Potential for unmeasured confounding </li></ul><ul><ul><ul><li>Monthly HCT and EPO dose </li></ul></ul></ul><ul><ul><ul><li>Unobserved clinical factors (iron level, blood pressure, nutritional status...) </li></ul></ul></ul><ul><ul><ul><li>EPO use in the hospital, route of EPO administration </li></ul></ul></ul><ul><li>Did not consider dynamic EPO dosing regimes </li></ul><ul><li>Restriction of study period and population </li></ul>
Conclusions <ul><li>Dose-response curve is S-shaped </li></ul><ul><li>HCT plateaus at 38.5% for average EPO doses greater than 20,000 units/week </li></ul><ul><li>Normal HCT target might not be achievable for dialysis population </li></ul><ul><li>Starting doses recommended by FDA are appropriate and are in the linear portion of the curve </li></ul>
The relationship between EPO dose and survival among hemodialysis patients Zhang Y, Thamer, Cotter D, Kaufman J, Hern á n MA Joint Statistical Meetings 2007 [Abstract]
Research goals To mimic an RCT in which subjects are randomly assigned to different arms, each receiving a different average dose of EPO To compare the survival in each arm
Previous research <ul><li>A plethora of observational studies have shown that higher hematocrit is associated with better survival for dialysis patients </li></ul><ul><li>However, results of clinical trials demonstrated that patients targeted to higher hematocrit levels did not show survival benefits </li></ul><ul><ul><li>led to a recent FDA black box warning </li></ul></ul><ul><li>The EPO dose-survival relationship has not been empirically determined </li></ul>
Study design <ul><li>20,580 incident hemodialysis patients </li></ul><ul><li>Eligibility criteria </li></ul><ul><ul><li>Age 65 and older </li></ul></ul><ul><ul><li>First ESRD service in 2003 </li></ul></ul><ul><ul><li>Attend freestanding facilities </li></ul></ul><ul><ul><li>Complete baseline (first 3 months of dialysis) data </li></ul></ul><ul><li>Exposure: cumulative average EPO dose </li></ul><ul><li>Outcome: death during months 4-12 </li></ul><ul><li>Censored if change of provider/modality, or loss to follow-up </li></ul>
Methods <ul><li>Estimated inverse probability weights to adjust for measured confounders, and then fit a weighted Cox model </li></ul><ul><li>Constructed survival curves for each EPO dose </li></ul><ul><li>Each curve shows the survival if subjects had been randomly assigned to that EPO dose </li></ul><ul><li>95% CI were based on bootstrap techniques </li></ul>
Mortality hazard ratios by EPO dose (quartiles)
Survival for EPO doses based on 3 different doses
Study limitations <ul><li>Potential for unmeasured confounding as always </li></ul><ul><li>Did not consider dynamic EPO dosing regimes </li></ul><ul><li>One-year survival only </li></ul>
<ul><li>Lowest mortality found for average EPO doses of 8,500-15,000 units per week </li></ul><ul><li>Treating all patients with higher EPO doses (>15,000 U/wk) might decrease average survival </li></ul>Conclusions
Relevance of research findings to FDA labeling decisions INITIAL DOSE In our study cohort, 61% of all incident elderly dialysis patients received an initial EPO dose higher than the FDA-approved 50-100 U/kg range DOSE-RESPONSE Based on our dose-response model, a population average EPO dose higher than 12,000 U/week would result in exceeding the FDA-approved HCT target of 36% RISK Based on our dose-survival model, a population average EPO dose higher than 15,000 U/week would result in progressively higher mortality risks HYPORESPONSIVE PATIENTS The risk of increased mortality is greatest among hyporesponsive patients who receive the largest EPO doses Return to Cotter