Efficacy and Safety of Epoietin Alfa in Critically Ill Patients NEJM 2007, 357: TMR Journal Club October 10, 2007 Katerina Pavenski, MD FRCPC TM Resident, McMaster University 1

Background Anemia in critically ill patients is common –95% of patients admitted to ICU have Hgb below normal by day 3 –42-50% of patients admitted to ICU will require RBC transfusion –85% of patients admitted to ICU for >13 days will require RBC transfusion 2 Zarychanski et al. CMAJ 2007

Background Etiology is multi-factorial –Hemodilution –Increased blood loss (bleeding, blood draws) –Shortened red blood cell survival –Impaired production Nutritional deficiencies Anemia of critical illness: decreased Epo production, blunted response to Epo, functional iron deficiency (upregulation of hepcidin, trapping of iron in macrophages, etc.) 3

Background Corwin et al Crit Care Med 1999, 27: –RCT, designed as superiority study (sample size not achieved because of poor accrual) –Setting: multidisciplinary ICU at 3 academic institutions –Study population: 80 patients per study group; excluded patients with recent thromboembolic disease (w/i 6 mos. of enrollment) –Outcomes: Cumulative blood transfusion requirement from day 1 Transfusion independence between Day 8 and 42 4

Background Corwin et al Crit Care Med 1999, 27: –Intervention: 300 U/kg OR placebo sc qd for 5 days and then alternate days until Hct>38% Iron supplementation (same as in 2007 study) RBC transfusion at discretion of attending physician; no transfusion protocol –Results Significantly less RBCs transfused in epo vs. placebo group (p<0.002) Final Hct concentration better in epo vs. placebo group (p<0.01) No significant differences in mortality or adverse events 5

Background Corwin et al, JAMA 2002, 288: –RCT, superiority with planned sample size of 1300 (90% power to detect absolute treatment difference of 10% vis-à- vis primary outcome) –Setting: medical, surgical or medico-surgical ICU at 65 US centres –Study population: 650 patients in Epo group and 652 in placebo group; did not exclude patients with thromboembolic events –Outcomes: Transfusion independence (% patients in each treatment group who received any RBC between study days 1 and 28) Cumulative RBC transfused per patient through day 28 Change in Hgb from baseline Time to first transfusion or death 6

Background Corwin et al, JAMA 2002, 288: –Intervention Epo 40,000 U OR placebo sc on day 3 and then weekly for three doses (4 th dose if in ICU on day 21) Iron supplementation (same as in 2007 study) RBC transfusion at discretion of treating physician; no transfusion if Hgb >9.0 g/dL or Hct >27% –Results Patients on Epo were less likely to undergo transfusion (60.4% vs. 50.5%, p<0.001, OR 0.67, 95% CI ) Reduction in the total # of RBCs transfused in Epo group and reduction in RBCs transfused per day alive Increase in Hgb from baseline was greater in Epo group Mortality and severe adverse events were not significantly different Post-hoc analysis: with Epo treatment, difference in MR in patient subgroups 7

Study Design Design: –Multi-centre, randomized, double-blind, placebo-controlled trial Hypothesis –Superiority with required sample size 1300 patients (power 80% to detect an absolute difference of 8% in the primary endpoint) –Blinded review led to increase in sample size to

Study Design Randomization between 48 and 96 hours after admission and stratified according to site and subgroup Randomization via computer-generated random numbers and concealed Intention-to-treat analysis Planned follow-up 140 days 9

Statistical Analysis Percentage of patients receiving RBC transfusion –Cochran-Mantel-Haenszel test –Stratified according to admission group –2 scenarios considered Numbers of RBC transfused –Wilcoxon-Mann-Whitney test –Transfusion rate – number of units that were transfused for a given patient divided by the total number of days the patient was alive Mortality –Kaplan-Meier method –Evaluated at days 29 and 140 –Cox regression model to evaluate interaction between study group and admission group 10

Study Population Inclusion criteria –Patients admitted to the participating ICUs and remaining in that ICU for 2 days; 18 years old or older; Hgb<12 g/dL; provided written informed consent Exclusion Criteria –Expected discharge from ICU w/i 48 hrs after the second day in the ICU; acute ischemic heart disease; stay more than 48 hours in the ICU of a transferring hospital; presence of LVAD; history of pulmonary embolus, DVT, ischemic stroke, other arterial or venous thrombotic event or a chronic hypercoagulable disorder; dialysis for any indication; uncontrolled hypertension after adequate antihypertensive therapy; new onset seizures within past 3 mos. Or seizures not controlled by medication; third degree burns on >20% BSA; pregnancy or lactation; diagnosis of acute, clinically significant GI bleeding on admission; transfusion at the time of planned enrollment; treatment with epoietin alfa within past 30 days; inability or unwillingness to receive blood products; participation in another study; hypersensitivity to epoietin alfa or any of its components 11

Study Population Setting –Medical, surgical and medico-surgical ICU at 115 centers Accrual –December 2003 to June

Study Design - Intervention Epoietin alfa 40,000 U OR placebo sc on days 1, 8, and 15 –Withhold study drug if hemoglobin >12g/dL on days 8 and 15 Elemental iron 150 mg po/NG starting on day 1 (or when first able to tolerate feeds) –Replace with parenteral iron if response inadequate (TS<20%, serum ferritin <100ng/ml) RBC transfusion on discretion of a treating physician –Target Hgb 7-9 g/dL 13

Study Design - Outcomes Primary –Percentage of patients receiving RBC transfusion between days 1 and 29 Secondary –# RBC transfused between days 1 and 42 –Mortality at days 29 and 140 –Change in hemoglobin concentration from baseline at day 29 14

Results: Study Flow 15

Results: Table 1 16

Results – Table 2 17

Results Hemoglobin concentration –At day 29, Hgb concentration from baseline was greater in epoietin arm vs. placebo arm (1.6+/-2.0 g/dL vs. 1.2+/1.8 g/dL, p<0.001) –At day 29, absolute Hgb concentration was greater in epoietin arm vs. placebo arm (11.2+/-1.8 g/dL vs /-1.7 g/dL, p<0.001) –By day 42, Hgb concentration was not statistically different in two groups 18

Results: Figure 2 19

Results At day 29, mortality was significantly lower in the epoietin group (8.5% vs. 11.4%, p=0.02) –In trauma group, mortality was also significantly lower in the epoietin group (3.5% vs. 6.6%, p=0.04) At day 140, mortality was similar in two groups –For all patients, 14.2% (epoietin) vs. 16.8% (placebo), p=0.08 –For trauma group, 6.0% (epoietin) vs. 9.2% (placebo), p=0.08 Cox model –Hazard ratios for mortality in the trauma patients were significant at day 29 and day

Results: Table 3 21

Adverse Events At least one adverse event –94.4% of patients on placebo, 94.8% of patients on epoietin Serious adverse event –43.5% of patients on placebo, 44.0% of patients on epoietin –Significantly increased incidence of thrombotic vascular events in epoietin group (16.5% vs. 11.5%, hazard ratio 1.41, 95% CI 1.06 to 1.86, p=0.008) Most apparent in those who received three doses of epoietin alfa vs. three doses of placebo 22

Adverse Events: Table 4 23

Results No significant difference between the epoietin and placebo groups in median length of ICU stay Similar number of ventilator-free days Similar median duration of mechanical ventilation At day 140, ventilation has been discontinued for 96.6% of patients on epoietin and 98.4% of patients on placebo (p=0.02) 24

Author’s Recommendations 25 Corwin CMAJ 2007

Critical Appraisal Are results valid? –Patients randomized? Yes –Randomization concealed? Yes –Was follow-up sufficiently long and complete? 140 days probably enough 109 patients (7.5%) lost to follow-up (7.6% in placebo group, 7.4% in Epo group) –Patients analyzed in the groups to which they were randomized? Yes (intention-to-treat) –Were patients and clinicians blinded? Yes –Were the groups treated equally, apart from experimental treatment? Yes –Were the groups similar at the start of the trial? Yes 26

Critical Appraisal Are the valid results of this randomized trial important? –Trial was powered for the primary outcome (% of patients receiving RBC transfusion) Conclusion that “epoietin alfa does not reduce the incidence of red cell transfusion among critically ill patients” is definitely trustworthy –Trial was not powered for the composite secondary outcomes (including mortality) –Overall mortality was not affected by Epo –However, pre-planned subgroup analysis showed mortality benefit in trauma subgroup

Critical Appraisal Therefore, the findings of mortality benefit should be viewed as hypothesis generating Potential mortality benefit of Epo –At day 29, RRR 47%, ARR 3.1%, NNT 30 –At day 140, RRR 35%, ARR 3.2, NNT 31 Likelihood of help versus harm –Unable to calculate – unclear how many TE were in the trauma group –Trauma patients are inherently pro-thrombotic (higher baseline risk of TE) –Real TE rate in this study may be higher than reported because there was no active surveillance for TE

Critical Appraisal Is apparent difference in treatment efficacy among subgroups likely? –Does it make biological and clinical sense? No Epo acts as anti-apoptotic agent and may protect cells from hypoxemia and ischemia Presumably this happens in all critical illnesses, so why would trauma patients have a particular benefit? (stats) –Is this difference clinically and statistically significant? Yes and yes –Was it hypothesized before the study began? Yes –Was it confirmed in other studies? Suggested by previous Corwin’s study –Was it one of just a few subgroup analyses carried out in the study? Yes 29

Critical Appraisal Are these valid results applicable to our patient? –Where the study patients similar to my patients? Yes –Is the treatment feasible in our setting? Yes –Were all clinically important outcomes considered? Yes –Are the likely treatment benefits worth the potential harm and costs? Unclear So where does this leave us?

Instead of Conclusion R. Zarychanski et al. CMAJ 2007, 177: –Meta-analysis identified 9 relevant RCTs (3326 patients) included 3 Corwin’s studies –Epoietin alfa (compared to placebo or no intervention) had no statistically significant effect on overall mortality (OR 0.86, 95% CI , I 2 0%) –Epo alfa (compared with placebo) significantly reduced odds of a patient receiving at least 1 transfusion (OR 0.73, 95% CI , I %) –Mean number of RBC transfused per patient decreased by 0.41 units in Epo group (95% CI , I %) 31