AJN314628.indd Fax +41 61 306 12 34 E-Mail karger@karger.ch www.karger.com Hot Topics in Nephrology: A Debate Am J Nephrol 2010;31:552–556 DOI: 10.1159/000314628 Should Hemoglobin Targets for Anemic Patients with Chronic Kidney Disease Be Changed? Ajay K. Singh  Renal Division, Brigham and Women’s Hospital and Harvard Medical School, Boston, Mass. , USA a higher Hb in dialysis patients. Aiming for an Hb of 1 9 g/dl as a minimum threshold is commensurate with the placebo arm of the TREAT study as well as the lower Hb arm of the Normal Hematocrit study. Having an up- per limit for Hb (i.e. a target range) is not supported by evidence and encourages targeting with ESAs to just be- low 12 g/dl and within the target range, but beyond the prevailing evidence with respect to safety. The focus needs to be either avoiding ESAs entirely where possible, or re- ducing exposure to high ESA dosage where necessary. Intervention in these four trials (comprising over 7,000 patients) involved ‘targeting a higher Hb concentration’. This ‘targeting’ of higher Hb embodied treatment with ESAs. In fact, in all of the four trials, an algorithm con- trolled ESA dosage. In addition, in all of the trials, achiev- ing a higher Hb concentration was associated with better outcomes compared to achieving a lower Hb level. In oth- er words, targeting a higher Hb with ESA, not the actual achieving, is the problem. Thus, the singular focus on an Hb range misses the point. There is no evidence to indi- cate increased mortality or cardiovascular risk with oth- er interventions in targeting a higher Hb – blood transfu- sions or iron. Indeed, dialysis patients with higher Hb because of high altitude have better outcomes than pa- tients with lower Hb [9] . Regardless of the achieved Hb, treatment with high dosages of epoetin independently predicts death and cardiovascular complications. Four randomized controlled trials (RCTs) have dem- onstrated that erythropoiesis-stimulating agents (ESAs) targeted to normalize hemoglobin (Hb) in chronic kidney disease (CKD) patients result in a higher rate of death and/ or cardiovascular complications. In the Normal Hemato- crit study, the point estimate of risk in the direction of harm was 30% [1] ; in CREATE, it was 22% (95% CI: 0.53– 1.14) [2] ; in CHOIR, it was 34% (95% CI: 1.03–1.74) [3] , and in TREAT it was 5% (95% CI: 0.94–1.17) [4] . Several ob- servational analyses [5–9] implicate exposure to high dos- ages of ESAs in explaining these adverse outcomes. Stud- ies in nonrenal settings confirm the direct risk conferred by ESAs. The case for ESA toxicity in explaining the risk of targeting a higher Hb in CKD patients is strong. The case is weak for a continued obsession with the current target range for Hb of 10–12 g/dl (the FDA recommenda- tion) or 11–12 g/dl (K-DOQI). The status quo must change. The anemia RCTs demonstrate that the Hb is a f lawed surrogate end point, fundamentally undermining the current focus on an Hb target range. Fleming and DeMets [10] emphasize that a valid surrogate should both corre- late with the true clinical outcome and fully capture the net effect of treatment on the clinical outcome. Hb does not do this because targeting a higher Hb in the trials is not associated with a reduction in mortality or a lower rate of cardiovascular complications. On the contrary, the RCTs demonstrate there is increased risk in targeting Published online: May 21, 2010 Nephrology American Journal of Ajay K. Singh, MD Renal Division Brigham and Women’s Hospital Boston, MA 02115 (USA) Tel. +1 617 732 5951, Fax +1 617 732 6392, E-Mail asingh   @   partners.org © 2010 S. Karger AG, Basel 0250–8095/10/0316–0552$26.00/0 Accessible online at: www.karger.com/ajn Debate: PRO Position http://dx.doi.org/10.1159%2F000314628 Debate: PRO Position Am J Nephrol 2010;31:552–556 553 Evidence that Targeting a Higher Hemoglobin with ESA Therapy Is Harmful The design characteristics of the four large RCTs are shown in table 1 , and have been discussed in detail else- where [11] . The Normal Hematocrit study [1] enrolled symptom- atic high-risk dialysis patients, who were randomized to either an Hb of 13–15 g/dl or an Hb of 9–11 g/dl. The mean epoetin dosage was 460 U/kg/week and 160 U/kg/ week, in the high versus low Hb arms, respectively. The Data Safety Monitoring Board halted the study for safety reasons. At 29 months, there were 183 deaths and 19 first nonfatal myocardial infarctions in the higher Hb versus 150 deaths and 14 nonfatal myocardial infarctions in the lower Hb group (RR: 1.3; 95% CI: 0.9–1.9). There was also a higher rate of vascular thrombosis and strokes in pa- tients in the higher Hb arm compared to patients ran- domized to the lower Hb arm. Three RCTs have evaluated nondialysis CKD patients: CREATE, CHOIR and TREAT ( table 1 ). All three trials demonstrated increased risk in targeting higher Hb with higher doses of ESAs. The CREATE study evaluated the effect of complete versus partial correction of anemia in 603 patients with CKD (12). The achieved Hb was 13.49 g/dl in the high Hb group versus 11.6 g/dl in the low Hb group. A median dose of 5,000 versus 2,000 units of epoetin- � per week was used in the higher versus lower Hb group, respec- tively. At 4 years, complete anemia correction was not as- sociated with a higher rate of the first cardiovascular event (HR: 0.78; p = 0.20), although there was a trend to- wards harm. There was a significantly higher risk of de- veloping end-stage renal disease in patients randomized to the higher Hb concentration. The CHOIR study [3] enrolled 1,432 patients with CKD anemia and compared the effect of raising Hb to high (13.5 g/dl) as compared to low (11.3 g/dl) levels on outcomes. The median epoetin dose used in the trial was 10,952 U/week in the high Hb group and 5,506 U/week in the low Hb arm. There were 125 composite events (death, myocardial infarction, congestive heart failure hospital- ization and stroke) in patients in the higher Hb group versus 97 events in the low Hb group (HR 1.337; p = 0.03). The higher rate of composite events was explained large- ly by a higher rate of death (48% higher risk; p = 0.07) and congestive heart failure hospitalization (41%; p = 0.07). The TREAT study [4] was a double-blind trial com- prising 4,038 subjects. Patients were randomized to either darbepoetin or placebo, with a target Hb of 13 g/dl in the darbepoetin treatment arm and an Hb above 9 g/dl in the placebo arm. A median dose of 176 � g/month was used in the darbepoetin-treated arm compared to 0 � g/month in the placebo rescue arm. The trial was neutral for the Table 1. D esign characteristics for anemia RCTs Normal Hematocrit CREATE CHOIR TREAT D esign randomized, open-label randomized, open-label randomized, open-label randomized, double-blind Sponsor/agent Amgen/Epogent� (epoetin-�) Amgen/Aranesp� (darbepoetin-�) J&J/Procrit� (epoetin-�) Amgen/Aranesp� (darbepoetin-�) Dosing unclear 2,000 weekly initiate 10,000 weekly when stable go to bi-weekly 0.75 mcg/kg/Q2W double dose when stable and go to monthly Dosing frequency 3 times weekly on dialysis de novo to weekly de novo to weekly to bi-weekly de novo to bi-weekly to monthly Hb target(s), g/l arm 1 9–11 13.0–15.0 13.0 13.0 arm 2 13–15 10.5–11.5 placebo (rescue for Hb <9.0) placebo (rescue for Hb <9.0) Regions USA global USA global Inclusion criteria Hb, g/l 9–11.0 11.0–12.5 <11.0 ≤11.0 eGFR/CrCl ESRD 15–35 15–50 20–60 Diabetes ;44% ;25% 48.5% 100% Singh Am J Nephrol 2010;31:552–556554 primary composite of death or a cardiovascular event (HR for darbepoetin vs. placebo: 1.05; p = 0.41), but there was a significantly higher rate of strokes in the darbepo- etin-treated patients (HR: 1.92; p ! 0.001). Death or end- stage renal disease occurred in 652 patients in the darbe- poetin- � group (32.4%) and in 618 patients in the placebo group (30.5%; HR for darbepoetin- � versus placebo: 1.06; 95% CI: 0.95–1.19; p = 0.29). A higher rate of both throm- boembolism and cancer-related deaths among patients with a history of cancer in the darbepoetin-treated pa- tients was also observed. Taken collectively, the anemia RCTs prove the inade- quacy of Hb as a valid surrogate end point and point to the targeting of a higher Hb with ESA as being the key problem. The next question is whether exposure to epo- etin, especially at high dosage levels, independently pre- dicts adverse outcome. In a secondary analysis of the CHOIR study [12] , the question of whether exposure to epoetin- � explained the higher risk of adverse events observed with anemia treat- ment was evaluated. Landmark analyses at 4 and 9 months was used to avoid some of the biases and confounding in- herent in post-hoc studies. In unadjusted analyses, both the inability to achieve target hemoglobin and the re- quirement of high-dose epoetin were significantly associ- ated with an increased hazard of the primary end point (p = 0.05 and 0.003, respectively). In adjusted models, the increased hazard associated with randomization to the high hemoglobin arm from the primary trial was no lon- ger significant (p = 0.49), while high-dose epoetin was as- sociated with a 57% increased hazard to the primary end point (HR: 1.57; 95% CI: 1.04–2.36; p = 0.03). Thus, expo- sure to high doses of epoetin and not the targeted Hb in- dependently predicted adverse outcomes in CHOIR. Observational Studies Support ESA Toxicity Several observational analyses have examined the re- lationship between epoetin exposure and adverse risk in treating anemia in the CKD population. Zhang et al. [5] studied the relationship between epoetin and all-cause mortality in 94,569 prevalent hemodialysis patients using Cox proportional hazard regression analysis with adjust- ment for baseline variables. For every hematocrit strata studied, patients administered higher doses of epoetin had significantly lower hematocrit values and greater mortality rates. Using the cubic spline function, a signif- icant nonlinear relationship between increased epoetin dose and mortality was found regardless of hematocrit (p  ! 0.0001), with the steepest increase in relative risk for death found after the 72.5 dose percentile. Bradbury et al. [6] explored a Fresenius North Amer- ica cohort of 22,955 prevalent hemodialysis patients using Cox proportional hazard models and time-dependent models fitted with time-varying log EPO and Hb concen- tration. In the unadjusted model, after adjustment for baseline patient characteristics, an increased mortality risk with increasing epoetin dose was observed (HR: 1.31 per log unit increase; 95% CI: 1.26–1.36). However, ad- justment for baseline patient characteristics resulted in attenuation of the mortality risk estimate (HR: 1.21; 95% CI: 1.15–1.28) that became more attenuated in lagged time-dependent analyses. Streja et al. [7] explored the relationship between epo- etin exposure, iron deficiency and thrombocytosis in 40,787 DaVita maintenance hemodialysis patients. A higher Hb 1 13 g/dl was associated with greater mortality (case-mix-adjusted death relative risk of 1.21; 95% CI: 1.02–1.44; p = 0.03) in the presence of thrombocytosis (platelet count 1 300,000/ � l), but not in the absence of thrombocytosis. However, there was an association be- tween epoetin exposure at very high doses of 1 20,000 units/week, and mortality over 3 years (relative risk of death: 1.59, 95% CI: 1.54–1.65; p ! 0.001). Servilla et al. [8] evaluated 12,733 epoetin-exposed in- cident hemodialysis patients. A proportional hazards modeling with time-varying covariates was used. Epoe- tin doses ! 8,000 U/week were associated with decreased risk. Higher epoetin doses were associated with increased mortality at Hb concentrations of 10–12.9 g/dl and with increased hospitalization at all Hb concentrations of 10 g/dl or greater. Higher epoetin doses were also associated with increased mortality and hospitalization within each tertile of serum albumin concentration. Winkelmayer et al. [9] used instrumental variable modeling to examine the relationship between epoetin and outcome in 269,717 subjects in 4,500 dialysis units in the United States. Mortality was low among patients with a low Hb exposed to high doses of epoetin. However, mortality rates were increased in centers that used larger ESA doses in patients with hematocrit between 33 and 35.9% (highest vs. lowest quintile of predicted dose, HR: 1.07; 95% CI: 1.03–1.12) and in those with hematocrit of 36% or higher (highest vs. lowest quintile of predicted dose, HR: 1.11; 95% CI: 1.07–1.15). Synthesizing the observational studies, the evidence suggests that there is indeed a relationship between ESA exposure and adverse outcome, but since confounding cannot be excluded, causality cannot be established. Debate: PRO Position Am J Nephrol 2010;31:552–556 555 Evidence of ESA-Associated Adverse Outcomes in Nonrenal Populations Demonstrating that ESA therapy used in nonrenal set- tings can be harmful lends further support to the case for ESA toxicity. ESA therapy has been used in a variety of nonrenal settings, including the treatment of cancer-in- duced anemia, anemia of critical illness and in prevent- ing blood transfusions prior to spine surgery. Bohlius et al. [13] did a meta-analysis of 53 trials com- prising 13,933 patients with cancer-induced anemia who received epoetin or darbepoetin plus red blood cell trans- fusion for treatment of anemia compared to patients re- ceiving only transfusion. High doses of ESAs were used (21,000–63,000 IU of epoetin or 100–157 � g of darbepo- etin per week for 8–52 weeks). ESAs increased all-cause mortality by 17% in all patients compared to control groups, and by 10% in patients undergoing chemotherapy compared to control groups. In another meta-analysis, Bennett et al. [14] evaluated 51 phase 3 trials comprising 13,611 patients with cancer (venous thromboembolism risk was evaluated in 8,172 patients with cancer from 38 phase 3 trials). The risk of mortality was also greater in ESA-treated patients (HR: 1.10; 95% CI 1.01–1.20). A sig- nificantly increased risk of venous thromboembolism in patients treated with ESAs (334 events /4,610 patients) versus control patients (173 events/3,562 control patients; RR: 1.57; 95% CI: 1.31–1.87) was observed. Furthermore, there was increased mortality among those with chemo- therapy-induced anemia (HR: 1.29; 95% CI: 1.00–1.67, p  = 0.05) and chemotherapy-associated anemia (HR: 1.09; 95% CI: 0.99–1.19). The most convincing evidence of a direct adverse ef- fect of ESA is from a prospective, multicenter, open-label, randomized, parallel group trial by Stowell et al. [15] . Subjects received either epoetin- � 600 U/kg subcutane- ously once weekly starting 3 weeks before spinal surgery plus standard of care for blood conservation or standard of care alone, regardless of baseline Hb concentration. There were 340 in each treatment group (n = 680): 16 sub- jects (4.7%) in the epoetin- � group and 7 subjects (2.1%) in the standard of care group had a diagnosis of deep vein thrombosis and 1.5% and 0.9%, respectively, had other clinically relevant thrombovascular events. In summary, evidence from ESA treatment in nonre- nal populations taken together with the CKD data impli- cates exposure to high doses of ESAs as the likeliest rea- son for the increased risk of adverse outcomes. Conclusion Targeting a higher Hb concentration with high dosage of ESA in CKD patients is associated with increased risk. Aiming for an Hb of 1 9 g/dl is commensurate with the placebo arm of the TREAT study as well as the lower Hb arm of the Normal Hematocrit study. However, rather than focusing on the Hb, which at best is an unreliable surrogate, attention needs to be directed at minimizing exposure to high ESA dosage. The evidence for ESA expo- sure particularly at high dosage is strong, albeit more cir- cumstantial. However, as Henry David Thoreau is fa- mously quoted saying, ‘Some circumstantial evidence is very strong, as when you find a trout in the milk’. References 1 Besarab A, Bolton WK, Browne JK, Egrie JC, Nissenson AR, Okamoto DM, Schwab SJ, Goodkin DA: The effects of normal as com- pared with low hematocrit values in patients with cardiac disease who are receiving he- modialysis and epoetin. N Engl J Med 1998; 339: 584–590. 2 Drueke TB, Locatelli F, Clyne N, Eckardt KU, Macdougall IC, Tsakiris D, Burger HU, Scherhag A; CREATE Investigators: Nor- malization of hemoglobin level in patients with chronic kidney disease and anemia. N Engl J Med 2006; 355: 2071–2074. 3 Singh AK, Szczech L, Tang KL, Barnhart H, Sapp S, Wolfson M, Reddan D; CHOIR In- vestigators: Correction of anemia with epo- etin alfa in chronic kidney disease. N Engl J Med 2006; 355: 2085–2098. 4 Pfeffer MA, Burdmann EA, Chen CY, Coo- per ME, de Zeeuw D, Eckardt KU, Feyzi JM, Ivanovich P, Kewalramani R, Levey AS, Lew- is EF, McGill JB, McMurray JJ, Parfrey P, Parving HH, Remuzzi G, Singh AK, Solo- mon SD, Toto R; the TREAT Investigators: A trial of darbepoetin alfa in type 2 diabetes and chronic kidney disease. N Engl J Med 2009; 361: 2019–2032. 5 Zhang Y, Thamer M, Stefanik K, Kaufman J, Cotter DJ: Epoetin requirements predict mortality in hemodialysis patients. Am J Kidney Dis 2004; 44: 866–876. 6 Bradbury BD, Danese MD, Gleeson M, Critchlow CW: Effect of epoetin alfa dose changes on hemoglobin and mortality in he- modialysis patients with hemoglobin levels persistently below 11 g/dl. Clin J Am Soc Nephrol 2009; 4: 630–637. 7 Streja E, Kovesdy CP, Greenland S, Kopple JD, McAllister CJ, Nissenson AR, Kalantar- Zadeh K: Erythropoietin, iron depletion, and relative thrombocytosis: a possible ex- planation for hemoglobin-survival paradox in hemodialysis. Am J Kidney Dis 2008; 52: 727–736. 8 Servilla KS, Singh AK, Hunt WC, Harford AM, Miskulin D, Meyer KB, Bedrick EJ, Rohrscheib MR, Tzamaloukas AH, Johnson HK, Zager PG: Anemia management and as- sociation of race with mortality and hospi- talization in a large not-for-profit dialysis or- ganization. Am J Kidney Dis 2009; 54: 498–510. Singh Am J Nephrol 2010;31:552–556556 9 Winkelmayer WC, Liu J, Brookhart MA: Al- titude and all-cause mortality in incident di- alysis patients. JAMA 2009; 301: 508–512. 10 Fleming TR, DeMets DL: Surrogate end points in clinical trials: are we being misled? Ann Intern Med 1996; 125: 605–613. 11 Singh AK: Does TREAT give the boot to ESAs in the treatment of CKD anemia? J Am Soc Nephrol 2010; 21: 2–6. 12 Szczech LA, Barnhart HX, Inrig JK, Reddan DN, Sapp S, Califf RM, Patel UD, Singh AK: Secondary analysis of the CHOIR trial epo- etin-alpha dose and achieved hemoglobin outcomes. Kidney Int 2008; 74: 791–798. 13 Bohlius J, Schmidlin K, Brillant C, Schwar- zer G, Trelle S, Seidenfeld J, Zwahlen M, Clarke M, Weingart O, Kluge S, Piper M, Rades D, Steensma DP, Djulbegovic B, Fey MF, Ray-Coquard I, Machtay M, Moebus V, Thomas G, Untch M, Schumacher M, Egger M, Engert A: Recombinant human erythro- poiesis-stimulating agents and mortality in patients with cancer: a meta-analysis of ran- domised trials. Lancet 2009; 373: 1532–1542. 14 Bennett CL, Silver SM, Djulbegovic B, Sama- ras AT, Blau CA, Gleason KJ, Barnato SE, El- verman KM, Courtney DM, McKoy JM, Ed- wards BJ, Tigue CC, Raisch DW, Yarnold PR, Dorr DA, Kuzel TM, Tallman MS, Trifilio SM, West DP, Lai SY, Henke M: Venous thromboembolism and mortality associated with recombinant erythropoietin and dar- bepoetin administration for the treatment of cancer-associated anemia. JAMA 2008; 299: 914–924. 15 Stowell CP, Jones SC, Enny C, Langholff W, Leitz G: An open-label, randomized, paral- lel-group study of perioperative epoetin alfa versus standard of care for blood conserva- tion in major elective spinal surgery: safety analysis. Spine (Phila Pa 1976) 2009; 34: 2479–2485.