Title:Trends in Anemia Management in US Hemodialysis Patients

Transcription

1 Author's response to reviews Title:Trends in Anemia Management in US Hemodialysis Patients Authors: Dana C Miskulin (dmiskulin@tuftsmedicalcenter.org) Jing Zhou (jzhou9@jhmi.edu) Navdeep Tangri (ntangri@sogh.mb.ca) Karen Bandeen-Roche (kbandeen@jhsph.edu) Courtney Cook (ccook10@jhmi.edu) Patti Ephraim (pephraim@jhsph.edu) Deidra C Crews (dcrews1@jhmi.edu) Julia S Scialla (jscialla@miami.med.edu) Stephen Sozio (ssozio@jhmi.edu) Tariq Shafi (tshafi@jhmi.edu) Bernard Jaar (bjaar@jhmi.edu) L. Ebony Boulware (ebony.boulware@duke.edu) Version:2Date:8 November 2013 Author's response to reviews: see over

2 Response to Reviewers Reviewer #1 Problems Section: Background This paper has 12 authors of whom 8 (at least) contributed to conception and design. This reviewer searched, in vain, for delineation of an hypothesis underlying the study. One of the authors should have ensured that the study was done with clear objective(s). We have stated the objective more clearly. WE have put the text below into the last paragraph of the introduction. We sought to determine the potential influence of the publication of clinical trials showing harm with targeting higher Hb levels and changes in safety regulations, with changes in reimbursement policies on anemia management practices in a representative US dialysis population. The only possible reason enunciated for submission of this manuscript is to study the influence of ESA use, harms, safety regulation and reimbursement policy changes on anemia management. Have these not been the subject of studies from USRDS and DOPPS, among others? And, does this study provide insights different from those of USRDS and of DOPPS? 1) USRDS does not have TSat and ferritin data and therein, does not report on these trends that have occurred over the past few years. 2) DOPPS describes serum ferritin and TSAt values between 2010 vs 2011 but does not attempt to tease out the effect of trial results and changes in safety regulations with

3 changes in reimbursement policy, as this current study does. Our purpose is to explicitly examine the influence of trials and labeling changes on ESA (occurred in 2007) as compared with changes in reimbursement (2010) and accordingly, we compare trends over three time periods and describe the magnitude of the changes. The effects of the publication of trial results/fda black box warning is seen in the difference between the slopes (Fig 1 and 2) during vs and the effect of policy change by comparing the slope during to 2010 (Fig 1 and 2). As stated in the Discussion (and I have newly added a sentence to the abstract), our data would suggest that practice is more strongly influenced by economic considerations, than trial data and safety concerns. The DOPPS papers simply describe trends over time. The quoted literature is a problem, in at least 3 respects: 1. The highly selected literature on ESA treatment, limited almost exclusively to Pharmaceutical company sponsored trials, and omission of an extensive literature on large observational studies. 2. Although mentioned, the failure to communicate clearly the important difference between target Hb and achieved Hb. The study addresses achieved Hb, while the literature reviewed does not. 3. The failure to recognize and address the very long known fact that adequate iron supplies are necessary when Hb production is to be stimulated. And the failure to appreciate that iron is essential for many fundamental physiologic functions. The naive reader might suppose that the only published literature relating to treatment with ESAs on the subject are References 1 & 2 and Refs 5, 6, 7, 9. Only one of these (#6 and perhaps #16) is

4 a report dealing with HD patients. #6, and the others, are notable for the following: They are drug house designed studies, designed apparently for marketing purposes. They are certainly not major as the authors suggest - at least not major scientific studies. They were designed to test the ability of the ESA to raise Hb to pre-determined targets when smaller doses failed to result in a pre-determined Hb rise. Although it has long been known that adequate iron is required for Hb production, these studies were done without attention to whether or not the patients were iron sufficient when the study started. Examination of TSAT and ferritin levels at study start shows that many were not. So, in all of them the patients were subject to ESA stimulation when they had or were soon to develop iron deficiency. The following large observational studies (among others) are not mentioned at all: Gilbertson DT, et al: Hemoglobin level variability: associations with mortality. Clin J Am Soc Nephrol 2008, 3: Ofsthun N, et al (2003) The effects of higher hemoglobin levels on mortality and hospitalization in hemodialysis patients. Kidney Int 63: Hampl H, et al (2005) Effects of Optimized Heart Failure Therapy and Anemia Correction with Epoetin Beta on Left Ventricular Mass in Hemodialysis Patients. Am J Nephrol 25: Regidor DL, et al (2006) Associations between Changes in Hemoglobin and Administered Erythropoiesis-Stimulating Agent and Survival in Hemodialysis Patients. J Am Soc Nephrol 17:

5 Portoles J, et al. (2007) A prospective multicentre study of the role of anaemia as a risk factor in haemodialysis patients: the MAR Study. Nephrol Dial Transplant 22: Fort J, et al (2010) Mortality in incident haemodialysis patients: time-dependent haemoglobin levels and erythropoiesis-stimulating agent dose are independent predictive factors in the ANSWER study. Nephrol Dial Transplant 25: The following two papers, also not cited, deal with reanalysis of previous reports, and show that while target Hb may have been associated with a poor outcome, achieved Hb was not. Besarab A, et al. (2009) What is so bad about a hemoglobin level of 12 to 13 g/dl for chronic kidney disease patients anyway? Adv Chronic Kidney Dis 16: Agarwal R (2010) Individualizing decision-making--resurrecting the doctor-patient relationship in the anemia debate. Clin J Am Soc Nephrol 5: There is important literature that shows that iron insufficiency is present in patients with CKD before they become dialysis dependent, making it virtually certain that patients will start dialysis iron deficient. Hsu CY, McCulloch CE, Curhan GC (2002) Iron status and hemoglobin level in chronic renal insufficiency. J Am Soc Nephrol 13: Van Wyck DB, et al. A randomized, controlled trial comparing IV iron sucrose to oral iron in anemic patients with nondialysis-dependent CKD. Kidney Int 2005, 68: Mircescu G, et al.: Intravenous iron supplementation for the treatment of anaemia in predialyzed chronic renal failure patients. Nephrol Dial Transplant 2006, 21: Gotloib L,et al: Iron deficiency is a common cause of anemia in chronic kidney disease and can often be corrected with intravenous iron. J Nephrol 2006, 19:

6 The paper of Gotloib is particularly important because these authors document the gold standard for clinical diagnosis of iron deficiency, viz., absence of stainable iron in bone marrow. It is not surprising, therefore, that iron insufficiency is highly prevalent in patients treated by HD, as the following show: Pollak VE, et al (2001) Unanticipated favorable effects of correcting iron deficiency in chronic hemodialysis patients. J Investig Med 49: Kalantar-Zadeh K, et al: A low serum iron level is a predictor of poor outcome in hemodialysis patients. Am J Kidney Dis 2004, 43: Pollak VE, et al. (2009) The importance of iron in long-term survival of maintenance hemodialysis patients treated with epoetin-alpha and intravenous iron: analysis of 9.5 years of prospectively collected data. BMC Nephrol 10: 6. Some of the physiologic and metabolic consequences of iron deficiency are referenced below: Lieu PT, et al: The roles of iron in health and disease. Mol Aspects Med 2001, 22:1-87. Ponka P: Cellular iron metabolism. Kidney Int Suppl 1999, 69:S2-11. Walter PB, et al.: Iron deficiency and iron excess damage mitochondria and mitochondrial DNA in rats. Proc Natl Acad Sci USA 2002, 99: Finch CA, et al: Iron deficiency in the rat. Physiological and biochemical studies of muscle dysfunction. J Clin Invest 1976, 58: Willis WT, et al: Iron deficiency: improved exercise performance within 15 hours of iron treatment in rats. J Nutr 1990, 120: Returning to the paper of Hampl (Am J Nephrol 25: ), we read Optimized CHF therapy, in combination with anemia correction to normal Hb targets, results in significant

7 reduction in left ventricular hypertrophy, increase in left ventricular ejection fraction, and that complete regression and prevention of ventricular hypertrophy is possible in HD patients - AND, after all, cardiac problems are a major cause of mortality in these patients. Without a clear hypothesis this paper simply documents certain findings in patients treated in 200 dialysis units in 3 time periods, relates the history of external events in the same time periods, and then speculates on what their. relationship might be. the authors rightly conclude that they cannot make causal inferences. This reviewer agrees. Further, the material does not substantially add to USRDS and DOPPS analyses. The reviewer has an extensive knowledge of the field, and we wholeheartedly agree with your critique of the clinical trials for their failure to manage iron deficiency, and the importance and primacy of managing iron in treating anemia in our patients. However, the purpose of this paper is not to review and critique the prior studies, but rather, to examine the influence of publications of major trials showing harm with targeting higher Hb, changes in safety regulations, and reimbursement policy change on practice between , a period of time over which dramatic changes in anemia management occurred. In terms of iron trials, there are many valuable physiological and observational trials that you have listed but the DRIVE Study, is a randomized clinical trial, was widely talked about and likely influenced practice, more than others, even though this 6-week trial did not assess the safety of administering IV iron over the long term. We agree that his paper had a big influence, and we have added it to the introduction. See p 4 last paragraph ref 7. The DRIVE Study was published around the same time as the 2006 CHOIR and CREATE trials, and therein this is part of the clinical trial data influence between as compared with policy change influence that happened much later in We do not reference or discuss the many past observational studies that have found associations of Hb levels with mortality or talk about achieved Hb (which effectively, is a metric consistent with an observational study) because these studies are likely biased. The large randomized clinical trials were published in high profile journals, which makes them more likely to influence practice during this time period plus as RCTs, they provide much higher quality data than these observational studies, and therein, constituted the major scientific influence on practice. Failing to achieve a targeted Hb level identifies patients who have

8 underlying illness/ inflammation that leads EPO resistance, and therein are destined for a poor outcome, irrespective of the Hb level. Randomizing a patient to Hb target should remove this bias if the number of patients is large enough such that the factors that determine EPO sensitivity are evenly balanced at the start of the study, and in these trials they are. This is why the observational data should be considered misleading, as these studies consistently show opposite results to the RCTs. While it is true that these RCTS were all sponsored by industry, there are no other trials available, and the authors state that the sponsors in no way contributed to the interpretation of results. We believe this is true because the trial conclusions in all cases, do not work in the favor of the sponsor. In this paper, we are interested in influences of data on practice and the key data that influenced practice, are these RCTs not the prior observational data or subgroup analyses of the RCTs. Reviewer #2 Minor Essential Revisions: - Why were excluded home HD patients? What is the percentage of home HD in USA? Home HD is a very uncommon practice in the US. Home HD represents <1% of the HD population. These patients also tend to use a lot less EPO and are quite a different and small subset of the US HD population. - It is reported that "DCI is a not-for-profit dialysis provider and it may respond differently to secular factors than for-profit dialysis providers". It would be important to inform the number of not-for-profit and for-profit dialysis in USA. DCI is the largest of the not-for-profit providers though it takes care of ~13,000 patients while the for-profit providers (Davita and Fresenius) take care of ~300,000 patients. The trends we describe here are seen in the full US dialysis population though, as I have stated in the discussion, reference the USRDS Atlas (paragraph 2 first 3 lines). Thus, we feel these results are representative of practice in the US.

9 Reviewer #3 1. Major compulsory revisions: The main criticism is that the trials chosen by the authors to justify the 3 periods that are compared (CREATE, CHOIR and TREAT -2009) are actually trials conducted in non-dialysis CKD patients and thus those results may not necessarily apply to patients on HD and alter decisions to manage anemia. This should be clearly specified in the ms and the interpretation and discussion should reflect this point. The reviewer makes a fair point about nephrologists not changing practice because they may not have believed results apply to HD patients. We have added this as an alternate interpretation as to why there was little change in practice between However, if this was the explanation, the TREAT trial, published in late 2009, would not have led to the large change in practice seen in 2010, as it was also conducted in CKD patients. We ve added this to the discussion p.14. Similar periods of comparisons can be chosen based on the FDA decisions specific to HD patients.if possible, analyses of time trends (figures 1 and 2) need to be run again without pre-specifying knots that are based on the publication dates of these trials. The knots were selected a priori and are not data driven. The FDA s issuance of a block box warning was released in Mar 2007, right after the publication of CHOIR and CREATE and thus, we would not change the timing of this knot (currently at Jan 2007) as it coincides with changes in FDA regulations. The knots are separated by 3 years, so moving them a couple of months forwards or backwards would not alter the results. 2. minor revisions: Subject and Methods (page 9): the TREAT trial did not show harm with treating to a higher HB target, but harm (in a secondary endpoint: stroke) with EPO administration compared to placebo (not a lower Hb target). This should precised in the manuscript. I agree the primary outcome was not met. I have edited the text to clarify this in the Introduction top of p.5 3. Discretionary Revisions please try to shorten parts of the results section. it is essentially a duplicate of data shown in tables and figures. I agree and have edited this section substantially. See the edited Results section.