You Are Here: AHRQHome > Clinical Information > Technology Assessments > The Impact of Pre-transplant Red Blood Cell Transfusions in Renal Allograft Rejection

The Impact of Pre-transplant Red Blood Cell Transfusions in Renal Allograft Rejection

Disposition of Comments

Project ID: RENT061-0

The Agency for Healthcare Research and Quality's (AHRQ) Technology Assessment (TA) Program supports and is committed to the transparency of its review process. Therefore, invited peer review comments and public review comments are publicly posted on the TA Program Web site at http://www.ahrq.gov/clinic/techix.htm within 3 months after the associated final report is posted.

This document presents the peer review comments and public review comments sent in response to the draft report, The Impact of Pre-transplant Red Blood Cell Transfusions in Renal Allograft Rejection, which was made available for comment on the AHRQ Web site. The final version of the report is available online.

Select for Table 1: Invited Peer Reviewer Comments
Select for Table 2: Public Review Comments
Select for Appendix: Public Review Comments Appendix: Response to Amgen
Select for Print version (PDF file, 315 KB). Plugin Software Help.

Table 1: Invited Peer Reviewer Comments

Reviewer¹	Section²	Reviewer Comments	Author Response³
Reviewer 1	General	The purpose of the Study is very appropriate. The research conducted by the University of Connecticut/Hartford has elucidated in great details the question about the possible beneficial effects of pre-transplant blood transfusion according to different immunosuppressive eras.	We would like to thank the reviewer for the comments.
Reviewer 1	Executive Summary	The summary gives a completed overview of the pre-transplant transfusion issues and advantages including the rationale for the possible beneficial effect of donor specific transfusions (DST) that led to this option in several transplant centers.	We would like to thank the reviewer for the comments.
Reviewer 1	Introduction/ Background	The Background is well written and offers a complete picture of the current knowledge of the role of HLA and its impact on the different kinds of rejection. It is well known that the use of recombinant erythropoietin has minimized the need for blood transfusions in patients with ESRD and anemia but the role of pre-transplant transfusions, in particular donor specific transfusions (DST) need to be revisited. The objectives of the study have been properly focused and addressed.	We would like to thank the reviewer for the comments.
Reviewer 1	Methods	The systematic literature search, conducted by 2 independent investigators is highly satisfactory. The data extraction, synthesis and analysis, including the tables reporting the different outcomes show a complete overview of the data available and an excellent summary of the conclusions.	We would like to thank the reviewer for the comments.
Reviewer 1	Results	Although the limitations related to the heterogeneity of the different studies retrospectively analyzed, the results are offering a better and clear understanding of the role of the pre-transplant transfusion, across different immunosuppressant eras, multiple variables and different methodological approaches.	We would like to thank the reviewer for the comments.
Reviewer 1	Discussion/ Conclusion	A study of such magnitude was needed in particular when the status of the art in kidney transplantation is more and more focused on tolerance induction. The study is also very relevant because offers an important background for designing additional multicenter studies that will be able to clarify unknown answers in the transplant immunology field. As a reviewer I am complimenting the authors for the enormous efforts that have been dedicated to this huge and demanding project.	We would like to thank the reviewer for the complementary thoughts.
Reviewer 1	Tables	The tables and figures are easy to understand and are very comprehensive.	We would like to thank the reviewer for the comments.
Reviewer 2	General	The overall report is well organized, comprehensive in scope, with appropriate methods, results and conclusions. However, I have three concerns: 1) the conclusions should be clearer; 2) the conclusions do not fully reflect the quantitative findings in the results; and 3) the recommendation for additional studies is not supported by the data nor by the discussion in the report itself.	Thank you for these comments. We worked hard to generate a report that met these facets. (1) & (2) – The conclusions were revised somewhat to reflect our responses to the reviewer’s comments. (3) – The recommendation for additional studies is based on the low/insufficient strength of the overall body of evidence of the literature. Future studies with better design (i.e. adequately powered, accounting for confounders, etc) may change the conclusion of this technology assessment. Whether one believes it is ethical to conduct such a study does not change the poor quality of the current body of evidence does not engender confidence in the results that they convey.
Reviewer 2	Executive Summary	The Tables in this section and the Results section should provide some sense of the range of Strength of Evidence for each group of studies in addition to the overall averages, which are all either low or insufficient. I assume that for some of the outcomes examined there are some sub-groups of studies that have moderate or high strength of evidence. Likewise, the final conclusions in the discussion should be a bit more granular to reflect the Results; i.e., that the magnitude of effects on some outcome measures are larger than others (albeit, all with low strength).	We provide an assessment of individual study quality ratings in the report (but not for each endpoint) as well as the rating of the strength of evidence. The strength of evidence, according to AHRQ Methods Guide for Effectiveness and Comparative Effectiveness Reviews state that the grading should be on the body of evidence. For each key question we do provide the references which can be cross-referenced with the individual study quality rating. We also pull out the good quality studies and talk about them separately which should help with this comment as well.
		There is also a trend in several outcome measures for a large beneficial to neutral effect, without evidence of a detrimental effect, which should be highlighted in the final summary. It is hard to determine from the final discussion that the transfusion effect on graft survival was consistently in the range of strongly beneficial to neutral. If the outcome was focused on a detrimental (vs beneficial) effect of transfusions, would the strength of evidence that transfusions are not detrimental still be low? I presume not, but this should be addressed explicitly in the methods, results, and discussion, and clarified in the conclusions.	We use the AHRQ Methods Guide for Effectiveness and Comparative Effectiveness Reviews in determining the strength of the body of evidence and the methodology is specified in the report. Strength of evidence is related to the confidence that we have that future well conducted studies/trials will not likely change the conclusions. So if the quality of the literature is overwhelmingly poor in nature, the strength of evidence will be low. We have altered the conclusions somewhat based on reviewer’s comments.
		This report is supposedly about the impact of transfusions, but it is written as if the report is simply judging the beneficial impact. I recommend it should be refocused to provide a balance of conclusions between the beneficial and detrimental effects on graft outcome, especially since this is one of the outcomes examined in results. This would lead to dual conclusions that 1) there is no substantial evidence for a detrimental effect (? strength), and 2) some evidence (low strength) for a beneficial effect in some cohorts. Conclusions should be provided in the context of the improved maintenance and anti-rejection diagnosis and treatment currently available, and current non-transplant risk-benefit aspects of transfusions.	We do already provide this level of granularity in the tables. However, in order to enhance the readability of the report, which is already large and cumbersome, we do not feel that describing the results in terms of the percentage of trials that shows benefits, has neutral effects, or has detrimental effects would supply additional information over and above assessing a beneficial to neutral effect together.
		The Future Research Directions section advocating additional studies is inconsistent with the results and the immediately preceding discussion. I presume that grammar in the last sentence of Discussion is intended to be that “Given the problems with internal validity with these individual studies and the heterogeneity contained within the studies, we only have a low or insufficient strength of evidence for any of these findings. As such, we have low confidence that the results of our report would not change upon publication of additional higher quality evidence.” That is, they have low confidence that additional studies would change their conclusions, rather than not change their conclusions. If so, their call for additional studies needs better justification. If the last sentence is written as intended, it is inconsistent with the discussion, and should be re-written without double negatives; i.e. “we have high confidence that the results of our report would change upon the publication of additional higher quality evidence.” I would disagree if that is indeed their conclusion.	We believe that future studies may very well alter the findings that we have come up with because the body of evidence has low quality and is primarily predicated on data that might not reflect contemporary practice. We now specify in the executive summary and body of the report that “We have low confidence that the evidence reflects the true effect. Further research is likely to change our confidence in the estimate of effect and is likely to change the estimate.”
Reviewer 2	Introduction/ Background	It would be helpful to provide a better context for the whole issue of transfusions in transplantation relative to sensitization and immunosuppression.	We expanded the discussion section substantially to reflect the issue raised by reviewer. We incorporated information obtained from the USRDS, UCLA, and UNOS in discussing this issue.
Reviewer 2	Methods	See comments above for Executive Summary	As above
Reviewer 2	Results	See comments above for Executive Summary	As above
Reviewer 2	Discussion/ Conclusion	See comments above for Executive Summary	As above
Reviewer 2	Tables	See comments above for Executive Summary – the tables in the Executive Summary should provide a balance between beneficial effects and detrimental effects, to more accurately reflect the tables in the results section.	As above
Reviewer 2	Figures	Fig 1 is appropriate. An added figure(s) would be helpful to summarize the outcomes examined, their relationship to each other, and the range of studies available for each.	As above
Reviewer 3	General	I will not provide a section by section review as in my opinion the review provides an excellent, well researched and well written summary of the data. The real question I found difficult to answer for myself was why this study was ever commissioned.	Thank you for the reviewer’s complementary comments. CMS would be in the best position to delineate why the report was commissioned but it may be because entities exist that believe that the literature base strongly suggests that transfusions are detrimental and they wanted an independent body to assess the literature.
		The authors provide very adequate criticisms of the limitations inherent to the study given the quality of the work available to them, but I think they overreach in suggesting further studies would be of use to the transplant community.	While we agree that there is improved management in the transplant population and future studies may require large populations, we need to suggest future research if the strength of the body of literature is weak. It may be that the question is not of great enough interest to justify the expenditures but in order to generate high quality data that would answer these key questions, we feel our recommendations would be helpful.
		The interest in this area, as can be seen if a “Bell Curve” is constructed of the dates of the publications, started as a consequence of a few papers from influential centers published in the late 1970’s when immunosuppression remained largely based on azathioprine and prednisone and when results were still quite poor making it possible for small studies to have some impact. In spite of this the evidence for the use of pre-transplant blood transfusions was never very strong and both this and peri-operative transfusion (not dealt with in this review) fell out of use as the effectiveness of newly approved immunosuppressive agents entered clinical practice, essentially overwhelming any small effects achieved by transfusion. In the 80’s, studies started when interest was quite high were fairly numerous, but the numbers tailed off over time, a good measure to my mind that the issue was considered largely irrelevant by the academic transplant community. Thus the number of publications as we move into the late 90’s and 2000’s is really too small to provide any useful analyses. Large database studies shed little more light than other studies, but do not suggest a significant effect from pre-operative transfusion, some even suggesting it may be harmful.	Thank you.
		The authors do attempt to correct for this in the section on Page 38 but they themselves point out the difficulties in so doing.	Thank you.
		There are a few further points of importance. The effect of blood transfusion, in as much as it ever had an impact was thought to either be a consequence of an undefined protective effect or the consequence of a winnowing effect whereby a subset of highly reactive patients were not transplantable following the elevation in their PRA following transfusion, with improved results in those that were transplanted. Few of the papers discuss the number of patients so affected.	Thank you.
		The authors correctly discuss the improvements in the characterization of pre-formed antibodies and the current context of defining specific antigens to which potential recipients are highly sensitized but this is not germane to the period from which most of the data is drawn.	Thank you for the comment.
		The section on the use of leukocyte depleted blood transfusions highlights another area of weakness. It is highly likely that many of the studies that do not specifically state this used leukocyte depleted blood as this was essentially the standard of care from at least the early 90’s if not before.	We agree.
		As an historical analysis of this area the review is excellent. However the improved and quite excellent outcomes of transplantation have impacted our ability to show differences in outcomes in trials of new immunosuppressive medications unless very large numbers of patients are entered into each arm of the study. For an effect as weak as that of blood transfusion the size and cost of any study would be prohibitive.	The true transfusion effects on renal allograft outcomes is still up for debate as the strength of overall body of evidence for the available literature on this topic is either low or insufficient. Although certain practice over time becomes commonplace, the general acceptance of this practice may not reflect the reality of the science behind the practice. Our position is that, in light of the poor quality of literature, more well-designed studies may shed more light on this area of practice.
Reviewer 4	General	Terminology should be updated to standard terms. For example, cadaveric is no longer considered an appropriate term, and has been replaced by deceased donor. Similarly, chronic rejection is not considered an entity per se and is, in its general form, sometimes referred to as chronic allograft nephropathy or the appropriate histopathological description of the process.	These terms were used in the literature. We used the same terms to maintain consistency with the literature included in the TA. We also modified the section that discusses chronic rejection to address this concern.
		The key questions are a bit misled, for several reasons: If the clinical question is "should there be a drive to transfuse patients while awaiting kidney transplantation" then the target timeframe should include the time awaiting transplantation in addition to the post-transplant time. Otherwise there is a survivor bias for those reaching transplantation, and the "endpoint" of achieving a transplant is also missed. For example, if transfusions raise the PRA so high that no compatible donors can be found, the patients will die faster while waiting for a transplant, and those that become sensitized to that level will never reach transplantation, and as such be excluded from the study cohort. The hypothesis is that transfusions will sensitize a patient to certain antigens. However, since donors and recipients are crossmatched prior to transplant, any antibodies that have developed in the patient as a result of transfusions will be detected during crossmatch testing, and most centers will then forego the transplant. In other words, those who are sensitized by previous transfusions who do undergo a transplant will have a negative crossmatch with their donor, and so the harmful effects of the transfusions will not be detected. The only exceptions to these situations would be patients knowingly transplanted across antibody barriers, but these transplants are quite rare, are clustered at a handful of centers, and are reported accordingly. Several major temporal trends exist. First, outcomes from the 80s are in no way comparable to outcomes from 2000's because of major advances in immunosuppression. Many would argue that every study prior to 1990 or even 1995 should not even be reported, because transplantation was a very different field at that time. Additionally, and quite importantly, crossmatch testing has changed over time, and centers have become progressively more sensitive in detecting antibody. Since a positive crossmatch will block a transplant, then when crossmatch tests are more sensitive, only the lowest antibody risk transplants can occur; when crossmatch tests were less sensitive, it is possible that some degree of low-level antibody existed and caused problems post-operatively.	We understand that you have some reservations about the key questions that were posed. These questions were posed by CMS and we were asked to answer them in a rigorous and unbiased manner. We were not charged with construction of the key questions. So while these are valid points, but they are tangential to the key questions that we are asked to assess or address. Hopefully our expanded discussion section will provide more transparency into some of these areas and flow better into our future research section.
		Because of the above points, one of the "outcomes" related to transplantation should have been degree of sensitization, or PRA / cPRA / whatever was used at the time. A more sensitized patient would have less opportunity for transplantation, so perhaps transplant rates, or pre-transplant death, could have been reasonable surrogates.	These outcomes were not included in the key questions that were posed by CMS. We cannot include new endpoints, add new key questions, or conduct a new literature search at this point. However, we do discuss this now in the discussion which should present the information in a reasonable fashion.
Reviewer 4	Executive Summary	This is clearly written and, for the most part (see nitpicky terminology comments above), quite informative, at least in the context of an essentially (and predictably) negative study. The limitations of the study design are discussed, although limitations noted in my "general comments" above should also be discussed. In ES-1, it is unclear to the reader at this point in the document what "Large beneficial to small impact on graft survival" means -- I understand what a "large beneficial impact" is, but what is a "large small impact"? The conclusions are too strong, however. The inference that "the literature… supports a neutral to positive effect resulting from transfusion and does not support a detrimental effect resulting from transfusion" is just plain misleading, for the reason specified above. Even to say "pre-transplant transfusion was not associated with worse outcomes among those who received transplants" is a bit strong without saying "although many transplant candidates might have become sensitized as a result of the transfusions and not had the opportunity to receive a transplant." But conclusions of this type, in their current form in this document, are far worse and are not appropriate. The future research section would be much stronger if suggestions were made that the true studies include pre-transplant time at risk, and that "transplantability" (i.e. transplant rates) is the appropriate outcome when considering sensitization resulting from something like transfusions.	Thank you for these kind thoughts. We changed the terminology to “large beneficial impact or small impact” to reflect the two different magnitudes (i.e. >10% or ±10%) evaluated in the TA. The statement quoted here is from the discussion section, and it reflects our findings based on the current literature, which evaluated the allograft outcomes in renal transplant recipients. In no way is this to be taken as a conclusive statement of this TA. That is clear when you see the strength of evidence for the body of literature. Both the statement and the strength of evidence need to be viewed together when assessing a systematic review. However, in light of the comments, we have revised the sentence, so it will not be mistaken as a conclusive statement. Our conclusions have already taken into account the reviewer’s points regarding sensitization and opportunity to receive a transplant. Hopefully our new discussion section also raises these issues and frames the discussion appropriately. We concur with these comments, and changes were made in the discussion section to flesh this out.
Reviewer 4	Introduction/ Background	As above, the concept of chronic rejection needs to be updated to modern terminology and understanding of this process. It is surprising that a major work investigating issues of sensitization resulting from blood transfusions fails to make any mention, in the introduction/background, of transplantation of sensitized patients, desensitization or incompatible kidney transplant protocols, and technologies for measuring the activity of and characterizing the specificity of antibodies.	We were not charged with assessing or addressing these issues. While they may be important issues, they are tangential to the key questions we are asked to address.
Reviewer 4	Methods	The systematic review methods are appropriate, albeit quite simple given the heterogeneity of the studies and thus the inability to pool estimates.	Thank you for the kind thoughts. We agree that this is probably the best that can be done given the literature base at hand.
Reviewer 4	Results	Text is acceptable and clear. Limitations in the nature of the defined outcomes are addressed above in my general comments.	Thank you.
		It is unclear what "rejection" refers to – acute cellular? acute humoral? hyperacute? Combining these endpoints is somewhat inappropriate, as mechanisms underlying these different types of rejections are differentially related to sensitization and immunosuppression regimen.	Due to the nature of the studies, the vast majority did not define the types of rejections. With the limited definitions from the literature, we concluded that the strength of evidence of the magnitude of rejection was insufficient to reflect this.
		It is unclear if "graft survival" was all-cause or censored for death (some studies consider death with a functioning graft to be "graft loss" and some do not).	We tried to account for many layers of heterogeneity within the trials. This is a yet another type of methodological heterogeneity in this body of literature. The problem really is that slicing and dicing the trials into smaller and smaller groups or defining every study qualitatively rather than grouping results in a large intelligible report. Because ultimately we have three categories here, those with all-cause, those censored for death, and the larger group that is not adequately specified. So we would in effect need three tables instead of one for each key question and numerous other tables to split out the other variables. At some point we had to make some generalizations.
		All of the DST studies are >15 years old (mostly >20 years old) and the relevance of these studies to current practice is unclear (probably none). Kidney transplantation is a very different field in 2010 than it was in the 1980's.	We agree that the relevance, or lack thereof, of trials from the past are hard to determine. We clearly delineate in Key Question 1 the impact of more recent trials on results. We do find a tendency toward less benefit and more tendencies towards a neutral effect in more recent trials. This is clearly specified in the report. We also pull out the DST trials into a subgroup and view them separately which should help address this comment.
		Ascertainment of number of transfusions can be tricky; it would be nice to have a description of how number of prior transfusions were ascertained (medical records, patient history, etc) and some quality measure of this with regards to potential bias.	We agree that it is tricky to ascertain transfusion exposure and this does impact the strength of evidence by introducing methodological heterogeneity. Like the graft survival example above, again we would need to subcategorize into numerous groups (those with one or more ways that transfusion exposure can be ascertained and the category where it is simply not defined). We do not feel that the large number of sub-categorizations necessary to tease these things out would yield useful information and would do something other than make it more difficult to read and understand. However, we now review trials of good quality separately which can help to alleviate your concerns.
		The PRA section is reasonable, but fails to draw attention to a significant issue, which is that PRA (or cPRA) quantifies the reactivity to the general pool of donors, but that is not a relevant issue when the patient is transplanted with a specific donor, which is what indeed happens. PRA is a great predictor of likelihood of transplantation, but is not per se in the causal pathway to rejection/allograft loss for a given transplant. The important piece for a given transplant is donor-specific antibody (DSA), in other words the breadth and strength of antibodies that a recipient has to the actual donor whose kidney he is receiving. DSA is NOT captured in PRA. While it is true that some patients with a very high PRA might be transplanted with some low-level DSA, this occurs at the discretion of the center, and there could just as easily be patients with lower PRA but higher-level DSA. Saying that a high PRA is causally associated with rejection or allograft loss is probably somewhat misleading; while it might be associated with it, for confounding reasons (donor choice, immunosuppression choice, etc), it is not causally associated with it; PRA is at most a surrogate for some other unmeasured factors which actually do influence outcomes.	We recognize that this is an important issue in the transplant world, and we have altered the discussion section of the final report to address this facet.
Reviewer 4	Discussion/ Conclusion	Limitations to inference as specified above in "general comments" and elsewhere apply here. Heterogeneity is appropriately described.	As above.

¹ Peer reviewers are not listed in alphabetical order.
² If listed, page number, line number, or section refers to the draft report.
³ If listed, page number, line number, or section refers to the final report.

Table 2: Public Review Comments

Reviewer Name¹	Reviewer Affiliation²	Section³	Reviewer Comments	Author Response⁴
Joshua Ofman, MD, MSHS	Amgen	General	Select for Public Review Comments Appendix: Response to Amgen
Joseph V. Bonventre, MD, PhD, FASN	American Society of Nephrology	General	Donor-Specific Transfusions have not been utilized for more than two decades. The true effect of Donor-Specific Transfusions remains unclear Modern single-antigen antibody tests have supplanted classic PRA analyses and are used to improve the efficiency of organ allocation ESAs are indicated to treat anemia; administration of ESAs prevents patients from risk of sensitization due to transfusions that would otherwise be necessary	No specific changes were made because the comments did not address any specific issues relating to this technology assessment. However, we do now look at DST versus no transfusions separately in a subgroup which should address part of the concern.
Karen E. Ryals	American Association of Kidney Patients	NA	The Technology Assessment draft document was for the most part (greater than 90%) a retrospective analysis of observational data. AAKP believes that the document is fundamentally flawed in terms of its ability to adequately inform potential policies or decisions based on the most recent scientific research. We firmly believe that policy-makers who rely on the study in its present form and its outdated information will make poorly reasoned decisions that will impact patients and their access to future care options.	The inclusion criteria for the literature reviewed in this TA was tailored specifically to answer the key questions posed by the CMS. The retrospective and older nature of the literature base shows that interest in this topic has diminished over the years, and we have no control over this. We were charged with summarizing and grading the strength of evidence for the literature base.
Karen E. Ryals	American Association of Kidney Patients	NA	The assessment failed to look at the highly relevant and current United States Renal Data System 2010 Annual Report that stated that the 3-year cumulative incidence of blood transfusions in patients on the transplant list with PRAs over 80% was about 41%. Note that in those who had no antibodies the incidence of blood transfusion was only 24 to 25%.	We now provide the USRDS annual report information germane to this area in the discussion along with data derived from registries. These data are not peer reviewed, have scant methodology, and generally do not account for confounders. Of note, the USRDS data, which used the OPTN-UNOS registry, do not collect quantitative/qualitative information on transfusions. These data are self-reported, and information on whether patients received transfusions are limited to discrete data (i.e. yes, no, unknown). It is not known if therapeutic transfusions are actually indicated or required.
Karen E. Ryals	American Association of Kidney Patients	NA	The HRSA Organ Procurement and Transplantation Network database further documents that patients who are sensitized must then wait at least one to three years longer on the list for a kidney transplant. If they do receive one, it does not last as long, and the patients have a higher complication rate, including a 19% higher risk of death. The worst part of this issue is that the patients who have high panel PRAs may never receive a kidney transplant. High PRA levels then block access to transplantation. This data compounds data that women who have been pregnant may also develop PRAs, and thus disproportionately limits access to women. In summary high PRA levels are a medical catastrophe for the kidney patient who seeks transplantation and it must be avoided in any way possible. This fact was not considered in the Technology Assessment because this unfortunate population was never studied.	The statements about waiting times and consequences of waiting for transplants are no longer valid since there are desensitization treatments available for sensitized patients, and patients who are highly sensitized can still receive transplants nowadays. The study by Montgomery et al provided alternative treatment options for patients with HLA sensitization, which provides some insight to the current issue. (NEJM 2011;365:318-26) The information presented here from the HRSA OPTN databases is not directly related to the key questions of the TA, and we have already adopted relevant information from the OPTN database in answering key question 2a in the draft report. We now highlight the data from the book chapters and USRDS annual report in the largely expanded discussion. We cannot change the key questions, as they were developed by CMS, and our task was to answer the questions that were asked.
Karen E. Ryals	American Association of Kidney Patients	NA	The Technology Assessment relied on research done before 1992. Only 10% of the studies were performed afterwards. These studies have no real bearing on modern transplantation for a multitude of reasons: That was still an era of intentional transfusions between donors and recipients. Immunosuppression has evolved extensively since then. (Mycophenolic acid, mycophenolate, sirolimus, everlimus, thymoglobulin, steroid withdrawal are all subsequent developments since 1992). Histocompatibility techniques are substantially more advanced -- to the point that we can now identify specifically which anti-HLA antibodies are present. The organ shortage is substantially worse. Transplantation is routinely being performed on much older and sicker patients. Multiple strategies for transplantation of high-PRA patients and incompatible pairs (plasmapheresis, IVIG, Rituxan, splenectomy) are contributing very complex circumstances to the broad national database -- and are often not sorted out in studies that look solely at these national data for outcomes.	We recognize that the relevance of the body of literature included in the TA is hard to determine. As specified in the draft report, we have clearly outlined in Key Question 1biv-v the impact of more recent trials on results. We did identify that a tendency toward less benefit and more tendencies towards a neutral effect was seen in more recent literature. This is in addition to selecting out DST transfusions into a subgroup and looking only at good quality trials.
Edward R. Jones, M.D.	Renal Physician Association	General	The scope of the questions was so narrowly defined that the broader, truer picture of the impact of ESA use in the care of Medicare beneficiaries and other patients with chronic kidney disease (CKD), both on dialysis and not on dialysis, is obscured.	The assigned scope of this TA did not include evaluating the impact of ESA use in CKD patients. It is commissioned to evaluate the overall strength of evidence of the literature on the impact of transfusion in kidney transplant population. The key questions were defined by CMS.
Glenn M. Chertow, MD, MPH	Standford University School of Medicine	NA	None of authors of the Technology Assessment Report were physicians or surgeons experienced in the care of kidney transplant recipients. While the objective evaluation of published evidence can be a valuable exercise, conclusions based on such evidence without the clinical context should not be used to inform or influence policy.	Aside from our group of pharmacists and physicians, we utilized the expertise of a Key Informant - J. Michael Cecka, Ph.D. who is Director of Clinical Research at the UCLA Immunogenetics Center.
		NA	The University of Connecticut/Hartford Hospital EPC should have engaged the assistance of one or more physicians or surgeons involved in the care of patients with advanced chronic kidney disease or end-stage renal disease and kidney transplant recipients.	I believe the impression is that if we had a higher utilization of transplant personnel as key informants that we would have devised other key questions. However, we did not derive the key questions. We were charged with answering the questions that were posed by CMS. We do not believe that having a transplant surgeon would have altered the results or increased the strength of the body of evidence for the questions we were charged with answering.
			While the EPC dutifully addressed the questions posed, they failed to recognize the narrow scope of the questions. Examination of the overriding question of whether blood transfusion is associated with prolonged or foreshortened allograft survival without considering the effects of blood transfusion on sensitization is the most critical flaw of the entire exercise. Had anyone on the EPC considered the effects of blood transfusion on sensitization and the effects of sensitization on transplant eligibility and outcomes, the conclusions of the Technology Assessment Report would have been vastly different.	Technology Assessments, in comparison to other AHRQ CERs have a rapid turnaround time. We were charged with answering the key questions that were posed by CMS in a scientifically vigorous, transparent, and unbiased manner.
			The statement made in the Technology Assessment Report (page ES-1) "In transplants, the practice of blood transfusion has been shifting back and forth for the last few decades" is grossly inaccurate. Owing to the adverse immunologic effects of blood transfusion (the development of anti-HLA and other antibodies), most transplant physicians and surgeons have aimed to assiduously avoid blood transfusions in all prospective transplant recipients, whether persons with advanced kidney, liver, heart, lung or malignant disease, where solid organ or bone marrow transplant might be a future therapeutic option.	We removed this sentence from the background section.
			Even in response to the impractically narrow scope of questions posed, the authors concluded that the evidence used to support their conclusions was either "weak" or "insufficient."	This is a very weak literature base from which to make healthcare decisions. Future studies may confirm our conclusions or may refute them, it is hard to engender confidence in the results when the studies have this number of limitations.
			The tallying of studies on one side of the "effect of transfusion on allograft function" effect or the other failed to account for relative differences in data quality (e.g., low versus very low quality of evidence), the magnitude and stability of the estimates and the sample size on which the study and its conclusions were based.	While we understand your concern, we graded the studies on multiple different factors, which included sample size. We rated most of the included studies as poor due to their poor study designs, and we then graded the overall strength of evidence for the body of literature as either low or insufficient (based on AHRQ Methods Guide for Effectiveness and Comparative Effectiveness Reviews, which does not recognize “very low” quality of evidence). Because the heterogeneous nature and the weak strength of the literature base, we believe that there will be no appreciable change to our conclusion.
			The authors of the Technology Assessment Report failed to distinguish the use of "donor-specific" blood transfusions and transfusions given for the management of severe anemia. Donor-specific transfusions were used with the hope of inducing "tolerance" - a state in which the recipient’s exposure to the donor's foreign antigen might attenuate the subsequent immune response after organ transplantation. This practice has been abandoned as unsuccessful. Both reasons for transfusion can result in sensitization and preclude or limit an individual's access to organ transplantation.	We certainly appreciate the different mechanisms underlying the use of DST and therapeutic transfusion. We have separated the DST and therapeutic transfusion analyses from Key Question 1 for subgroup analysis. The subsequent results are not significantly different, and do not change any of our conclusions.
		ES-12	The conclusion that "additional adequately powered studies (implying clinical trials) should be conducted" (ES-12) is unreasonable, and likely the result of 1) the narrow scope of the questions posed; and 2) the lack of hands-on clinical transplant experience of the EPC members. Understanding the effects of transfusion on sensitization and the effects of sensitization on access to organ transplant and subsequent organ longevity, no responsible physician or surgeon or prospective transplant recipient (or Institutional Review or Ethics Board) would allow a trial of transfusion versus no transfusion to be conducted.	It is a misunderstanding to conclude that our statement means one, and only one type of study design (clinical trials). Taking the whole paragraph into account, will demonstrate that we are only referring to better methodology of future studies (both clinical controlled and observational studies), and not to any specific study design, as the reviewer seems to have done.
			The authors of the Technology Assessment Report state "This (referring to immunosuppressive regimens) should be specifically evaluated to determine whether transplants need to be encouraged, avoided, or matched with certain regimens. Such evaluations should adhere to good study conduction practices." The authors must have meant to state "...to determine whether transfusions need to be encouraged, avoided or matched..."	Thank you for pointing this out. We have made the change accordingly in the final report.
			The authors of the Technology Assessment Report failed to adequately address residual confounding and selection bias in the observational data. I addressed the "denominator" problem during the discussion period of the MedCAC meeting. In other words, studies in which transfusion and transplantation were evaluated included only individuals who had been transfused and ultimately received an organ, a smaller denominator than the eligible population. Authors of the original manuscripts failed to account for factors associated with the provision of transfusion, and the EPC failed to consider these additional weaknesses.	The TA was commissioned to report the evidence of transfusion effect on renal transplant recipients, and we were not tasked with evaluating all the outcomes related to kidney transplant candidates. Noted that there are desensitization treatments available for patients with HLA sensitization, and these patients can still have the opportunities to receive transplants. (Montgomery et al, NEJM 2011;365:318-26) Although we have discussed this issue briefly in the discussion section of the original draft report, we have elaborated it to a greater extent in the discussion in the final report. We hope our much expanded discussion section will suffice here.
			There appears to be confusion with respect to "retrospective" and "prospective" studies, numerous typographical errors and errors in detail (e.g., considering studies with p-values >0.05 as "significant") that lessen my level of confidence in the overall integrity of the report.	Since this comment did not specify where the confusion appeared to be in our report, we were not able to address this comment. However, we have reviewed our data, and we do not believe that there were errors in interpreting the significant p-values. We have made corrections to the typographical errors in the final report.
			Officials at AHRQ and CMS should carefully consider the lives of persons with chronic kidney disease and in particular those with end-stage renal disease (ESRD). Dialysis sustains life, but rarely restores health. Patients on dialysis in the United States experience mortality rates in the range of 18 to 20% per year. Moreover, they experience marked functional impairment and relatively poor health-related quality of life.	Thank you. In our expanded discussion section, we now address this in much greater detail. We hope this helps because we cannot alter the key questions, they were provided to us by CMS.
			Even when considering the risks and costs of immunosuppressive therapy, outcomes associated with kidney transplantation are hugely superior to those on dialysis. Virtually all patients with ESRD would benefit from kidney transplantation if more donor organs (from living and deceased sources) were available.	While the reviewer believes that kidney transplantation may benefit all ESRD patients, this statement cannot be validated with the existing data.
			Restricted access to ESAs for Medicare beneficiaries would result in an unnecessary increase in the provision of transfusions, which would "pull the rug out from under" patients with ESRD, by increasing the likelihood of sensitization. Not only would this affect individual patient, but would result in an overall reduction of "life years with a functional graft" for the entire ESRD population. This would be a tragic outcome.	Thank you. We were not asked by CMS to evaluate ESA use in dialysis or transplantation. CMS set the key questions for this TA.
Mary Cushman, MD, MSc	American Society of Hematology	NA	The strength of evidence is low. This limits the usefulness of the data in the report. If this remains a clinically relevant question in kidney transplant practice, ASH is supportive of well-designed, multicenter studies to better determine benefits.	We agree with the comments. Hopefully our proposed trial ideas can be a place to start.
Geoffrey Land, PhD	American Society for Histocompatibility & Immunogenetics (ASHI)	NA	The questions in the assessment only addressed the impact of transfusion on transplant outcomes. They did not address or consider the significant impact of sensitization on access to transplantation, particularly for women, pediatric patients and transplant patients requiring second and third transplants	While the reviewer makes a valid point, it is outside the scope of the key questions set by CMS. Hopefully the newly expanded discussion section will help provide this data in an understandable manner even though it is outside the scope.
			The assessment was based on a literature review that was not comprehensive and which focused primarily on older articles where assessment of sensitization was based on PRA levels obtained from cell panels. They did not review or request Organ Procurement Transplant Network (OPTN) data and only included some Collaborative Transplant Study (CTS) data. In fact, the executive summary of the report rates the strength of evidence from the literature cited in the search as “low” or “insufficient” in all questions posed	Firstly, the OPTN database did not meet the inclusion criteria because it was not an original clinical trial, and the analyses were not peer reviewed.
				Secondly, we included data from CTS in key question 1. There were multiple publications by CTS, and some of these had overlapping populations. We tried our best to account for this. We have now added data from the CTS to our final report for key question 2b.
				The low or insufficient strength of evidence was based on the poor quality of the literature base (poor study design, study heterogeneity).
			The assessment did not address the fact that more recent studies detailing HLA-specific antibodies in transplant patients are based on much more accurate and sensitive solid-phase assays and, therefore, that the older data cannot be used to draw conclusions for current practices.	We understand that current practice has changed. We have accounted for these advances in KQ 1biv-v, where we clearly delineate in the impact of more recent trials on results.
			The assessment also included some papers that showed a beneficial impact of donor-specific transfusions (DST). They failed to note that the DSTs in these studies were given to recipients in live organ donor protocols designed to induce tolerance to donor HLA antigens. The transfusions in question in the current discussion are from random donors and are likely to induce sensitization	We now provide a subgroup analysis on DST versus no transfusion to try to alleviate these concerns.
			The assessment attempted to address whether or not changes in immunosuppression changed the impact of transfusions, concluding that prior to 1992 transfusion had a beneficial effect, but that after 1992 the transfusion impact was equivocal. In fact, immunosuppression has changed dramatically since this period, with expanding use of induction therapies, desensitization protocols and rescue therapies directed at preventing antibody-mediated graft loss and reducing chronic rejection of allografts.	Yes, we agree. You rightly point out that the strength of the evidence is low to insufficient which is what we believe as well.
			Finally, the overall conclusion of the report was that “transfusion has a beneficial to neutral effect on graft survival.” Clearly this is inconsistent with the current standard of care, where transfusion is avoided in order to reduce the possibility of increased sensitization to HLA antigens in this population of patients. Increased sensitization limits access to transplantation and increases the morbidity and mortality related to extended dialysis. It also does not address the needs of patients requiring chronic support for maintenance of adequate hemoglobin levels while on dialysis.	Our conclusions were based on evaluations of allograft outcomes. Furthermore, the strength of evidence must be taken into account when interpreting our conclusions.
				Sensitization was not part of the outcomes that this TA was commissioned to evaluate. Having a low strength of evidence means that we have very little confidence that future studies will not change the results.
Bryan N. Becker, MD	National Kidney Foundation	NA	Notably the contractor failed to explore the rich data sets available from the United States Renal Data System (USRDS) and the Scientific Registry for Transplant Recipients (SRTR). Both the USRDS and SRTR are federally-funded resources which capture a wealth of information about individuals with ESRD in the United States. (For example, the data referenced in the first paragraph of this letter were published in the 2010 Annual Report of the USRDS.) The data available from USRDS and the SRTR provide critical population-level information about kidney transplant candidates and recipients. Chapter Seven, Volume Two, of the 2010 USRDS Annual Report includes information about trends in pre-transplant transfusion status of transplanted patients, by gender; a graph on the effect of pre-transplant transfusion on the risk of elevated PRA in transplant patients, by gender, 2004-2008; and a graph depicting the association between transfusions & five-year mortality & transplant rates. As for SRTR data, please see Robert A. Wolfe, et al. “Analytical Approaches for Transplant Research,” American Journal of Transplantation, Volume 3, Issue Supplement s4, pages 103–113, April 2003. The availability of such datasets and their exclusion from analysis appears to leave an important gap in the technology assessment.	While we recognize the availability of these datasets, and agree that they offer great deal of information on this topic, we excluded them as they did not meet our predefined inclusion criteria as they were not published as original peer reviewed articles. The inclusion and exclusion criteria were specified in the report. We have now included the information from these datasets in the discussion section of the final report which includes the points that you just made. Rather than descriptive/numerical data included from these databases, databases that provide higher level of analyses (i.e. with control groups) can add value to future research. We now specifically say in the future research directions sections of the report “Data from UNOS and USRDS registries in particular could be used for future research but should be published in peer reviewed journals, have an adequate use and description of methods, and account for a myriad of confounders.”
			Traditionally, the transplant community has relied on smaller, center-specific reports to bolster changes in therapy and science. Basing policy conclusions on a literature search is problematic given the lack of large, multi-center, adequately powered studies in transplantation, and the failure to examine epidemiologic data.	I believe that the quality of the observation data (UNOS, UCLA, and USRDS) that we now include in our discussion section would still not rate as good since they have scant methods, no demographic data in the two comparison populations, do not, by and large account for confounders, and were not published in peer reviewed journals. That is not to say that are not ultimately correct in what they are saying just that the confidence that people can have that a future well designed clinical trial would not come up with a different conclusion is low.
			The relevance of Donor Specific Transfusion (DST) strategies and their impact on transplantation outcomes is difficult to accurately assess as the vast majority of DST studies were undertaken prior to the contemporary era of immunosuppression, e.g. tacrolimus and mycophenolate mofetil. Similarly, the assessment of DST as a strategy for effectively abetting transplantation is hampered by the contractor’s failure to analyze all DST recipients, especially those who became sensitized and thus, could not receive a transplant. Indeed, when analysis is present, e.g. Leone et al. J Urol 1990 or Otsuka et al. Nephron 2001, there is a documented rate of donor sensitization, ranging from 7.5-30%. Thus, patients enrolled in these protocols actually were denied the opportunity to receive their transplant, an outcome that obviously is not optimal. The University of Connecticut/Harford Hospital EPC should not have pooled the data from donor-specific and therapeutic blood transfusions in their analyses.	We have separated DST and other non-DST transfusion analyses, and evaluated them separately. We found that there was no material effect on our results, nor does it affect our conclusion. We were explicitly asked by CMS in Key Question 1 to do a transfusion of any kind versus no transfusion analysis and we concurred. We did then do several other analyses in the second part of key question 1.
				We excluded Leone et al. 1990 because the study is not related to the key questions of this TA as it compared DST to cyclosporine instead of comparing DST to no transfusion or to non-DST transfusion. We also excluded Otsuka et al. 2001 because this study evaluated a transfusion type that was not of interest to this TA as it evaluated buffy coat transfusion not a red cell/whole blood transfusion.
			In the contemporary era of immunosuppression (after 2002-4) there has been an increased prevalence of regimens including tacrolimus and/or mycophenolate mofetil and there have been a number of significant changes in induction therapy since the early 1990’s. As a result, current graft and patient survival rates are exceptional one year post-transplant. (See OPTN/SRTR annual report and the SRTR report on the state of transplantation published annually in the Am J Transplant). To assess impact of red blood cell transfusion based upon 1-year outcomes, therefore, could be misleading.	For the vast majority of the studies (including recently-published studies), 1-year graft survival was the most common time frame evaluated. Furthermore, besides this endpoint, we also evaluated max-duration graft survival.
			Over the last 20 years, there have been significant advances in immunosuppression with induction, calcineurin inhibition, evolution of anti-proliferative agents, and additional medications, e.g. mTOR inhibition. These advances define eras in transplantation and also confound inter-era comparisons, especially when the prevalence of red blood cell transfusions for transplant candidates and the availability of alternative anemia therapy as the result of the introduction of erythropoiesis stimulating agents were also changing in those eras. The fact that over half of the studies in the technology assessment were conducted before 1984, and less than ten percent of the studies were conducted after 1992 significantly undermines its relevance to current CMS policy considerations.	We understand that current practice has changed, but due to a myriad of reasons, the studies that ended up being included as part of the TA did not reflect those advances. We attempted to account for these advances in KQ 1biv-v, where we clearly delineate in the impact of more recent trials on results. We do find a tendency toward less benefit and more tendencies towards a neutral effect in more recent trials. This is clearly specified in the report.
			It may be important to distinguish the impact of transfusion prior to transplantation vs. the impact after transplantation in the data presented, especially if transfusions were part of the transplant surgery.	We added a section to discuss the impact of transfusion on sensitization, and its subsequent outcomes (i.e. transplanted, not transplanted, still awaiting for transplant, etc.) to address this comment.
			The definitions and diagnosis of rejection are not consistent over time. Rejection classification defined by Banff criteria in 1993 (Solez et al. Kidney Int) were an attempt to coalesce diverse definitions into a standardized format. Subsequent amendments and alterations to diagnostic criteria for rejection, e.g. Racusen et al. Kidney Int 1999; Racusen et al. 2003 Am J Transplant, demonstrate a shifting and indeed broader definition of rejection over time. Not only were there changes in nomenclature, with current practice nomenclature in that humoral rejection has been replaced by antibody-mediated rejection (either acute or chronic) and dispensing with chronic rejection as terminology but there were efforts towards greater histological definition with a focus on characteristics of the tissue, e.g. interstitial fibrosis/tubular atrophy (IF/TA). Such advances in tissue assessment confound the use of older studies that have used rejection as an endpoint. However, in the absence of standardizing criteria for assessing rejection as an outcome, use of this metric is very difficult to interpret.	We agree with the reviewer’s comment. The vast majority of the studies that reported rejection as an outcome did not classify the specific types of rejection, and there were no standard criteria to assess rejection as an outcome. Thus, we concluded the strength of evidence of the overall body of literature on this outcome was either low or insufficient.
Dr. Jerry Holmberg, PhD, The HHS Senior Advisor for Blood Policy and Executive Secretary for the Advisory Committee on Blood Safety and Availability	HHS	NA	To enhance stakeholder utility of the technology assessment, it would be useful to consider key research questions that are more linked to the chronic kidney disease (CKD) population. The questions addressed in the draft Technology Assessment are currently too narrow to evaluate the role of transfusions and the associated impact on renal transplants. As such, the addition of questions related to anemia management, transfusion avoidance, and the more general CKD population would help address these important considerations. For example, in the transplant candidate population, examining the impact of anemia management and transfusion status/exposure on overall outcomes (time to transplant), or death while awaiting a transplant. By focusing on the population that ultimately received a transplant, the true impact of transfusion in the transplant candidate population cannot be accurately evaluated.	Thank you for this comment. We have greatly expanded our discussion section to address real concerns such as this, even though they lie tangential to the questions we were asked to answer. This is indeed a complex issue and we recognized in the draft report that some participants in the trials did not end up going for transplant. In this revised version we now specify how many patients that entailed. This does not mean that our data shows those people would have had an adverse outcome because they were not transplanted but it does increase transparency for an important issue that the report had not adequately addressed previously. Know though that we cannot add new key questions or perform a totally new literature search at this point. These key questions were selected by CMS as being the most important for their needs and we answered those questions in the best manner we could. In our discussion we have a section dedicated to future research needs to elaborate on such areas which have been identified as a result of the TA.
			Consideration should be given to the potential differences in the impact of transfusion status/exposure on subpopulations such as women, older adults, and African Americans. Relevant outcomes in these subpopulations include: The impact of transfusions on panel reactive antibodies (PRA) The impact of sensitization on transplant wait times, given the longer wait times in these subpopulations The potential differential impact of a longer wait time due to sensitization in older populations or populations with a higher burden of comorbid conditions, e.g., mortality either awaiting a transplant or post-transplant.	Your comments on considering the potential differences in the impact of transfusion status/exposure on subpopulations are not firmly supported by the literature. Given current inconclusive data, future research is needed in this area. For the rest of the comments, refer to our response to your first statement.
			Another potential consideration not included in the key research questions is transfusion-associated infections or complications (eg. hepatitis C2, transfusion reactions7, transfusion associated lung injury6) which may preclude a patient from transplantation candidacy or be associated with considerable morbidity or mortality that may not be captured in the current assessment.	Thank you for this comment. The scope of the TA focuses on the impact of transfusions on renal transplant outcomes such as rejection, survival and patient survival. Transfusion related complications are already a known risk of transfusions and does not seem to warrant an additional key question here.
			Consideration should be given to the potential confounding in the observations of outcomes by transfusion status. Those transplant candidates that were transfused and ultimately transplanted may be less immunoresponsive/sensitized (manifested by lower PRA) than those that did not get transplanted. It may appear that transfusion has a positive impact on outcomes; however the more immunoresponsive (sensitized) transplant candidates would have been excluded, as they were not transplanted.	We agree with your comment and now evaluate the number/percentage of patients who had transplantation delayed or denied through the occurrence of sensitization in our included studies. Our expanded discussion highlights this even more.
Thomas J. De Deo, RPh, MBA, Manager, medical Information	Centocor Ortho Biotech Products, L.P.		In relation to Key Question 2b, sensitized transplant candidates, manifested by higher PRA, who were ultimately transplanted may have been treated by different immunosuppressive regimens than less sensitized transplant candidates. Is it the potential different immunosuppressive regimens /protocols that impact the outcomes and not transfusion status/PRA level? 1) What are the potential outcomes associated with the management of highly sensitized transplant candidates? More aggressive immunosuppressive regimens (induction and maintenance) used to treat the highly sensitized population may be associated with higher rates of complications such as infections, malignancy, and the likelihood of developing diabetes. The current analysis only looks at graft and patient survival.	Thank you for this comment. In a TA, we are asked by CMS to evaluate these key questions. We used a time related approach to look at more contemporary practices versus older practice but there simply is not data on whether some people received different regimens due to sensitization than other people. We agree that better research is needed to be able to evaluate this area but such data has not been generated. That is why the strength of evidence is so low. Additionally, we have a section dedicated to future research needs to elaborate on such areas which have been identified as a result of the TA.
			Consideration of the United States Renal Data System (USRDS) data would strengthen the completeness of the draft Technology Assessment. 3 Conclusions from 2010 USRDS report include: For transfused vs. non-transfused patients the Hazard Ratio was 4.04 for death and 0.72 for getting a transplant. Sensitized candidates wait longer for transplant, as 28% of wait-listed patients received a transfusion within 3 years of listing.	Thank you for this comment and the reference you provided. The USRDS data has been incorporated into the discussion section along with UCLA and UNOS data presented in book chapters in our final report. These data are not from peer reviewed sources, do not have adequate methods or demographics in both representative groups, and rarely account for confounders.
			Non-sensitized patients (PRA of 0% at listing) were as likely as mildly sensitized patients (PRA < 20%) to receive a transfusion. Highly sensitized patients (PRA 80%+) were more likely to receive a transfusion within 3 years of listing, 41% within 3 years of listing. Transfusion was associated with decreased likelihood of transplantation. PRA at transplant remains associated with adverse outcomes.	Of note, the USRDS data, which use the OPTN-UNOS registry, do not collect quantitative/qualitative information on transfusions. These data are self-reported, and information on whether patients received transfusions are limited to discrete data (i.e. yes, no, unknown). It is not known if therapeutic transfusions are actually indicated or required.
				Also noted that there are desensitization treatments available for patients with HLA sensitization, and these patients can still have the opportunities to receive transplants. (Montgomery et al, NEJM 2011;365:318-26)
			An additional consideration is, the venous access needed for transfusion support in the CKD population. The importance of the preservation of veins in CKD patients is a major consideration, as the National Kidney Foundation - Kidney Disease Outcomes Quality Initiative (NKF-KDOQI) Guidelines recommend that veins should be preserved for future vascular access for fistula creation.5 Transfusion use and its impact on the availability of suitable veins for vascular access for fistula creation should be considered. Coupled with the longer wait times associated with sensitization due to transfusion exposure and the resulting longer dependence on dialysis, this potential impact on long term vascular access availability to support dialysis needs to be considered.	Thank you for this comment. Vein preservation may be important in this population but evaluating that literature base would be outside the scope of the project we were charged with answering. At this point, we cannot add new key questions or conduct an entirely new literature search. The key questions were set by CMS.
Drs. Maryl Johnson and Michael Abecassis	American Society of Transplantation and American Society of Transplant Surgeons		In the document put forward for public comment entitled “The Impact of Pre-Transplant Red Blood Cell Transfusions in Renal Allograft Rejection”, it was concluded that “number of transfusions/transfused units versus no transfusion, or a smaller number of, transfusions/transfused units either resulted in either beneficial or small/null effects on rejection, graft survival, or patient survival. So the literature, weak as it is, supports a neutral to positive effect resulting from transfusion and does not support a detrimental effect resulting from transfusion of a larger number of transfusions”. We believe this statement to be inaccurate, and the data presented in the document to be limited in scope and applicability based on the following points:1) The data are predominantly drawn from publications in the 1980’s, many of which specifically examined the use of Donor Specific Transfusions (DST) prior to transplantation and their impact on graft survival. The conclusion does not take into consideration the effect of DST on access to transplantation and the development of positive cross matches preventing transplantation in up to 50% of cases. Indeed this is briefly mentioned in the document itself: “It should be noted that in some studies, patients who were candidates for transplantation were ultimately not offered the transplant due to high PRA levels. Some other studies did not disclose the number of patients who were ultimately not transplanted due to a high PRA as they focused on the population undergoing transplant.”	Given the key questions we were posed and the literature search we conducted, which meets AHRQs guidance for a comprehensive search, these are the results of the Technology Assessment. The quoted sentence “number of transfusions/transfused units versus no transfusion, or a smaller number of, transfusions/transfused units either resulted in either beneficial or small/null effects on rejection, graft survival, or patient survival. So the literature, weak as it is, supports a neutral to positive effect resulting from transfusion and does not support a detrimental effect resulting from transfusion of a larger number of transfusions” is from the discussion section, and it reflects our findings based on the literature. In no way is this to be taken as a conclusive statement of this TA. As it was pointed out in the comment, “Given the problems with internal validity with these individual studies and the heterogeneity contained within the studies, we only have a low or insufficient strength of evidence for any of these findings”, both the statement and the strength of evidence need to be viewed together when assessing a TA. However, in the light of the comments, we have revised the sentence to “So the literature, weak as it is, demonstrates at neutral to positive effect resulting from transfusion and does not reflect a detrimental effect resulting from transfusion.” in the discussion section, so it will not be mistaken as a conclusive statement.
			2) The extrapolation of DST in the immediate pre-transplant period to the general use of transfusion for chronic treatment of anemia pre-transplant is not based on scientific evidence and is unwarranted. The premise of the benefit of DST is that the transfusion of blood from the potential specific kidney donor results in the direct exposure to donor antigens prior to transplantation resulting in modulation of the immune responses toward those antigens. As noted above, in a significant percentage of cases, this resulted in the development of anti-donor antibodies thereby precluding transplantation from that donor, rather than facilitating graft acceptance. Accordingly, DST is significantly different to transfusion from multiple random donors which likewise exposes the potential recipient to multiple HLA antigens, increasing the risk of sensitization and thereby decreasing the potential candidate donor pool for a prospective recipient [1, 2].	We do identify several limitations for this literature base including the fact that some people are not given transplant or had transplantation postponed. We now make the percentage impacted in these studies specific to address your concerns. We also now pull out the DST data in Key Question 1 out in subgroup analysis and demonstrate that it really does not markedly alter the results and does not alter our conclusions. Previously, we analyzed the results over different time points which show the impact of relying on studies of older time points. As such, we do provide information that specifically addresses one of your concerns and will stay with our original analyses.
			3) The data used refer to Panel Reactive Antibodies, a measure of general reactivity toward the potential donor pool. This measure is used in the allocation of organs and does not indicate reactivity to the recipient’s specific donor. Current methodology is based on the identification of Donor Specific Antibodies (DSA), a better predictor for an adverse impact on graft survival through the development acute or chronic antibody mediated rejection [3-5]. In short, the review and conclusions are largely irrelevant to current practices and do not take into account current technology and immunosuppression. The difficulty with generating any broad conclusion based on the data presented is further emphasized by the following statement in the document itself: “Given the problems with internal validity with these individual studies and the heterogeneity contained within the studies, we only have a low or insufficient strength of evidence for any of these findings”. The strength of evidence presented for all articles referred to was ranked as low to insufficient data.	We agree largely with your final statement that the review and conclusions are markedly limited and the strength of evidence is low. As you say, our report makes it clear that “Given the problems with internal validity with these individual studies and the heterogeneity contained within the studies, we only have a low or insufficient strength of evidence for any of these findings”. Clearly, in order to truly determine the impact of transfusions on these outcomes, rigorous trials and studies will be needed.
			We would strongly urge that more recent data, which provide stronger evidence for the deleterious effects of sensitization and transfusion on transplant outcomes, and access to transplantation, be considered. A recent study examining the effect of DST and random pre-transplant blood transfusion (rPTF) on sensitization found that 25% of potential recipients did not receive a transplant following DST due to the development of DSA. Of those that received a rPTF, 27% developed anti-HLA antibodies and these were donor specific in 20.3% of cases. Synder et al presented data at the National meeting of the American Society of Nephrology in 2010 based on findings from the USRDS 2010 report. This data showed that patients with PRAs greater that 80% at the time of listing had a higher three year cumulative incidence of transfusion prelisting compared to those with PRAs less than 19% while patients with PRAs of 20 to 79% were intermediate. Patients that had a transfusion prior to listing had an increased adjusted hazard ratio of 4.04 risk of death on the transplant waiting list compared to those who had not received a transfusion, and a 28% reduction in the likelihood of transplantation. Fifty percent of waitlisted patients with a PRA of >80% die on the waiting list. Patients with a PRA >80% comprise 30% of the current renal transplant waiting list. The adjusted hazard ratio for death with a functioning graft was 1.41 for those with a PRA of 80%, compared to 1.21 for 20-79%, and 1.08 for 1-19%, with 0% as reference. The effect of transfusion is greatest for African Americans and women, with both groups being more likely to become sensitized [6]. This further disenfranchises two groups that already have decreased access to transplantation due to immunological and social factors. Kakaiya et al examine the prevalence of anti-HLA antibodies in blood donors that had had previous transfusions themselves using current methodologies [7]. They found that overall there was an increase in anti-HLA antibodies, and this was particularly so for parous females that received a transfusion compared to those that had not (OR 1.39). Similarly, Eikmans et al found that 35% of parous females that received a single transfusion developed sensitization. These are just a few of the more recent publications linking transfusion and sensitization [8].	Thank you for this comment and for the reference you provided. Although we recognize the availability of the USRDS data, the inclusion criteria allows only for the controlled studies (both clinical and observational) published in peer reviewed literature to be included. As such, future USRDS data from peer reviewed longitudinal cohort studies would meet the criteria. Additionally, the USRDS data is incorporated into the discussion section of the final report.
				Of note, the USRDS data, which use the OPTN-UNOS registry, do not collect quantitative/qualitative information on transfusions. These data are self-reported, and information on whether patients received transfusions are limited to discrete data (i.e. yes, no, unknown). It is not known if therapeutic transfusions are actually indicated or required.
				The second reference by Kakaiya et al 2010 examines the prevalence of anti-HLA antibodies in blood donors that had previous transfusions themselves. It does not evaluate the transfusion effect in renal transplant candidates or recipients. It is important to include a study that meets predefined inclusion criteria. Study inclusion and exclusion criteria are presented in the draft report.
				Similarly, the study by Eikmans et al 2010 examines the effect of blood transfusions but not in renal transplant population which is the scope of this TA.
				In regards to high PRA and the waiting times, one should note that there are desensitization treatments available for patients with HLA sensitization, and these patients can still have the opportunities to receive transplants. (Montgomery et al, NEJM 2011;365:318-26)
				In the future research sections of the report we now state that “Data from UNOS and USRDS registries in particular could be used for future research but should be published in peer reviewed journals, have an adequate use and description of methods, and account for a myriad of confounders.”
			There is limited evidence that treatment of post-transplant anemia is beneficial to anemic kidney transplant recipients since large randomized studies have not been performed in this population. The TREAT study examining ESAs versus placebo in diabetic patients with CKD demonstrated that 24.5% of patients in the placebo control arm required transfusion [9]. The impact on sensitization was not measured in this study but studies referred to above and others would suggest that this population would have a significant risk of developing anti-HLA antibodies. This is also inferred from data from the USRDS 2010 report which shows that the rate of pretransplant transfusion has decreased from 49% to 15% from 1991 to 2008, reflecting the increased use of ESAs in ESRD patients. At the same time the percentage of patients with 0% PRA on the waiting list has increased from 20% to over 40%. Vella et al showed that the number of transfusions in waitlisted patients decreased by 34% in the period before and after the introduction of ESAs [10]. Parallel with this the number of patients sensitized as a result of transfusion decreased from 63% to 28%, and this was associated with a significant reduction in the mean time to transplantation.	Noted
			The origin of anemia after transplantation is multifactorial. In the absence of ESAs renal transplant recipients would almost universally have anemia due to ESRD at the time of transplant. While recovery of erythropoesis occurs following transplant this is not immediate and fails to occur in up to 30% of patients. In the perioperative period anemia is worsened by the intraoperative loss of blood, the presence of delayed graft function, the initiation of immunosuppression (sirolimus and mycophenolate mofetil) and other medications, including ACEI. Perioperative anemia is recognized to contribute to perioperative complications [11]. In a study by Djamali et al, they found that in a population of patients on ESAs at the time of transplant, with hematocrits (Hct) ranging from 17% to 40% on post-op day 1, the average drop in Hct was 5.9?5.6% [12], Patients that had a Hct of less than 30%, representing 60% of the study population, had 17% incidence of acute cardiovascular events, significantly greater than those with a Hct > 30%. Increasing Hct was associated with a significant risk reduction for CV events. Death due to an acute CV event is the greatest cause of death with a functioning graft in the months post-transplant.	Noted
			One must also consider that from the practical perspective, reliance on blood transfusions for treatment of anemia prior to transplant would require referral to a hospital and potentially an admission, since most free standing dialysis units do not have the capabilities to transfuse patients in the unit. This has implications for cost and must be considered in the financial analysis overall.	Noted
			In conclusion, we strongly disagree with the statement that “number of transfusions/transfused units versus no transfusion, or a smaller number of, transfusions/transfused units resulted in either beneficial or small/null effects on rejection, graft survival, or patient survival”. The rationale leading to this conclusion is severely flawed for the following reasons: (1) the age and the weak scientific strength of the data considered; (2) failure to take into consideration newer data and techniques; and (3) failure to consider the overall impact of pre-transplant transfusion on access to transplantation. Use of these data in any decision-making process regarding the use of ESAs relative to transplantation would be misguided and has the potential to significantly impact our patients’ ability to get transplanted and their outcomes following transplantation.	When interpreting this statement, one should take into account the low to insufficient strength of evidence. The statement quoted from the draft report was based on the findings of the included studies, which were tailored to answer the key questions.
			We strongly urge that appropriate randomized control trials of the treatment of anemia in the transplant population be conducted, and that the results from these studies form the basis for any future decision-making regarding appropriate therapy in our patients. The AST and the ASTS stand ready to provide any needed assistance as this important issue is considered.	We also believe that these trials should be conducted and applaud your organization’s willingness to participate.

¹ Names are alphabetized by last name. Those who did not disclose name are labeled "Anonymous Reviewer 1," "Anonymous Reviewer 2," etc
² Affiliation is labeled "NA" for those who did not disclose affiliation
³ If listed, page number, line number, or section refers to the draft report.
⁴ If listed, page number, line number, or section refers to the final report

Select for Public Review Comments Appendix: Response to Amgen

Current as of April 2012

Internet Citation:

The Impact of Pre-transplant Red Blood Cell Transfusions in Renal Allograft Rejection. Disposition of Comments. April 2012. Rockville, MD: Agency for Healthcare Research and Quality. http://www.ahrq.gov/clinic/ta/comments/renaltrans/

The Impact of Pre-transplant Red Blood Cell Transfusions in Renal Allograft Rejection

Disposition of Comments

Contents

Table 1: Invited Peer Reviewer Comments

Table 2: Public Review Comments