AlloMap® Molecular Expression Testing For Detection of Rejection of Cardiac Allografts

November 12, 2020

VIA Electronic Mail to: CAGinquiries@cms.hhs.gov

Tamara Syrek Jensen
Director, Coverage and Analysis Group
Centers for Medicare & Medicaid Services
Mail Stop # S3-02-01
7500 Security Boulevard
Baltimore, MD 21244-1850

Re: CareDx Supports CMS’ Proposed Decision to retain coverage for AlloMap® Molecular Expression Testing through the local Medicare Administrative Contractors in lieu of a National Coverage Decision

Dear Ms. Jensen,

On behalf of CareDx, Inc. (“CareDx”), the developer of AlloMap® Molecular Expression Testing, we appreciate the opportunity to submit comments on the Proposed Decision Memo for AlloMap Molecular Expression Testing for Detection of Rejection of Cardiac Allografts (CAG-00455N). CareDx is a leading precision medicine solutions company focused on the discovery, development and commercialization of clinically differentiated, high-value healthcare solutions for transplant patients. CareDx offers products, testing services and digital healthcare solutions along the pre- and post-transplant patient journey and is the leading provider of genomics-based information for transplant patients. For over twenty years, CareDx has worked with CMS on the development and implementation of policies intended to facilitate appropriate patient access to high-quality clinical laboratory tests.

I. Background on AlloMap

The AlloMap Molecular Gene Expression Profiling Test (“AlloMap”) is an in vitro diagnostic multivariate index assay (IVDMIA) test service, performed by CareDx in our Brisbane California laboratory. AlloMap is intended to aid in the identification of heart transplant recipients with stable allograft function who have a low probability of moderate/severe acute cellular rejection (ACR) at the time of testing in conjunction with standard clinical assessment. The test is indicated for patients who are 15 years of age or older and at least two months (> 55 days) post-transplant. AlloMap was cleared by the Food and Drug Administration (FDA) on August 26, 2008 , has been covered by the Medicare program since 2006 and has been included in the International Society Heart and Lung Transplantation Guidelines beginning in 2010.

II. Existing Coverage Policies on AlloMap

AlloMap coverage decisions for Medicare beneficiaries currently exist with the local Medical Administrative Contractors (MACs). Beginning 2006, NHIC, Palmetto, Noridian, CGS and others began establishing positive LCDs for AlloMap based on their review of the analytical validity, clinical validity, and clinical utility data supporting the test. In addition, national payers established positive coverage policies based on similar reviews, technology assessments and the inclusion of AlloMap in prevailing clinical and transplant society guidelines.

In 2010, the International Society of Heart and Lung Transplantation issued guidelines for the care of heart transplant recipients including the use of AlloMap to rule out the presence of acute cellular rejection of grade 2R or greater in appropriate low-risk patients, between 6 months and 5 years after heart transplantation. AlloMap is broadly adopted as standard of care and is used by over 90% of US transplant centers in their standard protocol.

The primary publication establishing the clinical utility of AlloMap appeared in the New England Journal of Medicine in 2010. This landmark publication demonstrated that managing patients with AlloMap resulted in outcomes not inferior to those managed with biopsies; while significantly improving patient satisfaction. This publication was preceded by numerous AlloMap studies demonstrating the clinical validity, in particular the initial publication in the American Journal of Transplantation (Deng et al) in 2006. The heart transplant community continues to publish on the use of AlloMap, including a validation study in a European cohort (Crespo-Liero et al., European Heart Journal, 2016). Dr. Jeffrey Teuteberg, Associate Professor of Medicine at the Stanford University Medical Center and board certified in Cardiology and Heart Failure and Transplantation shared for comment “AlloMap has stood the test of time and has been included in International Guidelines (ISHLT)”. A full bibliography of AlloMap publications is included below.

Heart transplantation occurs predominately within the <65 patient population and as such 70% of all AlloMap claims are covered, processed and reimbursed by commercial payers. Since 2015, CareDx has submitted 30,500 claims for adjudication under Medicare Part B. During that timeframe, the MACs have approved and reimbursed 94.5% of these claims supporting that existing LCDs have done an excellent job ensuring utilization of AlloMap occurs when used in part with other factors to detect if an individual patient is at low risk of future allograft dysfunction so that one could consider reducing maintenance immunosuppression and the negative side effects associated with immunosuppression therapy.

The stakeholder who submitted the NCA request in 2013 asked for a national non-coverage policy for AlloMap due to perceived inadequacies in the test. Non-coverage NCDs may be appropriate if an item or service lacks the data needed to demonstrate its effectiveness. However, the FDA, Society Guidelines and multiple payers have reviewed the evidence supporting AlloMap and established positive approvals, recommendations and LCDs for the test. A national non-coverage NCD is not appropriate. CMS and the local MACs have reviewed the supporting evidence and unanimously agree it is appropriate to support coverage for Medicare beneficiaries. Reversing that long-standing history of coverage would unnecessarily restrict transplant patient access to this well-established service covered by Medicare since 2006.

We thank CMS for its diligence and thoughtfulness as it considered this request and urge CMS to finalize its decision as proposed and continue coverage for AlloMap through the local MAC coverage policies. Please don’t hesitate to contact me at dscelfo@caredx.com should you have any questions or if we can provide you with further information.

Sincerely,

Danielle Scelfo, MHSA
Vice President, Market Access and Health Policy
dscelfo@caredx.com ? 415-672-1170 ? www.CareDx.com

Appendix. AlloMap Bibliography

AlloMap Laboratory Services Guide. LQ-10004R2.

Moayedi Y, Fan CP, Miller R et al. Gene expression profiling and racial disparities in outcomes after heart transplantation. J Heart Lung Transplant. 2019 Aug;38(8):820-829. Moayedi Y, Foroutan F, Miller R et al. Risk evaluation using gene expression screening to monitor for acute cellular rejection in heart transplant recipients. J Heart Lung Transplant. 2019 Jan;38(1):51-58.

Deng M. The AlloMap™ genomic biomarker story: 10 years after. Clin Transplant. 2017 Mar;31(3).

Crespo-Leiro MG, Stypmann J, Schulz U et al. Clinical usefulness of gene-expression profile to rule out acute rejection after heart transplantation: CARGO II. Eur Heart J. 2016 Sep 1;37(33):2591-601.

Crespo-Leiro MG, Stypmann J, Schulz U et al. Performance of gene-expression profiling test score variability to predict future clinical events in heart transplant recipients. BMC Cardiovasc Disord. 2015 Oct 9;15:120.

Khush K, Pham MX, Teuteberg JJ et al. Gene expression profiling to study racial differences after heart transplantation. J Heart Lung Transplant. 2015 Jul;34(7):970-7.

Kobashigawa J, Patel J, Azarbal B et al. Randomized Pilot Trial of Gene Expression Profiling Versus Heart Biopsy in the First Year After Heart Transplant. Circ Heart Fail. 2015 May;8(3):557-64.

Deng MC, Elashoff B, Pham MX et al. Utility of gene expression profiling score variability to predict clinical events in heart transplant recipients. Transplantation. 2014 Mar 27;97(6):708-14.

Austin BA, Arnold PJ, Kao A. The Impact of Time Post Cardiac Transplant on Gene Expression Profile Scores, an Analysis of 32,043 Tests. J Cardiovasc Dis Diagn. 2013;1(3):1-4.

Costanzo MR, Dipchand A, et al. The International Society for Heart and Lung Transplantation Guidelines for the Care of Heart Transplant Recipients. J Heart Lung Transplant 2010;29: 914–956.

Deng MC, Alexander G, Wolters H et al. Low variability of intraindividual longitudinal leukocyte gene expression profiling cardiac allograft rejection scores. Transplantation. 2010 Aug 27;90(4):459-61.

Pham MX, Teuteberg JJ, Kfoury AG, et al. Gene-Expression Profiling for Rejection Surveillance after Cardiac Transplantation. NEJM 2010;20:1890–1900.

Miller JD, van der Most RG, Akondy RS, et al. Human effector and memory CD8+ T cell responses to smallpox and yellow fever vaccines. Immunity 2008;28:604–606.

Bernstein D, Williams GE, Eisen H et al. Gene expression profiling distinguishes a molecular signature for grade 1B mild acute cellular rejection in cardiac allograft recipients. J Heart Lung Transplant. 2007 Dec;26(12):1270-80.

Dedrick R. Understanding Gene Expression Patterns in Immune-Mediated Disorders. J Immunotoxicol. 2007 4:3, 201 – 207.

Fang KC. Clinical utilities of peripheral blood gene expression profiling in the management of cardiac transplant patients. J Immunotoxicol. 2007 Jul;4(3):209-17.

Goh SH, Josleyn M, Lee YT, et al. The human reticulocyte transcriptome. Physiol Genomics 2007;30:172–178.

Yamani MH, Taylor DO, Rodriguez ER et al. Transplant vasculopathy is associated with increased AlloMap gene expression score. J Heart Lung Transplant. 2007 Apr;26(4):403-6.

Deng MC, Eisen HJ, Mehra MR, et al. Noninvasive discrimination of rejection in cardiac allograft recipients using gene expression profiling. Am J Transplant 2006;6:150–160.

Starling RC, Pham MX, Valentine H et al. Molecular Testing in the Management of Cardiac Transplant Recipients: Initial Clinical Experience. J Heart Lung Transplant. 2006;25: 1389–1395.

McRedmond JP, Park SD, Reilly DF, et al. Integration of proteomics and genomics in platelets: a profile of platelet proteins and platelet-specific genes. Mol Cell Proteomics 2004;3:133–144.

Segal JB, Kasper EK, Rohde C, et al. Coagulation markers predicting cardiac transplant rejection. Transplantation 2001;72:233–237.

Bustin SA. Absolute quantification of mRNA using real-time reverse transcription polymerase chain reaction assays. J Molec Endocrinology 2000;25 169–193.

Ulich TR, del Castillo J, Guo KZ. In vivo hematologic effects of recombinant interleukin-6 on hematopoiesis and circulating numbers of RBCs and WBCs. Blood 1989;73:108–110.

Springer TA. Traffic signals for lymphocyte recirculation and leukocyte emigration: the multistep paradigm. Cell 1994;76:301–314.

Less

AlloMap remains a valuable non-invasive transplant test for rejection

This test is essential during this COVID crisis when patients are fearful to come to hospitals for a biopsy and when hospitals may not have the capacity to perform routine invasive testing. AlloMap has also been invaluable for patients who have no venous access or who have such severe scar tissue that biopsies become dangerous and very painful. In addition, noninvasive approaches to assess rejection prevent biopsy related complications which include injury to the tricuspid valve (common) that leads to right sided heart failure and cirrhosis as well as perforations that can lead to tamponade and death.

Less

I am going to begin this public comment with a simple thought experiment. Please envision an ideal test that provides a meaningful result to a patient. What do you envision as the positive predictive value (PPV) and negative predictive value (NPV) of this ideal test? Perhaps 100% PPV and 100% NPV? Now that is truly ideal and unlikely to be true of any test. So instead envision a very good test. What sort of predictive value are you willing to tolerate? 99%? 98%? 95%? Are you willing to accept that value for either the NPV or PPV or do you want that value for both? For the sake of argument, I will guess that you selected 98% and only need this to be true for the NPV or PPV of the test, which seems perfectly reasonable to me. I would further suppose that you would look to see with what value the test had when the predictive value crossed from below 98% to above 98% to assign the proper threshold value to the test.

For example, to diagnose a heart attack, a troponin must be above the 99th percentile of the upper reference limit, which for the University of North Carolina medical center’s test is a value >0.34 ng/ml. Below this value the positive predictive value is too low, but when it crosses that threshold, the test turns positive. Conversely, when a D-dimer test has a value = to 500 ng/ml it is considered normal with a negative predictive value of ~100% (per the Mayo Medical Laboratory). If the value is >500 it crosses a threshold where it does not become a positive predictor of a pulmonary embolism, but it can no longer rule it out. To me, that is how tests work, at some non-arbitrary value, the result becomes meaningful to indicate a disease or other finding.

The reason I requested that you do this thought experiment and refresh your knowledge of a couple of well-characterized studies that use either a positive or negative predictive value, is that the AlloMap does not work according to those standard test metrics. You already have data from the IMAGE study included in the initial National Coverage Determination request of 2013, but I will include below additional data from CareDx’s more recent study, the OAR Trial. And in both of these studies, you can see that you can pick essentially any of their gene expression profile (GEP) scores and they all have the same test characteristic. They all have a NPV between 97.39% and 100% and an irrelevant PPV (essentially below 5% at every value). Therefore, any GEP score provides the same NPV discriminatory value. A score of 16 has a NPV of 98.87 while a score of 36 has a NPV of 98.07. Clinically, an NPV of that value would be treated the same. Allow me to state this succinctly. It is my opinion, based purely on data, that the CareDx AlloMap test has no clinically discriminatory use for monitoring heart transplant patients. There is no threshold over which the test has a meaningful gain in NPV. Hopefully all of the other tests that CMS pays for, such as troponin values of D-dimers, have tests with threshold values that can distinguish affected patients.

Despite a consistent NPV at every GEP test value, it is useful to read descriptions of the NPV in the literature, as it seems to be the major data point suggested as validating the AlloMap test.

• From Deng, MC The AlloMap™ genomic biomarker story: 10 years after. Clin Transplant 2017 “the AlloMap test was designed as a rule out test with a high negative predictive value of >99% at the cost of a lower positive predictive value, that is, to avoid biopsy if test results were below threshold.”
• From Kobashigawa J, et al. Randomized pilot trial of gene expression profiling versus heart biopsy in the first year after heart transplant: early invasive monitoring attenuation through gene expression trial. Circ Heart Fail. 2015;8(3). “Clinical analyses found GEP to be valid from 55 days post transplant with GEP scores < 34 being associated with low risk of moderate-to-severe acute cellular rejection on EMB (ISHLT Grade ≥3A/2R). This resulted in negative predictive values of 98.1% in patients >2 to 6 months post transplant, 98.5% in patients 6 to 12 months post transplant, and 99.2% >1 year post transplant in stable outpatients.”

Essentially, the NPV is the major selling point of the test, but is always clinically the same at any GEP value. That should be very concerning to anyone who understands how clinical tests usually work. It is my understanding that the CMS has already looked into the NCD and decided it is a worthwhile test. The conclusions of the CMS review to allow MACS to cover this test are based on three studies listed here and which I will discuss below.

I would start by pointing out that each of the authors of these publications have clear conflicts of interest. Not only are these studies paid for by the company, which is reasonable, but every major author either works for the company or has received money from XDx. The bias in the conclusions of these reports is notable.

The first paper, Deng MC et al, reuses the data from the IMAGE trial (Pham MX et al NEJM, 2010). The IMAGE trial data is the flawed data upon which this NCD request was initiated and shows no discriminatory value to the GEP score. This paper adds nothing new and provides no value to the decision making of CMS.

The second paper, Kobashigawa et al, is deceptive. In the abstract they wrote “The primary end point included a composite of death/retransplant, rejection with hemodynamic compromise or graft dysfunction at 18 months post transplant. A coprimary end point included change in first-year maximal intimal thickness by intravascular ultrasound, a recognized surrogate for long-term outcome.” However in the methods they wrote (underline mine) “Given the relatively small patient population of heart transplant centers, a single-center study of de novo heart transplant recipients could not offer adequate power to detect differences in our composite primary end point of death/retransplant, rejection with hemodynamic compromise, or allograft dysfunction, or our coprimary IVUS end point of 12-month maximal intimal thickness compared with baseline. These 2 primary end points were chosen for clinical relevance, and we include statistical analyses for descriptive purposes to highlight the feasibility of this pilot study.”

I have personally never seen a paper claim what their endpoint would be in the abstract and yet state that the endpoints were impossible to obtain in the study and that the statistics were purely meant to be descriptive. Further, there are only 3 actionable rejection events among the entire population (N=60) in 18 months of follow up. It is not possible to draw any meaningful conclusions about GEP from this paper as a result of too few endpoints. This paper should not be included in the decision making of CMS.

The third paper, Crespo-Leiro et al. is a study to investigate if variability of 4 GEP scores can be used to predict a future composite endpoint of death from any cause, re-transplantation or graft failure that occur ≥ 3 years after the last GEP measurement. They demonstrate that increased GEP score variability is observed among patients with a composite endpoint and that a NPV on variability might suggest a more stable course 3 years in the future. This is certainly an interesting study, but it is not relevant to the use of an AlloMap GEP score to predict current heart transplant rejection. It is also important to note what they did not report. The authors did not report that a low GEP score was not associated with the composite endpoint. That would be the more logical measurement to promote, consistent with the low GEP score (<34) being the approved value to use (not variability). It is possible to likely that that data was not shown, because it was not significant. CMS should be able to ascertain if that postulate is correct. Furthermore, this study does not provide any correlation of GEP with acute rejection so it is not valuable for the decision making of CMS on the NCD.

Thus, the manuscripts upon which the CMS has made the NCD, are questionable to the task of making the NCD determination.

I would like to point out a study, from a group that does not claim to have any conflicts of interest with CareDx, for a different perspective on the test. Nikolaidis LA et al. Abstract 2312: Limitations of Molecular Expression Profiling (Allomap Score) in Predicting Endomyocardial Biopsy Findings in Cardiac Allograft Recipients: Two Year Follow-Up Surveillance Study Circulation (https://www.ahajournals.org/doi/abs/10.1161/circ.116.suppl_16.II_505-b). From the abstract, the authors note: Results: The mean Allomap score in the entire cohort was 29.9±0.4 (median: 31). The distribution was non-Gaussian with the majority of samples in the upper quartile [< 24: n=43 (18 %), 25–29: n=49 (20%), 30 –34: n=98 (40 %), > 35: n=52 (22 %)] Only 9 biopsy specimens (3.7%) were consistent with severe rejection (ISHLT 3A/3B or ≥ 2R per amended classification). Scores were not significantly different across the spectrum of ISHLT grades. Allomap scores of non-rejectors (ISHLT 0), mild rejectors (1-R) and significant rejectors (≥ 2R) were not significantly different (29.3±0.6 vs. 30.2±0.5 vs. 30.1±2.2, respectively). In predicting clinically significant (ISHLT ≥ 3A or 2-R) rejection, the NPV of Allomap was 95 %, yet PPV was < 10 %. Sensitivity (30 –55%) and specificity (37–78%) were highly dependent on the threshold used (≥30 vs. ≥35), thus ROC performance was suboptimal.

Additionally, I draw your attention to the OAR trial (Moayedi et al. Risk evaluation using gene expression screening to monitor for acute cellular rejection in heart transplant recipients. JHLT, 2018). Three study authors serve on the advisory board of CareDx and a fourth author works for CareDx. The OAR is an observational, prospective, multicenter study that included 1,504 patients. There were 2,456 matched endomyocardial biopsies (EMB) to GEP scores. An EMB value of 2R or greater is the value at which most clinicians treat the patient with increased immune suppression. If you recall, the original NEJM IMAGE paper set the GEP score threshold to 34, arbitrarily to reduce the number of biopsies. “[T]he protocol was amended to increase the threshold for a mandatory biopsy to a score of 34 in order to minimize the number of biopsies that would be needed in the gene-profiling group.” The AlloMap test was approved for patients ≥6 months post-transplant. However, the OAR study shows the Median interquartile range (IQR) is 32 from 6-12 months for a 2R result and an IQR of 31.5 for a 3R result (Table 2). Only in patients greater than one year post-transplant is the IQR for 2R and 3R at 35+. Per CareDx, a value <34 has a negative predictive value of >98% (2 in 100), yet the median 2R IQR in the 6-12 month range suggested that >50% of patients with rejection had a value less than 34! Their cutoff (which is an arbitrary value), if followed, would miss half of the patients who need an adjustment of their immune suppression. Also potentially worrisome in this study is that 61 patients had an EMB for symptoms, while 344 had one for an elevated GEP score. If the 61 patients had symptoms, but not an elevated GEP score, that could indicate a ~20% miss rate on the GEP score for identifying sick patients, which is dangerously high. Regardless, once again, NPV is lauded. “In the OAR cohort, GEP monitoring for acute rejection was associated with a high NPV, confirming its utility as a screening test.” The tables for NPVs from the OAR study are given below and closely mirror that of the IMAGE study, which again shows no discriminatory power across the GEP scores. It is an excellent NPV at every value. The conclusion of all of the new information is that unfortunately the CareDx AlloMap test, despite good intentions and a robust scientific pursuit for a blood-based test, is not adequate to the task.

I recognize the easiest course of action is to continue to approve the AlloMap test for transplant surveillance. I suspect that will be done. If you chose any other path you will endure the wrath of lobbyists and their political allies. And over what? A few million dollars of taxpayer dollars a year? This is an asymmetrical fight. However, it is absolutely clear, in my opinion and the opinion of the leading hospitals in the US that do not offer this test, that AlloMap offers no value to the patients who rely on it to keep them healthy. By supporting an inferior (and not cost-effective) test, CMS is not acting in the best interests of Medicare and Medicaid patients, and that should be the primary consideration.

Appendix 2A) OAR GEP test performance results in 0-6 months post transplantation period with timing from GEP to biopsy within 14 days

GEP Score	PPV (%)	NPV (%)	Sensitivity (%)	Specificity (%)
12	2.11	100	100	5.3
13	2.13	100	100	6.1
14	2.15	100	100	6.7
15	2.19	100	100	8.6
16	2.12	98.87	95	10.4
17	2.17	99.07	95	12.3
18	2.22	99.22	95	14.4
19	1.92	97.39	80	16.4
20	1.98	97.76	80	18.9
21	2.07	98.14	80	22.6
22	2.17	97.41	80	26.3
23	2.14	98.27	75	29.8
24	2.10	98.15	70	33.1
25	1.94	97.86	60	37.3
26	1.91	97.85	55	41.8
27	2.11	98.09	55	47.2
28	2.30	98.26	55	51.7
29	2.64	98.44	55	57.9
30	2.83	98.44	50	64.3
31	2.33	98.44	35	69.7
32	1.99	98.14	25	74.8
33	2.57	97.99	25	80.5
34	2.07	98.13	15	85.5
35	3.27	97.99	15	90.8
36	1.87	98.07	5	94.7
37	3.74	98.08	5	97.3
38	9.34	98.10	5	98.9
39	2.00	98.0	5	100

GEP: Gene Expression Profile, PPV: positive predictive value, NPV: negative predictive value. Nominal threshold highlighted in green (30)

Appendix 2B) OAR GEP test performance results in >6 months post transplantation period with timing of GEP to biopsy within 14 days

GEP Score	PPV (%)	NPV (%)	Sensitivity (%)	Specificity (%)
12	2.06	100	100	0.87
13	2.07	100	100	1.23
14	2.07	100	100	1.59
15	2.08	100	100	2.01
16	2.10	100	100	2.72
17	2.12	100	100	3.60
18	2.13	100	100
19	2.16	100	100	5.50
20	2.14	99.3	97.7	6.79
21	2.15	99.4	97.7	7.51
22	2.03	97.9	90.7	9.51
23	2.08	98.3	90.7	11.2
24	2.15	98.7	90.7	14.1
25	2.18	98.7	88.4	17.5
26	2.15	98.4	83.7	21.0
27	2.19	98.5	81.4	24.7
28	2.26	98.5	79.1	29.1
29	2.31	98.5	74.4	34.7
30	2.24	98.3	67.4	39.0
31	2.42	98.5	65.1
32	2.52	98.5	60.4	51.6
33	2.68	98.5	55.8	57.9
34		98.5	51.1	63.4
35	3.22	98.5	44.1	72.4
36	4.04	98.4	34.9	82.7
37	4.29	98.2	20.9	90.2
38	4.21	98.1	9.3	95.6
39	7.88	98.0	4.7	98.8

GEP: Gene Expression Profile, PPV: positive predictive value, NPV: negative predictive value.

[1] Deng MC, Elashoff B, Pham MX, et al. Utility of Gene Expression Profiling Score Variability to Predict Clinical Events in Heart Transplant Recipients. Transplantation. 2014;97(6).
[2] Kobashigawa J, Patel J, Azarbal B, et al. Randomized pilot trial of gene expression profiling versus heart biopsy in the first year after heart transplant: early invasive monitoring attenuation through gene expression trial. Circ Heart Fail. 2015;8(3).
[3] Crespo-Leiro MG, Stypmann J, Schulz U, et al. Performance of gene-expression profiling test score variability to predict future clinical events in heart transplant recipients. BMC Cardiovasc Disord. 2015;15(120).

Less