Response to Comments: MolDX: Gene Expression Profile Tests for Decision-Making in Castration Resistant and Metastatic Prostate Cancers

The following comment was submitted to CGS:

The development of new treatment paradigms has greatly enhanced options for many cancer patients. However, decision tools to individualize therapy selection have been sadly lagging behind. In this regard, any biomarker capable of predicting benefit of one therapy over another would be most welcome. Unfortunately, the evidence presented here does not support the use of the Decipher GC score or the PAM50 subtypes as predictive markers in the treatment of prostate cancer.

The data presented here are derived from two studies, the biomarker analyses of a subset of patients participating in the CHAARTED (Hamid et al. 2021) and SPARTAN trials (Feng et al. 2021), respectively. Addressing different prostate cancer patient groups and different treatment additions, both trials tested the benefit from combination therapy compared to ADT alone. Each biomarker study then attempts to establish a relationship between biomarker level and treatment benefit. A biomarker that is predictive of treatment benefit would be statistically significant in a biomarker-treatment interaction test.

The proposal cites data from both studies indicating that the hazard ratio point estimate for the combination treatment is lower for patients with high GC scores than the hazard ratio for patients with a low GC score. This is suggestive, but not definitive, as the confidence intervals on both estimates are wide, allowing for the possibility that the treatment benefit difference is much smaller or even negligible. Similarly, the hazard ratios for treatment efficacy in patients with PAM50 luminal vs basal tumor samples differ but confidence intervals are wide and overlapping. Not surprisingly, biomarker-treatment interaction tests in the SPARTAN study were not significant and were not even performed in the study by Hamid et al. (see Methods).

In addition to the lack of statistical evidence of an association between biomarkers and treatment benefit, the two studies differ in their categorization of the GC score. While the study by Feng et al. applies a predefined threshold, the study by Hamid et al. categorizes GC scores by quartiles in the analysis cohort. These newly defined thresholds would have to be validated.

In summary, data on the predictive value of GC score or PAM50 subtypes is suggestive, but not conclusive. The authors of both papers acknowledge this uncertainty and emphasize that further studies are needed to establish these biomarkers as predictors of treatment benefit in prostate cancer patients.

References were provided for review.

Thank you for your comment.

While the limitations mentioned are of concern, no study is perfect. Coverage for this type of service in prostate cancer management has been established for some time and the evidence has only continued to grow. On the whole the evidence supports the use of this service. Should additional evidence be presented demonstrating that these tests lack benefit, we would gladly review that and revisit this policy as necessary.

The following comment was submitted to Palmetto GBA:

I am writing in support of Medicare covering beneficiaries to undergo biomarker testing for relapsed, recurrent, advanced, and metastatic prostate cancer. As a genitourinary medical oncologist, I spend hours talking with patients about their prostate cancer and how to best move forward with treatment. Many of our discussions revolve around trying to understand how aggressive or tame the cancer is, and how effective we expect different treatments to be against the cancer. Biomarkers that can help us predict how aggressive the cancer is can help us determine how aggressive we need to be with treatment as we often use our more toxic therapies in settings in which the cancer poses a greater, more urgent threat. Biomarkers and molecular tests that can tell us whether a given treatment will help a specific patient or not can be even more useful, helping us understand whether we should use drug x or drug y in a given setting. This is critical in cancer care when the side effects of treatment can themselves be deadly or cause significant illness. Ensuring coverage for Medicare beneficiaries means that patients will have access to understanding their prostate cancer and understanding the potential that different treatments will help. This power can make all the difference when they are facing life-threatening illnesses, and can help them and their loved ones have the peace of making informed decisions to support their health.

Thank you for your support of this policy.

The following comment was submitted to Palmetto GBA:

I attach two manuscripts which suggest the utility of gene expression analysis and molecular classifiers in advanced prostate cancer. This is specifically relevant to men with high risk recurrent prostate cancer and a rising PSA after radical prostatectomy, and who are considering salvage radiotherapy. In one study, a retrospective analysis of the phase 3 randomized RTOG 9601 trial, the DECIPHER gene expression classifier was able to identify those men in this setting who benefited from androgen deprivation therapy (ADT) with salvage radiotherapy based on an elevated risk of recurrence, metastasis, or death long term which ADT was able to prevent. The second was a prospective clinical trial (STREAM) that demonstrated that successful potent hormonal therapy with radiation could be predicted by this same gene expression classifier, and identified a subset of men with adverse gene expression profiles that remain at high risk for recurrence and would benefit from additional or alternative approaches.

Together, this data indicates clinical utility for the Decipher genomic classifier in this specific disease state in recurrent advanced prostate cancer, where patient management can be favorably impacted and can result in improved long term outcomes, including durable remissions and prevention of metastatic disease or death.

References were provided for review.

Thank you for your support of this policy. We agree that there is utility of molecular classifiers in advanced prostate cancer.

The following comment was submitted to Palmetto GBA:

I am a Urologist and my practice focuses exclusively on prostate cancer. I am writing to express my support for the Proposed Local Coverage Decision (LCD): “Gene Expression Profile Tests for Decision-Making in Castration Resistant and Metastatic Prostate Cancers.”

The incidence of metastatic prostate cancer has been increasing. Thankfully, we have also seen the increase of novel therapies and therapeutic approaches that can extend survival for men. Choosing the correct treatment regimen however is nuanced and a balance between over and undertreatment. Knowledge of the molecular underpinnings of the tumors I treat is invaluable in making shared decisions regarding care with the men I treat. The Proposed LCD could provide me and my Medicare patients access the Decipher Prostate Genomic Classifier, a highly validated test which evaluates gene expression in routinely collected prostate cancer tissue, and would help guide superior decisions for my patient’s care.

While I am excited about the Proposed LCD, I would like to request a modification of the 3rd coverage criteria listed. Currently the 3^rd coverage criteria states:

“The patient has not been tested with the same or similar test for prostate cancer.”

I believe this statement may be too broad/vague as stated. I would like to bring up two points and suggest a modified statement.

First, over the last decade we have seen the development of several gene expression based tests for prostate cancer (i.e. Decipher Prostate Genomic Classifier, OncotypeDx Prostate, Prolaris Cell Cycle Progression). These tests are similar in that they measure gene expression from prostate tissue but they are not similar in the genes evaluated, nor are they similar in the level of evidence and validation (the Decipher Prostate Genomic Classifier is by far the most validated, and is the only test with evidence in the metastatic setting). In my practice and at my institution, Decipher is the only gene expression test utilized. That said, I would not want a patient denied Decipher testing on the basis that he was tested with a “similar” but fundamentally different gene expression test for his prostate cancer along his cancer journey.

Second, while most patients will present with metastatic disease and genomic testing will be done on their diagnostic biopsy, there are alternative scenarios where the same patient might benefit from being tested twice with a gene expression test through his cancer journey. One example could be a gentleman who initially presents with localized favorable intermediate risk prostate cancer. His genomics could be tested and a clinical decision could be made to place him on active surveillance. Years later his tumor might progress (with a change in histologic grade and stage) to metastatic disease. At this point I would want to utilize the most recent cancer tissue for genomic testing as it would be the most likely to capture his disease biology.

A modification could be:

“The patient has not been previously tested with the same test in this episode of care”

Thank you for your support of this policy. The evidence to-date does not support the use of more than one test for a given patient with prostate cancer. As stated in the policy, “Further research is needed to better define intra-individual genomic tumor heterogeneity and divergence as well as phenotypic plasticity that may occur as a result of pressures from systemic therapies.” Given this, it is expected that most patients will only require one such test for their prostate cancer indication.

Further, the most appropriate and evidence-based testing should be used for a patient’s given indication. We note that only validated tests that successfully complete a Technical Assessment review by this contractor may be covered services. Testing twice for the same intended use, just because two tests interrogate different genes, is considered duplicative and not reasonable and necessary.

We have revised the aforementioned criterion which now states “The patient has not been tested with the same or similar test for the same intended use.” Should evidence develop to support subsequent testing, the subsequent test would also have to meet ALL criteria of the policy including technical assessment review for that intended use.

The following comment was submitted to Palmetto GBA:

On behalf of Veracyte, Inc., I am writing to express our support for the Proposed Local Coverage Decision (LCD): “Gene Expression Profile Tests for Decision-Making in Castration Resistant and Metastatic Prostate Cancers.”

The treatment landscape for patients with metastatic prostate cancer continues to advance and evolve. Over the last several years, technology has emerged to identify metastatic disease earlier and practice-changing trials have led to the approval of novel drugs or expanded the indications for existing drugs, all which have led to the availability of multiple treatment options for patients with metastatic prostate cancer and a need for tools to navigate treatment decisions. Biomarkers, including gene signatures on Veracyte’s Decipher testing platform, have been studied and continue to be studied across the continuum of both localized and metastatic prostate cancer to help stratify patient risk and aid in treatment decision-making. Veracyte strongly supports this Proposed LCD, which provides Medicare beneficiaries and their physicians access to tools like the Decipher Prostate Genomic Classifier to help navigate their treatment journey and ultimately improve both their oncologic outcomes and quality of life. We encourage MolDX to finalize the Proposed LCD as expeditiously as possible.

We would respectfully request a single amendment to coverage criterion (3), which currently states coverage would only be afforded if:

“The patient has not been tested with the same or similar test for prostate cancer.”

Instead, we recommend the following revised language to allow for the appropriate use of gene expression testing to inform clinical decision making across the spectrum of prostate cancer:

“The patient has not been previously tested with the same or similar test for prostate cancer unless the test will be performed on a different sample or there are new clinical circumstances that make additional testing medically advisable (i.e., new tumor found, patient has experienced disease progression, etc.) and the result will aid in clinical decision-making.”

Our understanding is that MolDX’s intent regarding proposed criterion (3) is to limit testing on each sample with one test so that providers are not using 2 or 3 different tests during the same episode of care, which would indeed be a waste of healthcare resources. However, we are concerned that this criterion, as currently written, sweeps too broadly and would deny coverage for testing in scenarios where the test interrogates new cancer genetic content and will aid in clinical decision-making.

For instance, we believe that coverage would be appropriate in the following scenarios:

1. A patient is diagnosed by a urologist (Provider #1) using a systematic or template biopsy which finds a left- sided tumor (Sample #1) with Grade Group 2 disease (i.e., favorable clinical risk). Provider #1 then orders a genomic test on this sample, and the result shows “low risk of metastatic progression”, which is not in alignment with the patient’s clinical factors (e.g., PSA is 100 ng/mL). The same patient then visits with a new provider (Provider #2) at a comprehensive cancer center who conducts a targeted MRI-guided biopsy, which finds a right-sided tumor (Sample #2) with Grade Group 5 disease. Sample #1 and Sample #2 originated from two different lesions recovered from two different samplings of the prostate by different providers during two different episodes of care, which contain distinct information with regards to the tumor biology. In a situation such as this, Provider #2 should be able to use a genomic test to assess Sample #2, which is believed to be the “index” lesion and is more consistent with other clinical findings, the result of which will be used to aid in clinical decision-making. As proposed, coverage criterion (3) would appear to prohibit Provider #2 from testing the new sample, even if medically advisable to do so.
2. A patient is diagnosed with NCCN unfavorable intermediate risk disease, and his physician orders a genomic test other than Decipher Prostate to determine treatment intensity. As standard workup, the patient has a PSMA PET to evaluate the extent of his disease. The results show three small lesions, and the patient is then diagnosed with metastatic disease and is eligible for singlet, doublet, and triplet therapy according to PROS-12 in the NCCN The genomic test that was originally ordered is not validated for and the results are not applicable for navigating treatment decisions for patients with metastatic prostate cancer, so the physician orders Decipher Prostate to help them make the decision between androgen deprivation therapy (ADT) alone, ADT + abiraterone, and ADT + docetaxel + abiraterone. As currently written, coverage criterion (3) would appear to prohibit the physician from ordering Decipher Prostate testing for this patient, even if medically advisable to do so.
3. A patient is initially diagnosed with NCCN unfavorable intermediate risk disease and receives Decipher Prostate testing on his biopsy sample. His Decipher score is 0.65, and he receives a radical prostatectomy. Five years later, they find distant metastasis. His physician wants to perform a Decipher test on his post- prostatectomy sample to help decide between doublet and triplet therapy. As currently written, coverage criterion (3) would appear to prohibit the physician from ordering Decipher Prostate testing for this patient, even if medically advisable to do so.

We believe that the recommended modification would remove these potentially unintended consequences of the current criterion (3), and would instead allow MolDX the flexibility to determine the appropriate coverage criteria for each individual gene expression profile test through the Technical Assessment process. We note that our recommendations would align with the approach taken by CMS in NCD 90.2 for next-generation sequencing for cancer, which prohibits coverage of the same test “for the same cancer genetic content,” but allows for coverage of additional testing if the cancer genetic content is not the same. In establishing this standard, CMS “recognize[d] that specific clinical scenarios may necessitate repeat testing.” Given the above considerations, we believe the same approach is warranted in the finalized LCD.

We appreciate MolDX’s consideration of our comments, and encourage the expeditious finalization of Draft LCD with a modification to coverage criterion (3) to ensure that MolDX has the discretion to allow for more than one test of a patient at different time points when medically advisable.

Thank you for your support of this policy. Please see Response #4.