Response to Comments: MolDX: Lab-Developed Tests for Inherited Cancer Syndromes in Patients with Cancer (DL39040)

We are writing this letter to urge you to rescind the proposed local coverage determination (LCD) entitled “MolDX: Next Generation Sequencing Lab-Developed Tests for Inherited Cancer Syndromes” (“Draft LCD”) and the associated draft Local Coverage Article entitled “Billing and Coding: MolDX: Next Generation Sequencing Lab-Developed Tests for Inherited Cancer Syndromes” (“Draft Article”). The MolDX program and Medicare beneficiaries alike would be better served by a new draft LCD that addresses the numerous procedural and substantive deficiencies in the current draft. Among these are:

• Stakeholders have had to assemble the pieces of the coverage puzzle on their own – in addition to the Draft LCD and Draft Article, a MolDX Worksheet is integral to coverage for these services, yet it was not referenced or linked from the Draft LCD or Draft Article in any way.
• “Minimum genetic content” for test coverage would be included in associated MolDX documents outside of the Draft LCD, which could change at any time in the MolDX program’s sole discretion – without notice to stakeholders or any public comment.
• The Draft Article is a “work in progress” and is not ripe for meaningful stakeholder input and comment.
• The Draft LCD fails to explain its effect on other LCDs for hereditary cancer that currently are in effect and whose subject matters presumably also would be addressed by the Draft LCD.
• The MolDX Worksheet presents problematic “minimum genetic content” for certain cancer types and would lead to non-coverage for next generation sequencing (NGS) panel options currently offered by labs and supported by medical literature.

We discuss these issues in further detail below, to illustrate why the Draft LCD is not appropriate for stakeholder comment at this time and should be withdrawn in its entirety, along with the Draft Article and MolDX Worksheet.

A. The Draft LCD’s procedural deficiencies deny stakeholders the opportunity to provide meaningful comments on it.

An LCD is a determination by a Medicare Administrative Contractor (MAC) regarding whether or not a particular item or service is covered on a contractor–wide basis.¹ As written and presented, the Draft LCD lacks sufficient clarity to allow stakeholders to understand which NGS testing for inherited cancers would be covered and under what circumstances.

The Draft LCD must be considered in concert with the Draft Article, which contains billing and coding guidance that would implement the Draft LCD’s coverage policies. In this case, the Draft LCD also must be considered in concert with a MolDX worksheet entitled “Analytical Validity, Clinical Validation summary worksheet, NGS in Hereditary Germline testing, Cancer panels” (“Worksheet”), which includes “minimum genetic content” for test coverage and enumerates the required genes for seven hereditary cancer types.² The Worksheet is not referenced anywhere in the Draft LCD or Draft Article, yet it is the keystone of MolDX coverage policies for NGS testing for inherited cancers under the Draft LCD. Without notice from MolDX about the existence of this important piece of the “coverage puzzle” and no reference to it in the Draft LCD, some ACLA members did not become aware of it until last week. That short period of time is wholly insufficient for stakeholders to comment meaningfully on the Worksheet.

Even more troubling than the fact that stakeholders were unaware of the Worksheet is that, if it is used as proposed, it effectively would allow the MolDX program to alter coverage for NGS testing for inherited cancers unilaterally without any notice to or comment from stakeholders. Under “Criteria for Coverage,” the Draft LCD states that in order for a test to be eligible for coverage:

The assay performed includes at least the minimum genetic content (genes or genetic variants) required for clinical decision making for its intended use that can be reasonably detected by the test. Because these genes and variants will change as the literature and drug indications evolve, they are listed separately in associated documents, such as the MolDX TA forms.

While a MAC must engage in consultation, publication of proposed LCD, an open meeting concerning the proposed policy, and an opportunity for public comment in writing for a revised LCD that makes a substantive change or discretionary coverage update,³ there is no such requirement for a change to an “associated document” like the Worksheet.

We are very concerned about the lack of procedural safeguards to ensure that stakeholders may provide input on and have ample advance notice of changes to the “minimum genetic content” that would spell coverage or non-coverage of a given NGS test for inherited cancers. This is a fatal flaw in how coverage under the Draft LCD would be implemented, and the MolDX program should revise its approach so that the “minimum genetic content” is included in the body of the LCD and not “separately in associated documents.”

B. Overlap with current effective LCDs for hereditary cancers causes confusion and uncertainty for stakeholders.

The Draft LCD makes no mention whatsoever of two currently effective MolDX LCDs for hereditary cancer testing: BRCA1 and BRCA2 Genetic Testing and Genetic Testing for Lynch Syndrome. This has created considerable confusion about whether the existing MolDX LCDs would be replaced or remain in effect and, if they remain in effect, how to reconcile policy differences between the Draft LCD and existing LCDs.

One example of a potential inconsistency between the LCDs is the ICD-10 codes in the Draft Article, which raise questions about coverage of BRCA1 and BRCA 2 genetic testing. The Draft Article includes ICD-10 code Z80.3 (family history of malignant neoplasm of breast). It is not clear whether this indicates that when that ICD-10 code is included on a claim for a test represented by CPT code 81432 (genomic sequence analysis panel for hereditary breast cancer-related disorders), the claim would be covered. Under MolDX: BRCA1 and BRCA2 Genetic Testing, BRCA 1 and BRCA2 testing is non-covered when an individual has no personal history of cancer.

The interplay between the Draft LCD and other MolDX LCDs currently in effect is a fundamental question that must be addressed before stakeholders can comment on the Draft LCD meaningfully. Without information about the disposition of the existing LCDs for hereditary cancer testing, it is impossible to know the scope of the Draft LCD or for stakeholders to comment on it fully.

C. Contents of the “minimum genetic content” Worksheet is inconsistent with Medicare policy and current laboratory practice.

The MolDX program has provided no commentary or explanation of the Worksheet, other than it is an “associated document” for the Draft LCD that could be altered at will. The sources of the “minimum genetic content” for NGS testing for various cancer types is known only to the MolDX program. It is essential that the rationale for including genes or genetic variants in coverage of testing is transparent to stakeholders and adequately explained. Without any such information, it is not possible for stakeholders to provide substantive comment on the Draft LCD and its “associated documents” as a whole.

As presented, the Worksheet requires a minimum of 11 genes for coverage of a breast cancer assay. This would exclude from coverage currently-available NGS panel options that include a smaller number of high-risk genes but that do not include each and every one of the 11 genes in the Worksheet. It also would exclude from coverage a panel that includes only BRCA1 and BRCA2, which is contrary to current MolDX coverage policy.⁴ Similarly, a minimum of 12 genes are required for coverage of an NGS panel for inherited colon cancer, which would exclude from coverage a panel that includes only the genes associated with Lynch syndrome, which is contrary to current MolDX coverage policy.⁵ Particularly because the Worksheet contains information that effectively may eliminate Medicare coverage for currently-covered tests, it is critical that stakeholders have more information from the MolDX program about the Worksheet’s content and development.

D. Conclusion

For the foregoing reasons, we urge the MolDX program to rescind the Draft LCD, Draft Article, and Worksheet in their entirety and engage with stakeholders to improve the coverage paradigm for NGS testing for inheritable cancers. We also urge you to follow such rescission with a rationale for why a new draft LCD is needed and information about the disposition of current MolDX LCDs for hereditary cancer testing.

If the MolDX program does not rescind the documents, at a minimum, you must extend the Draft LCD comment period for 60 days so that stakeholders have adequate time to provide the MolDX program with meaningful feedback and comment on both the form and the substance of the Draft LCD and accompanying documents. Thank you for your consideration of our request.

References were provided for review.

The policy's scope is NGS germline testing, and exclusive of acquired-cancer uses. Tests are covered per NCD 90.2, with a cancer diagnosis, clinical indication, and risk factor. The test must complete a TA and must "include at least the minimum genetic content," not here stated.

1. POSITION OF DRAFT VS LYNCH AND HBOC LCDs UNCLEAR

My intuition was initially that this umbrella LCD would replace HBOC and LYNCH LCDs.

However, this is not stated, and stakeholders have expressed considerable confusion. It is very difficult to comment on the LCD give this vagueness and given the large areas of unspecified content in the LCD. It may be better to retract this version promptly and return with a clearer version in September.

2. BURDENSOME TASK ENVISIONED

The LCD does not refer to any specific MolDX document, or example document, for a list of "the minimum specified content." I have heard there is such a draft list circulating but could not find it on the MolDX or Palmetto websites.

If indeed the plan is a spreadsheet mapping cancers to genes, this may be ill conceived. MolDX already has a great deal of challenge meeting submission volume and deadlines and biting off all cancers and all genes in hereditary disease with constant updating seems burdensome. Indeed, the margins of necessity are always in flux. It may be better to point to major guidelines (such as NCCN), an approach used in the MolDX pharmacogenetics LCD. Outside of MolDX, the CMS NGS MAC HBOC LCD simply points to NCCN indications for HBOC testing. Trying to control all aspects of all uses of genomics in healthcare may be a larger task than MolDX needs to take on. It is unclear there is, at any time, a bright line between medically necessary and other genes. Given this, allowing clinical judgment may be appropriate.

In addition, Medicare germline risk coverage has traditionally included (a) gene name, (b) cancer name, and (c) a range of clinical factors. These (a,b,c) do not easily map to the 2D spreadsheet format.

3. LCD DOES NOT FULFILL ROLE OF INFORMING PHYSICIANS AND PATIENTS

Because all key decisions are left to other documents unnamed, it is difficult for physicians and patients to use this LCD to determine coverage. This is a fundamental purpose of LCDs. They also allow coverage to be audited by third parties (e.g. RACs) or to be applied by third parties (e.g. Medicare Advantage plans).

4. OVERUSE OF NON ARCHIVED DOCUMENTS IS BAD

MolDX appears to plan to rely heavily on extrinsic documents like Excel spreadsheets. Historically, these are not archived and are often updated (e.g. "version 14"). This makes it impossible to trace what was covered at a particular time point. This makes it impossible for auditors (e.g. CERT, RAC) to apply rules. Unlike LCDs, updates rarely or never include a punch list of updates. Sometimes users can make their own redline, but this is a Sherlock Holmes exercise that should not be required.

5. WOULD BE VERY HELPFUL TO DEFINE "CLINICAL INDICATION" and "RISK FACTOR."

We assume MolDX is following the definitions of these terms as given in the NCD, but it's not clear this is the case. For example, MolDX might feel obligated to follow rules but not NCA remarks.

In any case, while maintaining limited quotation from the NCD, it would be very helpful to define these terms in the LCD, because the LCD is cryptic without understanding these terms. In fact, these are not defined "in the NCD" but in "the decision memo" or NCA. Clinical indications are different from risk factors. Clinical indication is: a sign, symptom, test, or condition. A risk factor is: a variable associated with a disease, including age, gender, or family history. A risk factor is not a sign or symptom.

It appears that a personal history of cancer IS a clinical sign or symptom, while risk factor can be interpreted more broadly than family history, including as stated by CMS, age or gender. This is notable because recent studies have shown a closely similar prevalence of risk genes in cancer patients > 65 years with or without a family history. Indeed, not half but most patients may be missed by applying arbitrary (poorly proven) family history rules. Therefore, a personal history of cancer (or promise of potential interventions) could be the clinical indication per NCD, while age > 65 years could be the newer LCD rational risk factor per NCD (replacing family history risk factor).

6. PARALLEL LAB PROCESS FOR RNA ANALYSIS ADDS VALUE

Recent data shows that parallel processes (an independent lab workstream) for RNASEQ can add significant value in hereditary cancer testing. Traditionally, mutations were found by DNASEQ (say, 5% for example) and more pathogenic lesions by DUP DEL analysis (say, 2% for example). The total yield was a small proportion of patient eligible for testing (here, for example, 7%). It was not feasible to sequence all intronic or non coding DNA, because it yields countless variants (SNPs) to which prediction rules can't currently be applied. However, one can "jump over" the problem of interpreting noncoding DNA by documenting major RNA lesions (such as truncation mutations or exon skipping). Data shows these can be as numerous as DNA Dup Del lesions. RNASEQ can also resolve VUS in some cases. By documenting as many lesions as DUP DEL, which is covered, RNASEQ should be considered an additional resource that is medically necessary when performed.

References provided for review.

Thank you for the opportunity to submit comments on Palmetto’s Proposed Local Coverage Determination: MolDX: Next-Generation Sequencing Lab-Developed Tests for Inherited Cancer Syndromes. Invitae is the fastest-growing clinical genetics company in the United States. Our mission is to bring comprehensive genetic information into mainstream medicine to improve healthcare for patients while at the same time make testing affordable and accessible to those who can benefit from it.

Criteria for Coverage:

All the following must be present for coverage eligibility:

• Meets the criteria set in NCD 90.2 for coverage, including that the patient has:
  • Any cancer diagnosis
  • AND a clinical indication for germline (inherited) testing for hereditary cancer
  • AND a risk factor for germline (inherited) cancer
  • AND has not been previously tested with the same germline test using NGS for the same germline genetic content.
• The test has satisfactorily completed a Technical Assessment (TA) by MolDX for the stated indications of the test.
• The assay performed includes at least the minimum genetic content (genes or genetic variants) required for clinical decision making for its intended use that can be reasonably detected by the test.
  • Because these genes and variants will change as the literature and drug indications evolve, they are listed separately in associated documents, such as the MolDX TA forms
  • A single gene may be tested if it is the only gene considered to be reasonable and necessary for a cancer type.
  • If a previous NGS test was performed with a similar/duplicative intended use, a subsequent test is only reasonable and necessary if the non-duplicative genetic content of the second test is reasonable and necessary.

Response and Comment:

We fully support the requirement that tests that have not undergone FDA review should undergo a completed Technical Assessment by MolDX and that there be a reasonable expectation that the tests’ reportable ranges will contain the minimum number of genes and variants required for clinical decision making. Tests that have been authorized by the FDA either through premarket approval or 510(k) clearance should be exempt from completing the Technical Assessment process. We do, however, disagree with the CMS requirements that there be

1. a clinical indication for germline (inherited) testing for hereditary cancer,

AND

2. a risk factor for germline (inherited) cancer

NCD 90.2 defines “clinical indication” as typically a personal history of cancer, and “risk factor” as contributory history (usually family history, but sometimes multiple cancers in the patient.) While fully recognizing that both requirements are national binding criteria in the NCD for genetic testing, which MolDX can't unilaterally change, we felt it important to use this comment opportunity to document why these criteria are outdated and in fact potentially harmful.

We will first discuss requirement #2, that there be “a risk factor for germline (inherited) cancer” to justify germline testing. Historically, germline testing for Hereditary Cancer Predisposition Syndromes (HCPS) was adopted for BRCA1/2 testing and, soon afterwards, for Lynch syndrome testing. The first genes were discovered through the study of families with multiple affected relatives and autosomal dominant inheritance and, as a consequence, it was originally thought that all high-risk HCPS had a clear Mendelian pattern of inheritance. As a result, early onset of disease, bilaterality, and a strong family history came to be seen as sine qua non “risk factors” that were to be used to determine if a cancer predisposition syndrome was likely or not. However, we now know that these risk factors are often not present in individuals who carry a pathogenic HCPS gene variant. Even high-risk genes can have incomplete penetrance in a family or present as a de novo mutation.^1-3 Dozens of studies show that risk factors have a high cost in sensitivity, causing many patients to be missed, and may or may not meaningfully enrich the testing yield.

As a consequence, testing guidelines deprive a significant fraction of individuals of a molecular diagnosis of a HCPS caused by pathogenic variants in genes, many of which have accepted clinical actionability (discussed in later section).

Medicare’s medical necessity criteria for clinical genetic testing for patients with cancer needs to be based on risk to benefit ratio. By including patients with cancer, testing provides clinically actionable results to many more patients (to be discussed), while the risk is primarily the cost of genetic testing, which has fallen precipitously in recent years, and continues to decrease. On the other hand, the risks of restrictive testing criteria have a much greater negative impact on health care than might at first appear to be the case if one simply focuses on the higher percentage of patients with pathogenic germline variants (if any) among the group that meet testing criteria versus those who do not. This is because the number of patients not meeting testing criteria is greater than those who do and therefore the total number of patients with pathogenic germline variants is also greater among the group that does not meet criteria.

• For example, 10% of all prostate cancer patients first present with castration-resistant metastatic prostate cancer.⁴⁴
• If only those patients, in whom the diagnostic yield for homologous recombination repair gene defects is ~11%,³² are tested and the remaining 90% without castration-resistant metastatic prostate cancer, in whom there is a ~5% diagnostic yield,³² are not, ~80% of all patients with pathogenic HRR defects would be missed.

Medicare’s requirement #1, that there be a “clinical indication for germline (inherited) testing”, will now be addressed. The primary clinical indication for germline testing is that it identifies cancer patients for whom knowing a molecular diagnosis may be clinically actionable. The recent prospective, multisite INTERCEPT study (INTERrogating Cancer Etiology using Proactive genetic Testing) enrolled patients with solid cancers.⁴ Patients were unselected for cancer type, stage of disease, family history of cancer, and demographic characteristics. Among the 149 patients with a high-penetrance pathogenic germline variant, 42 (28.2%) had documented modifications to their treatment and management because of the findings.⁴

There are two main drivers of actionability. First, many HCPS put individuals at risk for multiple cancers. For example, making a diagnosis of Lynch syndrome in a female colorectal cancer patient is a clear indication for the need for increased surveillance and/or surgical prophylaxis for uterine and ovarian cancer. Second, germline pathogenic variants make a patient eligible for precision therapy, such as PARP inhibitor therapy. Ready examples come from breast cancer and colorectal cancer. In several large cohorts of breast cancer patients, 83.3% to 91.5% of patients with a pathogenic variant had findings that could change medical management.^6,9,11-12 Clinically actionable findings in colorectal cancer (CRC) ranged from 10.7% to 12.7% of patients,^23,25 but has also reached nearly 95% of patients in one study.²⁴ This extends to other gastrointestinal cancers as well. In a study of pancreatic cancer patients, 50% of patients with pathogenic germline variants were eligible for targeted therapy.²⁹ Studies in other gastrointestinal cancers have also demonstrated that clinical management of patients changed following the return of a pathogenic germline variant from germline genetic testing.³⁰ This finding also holds for urogenital cancers. In a recent presentation at the Annual Meeting of the American Society of Clinical Oncology, Esplin and colleagues provided an interim report on the first 615 prostate cancer patients who underwent genetic testing unrestricted by stage, family history, tumor grade, 29% and 71% of patients with any pathogenic germline variant had the variant in a gene conferring potential eligibility for precision therapies or clinical treatment trial enrollment, respectively. Overall, 77% of the patients with any pathogenic germline variant had the variant in a cancer predisposition gene that was associated with clinical management recommendations, eligibility for precision therapies and/or enrollment in clinical trials.³⁶ Among other urogenital cancers, from 10% to 18.8% of patients had pathogenic germline variants in clinically actionable cancer predisposition genes.^39-41

The rationale for maintaining guidelines to determine which cancer patients should undergo germline testing is becoming obsolete, even as evidence is accumulating that they have a particularly poor predictive value for underrepresented populations.³⁶ The body of evidence demonstrating the utility of universal germline genetic testing has already changed guidelines to recommend testing to all diagnosed patients with pancreatic and ovarian cancer. Prospective studies have now firmly shown similar rates of pathogenic germline variants in breast cancer patients chosen without regard to legacy guidelines. This evidence led the American Society of Breast Surgeons (ASBrS) to recommend genetic testing for all patients.^12,45 Health insurance policies have also begun to change, with United Healthcare releasing a policy in January 2020 stating that germline multigene panel testing is proven and medically necessary for all patients diagnosed with colorectal, endometrial, gastric, ovarian, pancreatic, ureteral and renal pelvis, brain, biliary tract, and small intestinal cancers (namely, those associated with Lynch syndrome).⁴⁶

Major health systems have also begun to initiate programs that open germline genetic testing to all cancer patients. Beginning in 2019, Intermountain Healthcare began offering germline testing to all newly diagnosed cancer patients.⁴⁷ More recently, based on the findings from the INTERCEPT study,⁴ the Mayo Clinic has formed the Project TARGET task force to develop strategies for implementing universal germline genetic testing to their patients.

In summary, guidelines that restrict germline testing for HCPS only to patients meeting certain criteria such as positive family history and early age at onset are demonstrably insensitive and miss the majority of cancer patients who carry pathogenic variants in the tested HCPS genes. These are germline variants that are actionable for the management and treatment of a cancer patient who has them. Together with cost analyses suggesting that universal germline genetic testing should be considered,^48-50 the potential benefit of broadening testing is already changing guidelines, health insurance policies, and implementation within health systems. For patients who would benefit from testing but fall outside coverage criteria and are not covered by their insurance, financial barriers related to out-of-pocket costs could prohibit testing. In fact, clinicians and patients alike cite cost- and insurance-related issues as barriers to genetic testing for hereditary cancer.^51-57 Restrictive guidelines indicating whom to test and whom not to test prevent many out of criteria patients from having the benefits of a molecular diagnosis and deprive them of learning of their actionable findings. We recommend that universal germline testing for pathogenic variants in HCPS be made available to all cancer patients.

Thank you for the opportunity to submit these comments for your consideration.

One table and references were provided for review.

Ambry Genetics, A Konica Minolta Company, is writing to provide public comments on MolDX’s proposed coverage policy for next generation sequencing LDTs for hereditary cancer. We appreciate the opportunity to review the draft LCD and comment.

We appreciate the consideration and rationale put forth into the proposed LCD with the intent to positively impact patient care. After Ambry’s review of the draft LCD as written, we ask that to consider and respond to the following comments:

• MolDX should accelerate updates to its LCDs for HBOC testing and Lynch syndrome testing, as they contain outdated and unnecessarily complex language. The primary point of this update is that gene panel tests are widely recognized as acceptable and even preferable for initial testing.
• While MolDX does not want to quote excessively from the NCD, it would be worth noting clearly the difference between "clinical indication" and "risk factor" as defined in the NCD. Without knowing the specific CMS special definitions of these terms, it is hard to read and understand the NCD. A risk factor is a variable associated with increased risk, such as age or family history. A risk factor is not a sign or symptom. A clinical indication is a sign, symptom, or test, that leads to a procedure or treatment. Clarity would be warranted.
  • From these CMS definitions, it is not clear that a sign or symptom other than the personal history of cancer is required, although the different bullets in the LCD might indicate that.
  • From the CMS definitions, either "age" or "family history" could be a risk factor. Family history is NOT required by the CMS definition but is defined as one of several ways to meet the risk factor criterion.
• The main purpose of LCDs is to inform physicians and patients what are covered Medicare services, so they can select diagnostics and treatments that are covered. The lack of any specifics in this LCD make that purpose difficult.
• Historically, MACs including MolDX MACs have referred to current NCCN guidelines. The NGS MAC LCD for HBOC, for example, only points to current NCCN guidelines for HBOC testing. This would be an appropriate baseline for MolDX.
• We believe that MolDX has taken on an impossible and even unnecessary task to continually review all genes for all cancers in separate guidelines.
• For coverage, providers must complete a TA and submit complete TA materials by the original effective date of the policy or coverage will be denied. However, MolDX TA Form Clinical Validation Worksheet, there is a gene set with line 41 that states “Required Genes”. Does that mean ANY panel that does not have that gene set would not be approved, and thus not covered?
  • Would that gene set be required on the effective date of the LCD, or would there be any retroactive denial?
  • How often and with what oversight would/could this gene set be modified? There seems to be no oversight on what genes could be added and/or removed, thus making current panels fall in/out of LCD scope of coverage.
  • The resulting spreadsheets will be voluminous and hard to update. MolDX already finds it very difficult to keep up with deadlines and this extra task will be a further overload.
  • How would changes be communicated?
  • As MolDX acknowledges, labs generally prefer to run uniform large panels and cannot fine tune each panel to a patient's history. Similarly, it is not necessary to try to fine tune large spreadsheets for every possible gene combination and patient history.
• New technologies like RNA technology are acknowledged in the LCD. These are important for reducing VUS and for identifying occult mutations in non-coding DNA that result in large and pathogenic disruptions in transcriptions such as exon skipping or truncations. Since these detect as many pathogenic variants as Dup/Del testing, RNA testing is medically reasonable and necessary.

As a national clinical laboratory, Ambry Genetics, a Konica Minolta Company, excels at translating scientific research into clinically actionable test results based upon a deep understanding of the human genome and the biology behind genetic disease. Our unparalleled track record of discoveries over 20 years, and growing database that continues to expand in collaboration with academic, health plan/government and pharmaceutical partners, means we are first to market with innovative products and comprehensive analysis that enable clinicians to confidently inform patient health decisions.

Please accept the comments of the American Clinical Laboratory Association on the above-referenced draft local coverage determination (Draft LCD) and its associated draft article (Draft Article) and technical assessment (TA) worksheet. We continue to believe that the MolDX program and Medicare beneficiaries would be better served if the Draft LCD was rescinded in its entirety, with a new LCD proposed in its place that addresses issues with the current draft. Nevertheless, should the MolDX program proceed with the current draft, we offer our comments and constructive suggestions on it.

A. Overlap with current effective LCDs for hereditary cancers causes confusion and uncertainty for stakeholders.

The Draft LCD makes no mention whatsoever of currently-effective MolDX LCDs for hereditary cancer testing: BRCA1 and BRCA2 Genetic Testing, Genetic Testing for Lynch Syndrome, and APC and MUTYH Gene Testing. This creates considerable confusion about whether the existing MolDX LCDs would be replaced or remain in effect and, if they remain in effect, how to reconcile policy differences between the Draft LCD and existing LCDs.

If a laboratory performs BRCA1 and BRCA2 testing by NGS, is the test subject to the criteria in existing LCD or is it subject to the Draft LCD? The “minimum genetic content” set forth in the TA worksheet exceeds BRCA1 and BRCA2 for breast, ovarian, pancreatic, and prostate cancers. The Draft LCD effectively would exclude from coverage an assay that includes only BRCA1 and BRCA2, which is contrary to current MolDX coverage policy.¹ Also, the Draft Article includes ICD-10 code Z80.3 (familial history of malignant neoplasm of breast). It is not clear whether this indicates that when the ICD-10 code is included on a claim for an NGS test represented by CPT code 81432 (genomic sequence analysis panel for hereditary breast cancer-related disorders), the claim would be covered. Under MolDX: BRCA1 and BRCA2 Genetic Testing, BRCA1 and BRCA1 testing is non-covered when an individual has no personal history of cancer.

Similarly, if a laboratory performs Lynch syndrome testing using NGS methodology, is the test subject to the criteria in existing LCD or is it subject to the Draft LCD? MolDX: Genetic Testing for Lynch Syndrome provides coverage for Lynch syndrome-associated genes (MLH1, MSH2, MSH6, PMS2) in patients meeting the LCD’s criteria, either as individual gene tests as indicated by tumor results or as a multi-gene panel.² Also, the billing article associated with LCD includes codes for individual genes. Yet the Draft Article does not include individual genes, and the TA worksheet lists a minimum of 12 genes for colon cancer testing, rather than the four specified in a covered NGS panel in existing LCD.

MolDX: APC and MUTYH Gene Testing covers APC and MUTYH gene testing for individuals suspected to have familial adenomatous polyposis (FAP), attenuated FAP (AFAP), or MYH-associated polyposis, who have a personal history of 20 or more adenomas over a lifetime. The policy is method-agnostic and covers APC- and MYH-specific CPT codes, as well as certain genomic sequencing panels. But the Draft Article does not include individual gene CPT codes, indicating that the Draft LCD would not cover what is covered under LCD for patients with colon cancer.

The existing LCDs described above specify the risk factors for inherited cancer that make a patient eligible for specific tests. While the Draft LCD requires that a patient has “a risk factor for germline (inherited) cancer” to be eligible for coverage, specific risk factors are not delineated. The Draft LCD goes on to state within the Summary of Evidence that the National Comprehensive Cancer Network (NCCN) has defined such criteria for a variety of inherited cancers. Further direction should be provided within the Criteria for Coverage section that will allow a health care provider to know if a particular patient will be eligible for testing.

The interplay between the Draft LCD and other MolDX LCDs currently in effect is a fundamental question that must be addressed before stakeholders can comment meaningfully on aspects of the Draft LCD. Without information about the disposition of the existing LCDs for hereditary cancer testing, it is impossible to know the scope of the Draft LCD or for stakeholders to comment on it fully. As you know, existing coverage policies cannot be materially changed without notice and public consultation, so finalization of the Draft LCD would result in the MolDX program having conflicting policies until these issues are resolved.

B. The current process for developing and communicating “minimum genetic content” lacks much-needed transparency and stakeholder input.

The MolDX program has provided no commentary or explanation of the TA worksheet, other than it is an “associated document” for the Draft LCD that could be altered at will. The sources of the “minimum genetic content” for NGS testing for various cancer types is known only to the MolDX program. It is essential that the rationale for including genes or genetic variants in coverage of testing is transparent to stakeholders and adequately explained.

If used as proposed, the TA worksheet effectively would allow the MolDX program to alter coverage for NGS testing for inherited cancers unilaterally without any notice to or comment from stakeholders outside of the MolDX program’s subject matter experts. Under “Criteria for Coverage,” the Draft LCD states that in order for a test to be eligible for coverage, “The assay performed includes at least the minimum genetic content (genes or genetic variants) required for clinical decision making for its intended use that can be reasonably detected by the test. Because these genes and variants will change as the literature and drug indications evolve, they are listed separately in associated documents such as the MolDX TA forms.”

We are very concerned about the lack of procedural safeguards to ensure that stakeholders may provide input on and have ample advance notice of changes to the “minimum genetic content” that would spell coverage or non-coverage of a given NGS test for inherited cancers. The TA worksheet includes a notation about what “version” it is, but there is not a “history” indicating the changes that were made on a version. If a laboratory starts a TA worksheet in February and MolDX updates the “minimum genetic content” in April, there is no way for the laboratory to know what was changed. If the “minimum genetic content” information remains in a TA worksheet, it should include a history that would allow an analysis of what changes were made and when.

For other MolDX LCDs in which a list of “minimum genetic content” is located in a separate TA document, the MolDX program, via the associated Response to Comment articles has made vague promises about working with stakeholders to determine how much advance notice will be necessary to minimize disruptions to lab operations, yet it has no articulated process for receiving feedback on “minimum genetic content” or for apprising stakeholders of changes. We fear the same result with changes to “minimum genetic content” called for in the Draft LCD.

C. MolDX must develop a rational, evidence-based method for determining a test’s “minimum genetic content”.

The Draft LCD includes references to the Clinical Genome Resource (ClinGen), NCCN, American Society of Clinical Oncology (ASCO), American College of Medical Genetics (ACMG), and Association for Molecular Pathology (AMP) as experts that determine what gene content is relevant for a test’s indicated use. Yet there is no consensus among these experts about a list of genes that always should be tested for a specific hereditary cancer type, and these experts do not necessarily agree with the MolDX program about “minimum genetic content” for certain hereditary cancers. Further, each of these organizations have their own limitations as expert sources, in terms of the types of hereditary cancer for which they make recommendations, the strength and clarity of their relevance classifications for genes, and how often they update their recommendations.

If the MolDX program intends to publish a final version of the current Draft LCD that refers to a TA worksheet with lists of “minimum genetic content” for different hereditary cancers, the lists must be based on reputable third-party sources, including but not limited to professional societies and peer-reviewed publications, that have a well-understood approach for their recommendations. Respectfully, the MolDX program should not be the final arbiter of gene-disease validity. It must be clear on which third-party source’s recommendation the “minimum genetic content” for each hereditary cancer is based; it is not sufficient that a document includes a list of third-party sources without stating which “minimum genetic content” was recommended by which specific source. Also, it is essential that a TA worksheet that includes “minimum genetic content” is associated with/linked from an LCD and/or an article associated with the LCD, so that any stakeholder can find and access it easily.

As a possible approach, when ClinGen has an expert curation panel for a hereditary cancer type (e.g., HBOC, colon cancer), the Draft LCD’s “minimum genetic content” should be based on genes classified by the organization as definitive, as ClinGen’s approach to scoring gene associations with specific phenotypes is well-defined. For other cancer types, if it truly is necessary to provide a list of “minimum genetic content,” other sources could be used, as long as the “minimum genetic content” is attributed to a specific source. And the MolDX program should make any such lists available for stakeholder comment before finalizing them.

D. Conclusion

Thank you for your consideration of our comments. We look forward to continuing our collaborative working relationship with the MolDX program.

References were provided for review.