Response to Comments: MolDX: DecisionDx-Melanoma

DecisionDx-Melanoma® test (Castle Biosciences Inc., Friendswood, TX), is a gene expression profile test that is a signature 31 genes, 28 discriminating genes and 3 control genes. The test is used to measure risk of metastasis in patients with stage I or II cutaneous melanoma and classifies tumors into two groups of risk or metastasis. The test is purported to give an independent prediction of risk of tumor metastases, independent of currently used metrics of risk assessment such as Breslow’s thickness, ulceration status (present or absent), mitotic rate, American Joint Committee on Cancer (AJCC) stage, and sentinel lymph node biopsy status. It is suggested this information would add to a comprehensive baseline evaluation and determination of the initial surveillance and treatment of individual with high risk stage I or II disease than they would have otherwise received.

• Berger et. al. (2016) published a retrospective study of 156 consecutive patients from six institutions who had cutaneous melanoma and were evaluated with DecisionDX-Melanoma test. This study used chart review to describe changes in management, and examined whether management changes were associated with DecisionDX-Melanoma results. The frequency of clinic visits, imaging tests, referrals, and blood work was measured before and after results of DecisionDX-Melanoma. For patients with class 1 results, it showed reduced utilization in 40 of 42 patients; for patients with class 2 results, it showed increased utilization for 74 of 79. The difference in management changes by test class was statistically significant (p<0.001). However, follow-up data was not collected for this patient cohort. This study is limited regarding the impact of gene expression profiling based on management changes , healthcare resource utilization and patient outcomes. In summary, the clinical utility has not been substantiated. Future studies would benefit from the collection of follow-up data to show the impact of clinical practice adjustments on patient outcomes. Berger AC, Davidson RS, Poitras JK, et. al. Clinical impact of a 31 gene expression profile test for cutaneous melanoma in 156 prospectively and consecutively tested patients. Curr Med Res Opin. Jun 3 2016:1-6 PMID 27210115
• The clinical validity of the DecisionDx-Melanoma test was evaluated in a prospective, multicenter study of class 1 cutaneous melanoma tumors which analyzed microarray expression data to identify a prognostic 28-gene signature to predict risk of metastasis (Gerami, 2015). Based on modeling analysis of cohorts of primary cutaneous melanoma tumor tissue and Kaplan-Meier analysis, the study reported the 5-year disease-free survival (DFS) rates in the development set were 100% and 38% for predicted classes 1 and 2 tumors, respectively (p<0.0001). Disease free survival rates for the validation set were 97% and 31% for predicted classes 1 and 2 tumors, respectively (p<0.0001). The investigators suggested their preliminary analysis indicates the 28-gene signature is an independent predictor of metastasis risk in the studied cohort of cutaneous melanoma tumors. Gerami P, Cook RW, Wilkinson J, et. al. Development of a prognostic genetic signature to predict the metastatic risk associated with cutaneous melanoma. Clin Cancer Res. Jan 1 2015;21(1):175-183. PMID 25564571
• Gerami and colleagues (2015) assessed the prognostic accuracy of gene expression profiling for molecular staging of cutaneous melanoma in a multicenter cohort study of 217 individuals undergoing sentinel lymph node biopsy (SLNB). The prognostic accuracy of each test was determined using Kaplan-Meier and Cox regression analysis of disease-free, distant metastasis-free, and overall survival. For individuals with a negative SLNB and a class-2 gene expression profile signature (that is, a high risk outcome), Kaplan-Meier 5-year disease-free, distant metastasis-free, and overall survival rates were 35%, 49%, and 59%, respectively; however, there was no statistical difference in disease-free survival, or overall survival rates for individuals with class-2 gene expression profile signature and a negative SLNB result and individuals with a class-2 gene expression profile score and a positive SLNB result.
• A limitation of the Gerami study is the lack of data obtained from a randomized sample of cases, which the authors conclude as resulting "in a higher rate of distant metastasis than commonly observed or reported in the SLNB-negative group." Additional study is needed within a randomized sample of individuals to determine how gene expression profiling combined with SLNB would contribute to the accurate staging and treatment planning of individuals with cutaneous melanoma. Gerami P, Cook RW, Russell MC, et. al. Gene expression profiling for molecular staging of cutaneous melanoma in patients undergoing sentinel lymph node biopsy. J Am Acad Dermatol. May 2015;72(5):780-785 e783. PMID 25748297

In summary, based on review of the peer reviewed medical literature, there is insufficient evidence to evaluate the clinical validity and clinical utility of gene expression profile (GEP) testing (Decision Dx-Melanoma). Additional studies are needed to further validate if gene expression profiling of cutaneous melanoma will accurately identify which individuals are likely to have aggressive disease and how the results of this testing would alter treatment plans and improve health outcomes in the surveillance and treatment of high-risk cutaneous melanoma.

Moreover, NCCN guidelines Version 2.2018 for Melanoma state gene expression profiling is not recommended outside of clinical trials. In conclusion, the evidence is insufficient to determine the effects of this testing on net health outcomes and therefore Decision DX -Melanoma is considered investigational.

The commenter in summary suggests that the evidence is not sufficient to support coverage of the DecisionDx test. MolDX disagrees with this assessment. We agree that there is a limited amount of evidence available, but the available evidence is consistent. We agree that more evidence and data from randomized trials would further strengthen our knowledge about this test, but presently we believe that the evidence is sufficient for Medicare coverage. We will continue to monitor the evidence and reassess this coverage decision if relevant new evidence emerges.

We have also reviewed the NCCN guidelines, which do not recommend the use of genomic testing outside of a clinical trial. However, the NCCN guideline on melanoma treatment does not appear to review any particular genetic test in detail for consideration of whether or not any particular test does or does not have a role in treatment planning. Rather, the guideline considers the issue of genetic testing in general (reviewing evidence on the DecisionDX Melanoma test as one of a number of examples of available tests) and concludes that there is substantial heterogeneity among not only studies but also the particular genetic tests for melanoma. The guideline does not review the full evidence regarding any particular test. It is the opinion of MolDX that this guideline highlights the need to carefully consider each individual genetic test based on its own merits, as we have done with the case of the DecisionDX test.

The advancement of treatment for patients with melanoma is an important mission for the division of Dermatology and Dermatologic Surgery at MUSC. The policy you posted for DecisionDx is an important step forward in managing patients with cutaneous melanoma.

I use this test to guide therapy in many patients and having it available to use in my Medicare patients is quite helpful. I fully support the use of this test as identified in your policy. Its clinical usefulness and success has been documented in peer reviewed literature.

One thing that I also find it useful for is the transected deep margin diagnostic biopsy+ melanoma –even if 1a as it may prevent exhaustive work ups (imaging or SLNB) if the gene expression is favorable. Adding this group of melanoma patients to this policy would provide a more comprehensive coverage for appropriate patients who may benefit clinically from this test.

I appreciate your time in considering my comments to this policy and please contact me should you have any questions regarding my use of this test.

We agree in part with the comment. We will be expanding coverage to include a small subset of 1a melanomas. We will expand coverage to include T1a melanomas when there is significant uncertainty about the adequacy of microstaging or where there are adverse features since sentinel lymph node biopsy would also be considered in these patients as well.

By way of background, I am a Clinical Professor of Dermatology at New York University, a former President of the American Academy of Dermatology, American Society of Dermatologic Surgery and the American Dermatological Association. My practice is limited to skin cancer with a focus on melanoma. I have 30+ years experience with Melanoma with over 200 publications in this area and am the lead author on the major textbook in the field, Cancer of the Skin.

I am writing this letter today to convey to you my strong support for the use of the DecisionDx-Melanoma test as an adjunctive tool in assessing prognosis for patients with cutaneous malignant melanoma. I wholeheartedly support the utilization of this test in patients with Stage I and Stage II melanoma, and it is a regular part of my clinical management strategy in this population. Unfortunately, the clinical features that figure prominently in the American Joint Committee on Cancer (AJCC) melanoma staging system do not often supply the degree of granularity required to accurately and precisely stratify patients based on prognosis, as evidenced by the fact that approximately two-thirds of melanoma-related deaths occur in patients initially diagnosed with Stage I or Stage II disease.1 This fact underscores the need for a complementary and synergistic method of further risk-stratifying patients, as prognostic information features prominently in subsequent disease management and has important bearing on patient outcomes and overall healthcare costs.

My familiarity with the literature surrounding the DecisionDx test and my personal clinical experience with the assay has shown me that this test has all of the attributes necessary to help fill this void. First and foremost, this gene profiling system has been shown in multiple independent studies to accurately and efficiently categorize patients based on risk for subsequent metastasis and local recurrence.2,3 Furthermore, the test has been shown to work synergistically with the existing AJCC guidelines, with identification of individuals with high-risk tumors improved when the two methods are combined.4 However, a test is only useful clinically if it has an appropriate impact on management, and studies have demonstrated that the DecisionDx assay achieves this. Studies that I have been involved with, as well as others, have demonstrated that the results of this test correctly alters the direction and intensity of clinical management in roughly half of cases.5-7 The net result is that the test efficiently differentiates low-risk patients from high-risk patients, allowing more invasive and costly follow-up procedures to be reserved for the high-risk patients who will benefit the most. My own experience is that this leads to more efficient and cost-effective care of melanoma patients.

For these reasons, I wholeheartedly support the use of the DecisionDx test in the clinical management of cutaneous melanoma, and I hope you will develop a local coverage policy for this important technology that will benefit patients.

Thank you for the comment.

This letter is in support of your draft LCD for DecisionDx-Melanoma. This policy is an important advancement for patients with melanoma. As I discussed in an article published January 22, 2018 in Dermatology News there are significant patient benefits to utilizing a genetic test prior to recommending sentinel lymph node biopsy (SLNB) for patients with early stage melanoma. In these patients the benefits of performing a sentinel lymph node biopsy do not outweigh the complication rates nor justify the cost of this procedure. On behalf my Medicare patients I thank you for developing this draft policy.

Nearly 85% of all patients who undergo a SLNB have a negative result, which means most patient undergo surgical and anesthesia risk without benefit, and with the possibility of developing adverse effects from the procedure such as wound infections and lymphedema. Recent evidence has shown that this procedure with or without completion lymph node dissection, does not improve survival (Morton et al. NEJM 2014; Faries et al. NEJM 2017). I believe we can do better to reduce the number of patients subjected to this invasive procedure and I do believe this policy achieves this goal without putting patients at harm.

The DecisionDx-Melanoma test has been shown to have near identical sensitivity and specificity for developing metastatic disease as a sentinel node and has been validated extensively in the literature in both retrospective and prospective studies (Gerami et al. Clin Cancer Res 2015; Gerami et al. JAAD 2015, Hsueh et al. J Hematol Oncol 2017; and Zager et al. BMC Cancer 2018). Furthermore, Dr. John Vetto, a surgical oncologist from Oregon Health Sciences University Medical Center, shared with us the study results supporting the use of the test for SLNB guidance at the American Academy of Dermatology in February of this year. This evidence provides me and my colleagues confidence in the results of this test and it is why we use it to guide patient management today.

I thank you in advance for consideration of my comments to this policy and please feel free to contact me should you have any questions.

Thank you for the comment.

Thank you for developing and posting the draft LCD policy for DecisionDx-Melanoma (DL37725: MolDX: DecisionDxMelanoma) and providing access to this test for Medicare beneficiaries with cutaneous melanoma. After reviewing the policy, we would like to comment on a few items in the LCD that we request you consider modifying. We believe these changes may result in an LCD that will ensure the best care for Medicare beneficiaries and efficient processing of claims. Our comments are based both on our experience with performing this test plus input from the medical specialists who manage Medicare beneficiaries with cutaneous melanoma. Following are sections of the policy with our suggested modifications.

1) Coverage indications, limitations, and/or Medical Necessity (page 2)

• Text in draft LCD:

“The DecisionDx-Melanoma test will be covered only when the following clinical conditions are met:

Patients diagnosed with pathologic stage sentinel lymph node biopsy (SLNB) eligible T1b and T2 cutaneous melanoma tumors (as defined in AJCC Staging Manual v8, 2017) with clinically negative sentinel node basins who are being considered for SLNB to determine eligibility for adjuvant therapy. (Per current NCCN and ASCO guidelines, SLNB eligible patients are defined as:

• Patients with T1b tumors (≥ 0.8 mm or < 0.8 mm with ulceration)
• Patients with T2 tumors”

Physicians who routinely care for cutaneous melanoma patients have indicated that a substantial proportion of their SLNB eligible patients are pathologic stage T1a, as per current clinical guidelines. Patients with T1a tumors and specific adverse features are considered SLNB eligible under current guidelines. NCCN v2.2018 guidelines include consideration of sentinel lymph node biopsy for T1a tumors in presence of:

• Uncertain microstaging (generally due to uncertain Breslow Thickness from a biopsy with transected base/positive deep margins), OR
• Other adverse features (high mitotic rate, lymphovascular invasion, etc)

We analyzed data from the first 2,753 Medicare beneficiaries that were clinically tested with DecisionDx-Melanoma:

• Of these, 1,168 were patients with a T1a tumor (Breslow thickness <0.8mm without ulceration)
• Of these, 31% would have been considered for SLN biopsy due to either uncertain Breslow thickness or high mitotic rate

In addition, analysis of SEER data (2017 release) shows that clinicians are, in fact, performing SLN biopsies in these T1a patients. Specifically, SEER data indicates that 8.7% of T1a patients do undergo a SLN biopsy procedure and these represent 22% of SLNBs in T1-T2 patients ≥65 years of age.

Although thin, non-ulcerated tumors inherently have a low likelihood of SLN positivity, patients with uncertain microstaging and/or other adverse features can be considered under current guidelines. Enabling coverage of DecisionDx-Melanoma for T1a patients with adverse features can help reduce unnecessary SLN biopsy procedures in this patient population and is consistent with the intent of the LCD, avoiding surgical complications which are not insignificant in the Medicare population, and has significant health system cost savings.

We suggest the addition of a select T1a population as per NCCN guidelines v2.2018:

• T1a with Breslow depth <0.8 mm and uncertain microstaging or with other adverse features (e.g. Very high mitotic index (≥2/mm² [particularly in the setting of young age], lymphovascular invasion, or a combination of these factors)

Suggested wording (updated text in bold):

“The DecisionDx-Melanoma test will be covered only when the following clinical conditions are met:

• Patients diagnosed with pathologic stage sentinel lymph node biopsy (SLNB) eligible T1 and T2 cutaneous melanoma tumors (as defined in AJCC Staging Manual v8, 2017) with clinically negative sentinel node basins who are being considered for SLNB to determine eligibility for adjuvant therapy. Per current NCCN and/or ASCO guidelines, SLNB eligible patients are defined as:
  • Patients with T1a tumors:
    • in whom there is significant uncertainty about the adequacy of microstaging (positive deep margin), or
    • with Breslow depth <0.8 mm and with other adverse features (eg. very high mitotic index [≥2/mm2], lymphovascular invasion, or a combination of these factors)
• Patients with T1b tumors (≥ 0.8 mm or < 0.8 mm with ulceration)
• Patients with T2 tumors”

2) Summary of Evidence (pages 2-3) and Clinical Performance: Utility (Page 4)

1. These sections currently contain data from the fall of 2017 (n=481) when initial clinical performance data for SLNB guidance was provided for MolDX review. There is updated data on an expanded, prospective, multicenter cohort totaling 1,421 patients, which was presented at the 2018 Meeting of the American Academy of Dermatology.
2. For this reason, we recommend updating the “Summary of Evidence“ section paragraphs with the expanded dataset as follows (updated text in bold):
• “In general, a 5% likelihood for a positive SLN is recommended as a threshold for performing this procedure in a patient population [18]. Analysis of Castle’s retrospective cohort (n=782) identified that DecisionDx-Melanoma Class 1 patients (probability score <0.5) with tumors who had a Breslow thickness (BT) of ≤2.0 mm (AJCC T1-T2) achieved a positive SLNB rate below 5% [26]. To validate clinical use of the Class 1A cut-point for the DecisionDx-Melanoma test, two contemporary, multi-center, prospective study cohorts were tested: a 584 patient cohort from two published prospective studies (overall 14% SLN positive rate) [6, 27] and a 837 patient cohort from prospectively tested patients at 5 large academic institutions (overall 12% SLN positive rate) [26]. The rate of SLN positivity in both prospective study cohorts aligns with the SLN positivity rate in the general population of melanoma patients who have undergone SLNB. The results show that in patients from the Medicare-eligible population (65 years old and over) who had a SLNB performed, those patients who had a DecisionDx-Melanoma Class 1A, T1 or T2 tumor, have a 98.4% probability of being SLN negative. Thus, in this patient group, the DecisionDx-Melanoma test achieves a NPV that is comparable to other molecular tests used to guide surgical management decisions [26, 29-32]
• Analysis of the two prospective cohorts combined (n=1,421 patients) shows that SLN positivity rates are enriched from 12.8% using current SLNB criteria to 18.6% in those patients who would undergo the SLN biopsy procedure (T1/T2 with a Class 1B, 2A or 2B DecisionDx-Melanoma test result and T3/T4 patients). The rate of SLN positivity in the 65 years and older population from Castle’s prospective cohort (n=629) is 1.6% of Class 1A patients and 11.9% for Class 2B patients. SLNB positivity rates for T3 tumors is 8.7%% for Class 1A compared to 18% for Class 1B-2B; for T4 tumors it is 17% and 17% respectively. Importantly, the 5-year melanoma specific survival (MSS) rate for T1/T2 Class 1 patients remains favorable; with 99% MSS, comparable to that observed in T1a tumors and for which current guidelines do not recommend SLNB [7, 26, 33]. Furthermore, T1/T2 Class 1 patients show 5-year overall survival (OS) of 97% and distant metastasis free survival (DMFS) of 93% [26]. The MSLT-II study demonstrated that a delay in lymph node dissection does not adversely affect survival, thus clinical follow up of Class 1 patients and lymphadenectomy for those few who develop clinically detectable nodal disease should achieve similar outcomes to those who currently undergo a planned SLNB [9]. Thus, the test identifies a patient population with <5% likelihood of a positive SLN and high survival rates and therefore has utility in guiding SLNB decisions in patients 65 years-old and over with T1-T2 CM tumors. In this population, the test could potentially reduce the rate of SLNB by up to 78% while still maintaining a MSS survival rate of 99% in those patients with low-risk tumor biology who can safely avoid the procedure.”
3. This large cohort demonstrating that the DecisionDx-Melanoma is able to identify a population with <5% probability of a positive SLN provides strong support to this LCD. We recommend updating the tables in the “Clinical Performance: Utility” section with the expanded dataset as follows:

Summary of DecisionDx-Melanoma performance in SLNB guidance in T1/T2 melanoma patients ≥65 years old, eligible for SLNB biopsy^a

We will modify the existing draft LCD to expand coverage to include T1a patients with uncertain microstaging or with adverse features. We will also include the relevant updated evidence.

The DecisionDx-Melanoma test is a 31-Genes Expression Profile (GEP) that determines the risk for metastatic disease in Cutaneous Melanoma (CM) patients. The primary scope of the test, as documented in Castle Biosciences studies on the performance of the GEP signature, is purely prognostic, namely to classify the patients as Class 1, having a tumor with low prognostic risk for developing metastasis within 5 years of diagnosis, or Class 2 having a high prognostic risk for developing metastasis within 5 years of diagnosis.

Since the patients with a Class 1 tumor profile also have a low likelihood of being Sentinel Lymph Node (SLN) positive, Castle Biosciences is proposing, in the draft LCD (DL37725), to use the DecisionDx-Melanoma test to guide the decision making for the Sentinel Lymph Node Biopsy (SLNB): more specifically, low risk Class I patients would forgo the SLNB, whereas high risk Class II would still have to endure such a procedure with the associated comorbidities. 

As elucidated in the section “Coverage Indications, Limitations, and/or Medical Necessity” of the proposed/draft LCD (DL37725), the intended use population of the DecisionDx-Melanoma test is defined by two criteria:

• Patients diagnosed with pathologic SLNB eligible T1b and T2 cutaneous melanoma tumors (as defined in AJCC Staging Manual v8, 2017) with clinically negative regional lymph node basins; basins who are being considered for SLNB to determine eligibility for adjuvant therapy.
• Patients 65 years old and over.

There is clearly a strong clinical need for a molecular test able to discriminate SLN-negative from SLN positive patients: such a test would spare the large majority of the SLN-negative patients and the healthcare system the burden and the costs of an SLNB procedure, while retaining its advantages in terms of staging, prognostics and patient/therapy management.

Unfortunately, the DecisionDx-Melanoma does not address such a clinical need, since the level of evidence provided is weak and contradictory, and its utility is seriously undermined by an erroneous and misleading interpretation of the results:

1. Several cohorts are presented as supportive evidence, but none represents exclusively the intended use population. The results from several clinical studies are discussed in the draft LCD (DL37725):
   • Three multi-center, prospectively designed, archival tissue studies for a total of n= 782 patients;
   • Two contemporary multi-center prospective study cohorts with n= 322 and n=159 patients respectively. Out of these 481 patients (=322+159), there are
     • n=190 patients 65 and older from stages T1,T2, T3,T4
     • n=134 patients 65 and older from stages T1,T2; of these 134 eligible patients only n=6 patients present with a positive sentinel lymph node biopsy (4.8 %)
   • An interim analysis of two prospective registries (n=322)
   • A prospective independent study with n=510 patients.

However, none of these cohorts comply with the definition of the intended use population: patients 65 years old and over and with T1b-T2 CM tumors. While the 134 patient cohort is “closest” to the intended use population, since it is restricted to patients 65 years-old and over, it also includes T1a melanomas that are not part of the intended use definition. Consequently, this cohort has a SLN positivity rate that is significantly lower (4.8%) than expected for the intended use population: T2 CM tumors have an expected SLNB positivity rate of 12% (Cancer 15.116.6 (2010): 1535- 44); T1b CM patients selected for SLNB have an expected SLNB positivity rate of 6- 8.5% (Journal of Clinical Oncology 31.35 (2013): 4387-4393; Melanoma Research 25.2 (2015):157-63). Because of the low event rate, the number of samples evaluated in the cohort is very low; in particular, the number of SLNB positive samples (6) is critically low. Therefore, there is not enough evidence to draw any meaningful conclusion, and to generalize the results.

2. The considered population is low risk (SLN positivity rate <5%). As reported in the section “Clinical Performance: utility” of the proposed/draft LCD (DL37725) and in the table here reproduced,

Summary of DecisionDX-Melanoma performance in SLNB guidance T1/T2 melanoma patients ≥65 years ol

Data from two contemporary, prospective, multicenter cohorts (N=481) [6, 27]

the SLN positivity rate is 4.8%, therefore, lower than the 5% threshold suggested in the literature for performing the SLNB procedure for melanomas ticker than 0.75 mm (Han, Dale, et al. "Clinicopathologic predictors of sentinel lymph node metastasis in thin melanoma." Journal of Clinical Oncology 31.35 (2013): 4387-4393). So, in principle, this entire population could possibly forgo the SLNB procedure, in virtue of its low risk.

3. The DecisionDx-Melanoma test cannot be used as rule-out for SLNB. In the section “Clinical Performance: utility” of the proposed/draft LCD (DL37725), the performance of the DecisionDx-Melanoma test are summarized for the cohort of 134 patients, 65 years old and over, with T1-T2 CM tumors. The reported negative predictive value (NPV) is 96%: this means that if the test outcome for a patient is negative (Class I), there is still a 4% probability that the SLN is positive. But the prior probability of SLN positivity, before taking the test, is 4.8%, so the “gain” of the test is to move down the risk of a positive SLN from 4.8% to 4%. For all practical purposes, this 0.8% difference is absolutely not meaningful, and would not be actionable (see point 2).
4. The DecisionDx-Melanoma test is a poor rule-in for the SLNB. In the section “Clinical Performance: utility” of the proposed/draft LCD (DL37725), the performance of the DecisionDx-Melanoma test are summarized for the cohort of 134 patients, 65 years old and over, with T1-T2 CM tumors. It is reported that the SLN positivity rate in the patients that are labeled as Class II by the DecisionDx-Melanoma test is 7.1%. So, after taking the DecisionDx-Melanoma test, the risk of a positive SLN moves up from 4.8% (prevalence of SLN positivity in the general population) to 7.1%, so above the 5% threshold suggested in the literature for performing the SLNB procedure. However, DecisionDx-Melanoma serves as a poor rule in test: after taking the test and getting a positive outcome (Class II), there is still a staggering 92.9% probability that the actual SLNB is negative. We can draw similar conclusions if we reason in terms of absolute numbers of patients. The DecisionDx-Melanoma test labels 26% of the patients as Class II and 74% as Class I (these percentages can be easily derived from a simple calculation, knowing that the SLN positivity rates in the overall 134 population, in Class I and in Class II are respectively 4.8%, 4%, 7.1%). So out of the 6 (=134*4.8%) patients with a positive SLN, the DecisionDx-Melanoma will rule in for SLNB procedure only 2 (=134*26%*7.1%) and miss the other 4 (this correspond to a very low sensitivity of 33%). Current expert guidelines suggest that absent contraindications (significant comorbidities affecting life expectancy, extreme frailty or unsuitability for anesthesia) melanoma patients age 65 and older should be offered sentinel lymph node biopsy based on the same criteria as younger patients, not chronologic age alone, as nodal status and treatment influences both disease-free and melanoma-specific survival (NCCN melanoma guidelines v2.2018; Surgery 150.4 (2011):828-35; Eur J Surg Oncol 41.1 (2015):157-64).
5. DecisionDx-Melanoma test cannot effectively guide the SLNB procedure. Castle Biosciences’ own research (Berger, Adam C., et al. "Clinical impact of a 31-gene expression profile test for cutaneous melanoma in 156 prospectively and consecutively tested patients." Current medical research and opinion 32:9 (2016): 1599-1604) has shown that the outcome of the DecisionDx-Melanoma test would not significantly affect the clinical decision making about the SLNB procedure, since the physicians tend to comply with the NCCN guidelines: “An important finding of the study was the minimal effect of the 31 GEP results upon the use of the SLN biopsy procedure, and that test results were impactful following SLNB.” In other words, the DecisionDx-Melanoma test was taken into account by the physicians only in combination with the results of the SLNB biopsy, and not as a guide to selectively perform SLN surgery.

Conclusions

1. Several cohorts are presented as supportive evidence, but none represents exclusively the intended use population.
2. The DecisionDX-Melanoma test is not well suited to select patients for the SLNB procedure, since it does not significantly change the predicted risk of a positive SLNB, as dictated by the prevalence of SLNB positivity in the considered population.
3. The DecisionDX-Melanoma test may rule out the SLNB procedure for a substantial number of patients who do harbor metastases (due to its low sensitivity). This may delay therapy and negatively impact overall survival.
4. Any CMS decision first and foremost needs to consider the needs of the patient. Although there is a clear clinical need for a test that can guide/support the SLNB decision making, coverage for a molecular testing that is of questionable clinical utility should be reconsidered.
5. Our view is supported by current NCCN guidelines which state that while “there is interest in newer prognostic molecular techniques such as gene expression profiling to differentiate melanomas at low versus high risk for metastasis, routine (baseline) prognostic genetic testing of primary cutaneous melanomas (before or following SLNB) is not recommended outside of a clinical study (trial)”.

We appreciate the very thoughtful comments and analysis of the data presented. We will address each of the points individually.

Our review of the data submitted indicates that the test is to stratify the risk of sentinel lymph node positivity in a sample that adequately represents the intended use population (which is being modified to include T1b positive tumors). While the samples with data presented in the LCD had relatively low risks of a positive lymph node, the risk of a positive sentinel lymph node was nearly double for those classified as class 1B/2B compared with those considered class 1A by the test.

With absolute risks of a positive sentinel lymph node of around 10%, we recognize that this test is not sufficient to positively diagnose the presence of a sentinel lymph node; rather we believe that the risk stratification may be helpful for clinicians and patients who are trying to decide whether or not to pursue a sentinel lymph node biopsy.

MolDX interprets the evidence and a test’s clinical utility within the context of existing treatment approaches and the existing diagnostic approaches. While improved accuracy would no doubt be desirable in a genetic test for melanoma, the information that this test does provide has the potential to allow clinicians and patients to be more informed in their decisions on whether or not to pursue sentinel lymph node biopsy than they typically are in the absence of a test such as this.

The AADA is committed to excellence in medical and surgical treatment of skin disease; advocating high standards in clinical practice, education, and research in dermatology and dermatopathology; and supporting and enhancing patient care to reduce the burden of disease.

On behalf of the nearly 13,000 U.S. based members of the American Academy of Dermatology Association (AADA), we are writing to share our support for proposed LCD DL37748 MolDX: DecisionDx-Melanoma. However, we have some concerns, listed below, that should be addressed before the final LCD is released.

This Local Coverage Determination (LCD) proposes to cover the DecisionDx-Melanoma test that provides prognostic information for patients with cutaneous melanoma (CM). DecisionDx-Melanoma is a robust, gene expression profile test that demonstrates strong reproducibility between experiments and has high technical reliability on clinical samples.

From March 2013 through June 2016, the gene expression test was performed on 8244 CM tumors. De-identified data from Pathology Reports were used to assess technical success. Robust sample and reagent stability was observed. Inter-assay concordance on 168 specimens run on 2 consecutive days was 99% and matched probability scores were significantly correlated (R2 = 0.96). Inter-instrument concordance was 95%, and probability scores had a correlation R2 of 0.99 (p < 0.001).

From 8244 CM specimens submitted since 2013, 85% (7023) fulfilled pre-specified tumor content parameters.

Using formalin-fixed paraffin-embedded primary tumor tissue, the RT-PCR-based test classifies patients into a low- (Class 1) or high-risk (Class 2) category for recurrence based on expression of 31 genes.

Inter-assay, inter-instrument, and inter-operator studies were performed to evaluate reliability of the 31-gene expression test results, sample stability and reagent stability.

In these samples with sufficient tumor requirements, the technical success of the test was 98%.

We are in agreement with the Analysis of Evidence and approval of the policy, with the following concerns:

1. Castle Biosciences recommends the DecisionDx-Melanoma test for Stage I and II melanomas, recognizing that the test results can be useful in these stages. The confusion may lie in two descriptive pathologic/clinical classifications for melanoma:
   • Stage I = T1a, T1b and T2a
   • Stage II = T2b, T3a, T3b, T4a, T4b

2. Will a sentinel lymph node biopsy be considered medically necessary in the face of a Class I test result in a T1b patient?
   • Standard of care is to recommend a sentinel lymph node biopsy for T2 tumors. There is no clear data guidance as to what to do for T2 tumors that are Class 1.

We request that the sentinel lymph node biopsy be a covered service for T2 tumors regardless of the Class1/Class 2 result.

3. Medicare should provide guidance stipulating medical record documentation requirements for the TNM melanoma stage e.g., T1b; T2a, etc. in order to support coverage and medical necessity. For example, the patient’s record should specify the T staging of the melanoma.
4. We agree with the LCD’s T1 and T2 coverage but question if there can be coverage allowance for T3 and T4 (pathological Stage II tumors).
   • From a clinical decision-making standpoint, DecisionDx yields two practical, clinically relevant results:
     • Whether a melanoma is low risk (Class 1) or high risk (Class 2) for metastasis;
     • Based upon the above result, whether a sentinel lymph node biopsy and/or closer surveillance for metastases should be done

We encourage JF Noridian Healthcare Solutions, LLC to consider this draft DL 37748. We look forward to further discussing this topic should additional information be requested or needed.

Thank you for the comment. This coverage policy strictly provides coverage criteria regarding the testing of a skin lesion using the DecisionDx-Melanoma test. This LCD does not provide any coverage restrictions on how the information obtained from the test should be used by the treating clinician, but there is an expectation that the information from the test will be used by the treating clinician and patient as one consideration possibly among many to make an individualized decision regarding sentinel lymph node biopsy.

We will be restricting coverage to T1 and T2 melanomas at this point based on the studies used in this coverage decision. If there is sufficient evidence to expand coverage to more advanced tumors, we would be willing to review the evidence and reconsider it.

I am writing to support and request consideration of a suggested addition to the draft local coverage determination for DecisionDx-Melanoma, which is a test that is used routinely in my practice for the care of patients with cutaneous melanoma.

This test is an important tool I use to evaluate and manage patients with cutaneous melanoma the development of this draft policy is appreciated. The coverage guidance and intended use section of the draft describes appropriately that the test results can be used to guide sentinel lymph node biopsy (SLNB) as recently reported at the American College of Mohs Surgery meeting. As you know, about 85% of patients who undergo a SLNB procedure are negative, and this is over 90% if we look at just T1 tumors, which are the majority of patients I see. By applying the test as per the LCD’s intended use, we’ll be able to reduce unnecessary procedures without affecting outcomes in these patients. This should also translate to important cost savings for the healtcare system.

One modification I’d like to you consider, is adding a group of patients that per guidelines are considered for sentinel lymph node biopsy, T1a patients with unknown microstaging or other high risk features. I believe leaving out this group would unnecessarily limit the benefit of this policy and exclude a group of patients that would benefit significantly from this approach.

Thank you for providing the draft LCD for Medicare patients with cutaneous melanoma and the opportunity to provide comments. As a physician on the front line of melanoma care, it is important that our experience is taken into account.

Thank you for the comment. We agree with the comment; given that some T1a melanomas are managed in the same way as T1b melanomas, we are modifying the LCD to expand coverage to include these melanomas.

I am writing to support the Draft LCD DL37748 for coverage of the DecisionDx-Melanoma test. I routinely use this test to manage my melanoma patients and find it a reliable prognostic marker that helps me determine a management program that is specific based on my patient’s risk. This is based on the evidence that has been published. Coverage for this test is an important decision that will improve care of Medicare patients with cutaneous melanoma.

In my practice, I see a good number of patients who would be considered for sentinel lymph node biopsy based on staging information alone but that have excellent survival based on gene expression profiling. The additional information provided by this test will be extremely helpful in guiding whether we should perform this procedure in patients that have very low chance of being SLN positive and that have low risk of recurrence.

On behalf of my patients, I thank you for providing this draft for public comment.

Thank you for the comment.

I am writing to support the draft local coverage determination for DecisionDx-Melanoma, which is a test that is used routinely in my clinic to inform management of patients with cutaneous melanoma.

The coverage guidance and intended use section of the draft describes appropriately that the test results can be used to guide decisions regarding sentinel lymph node biopsy as recently reported at the American Academy of Dermatology and the American College of Mohs Surgery conferences. The use of this test is supported by strong evidence demonstrating that it provides prognostic information that is independent from standard histopathologic factors. As a dermatologist and dermatopathologist, I am well aware of how analyzing tumor genetics can help us move beyond morphology and identify patients at both low and high risk of metastatic disease. Understanding this risk helps me avoid intense management of patients who are at low risk for progression or a positive sentinel lymph node, as well as directs my time and resources to patients who are high risk and need a more attention, including performance of the SLNB procedure.

Thank you for opening access for this test for Medicare patients with cutaneous melanoma and for the opportunity to provide some feedback.

Thank you for the comment.

I would like to add my support to the Castle Bio Science Melanoma molecular test.

We have been using this test extensively for our invasive melanoma patients with commercial insurance.

It has proven very useful in giving our patients more information and has helped guide us with future follow up, imaging, and usefulness of sentinal node biopsy.

As the board certified dermatologist on the committee I would strongly recommend this be covered by Medicare.

Thank you for the comment

The proposed Noridian policy discourages SLN biopsies in stage IB-III patients. I would like to comment that this proposed policy contradicts current NCCN guidelines for SLN biopsy. There has never been prospective randomized trial to show Melanoma Decision DX testing can supersede staging information from SLN biopsy. In fact all data to support Melanoma Decision DX testing is single institution data and much of it is retrospective tissue analysis (less rigorous). Manufacturer of test has declined to perform randomized prospective testing of assay to date. My fear is that this deviation from recommended staging approach may lead to increased risk for recurrences in less than adequately staged patients, and some of these patients will risk relapse with systemic rather than regional disease.

Thank you for the comment. This policy is not intended to encourage or to discourage sentinel lymph node biopsies. NCCN guidelines indicate that for some T1a, T1b, and T2 melanomas, sentinel lymph node biopsy is to be considered. The intent of this LCD is to allow clinicians to order a test that may provide one piece of information that providers and patients can use to make an individualized decision regarding whether or not a sentinel lymph node biopsy should be pursued.

Our institution has incorporated the DecisionDx-Melanoma test in our clinical pathway to manage melanoma patients.

We recently reported our experience using the test, focusing on the management changes resulting from the test, in the peer-reviewed literature (Schuitevoerder et al, J Drugs Dermatol 2018).

I have also been involved in research that supports the clinical utility of this test for sentinel lymph node biopsy (SLNB) guidance, as outlined in the draft LCD. The results from these studies show that patients ≥55 years old who had a Class 1A GEP in a T1-T2 (≤2mm) tumor had a probability of being SLN positive that falls below 5%, the current NCCN cutoff for recommending the procedure. Patients in the Medicare population (≥65) years old with a Class 1A, T1-T2 tumor had a particularly low rate of SLN positivity (1.6%; i.e. well within the NCCN “Do Not Perform” category).

To validate this approach, two contemporary, multi-center study cohorts totaling 1,421 prospectively and consecutively tested patients, representative of the general population of melanoma patients who are considered for SLNB, were used (Vetto et al, submitted manuscript). It is important to note that these cohorts include T1a patients, as a subset of this group is currently eligible and contribute to a substantial number of patients who undergo the procedure.

The results from this study show that it is possible to identify CM patients who have a very low likelihood of having a positive SLN (

Thus, I would like to thank you for issuing this draft LCD and convey my support for it. As mentioned above, I find that T1a patients with uncertain thickness represent a group that I often see in my practice and who would benefit from being included in the final LCD.

Thank you for the comment.

On behalf of the nearly 13,0QO U.S. based members of the American Academy of Dermatology Association (AADA), we are writing to endorse the approval of Proposed LCD DL37748 MolDX: DecisionDX:.Melanoma. This Local Coverage Determination (LCD) proposes the coverage of the DecisionDX-Melanoma test for Melanoma patients, diagnosed with SLNB eligible Tlb and TZ tumors, who are being considered for SLNB.

Based on the studies referenced in the LCD, the DecisionDx-Melanoma test can help identify patients at higher risk for developing metastatic disease and thereby, guide clinicians toward proceeding with SLNB in these higher risk patients. Conversely, low risk Class 1 patients could be spared the risks of SLNB, by electing to avoid SLNB which would provide a low yield of positivity in this group.

Hopefully, the DecisionDx-Melanoma test will prove to be another useful tool in providing important prognostic information and clinical guidance in the management of Tla-TZ Melanoma patients.

For these reasons, we encourage JF Noridian Healthcare Solutions, LLC to approve draft LCD DL 37748.

Thank you for the comment.