Risk Factors

Exposure to solar ultraviolet radiation is a major risk factor for melanoma. Most melanomas occur on the sun-exposed skin, particularly those areas most susceptible to sunburn. Likewise, features that are associated with an individual’s sensitivity to sunlight, such as light skin pigmentation, red or blond hair, blue or green eyes, freckling tendency, and poor tanning ability are well-known risk factors for melanoma.^3,4, There is also a strong association between high total body nevus counts and melanoma.^5,

Several genes appear to contribute to melanoma predisposition such as tumor suppressor gene CDKN2A, melanocortin-1 receptor (MC1R) gene, and BAP1 variants.^6,7,8, Individuals with either familial or sporadic melanoma have a two to three times increased risk of developing a subsequent primary melanoma.^9, Several occupational exposures and lifestyle factors, such as body mass index and smoking, have been evaluated as possible risk factors for melanoma.^10,

Gene Expression Profiling

GEP measures the activity of thousands genes simultaneously and creates a snapshot of cellular function. Data for GEP are generated by several molecular technologies including DNA microarrays that measures activity relative to previously identified genes and RNA-Seq that directly sequences and quantifies RNA molecules. Clinical applications of GEP include disease diagnosis, disease classification, prediction of drug response, and prognosis.

Regulatory Status

Clinical laboratories may develop and validate tests in-house and market them as a laboratory service; laboratory-developed tests must meet the general regulatory standards of the Clinical Laboratory Improvement Amendments. The Pigmented Lesion Assay, myPath Melanoma, and DecisionDx-Melanoma tests are available under the auspices of the Clinical Laboratory Improvement Amendments. Laboratories that offer laboratory-developed tests must be licensed by the Clinical Laboratory Improvement Amendments for high-complexity testing. To date, the U.S. Food and Drug Administration has chosen not to require any regulatory review of this test.

Related Policies

• None

Experts recommend formal genetic counseling for patients who are at risk for inherited disorders and who wish to undergo genetic testing. Interpreting the results of genetic tests and understanding risk factors can be difficult for some patients; genetic counseling helps individuals understand the impact of genetic testing, including the possible effects the test results could have on the individual or their family members. It should be noted that genetic counseling may alter the utilization of genetic testing substantially and may reduce inappropriate testing; further, genetic counseling should be performed by an individual with experience and expertise in genetic medicine and genetic testing methods.

Medicare Coverage:
There is no National Coverage Determination (NCD) for Gene Expression Profiling for Cutaneous Melanoma. In the absence of an NCD, coverage decisions are left to the discretion of Local Medicare Carriers. Novitas Solutions, Inc, the Local Medicare Carrier for jurisdiction JL, has not issued a determination for this service.

PROPRIETARY LABS (Labs that are the sole source for the diagnostic lab test):
For labs which are proprietary (that is, the sole source for the diagnostic lab test involved), Medicare Advantage Products will follow the Medicare Local Coverage Determination of the State where the proprietary lab is located.

Pigmented Lesion Assay is a proprietary test. Per Noridian Healthcare Solutions, LLC, (Jurisdiction J-E,) FUTURE Local Coverage Determination (LCD): MolDX: Pigmented Lesion Assay (DL38151), effective 6/07/20, Pigmented Lesion Assay (CPT code 0089u) will be covered for use on melanocytic skin lesions with one or more clinical or historical characteristics suggestive of melanoma, including one or more ABCDE criteria (Asymmetry, Border, Color, Diameter, Evolving) when a clinician trained in the clinical diagnosis of skin cancer is considering the need for biopsy to rule out melanoma when Future LCD DL38151 and Future Article A58052 criteria are met.

For additional information and eligibility, refer to: FUTURE Local Coverage Determination (LCD): MolDX: Pigmented Lesion Assay (DL38151) and FUTURE Local Coverage Article: Billing and Coding: MolDX: Pigmented Lesion Assay (A58052). Available to be accessed at CMS LCD by State Index search: https://www.cms.gov/medicare-coverage-database/indexes/lcd-state-index.aspx?bc=AgAAAAAAAAAA&

MyPath® Melanoma is a proprietary test. Noridian Healthcare Solutions, LLC, (Jurisdiction J-F) has provided limited coverage for proprietary lab MyPath® Melanoma, Myriad Genetic Laboratories, when used for the diagnosis or exclusion of melanoma from a biopsy when LCD LCD L37881 and Article A57627 criteria are met including all of the following:
• The test is ordered by a board-certified dermatopathologist and;
• The specimen is a primary cutaneous melanocytic neoplasm for when diagnosis with a clear distinction between benign or malignant cannot be achieved using clinical and/or histopathological features alone and;
• The individual may be subjected to additional intervention, such as re-excision and/or sentinel lymph node biopsy, as a result of the diagnostic uncertainty.

For additional information and eligibility, refer to Noridian Healthcare Solutions, LLC , (Jurisdiction J-F), LCD L37881 and Article A57627. Local Coverage Determination (LCD): MolDX: myPath Melanoma Assay (L37881) and Noridian Healthcare Solutions, LLC , (Jurisdiction J-F) Local Coverage Article: Billing and Coding: MolDX: myPath Melanoma Assay (A57627). Available to be searched at Local Coverage Determinations (LCDs) by State Index: https://www.cms.gov/medicare-coverage-database/indexes/lcd-state-index.aspx.

DecisionDx-Melanoma is a proprietary test. Noridian Healthcare Solutions, LLC, ( Jurisdiction J-F), Proposed Local Coverage Determination (LCD):MolDX: DecisionDx-Melanoma (DL37748) proposes coverage of the DecisionDx-Melanoma test for certain individuals diagnosed with cutaneous melanoma. However, the LCD is in the comment review period and has not been finalized and is not effective at this time. For additional information, refer to Noridian Healthcare Solutions, LLC, ( Jurisdiction J-F), Proposed Local Coverage Determination (LCD):MolDX: DecisionDx-Melanoma (DL37748). Available to be searched at Local Coverage Determinations (LCDs) by State Index: https://www.cms.gov/medicare-coverage-database/indexes/lcd-state-index.aspx.


[RATIONALE: This policy was created in 2018 with a search of the PubMed database. The literature update was performed through March 20, 2020.

Evidence reviews assess whether a medical test is clinically useful. A useful test provides information to make a clinical management decision that improves the net health outcome. That is, the balance of benefits and harms is better when the test is used to manage the condition than when another test or no test is used to manage the condition.

The first step in assessing a medical test is to formulate the clinical context and purpose of the test. The test must be technically reliable, clinically valid, and clinically useful for that purpose. Evidence reviews assess the evidence on whether a test is clinically valid and clinically useful. Technical reliability is outside the scope of these reviews, and credible information on technical reliability is available from other sources.

Gene Expression Profiling to Guide Initial Biopsy Decisions
Clinical Context and Test Purpose

Primary care providers evaluate suspicious pigmented lesions to determine who should be referred to dermatology. Factors considered include both a patient’s risk for melanoma as well as a visual examination of the lesion. The visual examination assesses whether the lesion has features suggestive of melanoma.

Criteria for features suggestive of melanoma have been developed. One checklist is the ABCDE checklist^11,:

• Asymmetry;
• Border irregularities;
• Color variegation;
• Diameter ≥ 6 mm;
• Evolution.

Melanoma is difficult to diagnose based on visual examination, and the criterion standard for diagnosis is histopathology. There is a low threshold for excisional biopsy of suspicious lesions for histopathologic examination due to the procedure’s ease and low-risk as well as the high probability of missing melanoma. However, the yield of biopsy is fairly low. The number of biopsies performed to yield one melanoma diagnosis has been estimated to be about 15 for U.S. dermatologists.^13, Therefore a test that could accurately identify those lesions not needing a biopsy (ie, a rule-out test for biopsy) could be clinically useful.

The purpose of GEP in patients who have suspicious pigmented lesions being considered for biopsy is to inform a decision about whether to biopsy.

The question addressed in this section of the policy is: Does GEP improve the net health outcome in individuals with suspicious pigmented lesions?

The following PICO was used to select literature to inform this policy.

Patients

The relevant population of interest is patients with suspicious pigmented lesions being considered for referral for biopsy, specifically those lesions meeting one or more ABCDE criteria.

Interventions

The test being considered is the DermTech Pigmented Lesion Assay (PLA). The PLA test measures expression of six genes (PRAME, LINC00518, CMIP, B2M, ACTB, PPIA). The PRAME (PReferentially expressed Antigen in MElanoma) gene encodes an antigen that is preferentially expressed in human melanomas, and that is not expressed in normal tissues (except testis).^14, LINC00518 (Long Intergenic Non-protein Coding RNA518) is a regulatory RNA molecule. The other four genes provide normalization values.^15, The feasibility of a test like PLA was first described in Wachsman et al (2011) and Gerami et al (2014).^16,17, and development of the specific PLA test was described in Gerami et al (2017).^18,

The test is performed on skin samples of lesions at least 5 mm in diameter obtained via noninvasive, proprietary adhesive patch biopsies of a stratum corneum specimen. The test does not work on the palms of hands, soles of feet, nails, or mucous membranes, and it should not be used on bleeding or ulcerated lesions.^15,

The PLA test report includes two results. The first result is called the PLA MAGE (Melanoma Associated Gene Expression), which indicates low-risk (neither PRAME nor LINC00518 expression was detected), moderate-risk (expression of either PRAME or LINC00518 was detected), or high-risk (expression of both PRAME and LINC00518 was detected). The second result is as an algorithmic PLA score that ranges from 0 to 100, with higher scores indicating higher suspicion of malignant disease.^15,

It is not clear whether the PLA test is meant to be used as a replacement, triage, or add-on test with respect to dermoscopy. The PLA sample report states that for low-risk lesions, physicians should “consider surveillance,” while for moderate- and high-risk lesions, physicians should “recommend a biopsy.” It does not state whether lesions with negative results should be further evaluated with dermoscopy or other techniques to confirm the lesion should not be biopsied. Therefore, this policy evaluates the test as a replacement for dermoscopy. As mentioned previously, there is a low threshold for biopsy of suspicious lesions. As such, tests that can rule-out need for biopsy could be useful and thus sensitivity and negative predictive value are the performance characteristics of most interest.

Comparators

After a referral from primary care to dermatology settings, dermatologists use visual examination as well as tools such as dermoscopy to make decisions regarding biopsy of suspicious lesions. A meta-analysis of 9 studies (8487 lesions with 375 melanomas) compared dermoscopy with visual examination alone for the diagnosis of melanoma; it reported that, for clinicians with training in dermoscopy, adding dermoscopy to visual examination increased the sensitivity from 71% to 90%. The specificity numerically increased from 80% to 90%, but the difference was not statistically significant.^19, Although dermoscopy is noninvasive and may aid in decision making regarding biopsy, it is only used by approximately 50% to 80% of dermatologists in the U. S. due to lack of training, interest, or time required for the examination.^20,21,

The reference standard for diagnosis of melanoma is histopathology.

Outcomes

The beneficial outcomes of a true-positive test result are appropriate biopsy and diagnosis of melanoma. The beneficial outcome of a true-negative test result is potentially avoiding unnecessary biopsy.

The harmful outcome of a false-positive result is having an unnecessary biopsy. The harmful outcome of a false-negative result is potential delay in diagnosis and treatment.

The time frame of interest for calculating performance characteristics is time to biopsy result. Patients who forgo biopsy based on test results could miss or delay diagnosis of cancer. Longer follow-up would be necessary to determine the effects on overall survival..

Technically Reliable

Assessment of technical reliability focuses on specific tests and operators and requires review of unpublished and often proprietary information. Review of specific tests, operators, and unpublished data are outside the scope of this policy, and alternative sources exist. This policy focuses on the clinical validity and clinical utility.

Clinically Valid

A test must detect the presence or absence of a condition, the risk of developing a condition in the future, or treatment response (beneficial or adverse).

Determining whether a test can guide biopsy decisions is not based only on its sensitivity and specificity, but also on how the accuracy of the existing pathway for making biopsy decisions is changed by the test. Therefore, the appropriate design for evaluating performance characteristics depends on the role of the new test in the pathway for making biopsy decisions. New tests may be used as replacements for existing tests, to triage who proceeds for existing tests or add-on tests after existing tests. For replacement tests, the diagnostic accuracy of both tests should be concurrently compared, preferably in a paired design (ie, patients receive both tests), and all patients receive the reference standard. For a triage test, a paired design is also needed, with the reference standard being performed preferably on all patients but at least for all discordant results. For an add-on test, the included patients can be limited to those who were negative after existing tests with verification of the reference standard in patients who are positive on the new test.^22,

Study Selection Criteria

For the evaluation of clinical validity of the PLA test, studies that meet the following eligibility criteria were considered:

• Reported on a validation cohort that was independent of the development cohort;
• Reported on the accuracy of the marketed version of the technology;
• Included a suitable reference standard (histopathology);
• Patient/sample clinical characteristics were described
• Patient/sample selection criteria were described.

The validation cohort from the Gerami et al (2017) publication was included.^18, The study characteristics are described in Table 1. The report stated that included lesions were selected by dermatologists experienced in pigmented lesion management from 28 sites in the U. S., Europe, and Australia; therefore, the samples were likely not consecutive or random. Information regarding the previous testing was not provided. The flow of potential and included samples was not clear, and whether the samples were all independent or, multiple samples from the same patient were not described. Diagnosis of melanoma was based on consensus among a primary reader and three expert dermatopathologists. The report did not state whether the histopathologic diagnosis was blinded to the results of the PLA test but did state the diagnosis was “routinely” assessed. Interpretation of the PLA result does not depend on a reader, so it is blinded to histopathologic results. In 11% of cases originally selected, a consensus diagnosis was not reached, and these samples were not included in the training or validation cohorts. Dates of data collection were not reported. Sex and anatomic location of biopsy were reported, but other clinical characteristics (eg, risk factors for melanoma, presenting symptoms) were not. Study results are shown in Table 2. The study training cohort included 157 samples with 80 melanomas and 77 non-melanomas. The study validation cohort included 398 samples with 87 melanomas and 311 non-melanomas. Study relevance, design, and conduct gaps are in Tables 3 and 4.

Section Summary: Pigmented Lesion Assay Clinical Validity

Multiple high-quality studies are needed to establish the clinical validity of a test. The PLA test has one clinical validity study with many methodologic and reporting limitations. Therefore, performance characteristics are not well-characterized. Also, the test has not been compared with dermoscopy, another tool frequently used to make biopsy decisions.

Table 1. Clinical Validity Study Characteristics of the PLA Test for Diagnosing Melanoma

Study	Study Population	Design	Reference Standard for Dx of Melanoma	Threshold Score for PLA Test	Timing of Reference and PLA Tests	Blinding of Assessors
Gerami et al (2017)18,	Adults Suspicious pigmented lesion ≥4 mm in diameter Without obvious or suspicious nodular melanoma 24% from extremities, 13% from head and neck, 62% from trunk 55% of samples from men Median age, 49 y (range, 19-97 y)	Retrospective Not consecutive or random	Histopathology; consensus diagnosis	Quantitative PCR yielded an amplification curve and a measurable cycle threshold value Either LINC00518 or PRAME detected	PLA patch before surgical biopsy; timing between patch and surgical biopsy unclear	Not clear

Dx: diagnosis; PCR: polymerase chain reaction; PLA: Pigmented Lesion Assay.

Table 2. Clinical Validity Study Results of the PLA Test for Diagnosing Melanoma

Study	Initial N	Final N	Excluded Samples	Melanoma Prevalence	Sensitivityb	Specificityb	PPVb	NPVb
Gerami et al (2017)18,	398a	398	Before allocation to training and validation cohorts, 11% of original samples excluded due to lack of consensus diagnosis	22%	91 (83 to 96)	69 (64 to 74)	45 (38 to 53)c	96 (93 to 98)c

NPV: negative predictive value; PPV: positive predictive value; PLA: Pigmented Lesion Assay.

^a 398 samples were included in the validation cohort; the number of independent patients is unclear.
^b Values are percentages with 95% confidence interval.
^c Confidence intervals provided in the report; calculated from data provided.

Study	Populationa	Interventionb	Comparatorc	Outcomesd	Duration of Follow-Upe
Gerami et al (2017)18,	3. Study population characteristics not adequately described		3. No comparison to dermoscopy	3. Predictive values were not reported but were calculated based on data provided

^a Population key: 1. Intended use population unclear; 2. Clinical context is unclear; 3. Study population is unclear; 4. Study population not representative of intended use.
^b Intervention key: 1. Classification thresholds not defined; 2. Version used unclear; 3. Not intervention of interest.
^c Comparator key: 1. Classification thresholds not defined; 2. Not compared to credible reference standard; 3. Not compared to other tests in use for same purpose.
^d Outcomes key: 1. Study does not directly assess a key health outcome; 2. Evidence chain or decision model not explicated; 3. Key clinical validity outcomes not reported (sensitivity, specificity and predictive values); 4. Reclassification of diagnostic or risk categories not reported; 5. Adverse events of the test not described (excluding minor discomforts and inconvenience of venipuncture or noninvasive tests).
^e Follow-Up key: 1. Follow-up duration not sufficient with respect to natural history of disease (true-positives, true-negatives, false-positives, false-negatives cannot be determined).

Study	Selectiona	Blindingb	Delivery of Testc	Selective Reportingd	Completeness of Follow-Upe	Statisticalf
Gerami et al (2017)18,	1,2. Not clear what criteria used to select samples but it does not appear to have been random or consecutive	1. Blinding of histopathology readers not described	1. Patch biopsy administered before surgical biopsy but timing between procedures not described	1. No registration reported

^a Selection key: 1. Selection not described; 2. Selection not random or consecutive (ie, convenience).
^b Blinding key: 1. Not blinded to results of reference or other comparator tests.
^c Test Delivery key: 1. Timing of delivery of index or reference test not described; 2. Timing of index and comparator tests not same; 3. Procedure for interpreting tests not described; 4. Expertise of evaluators not described.
^d Selective Reporting key: 1. Not registered; 2. Evidence of selective reporting; 3. Evidence of selective publication.
^e Follow-Up key: 1. Inadequate description of indeterminate and missing samples; 2. High number of samples excluded; 3. High loss to follow-up or missing data.
^f Statistical key: 1. Confidence intervals and/or p values not reported; 2. Comparison to other tests not reported.

A test is clinically useful if the results inform management decisions that improve the net health outcome of care. The net health outcome can be improved if patients receive correct therapy, or more effective therapy, or avoid unnecessary therapy, or avoid unnecessary testing.

Direct Evidence

Direct evidence of clinical utility is provided by studies that have compared health outcomes for patients managed with and without the test. Because these are intervention studies, the preferred evidence would be from randomized controlled trials (RCTs).

No direct evidence of clinical utility was identified.

Chain of Evidence

Indirect evidence on clinical utility rests on clinical validity. If the evidence is insufficient to demonstrate test performance, no inferences can be made about clinical utility through a chain of evidence.

A decision-impact study by Ferris et al (2017) assessed the potential impact of PLA on physicians’ biopsy decisions in patients.^24, Forty-five dermatologists evaluated 60 clinical and dermoscopic images of atypical pigmented lesions (8 melanoma, 52 nonmelanoma). In the first round, dermatologists did not have PLA test results and, in the second round, dermatologists had access to PLA test results with the order of cases being scrambled. The dermatologists were asked whether the lesions should be biopsied after each round. Therefore, the corresponding number of biopsy decisions should be 45×60×2=5400. Data were collected in 2014 and 2015. Results were reported for 4680 decisions with no description of the disposition of the remaining decisions. Of the 4680 reported decisions, 750 correct biopsy decisions were made without PLA results while 1331 were made with PLA results and 1590 incorrect biopsy decisions were made without PLA results while 1009 incorrect biopsy decisions were made with PLA results.

Section Summary: Clinically Useful

There is no direct evidence of clinical utility. A chain of evidence for clinical utility cannot be constructed due to lack of robust evidence of clinical validity.

Gene Expression Profiling for Diagnosing Lesions with Indeterminate Histopathology
Clinical Context and Test Purpose

The diagnosis of melanoma was described in the previous section. The diagnosis of melanoma is histopathologic and when the histopathologic diagnosis is straightforward, ancillary methods such as comparative genomic hybridization, florescence in situ hybridization (FISH), and GEP are not recommended. Therefore the usefulness of an ancillary test is its ability to predict biologic behavior (metastasis) of lesions that are indeterminate by histopathology.

The purpose of GEP in patients whose melanocytic lesion is indeterminate after histopathology is to aid in the diagnosis of melanoma and decisions regarding treatment and surveillance.

The question addressed in this section of the policy is: Does GEP improve the net health outcome in individuals with indeterminate melanocytic lesions?

The following PICO was used to select literature to inform this policy.

Patients

The relevant population of interest is patients whose melanocytic lesion is indeterminate based on clinical and histopathologic features.

Interventions

The test being considered is the Myriad myPath Melanoma test. The myPath test measures expression of 23 genes using quantitative reverse-transcription polymerase chain reaction. Fourteen genes are involved in melanoma pathogenesis and are grouped into three components related to cell differentiation, cell signaling, and the immune response, and nine housekeeper genes are also included. The test is performed on five standard tissue sections from an existing formalin-fixed, paraffin-embedded biopsy specimen.

The myPath test report includes an algorithmic myPath score ranging from -16.7 to 11.1, with higher, positive scores indicating higher suspicion of malignant disease.^25, The myPath report also classifies these scores: -16.7 to -2.1 are “benign”; -2.0 to -0.1 are “indeterminate”; and 0.0 to +11.1 are “malignant”. Development of the test has been described by Clarke et al (2015).^26,

The myPath test is meant as an add-on test to standard histopathology. Studies have evaluated the performance characteristics of the test when histopathology is used as the reference standard.^26,27,28,but are not the focus of this policy given that the test's potential usefulness is in evaluation of indeterminate lesions.

No recommendations for treatment or surveillance are given on the report.

Comparators

The reference standard for diagnosis of melanoma is histopathology. However, in cases of indeterminate histopathology, long-term follow-up is needed to evaluate the clinical outcome, specifically metastasis.

Comparative genomic hybridization and FISH are also used to diagnosis indeterminate lesions although neither has been fully validated. FISH has been evaluated as a tool to aid in the diagnosis of lesions that are indeterminate, following histopathology in two studies that included histologically ambiguous lesions and a clinical, long-term follow-up. One study reported by Gaiser et al (2010) included 22 melanocytic lesions (12 indeterminate) followed for a mean of 65 months (range, 10-156 months) and reported a FISH sensitivity of 60% and a specificity of 50% for development of metastases during follow-up.^29, A second study, reported by Vergier et al (2011), included 90 indeterminate melanocytic lesions of which 69 had no recurrence for at least 5 years of follow-up (mean, 9 years; range, 5-19 years) and 21 lesions that exhibited metastases. The sensitivity and specificity rates of the histopathologic review combined with FISH for the clinical outcome were 76% and 90%, respectively.^30,

Outcomes

The beneficial outcomes of a true-positive test result are a diagnosis of melanoma and corresponding appropriate treatment and surveillance. The beneficial outcome of a true- negative test result is avoiding unnecessary surgery.

The harmful outcome of a false-positive result is having an unnecessary surgery and surveillance. The harmful outcome of a false-negative result is a delay in diagnosis and treatment.

The National Comprehensive Cancer Network guidelines state that even in the presence of node metastasis, indeterminate neoplasms can demonstrate benign biologic behavior, making it difficult to define a fully malignant lesion and also states that events in the group of indeterminate lesions tend to occur late. Therefore, the guidelines suggest that long-term follow-up is necessary to validate a test for this purpose.

Recurrence and metastases can occur may years after treatment of melanoma. In the two studies evaluating long-term outcomes of FISH (described above), the mean follow-up was approximately 5.5 and nine years.^29,30, In Vergier et al (2011), metastases in the FISH-negative group generally occurred by 5 years.^30,

For this section of the review, at least five years of event-free follow-up is required to confirm negative tests. The event of interest is metastasis.

Technically Reliable

Assessment of technical reliability focuses on specific tests and operators and requires review of unpublished and often proprietary information. Review of specific tests, operators, and unpublished data are outside the scope of this policy, and alternative sources exist. This policy focuses on the clinical validity and clinical utility.

Clinically Valid

A test must detect the presence or absence of a condition, the risk of developing a condition in the future, or treatment response (beneficial or adverse).

Study Selection Criteria

For the evaluation of clinical validity of the myPath test, studies that meet the following eligibility criteria were considered:

• Reported on a validation cohort that was independent of the development cohort;
• Reported on the accuracy of the marketed version of the technology;
• Included a suitable reference standard (clinical outcome with at least five years of follow-up for negatives);
• Patient/sample clinical characteristics were described
• Patient/sample selection criteria were described.

The Ko et al (2017) clinical validity study met selection criteria.^34, The study characteristics are described in Table 5. In Ko et al (2017), archived melanocytic neoplasms were submitted for myPath testing from university clinics in the United States and United Kingdom with additional samples acquired from Avaden BioSciences.^34, Stage I, II, and III primary cutaneous melanomas that produced distant metastases subsequent to the diagnosis and benign lesions with clinical follow-up and no evidence of recurrence of metastases were included. For benign samples, a disease-free time of at least five years was recommended. Information on the previous testing was not provided. It is not clear if any of the samples originally had indeterminate histopathology results. Dates of data collection were not reported. Sex, age, Breslow depth, and anatomic location were described; presenting symptoms were not reported. A total of 293 samples were submitted; of these 53 did not meet inclusion criteria and 58 (24% of those tested) failed to produce a valid test score. An additional seven samples with indeterminate results were excluded from the calculations of performance characteristics.

Study results are shown in Table 6. Study relevance, design, and conduct gaps are in Tables 7 and 8.

Table 5. Clinical Validity Study Characteristics of the myPath Test for Predicting Metastasis

Study	Study Population	Design	Reference Standard	Threshold Score for Positive myPath Test	Timing of Reference and myPath Tests	Blinding of Assessors
Ko et al (2017)34,	Primary cutaneous melanomas or benign melanocytic nevi Mean age, 53 y 55% of samples from men	Retrospective Not consecutive or randomly selected	Positive: malignant lesions that produced distant metastases Negative Event-free follow-up, recommended 5 y (median, 6.2 y)	Scores from 0.0 to 11.1 (ie, “malignant”)	Final clinical diagnosis established before  myPath  test Length of time between biopsy and myPath test unclear	Yes

Table 6. Clinical Validity Study Results of the myPath Test for Predicting Metastasis

Study	Initial N	Final N	Excluded Samples	Melanoma Prevalence	Sensitivitya	Specificitya	PPVa	NPVa
Ko et al (2017)34,	240	175	58 failed to produce test result 7 with indeterminate results	54	94 (87 to 98)b	96 (89 to 99)b	97 (91 to 99)b	93 (85 to 97)b

NPV: negative predictive value; PPV: positive predictive value.

^a Values are percentages with 95% confidence interval.
^b Confidence intervals not provided in the report; calculated from data provided.

Study	Populationa	Interventionb	Comparatorc	Outcomesd	Duration of Follow-Upe
Ko et al(2017)34,	4. Study population is not limited to lesions that are indeterminate following histopathology				None noted

^a Population key: 1. Intended use population unclear; 2. Clinical context is unclear; 3. Study population is unclear; 4. Study population not representative of intended use.
^b Intervention key: 1. Classification thresholds not defined; 2. Version used unclear; 3. Not intervention of interest.
^c Comparator key: 1. Classification thresholds not defined; 2. Not compared to credible reference standard; 3. Not compared to other tests in use for same purpose.
^d Outcomes key: 1. Study does not directly assess a key health outcome; 2. Evidence chain or decision model not explicated; 3. Key clinical validity outcomes not reported (sensitivity, specificity and predictive values); 4. Reclassification of diagnostic or risk categories not reported; 5. Adverse events of the test not described (excluding minor discomforts and inconvenience of venipuncture or noninvasive tests).
^e Follow-Up key: 1. Follow-up duration not sufficient with respect to natural history of disease (true positives, true negatives, false positives, false negatives cannot be determined).

Study	Selectiona	Blindingb	Delivery of Testc	Selective Reportingd	Completeness of Follow-Upe	Statisticalf
Ko et al(2017)34,	2. Samples not consecutive or random		1. Unclear how much time elapsed between biopsy and myPath test	1. No registration reported	2. More than 25% of samples tested did not produce results or produced indeterminate results	1. CIs for sensitivity and specificity not reported but were calculated based on data provided. NPV, PPV were not reported

^a Selection key: 1. Selection not described; 2. Selection not random or consecutive (ie, convenience).
^b Blinding key: 1. Not blinded to results of reference or other comparator tests.
^c Test Delivery key: 1. Timing of delivery of index or reference test not described; 2. Timing of index and comparator tests not same; 3. Procedure for interpreting tests not described; 4. Expertise of evaluators not described.
^d Selective Reporting key: 1. Not registered; 2. Evidence of selective reporting; 3. Evidence of selective publication.
^e Follow-Up key: 1. Inadequate description of indeterminate and missing samples; 2. High number of samples excluded; 3. High loss to follow-up or missing data.
^f Statistical key: 1. Confidence intervals and/or p values not reported; 2. Comparison to other tests not reported.

Multiple high-quality studies are needed to establish the clinical validity of a test. The myPath test has one clinical validity study including long-term follow-up for metastasis as the reference standard. However, it is not clear whether the study population included lesions that were indeterminate following histopathology and the study had other methodologic and reporting limitations. Therefore, performance characteristics are not well-characterized.

Clinically Useful

A test is clinically useful if the results inform management decisions that improve the net health outcome of care. The net health outcome can be improved if patients receive correct therapy, or more effective therapy, or avoid unnecessary therapy, or avoid unnecessary testing.

Direct Evidence

Direct evidence of clinical utility is provided by studies that have compared health outcomes for patients managed with and without the test. Because these are intervention studies, the preferred evidence would be from RCTs.

No direct evidence of clinical utility was identified.

Chain of Evidence

Indirect evidence on clinical utility rests on clinical validity. If the evidence is insufficient to demonstrate test performance, no inferences can be made about clinical utility.

Two decision-impact studies assessed the potential impact of myPath on physicians’ treatment decisions in patients with diagnostically challenging lesions.^35,36, Given the lack of health outcomes, it is not known whether any treatment changes were clinically appropriate.

Section Summary: Clinically Useful

There is no direct evidence of clinical utility. A chain of evidence for clinical utility cannot be constructed due to lack of robust evidence of clinical validity.

Gene Expression Profiling to Guide Management Decisions in Melanoma
Clinical Context and Test Purpose

Many treatments and surveillance decisions are determined by a patient’s prognostic stage group based the American Joint Committee on Cancer tumor, node, metastasis staging system.^37,The prognostic groups are as follows: stage I, T1a through T2a primary melanomas without evidence of regional or distant metastases; stage II, T2b through T4b primary melanomas without evidence of lymphatic disease or distant metastases; stage III: pathologically documented involvement of regional lymph nodes or in transit or satellite metastases (N1 to N3); stage IV: distant metastases. Patients may also undergo sentinel lymph node biopsy to gain more definitive information about the status of the regional nodes.

Wide local excision is the definitive surgical treatment of melanoma. Following surgery, patients with American Joint Committee on Cancer stage I or II (node-negative) melanoma do not generally receive adjuvant therapy. Patients with higher risk melanoma receive adjuvant immunotherapy or targeted therapy. Ipilimumab has been shown to prolong recurrence-free survival by approximately 25% compared with placebo at a median of 5.3 years in patients with resected, stage III disease.^38, Nivolumab has been shown to further prolong survival compared with ipilimumab by approximately 35% at 18 months.^39, For patients who are BRAF V600 variant-positive with stage III melanoma, the combination of dabrafenib plus trametinib has been estimated to prolong relapse-free survival by approximately 50% over 3 years.^40,

Patients with stage I and IIA disease should undergo an annual routine physical and dermatologic examination. These patients typically do not receive surveillance imaging. Patients with stage IIB - III melanoma may be managed with more frequent follow-up and imaging surveillance following therapy. However, follow-up strategies and intervals are not based on rigorous data, and opinions vary regarding appropriate strategies.

The purpose of GEP in patients with melanoma is to identify low and high-risk patients classified as stage I or II according to the AJCC criteria. Current guidelines do not recommend adjuvant therapy for AJCC stage I or II patients following surgery. Patients initially staged as I or II who have positive lymph nodes following sentinel lymph node biopsy (SLNB) are then eligible to be treated with adjuvant therapy as stage III patients.

At least three uses for the test have been suggested. One clinical validity study (described below), the authors stated that “high-risk patients with stage I and II disease may benefit from adjuvant therapy and/or enhanced imaging protocols to allow for early detection of metastasis.”^41, In another clinical validity study, the authors concluded that the test’s “role in consideration of patients for adjuvant therapy should be examined prospectively.”^42, This use of the test would be as a replacement for SLNB since SLNB is currently used to identify patients clinically diagnosed as stage I and II who have node involvement and are candidates for adjuvant therapy.

The manufacturer’s website has suggested that physicians can use DecisionDx-Melanoma information to guide decisions regarding:

1. "Whether to perform a sentinel lymph node biopsy surgical procedure for eligible patients 55 years of age and older who have tumors less than 2 mm deep (T1-T2)"
2. "Deciding what level of follow-up, imaging, and referrals are appropriate for any patient with a tumor at least 0.3 mm deep."

The question addressed in this section of the policy is: Does GEP improve the net health outcome in individuals with AJCC stage I or II melanoma?

The following PICO was used to select literature to inform this policy.

Patients

To select patients for adjuvant therapy and/or enhanced surveillance, the relevant population of interest are patients with AJCC stage I/II cutaneous melanoma.

To select patients who can avoid SLNB, the relevant population of interest are patients with AJCC stage I or II cutaneous melanoma who are being considered for SLNB. The manufacturer website says the test is for ' eligible patients 55 years of age and older who have tumors less than 2 mm deep (T1-T2)'

Interventions

The test being considered is the Castle Biosciences DecisionDx-Melanoma test. The DecisionDx test measures expression of 31 genes using quantitative reverse-transcription polymerase chain reaction. The test includes 28 prognostic gene targets and 3 endogenous control genes. The test is performed on standard tissue sections from an existing formalin-fixed, paraffin-embedded biopsy or wide local excision specimen.

Development of the test was described in Gerami et al (2015).^41,To develop the DecisionDx-Melanoma gene panel, Gerami et al (2015) conducted a meta-analysis of published studies that identified differential gene expression in metastatic vs nonmetastatic primary cutaneous melanoma. Of 54 identified genes, investigators selected 20 for further polymerase chain reaction analysis based on chromosomal location. Five genes from Castle Biosciences’ DecisionDx-UM gene panel were added based on analysis of metastatic and nonmetastatic primary cutaneous melanoma, and two probes of the BRCA1-associated protein 1 gene, BAP1, which has been associated with the metastatic potential of uveal melanoma, also were added. Finally, four genes with minimal variation in expression level between metastatic and nonmetastatic primary cutaneous melanoma were added as controls. Patients had a minimum follow-up of five years unless there was a well-documented metastatic event, including positive SLNB. Information about treatments received was not provided.

The DecisionDx test report provides a 'class' which stratifies tumors as class 1 or class 2,. According to the sample report available on the manufacturer website: "The DecisionDx-Melanoma algorithm generates a value between 0 and 1 with a crossover point of 0.5. Subclassification (A or B) is based on proximity of this value to the crossover point."

Comparators

Treatment and surveillance recommendations are based on AJCC staging. SLNB may be used to get more definitive information about the status of the regional nodes compared with a physical examination. The American Society of Clinical Oncology and National Comprehensive Cancer Network have similar but not identical recommendations regarding which patients should undergo SLNB based on thickness and other high-risk features.

SLNB has a low rate of complications; in the Sunbelt Melanoma Trial, a prospective multi-institutional study of SLNB for melanoma reported by Wrightson et al (2003), less than 5% of the 2120 patients developed major or minor complications associated with SLNB.^43,

Online tools are available to predict prognosis based on the AJCC guidelines. The original AJCC tool was developed by Soong et al (n.d.).^44, Callender et al (2012) incorporated SLNB results into a revised tool (http://www.melanomacalculator.com/).^45,

Outcomes

Regarding selecting patients for adjuvant therapy and/or enhanced surveillance:

If the test is used to 'rule-in' AJCC stage I or IIA patients, a negative DecisionDx (class 1) test result would not change outcomes. Per guidelines, the patients would not receive adjuvant therapy or enhanced surveillance, just as without the DecisionDx test. A positive DecisionDx (class 2) test result would indicate that a patient might benefit from adjuvant therapy or enhanced surveillance. Therefore, the potential beneficial outcomes of a true positive result are additional treatment and surveillance and potentially prolonged survival. The potential harmful outcomes of a false-positive result are unnecessary adverse effects and burdens of adjuvant therapy and enhanced surveillance.

If the test is used to 'rule-out AJCC stage IIB - III patients, a negative DecisionDx (class 1) test result would indicate that a patient might be able to avoid enhanced surveillance. Therefore, the potential beneficial outcomes of a true negative result are avoiding burdens of surveillance and potential overtreatment. The potential harmful outcomes of a false-negative result are potentially prolonged survival with enhanced surveillance.

Regarding selecting patients who can avoid SLNB:

For patients meeting guideline-recommended criteria for SLNB, a positive DecisionDx (class 2) test result would not change outcomes. The patients would proceed to SLNB, as they would have without the DecisionDx test, and treatment and imaging decisions would depend on SLNB results. A negative DecisionDx (class 1) test result in patients 55 years of age and older who have tumors less than 2 mm deep (T1-T2) would indicate that a patient could avoid an SLNB. Therefore, the potential beneficial outcomes of a true-negative result are avoidance of an SLNB and related adverse effects and burdens. The potential harmful outcomes of a false-negative result are reduced time to recurrence due to not identifying node-positive patients that would be eligible for beneficial adjuvant treatment and potentially reduced survival.

The risk of recurrence decreases over time but does not reach zero. In a study of 1568 patients with stage I melanoma, Dicker et al (1999) found that 80% of the recurrences occurred within the first 3 years.^46, A prospective study by Garbe et al (2003) reported that, for stage I and II patients, the risk of recurrence was low after 4.4 years.^47, Among 4731 patients treated for more than 10 years at 1 institution, Faries et al (2013) found the majority of recurrences occurred in the first 5 years.^48, However, 7% of patients experienced recurrence after 10 years (median, 16 years). Even among stage I/II patients, recurrence after 10 years occurred in 2% of patients. Five-year recurrence-free survival (RFS) is the outcome and time-point of interest.

Technically Reliable

Assessment of technical reliability focuses on specific tests and operators and requires review of unpublished and often proprietary information. Review of specific tests, operators, and unpublished data are outside the scope of this policy, and alternative sources exist. This policy focuses on the clinical validity and clinical utility.

Clinically Valid
Study Selection Criteria

For the evaluation of clinical validity of the DecisionDx test, studies that meet the following eligibility criteria were considered:

• Reported on a validation cohort that was independent of the development cohort;
• Reported on the accuracy of the marketed version of the technology;
• Included a suitable reference standard (five-year RFS or five-year metastasis-free survival[MFS]);
• Patient/sample clinical characteristics were described
• Patient/sample selection criteria were described.

Four independent clinical validity studies meeting eligibility criteria have been conducted. Characteristics and results are summarized in Tables 9 and 10 and briefly in the paragraphs that follow.
Table 9. Clinical Validity Study Characteristics of the DecisionDx Test for Diagnosing Melanoma

Study	Study Population	Design	Reference Standard / Outcome Measure	Threshold Score for Positive DecisionDx Test	Timing of Reference and DecisionDx Tests	Blinding of Assessors
Gerami et al (2015)41,; Validation subset	Adults Stage I-IV cutaneous melanoma (87% stage I/II)At least 5 y of FU (median, 7.0 y) Median Breslow thickness, 0.8 mm (nonmetastasis) and 3.99 mm (metastasis) SLN positivity NR	Retrospective Not consecutive or randomly selected	5-y RFS	Class 2 is high-risk Risk threshold not provided	Patient diagnosed between 1998 and 2009 Timing of Decision Dx not described	Yes
Zager et al (2018)42,	Stage I-III cutaneous melanoma (68% stage I/II)At least 5 y of FU (median, 7.5 y) Median Breslow thickness, 1.2 mm 30% SLN positive	Retrospective Not consecutive or randomly selected	5-y RFS	Class 2 is high-risk Class 1: probability score 0 to 0.49 Class 2: probability score 0.5 to 1	Patients diagnosed between 2000 and 2014 Timing of DecisionDx not described	Yes
Greenhaw et al (2018)55,	Patients who were treated for primary invasive CM of any Breslow depth within the last 5 years and had had GEP testing (86% stage I, 14% stage II) Mean follow-up of 23 months; only 20 patients had 5-year follow-up	Retrospective Consecutive	5-y MFS	Commercial test cutoffs used	Institution offered DecisionDx testing to newly diagnosed and those treated within the previous five years	Yes
Keller et al (2019)56,	Patients had CM (91% stage I/II), opted for GEP testing and underwent SNB and wide excision of primary tumor. Median follow‐up time, 3.5 years Median Breslow thickness, 1.4 mm 9% SLN positive	Prospective	3-yr RFS	Commercial test cutoffs used	Patients diagnosed between 2013 and 2015 GEP reported to be performed concurrently with SNB	Yes

FU: follow-up; RFS: recurrence-free survival; MFS: metastasis-free survival; GEP: gene expression profiling; CM: cutaneous melanoma.

Table 10. Clinical Validity Study Results of the DecisionDx Test for Diagnosing Melanoma

Study	Initial / Final N	Excluded Samples	Events and Kaplan-Meier 5-Year RFS or MFSa		Sensitivitya	Specificitya	PPVa	NPVa
			Class 1	Class 2
Gerami et al (2015)41,; Validation subset		Samples excluded if melanoma dx not confirmed,  dissectible area not acceptable
Overall	Unclear/104		4 events RFS=97 (NR)	31 events RFS=31 (NR)p<0.001 vs class 1	89 (73 to 97)b	83 (72 to 91)b	72 (56 to 85)b	93 (84 to 98)b
AJCC stage I and II	Unclear/78		3 events RFS=98 (NR)	18 events RFS=37 (NR)· p<0.001 vs class 1	86 (64 to 97)b	84 (72 to 93)b	67 (46 to 83)b	94 (84 to 99)b
Zager et al (2018)42,		Did not meet analytic quality control thresholds
Overall	601 / 523		42 events RFS=88 (85 to 92)	100 events RFS=52 (46 to 60)	70 (62 to 78)	71 (67 to 76)	48 (41 to 55)	87 (82 to 90)
AJCC stage I	Unclear / 264		11 events RFS=96 (94 to 99)	6 events RFS=85 (74 to 97)	35 (14 to 62)b	87 (82 to 91)b	15 (6 to 31)b	95 (91 to 98)b
AJCC stage II	Unclear / 93		9 events RFS=74 (60 to 91)	30 events RFS=55 (44 to 69)	77 (61 to 89)b	43 (29 to 57)b	49 (36 to 62)b	72 (53 to 86)b
Greenhaw et al (2018)55,	256 / 256	None excluded but only 20 had 5-year follow-up	3 events MFS=93 (82 to 100)	8 events MFS=69 (52 to 90)	77 (46 to 94)	87 (82 to 91)	24 (13 to 40)	99 (96 to 100)
Keller et al (2019)56,	159 / 174	15 patients had insufficient tumor for GEP testing	events unclear at year 5 RFSc ~ 97 (NR)	events unclear at year 5 RFSc ~ 40 (NR)	NR	NR	NR	NR

AJCC: American Joint Committee on Cancer; Dx: diagnosis; NPV: negative predictive value; NR: not reported; PPV: positive predictive value; RFS: recurrence-free survival; MFS: metastasis-free survival

a Values are percentages with 95% confidence interval.
b Confidence intervals not provided in the report; calculated from data provided.
c RFS at 5 years not provided in text but estimated from a figure

Table 11. Reclassification of Patients Based on AJCC Stages to DecisionDx Classes in the Gerami Validation Cohort

AJCC Stage	DecisionDx Class
	Class 1 (Low-Risk), N (row %)	Class 2 (High-Risk), N (row %)	Total
0	0	0
Total stage I	50 (89%)a	6 (11%)	56
IA	37	1
IB	10	5
Total stage II	10 (29%)	24 (71%)	34
IIA	5	8
IIB	5	12
IIC	0	4
Total stage III	1 (8%)	11 (92%)	12
Total stage IV	0 (0%)	2 (100%)	2
Total	61	43	104

Adapted from Gerami et al (2015).^41,
AJCC: American Joint Committee on Cancer.

^a The subclass for n=3 class 1 samples are not reported.

Relevance, design, and conduct gaps are summarized in Tables 13 and 14.

Table 12. Reclassification of Patients Based on SLNB Status to DecisionDx Classes

SLNB	DecisionDx Class 1 (Low-Risk)			DecisionDx Class 2 (High-Risk)			Total
	n (%)	Events	5-Year RFS (95% CI), %	n (%)	Events	5-Year RFS (95% CI), %
Negative	103 (65)	15	87 (81 to 94)	77 (43)	28	67 (57 to 79)	180
Positive	56 (35)	22	61 (49 to 76)	101 (57)	60	37 (28 to 49)	157
Total	159			178			337a

Adapted from Zager et al (2017).^42,
CI: confidence interval; RFS: recurrence-free survival; SLNB: sentinel lymph node biopsy.

^a 337 patients had DecisionDx results and SLNB results.

Keller et al (2019) reported results of a validity study including 159 patients (ages 26 to 88) diagnosed with melanoma 2013 and 2015 who underwent SNB and concurrent GEP testing.^56, Study characteristics and results were reported in the preceding Tables 9 and 10. 117 patients were classified as class 1 (91 subclass 1A and 26 subclass 1B) and 42 were classified as Class 2 (12 subclass 2A and 30 subclass 2B). 78% of the tumors were AJCC stage I, 13% were stage II, and 9% were stage III. Five-year RFS was reported only in a figure and sample sizes at year 5 and precision estimates were not included. There were 6 recurrent events (n=117) in class I patients by 3 years (3 year RFS, 97% [95% CI, 93 to 100])). There were 23 recurrent events (n=42) in class 2 patients (3 year RFS, 47% [95% CI, 34 to 65]). GEP class was significantly associated with RFS in multivariate analysis controlling for age, Breslow thickness, ulceration and SNB results.

Table 13. Clinical Validity Study Relevance Limitations of the DecisionDx Test

Study	Populationa	Interventionb	Comparatorc	Outcomesd	Duration of Follow-Upe
Gerami et al (2015)41,; Validation subset	4. Study population includes AJCC stage III/IV lesions (13%), although analysis for only stage I/II was provided	1. Risk threshold for classification into class 1 or 2 not provided.	3,4. Multivariate models included only control for AJCC stage. Reclassification not provided	2. Evidence-based treatment or surveillance pathway using the test is not described
Zager et al (2018)42,	4. Study population includes AJCC stage III lesions (32%), although analysis for only stage I/II was provided			2. Evidence-based treatment or surveillance pathway using the test is not described
Greenhaw et al (2018)55,			3. Not compared to other prediction tools	2. Evidence-based treatment or surveillance pathway using the test is not described	1. Only 20 patients had 5-year follow-up
Keller et al (2019)56,				2: Evidence-based treatment or surveillance pathway using the test is not described	1. Unclear how many patients had 5 year follow-up

The evidence limitations stated in this table are those notable in the current review; this is not a comprehensive gaps assessment.
AJCC: American Joint Committee on Cancer.

^a Population key: 1. Intended use population unclear; 2. Clinical context is unclear; 3. Study population is unclear; 4. Study population not representative of intended use.
^b Intervention key: 1. Classification thresholds not defined; 2. Version used unclear; 3. Not intervention of interest.
^c Comparator key: 1. Classification thresholds not defined; 2. Not compared to credible reference standard; 3. Not compared to other tests in use for same purpose.
^d Outcomes key: 1. Study does not directly assess a key health outcome; 2. Evidence chain or decision model not explicated; 3. Key clinical validity outcomes not reported (sensitivity, specificity and predictive values); 4. Reclassification of diagnostic or risk categories not reported; 5. Adverse events of the test not described (excluding minor discomforts and inconvenience of venipuncture or noninvasive tests).
^e Follow-Up key: 1. Follow-up duration not sufficient with respect to natural history of disease (true-positives, true-negatives, false-positives, false-negatives cannot be determined).

Study	Selectiona	Blindingb	Delivery of Testc	Selective Reportingd	Completeness of Follow-Upe	Statisticalf
Gerami et al (2015)41,; Validation subset	2. Not consecutive or random		1. Time between collection of biopsy and DecisionDx not described	1. No registration reported	1. No description of number of samples (if any) that failed to produce results or were indeterminate	1. CIs not reported but were calculated based on data provided
Zager et al (2018)42,	2. Not consecutive or random		1. Time between collection of biopsy and DecisionDx not described	1. No registration reported	1. No description of number of samples (if any) that failed to produce results or were indeterminate
Greenhaw et al (2018)55,			1. Some samples collected after treatment	1. No registration reported
Keller et al (2019)56,				1. No registration reported		1. Estimates and CIs at year 5 were not provided.

^a Selection key: 1. Selection not described; 2. Selection not random or consecutive (ie, convenience).
^b Blinding key: 1. Not blinded to results of reference or other comparator tests.
^c Test Delivery key: 1. Timing of delivery of index or reference test not described; 2. Timing of index and comparator tests not same; 3. Procedure for interpreting tests not described; 4. Expertise of evaluators not described.
^d Selective Reporting key: 1. Not registered; 2. Evidence of selective reporting; 3. Evidence of selective publication.
^e Follow-Up key: 1. Inadequate description of indeterminate and missing samples; 2. High number of samples excluded; 3. High loss to follow-up or missing data.
^f Statistical key: 1. Confidence intervals and/or p values not reported; 2. Comparison to other tests not reported.

Ferris et al (2017) compared the accuracy of DecisionDx-Melanoma with the web-based AJCC Individualized Melanoma Patient Outcome Prediction Tool.^58, The study included 205 patients who appear to overlap with the patients in the second Gerami et al (2015) study described above. AJCC-predicted 5-year survival for each patient was categorized into low and high-risk based on both a 68% predicted 5-year survival and a 79% predicted 5-year survival. The 68% and 79% cutpoints were reported to correspond to 5-year survival in patients with stage IIA and IIB, respectively, although it is unclear whether those cutpoints were prespecified, whether they were based on internal or external estimates of risk, or whether they are commonly used in practice. The prognostic sensitivity and specificity for death (median follow-up, 7 years) of the Decision-Dx Melanoma were 78% and 69%, respectively (CIs not reported). The sensitivity and specificity for the AJCC calculator with the 79% cutpoint were 60% and 74%, respectively. The combination of the DecisionDx-Melanoma and AJCC tools had a sensitivity of 82% and specificity of 62%. The cross-classification for the DecisionDx-Melanoma and AJCC tools for 5-year OSis shown in Table 15.

Table 15. Cross-Classification for the DecisionDx-Melanoma and AJCC Tool (79% Cutpoint) for 5-Year Overall Survival

Risk Classification (DecisionDx-Melanoma vs AJCC)	N	No. of Events	5-Year Overall Survival, %
Low/low	105	9	96
Low/high	13	2	83
High/low	30	11	71
High/high	57	28	44

Adapted from Ferris et al (2017).^24,
AJCC: American Joint Committee on Cancer.

Section Summary: Clinically Valid

To use prognostic information for decision-making, performance characteristics should be consistent and precise. Two independent studies, using archived tumor specimens, have reported 5-year RFS in AJCC stage I or II patients.

If the test is to be used to select stage I and II patients for adjuvant therapy or enhanced surveillance then it should identify a group with high-risk of recurrence. Gerami et al (2015) reported RFS rates of 37% for DecisionDx class 2 (high-risk) in patients in AJCC stage I and II patients. However, Zager et al (2018) reported RFS rates of 85% (95% CI, 74% to 97%) for DecisionDx class 2 patients in AJCC stage 1 and 55% (95% CI, 44% to 69%) for DecisionDx class 2 in AJCC stage II disease. In addition, to 'rule-in' patients for additional treatment or surveillance, the test should have specificity and PPV. Zager et al (2018) and Greenhaw et al (2018) the specificities were 71% and 87% respectively while the PPV were only 48% and 24%, respectively. The low PPV suggests that the majority of patients identified as high-risk by the DecisionDx test would not develop metastasis and would be unnecessarily subjected to additional treatment or surveillance. Five-year RFS data are not available for the subgroup of patients for whom a 'rule-out' test would be relevant (class IIB through III).

If the test is to be used to select stage I and II patients who can avoid SLNB, then it should identify a group who are eligible for SLNB but have low-risk of recurrence. Gerami et al (2015) reported RFS rates of 98% in DecisionDx class 1 (low-risk) without CIs in AJCC stage I or II patients. Zager et al (2018) reported RFS rates of 96% (95% CI, 94% to 99%) for DecisionDx class 1 in patients with AJCC stage I disease and RFS rates of 74% (95% CI, 60% to 91%) for DecisionDx class 1 n patients with AJCC stage II disease. Although CIs were not available for the first study, RFS does not appear to be well-characterized in either DecisionDx risk group as evidenced by the variation in estimates across studies. These studies do not report 5-year RFS in the specific population in which the manufacturer is suggesting utility for guiding SLNB (i.e., Class 1A patients ≤55 years old who have tumors less than 2 mm deep [T1-T2]). Data on five-year RFS is not available for this target population outside of the Vetto (2019) retrospective cohort that was used to develop the target population.

Zager et al (2017) also reported that 56 of 159 (35%) patients who were DecisionDx class 1 (low-risk) were SLNB-positive and in those patients 22 recurrences (39%) occurred over 5 years.^42, If the DecisionDx test were used as a triage for SLNB, these patients would not undergo SLNB and would likely not receive adjuvant therapy, which has shown to be effective at prolonging time to recurrence in node-positive patients.

Greenhaw et al (2018) also reported that in 219 AJCC stage I patients, 201 had DecisionDx class 1 (low-risk) scores and 18 had DecisionDx class 2 (high-risk) scores. The only metastasis in stage I patients occurred in a patient with a DecisionDx class 1 score. Therefore, with respect to the proposed uses of identifying higher risk patients that should receive adjuvant therapy or enhanced surveillance, none of their stage 1 patients benefited from DecisionDx testing but 18 (8%) were incorrectly identified as high-risk for metastasis and could have received unnecessary treatment or surveillance.

Clinical Useful

A test is clinically useful if the results inform management decisions that improve the net health outcome of care. The net health outcome can be improved if patients receive correct therapy, or more effective therapy, or avoid unnecessary therapy, or avoid unnecessary testing.

Direct Evidence

Direct evidence of clinical utility is provided by studies that have compared health outcomes for patients managed with and without the test. Because these are intervention studies, the preferred evidence would be from RCTs.

No direct evidence of clinical utility was identified.

Chain of Evidence

Indirect evidence on clinical utility rests on clinical validity. If the evidence is insufficient to demonstrate test performance, no inferences can be made about clinical utility.

Four decision-impact studies have been published reporting on the impact of DecisionDx on physicians’ management decisions.^59,60,61,62, Given the lack of established clinical validity and no reported long-term outcomes of the test used to select patients for active surveillance, it is not known whether any management changes were clinically appropriate.

For the proposed use of the test as a triage for SLNB (identify patients who can avoid SLNB), performance characteristics are not well-characterized.

For the proposed use of the test as a replacement for SLNB (identify patients who are AJCC stage I/II who should receive adjuvant therapy), performance characteristics are also not well-characterized. In addition, an evidence-based management pathway would be needed to support the chain of evidence. The existing RCTs demonstrating that adjuvant therapy reduces recurrence included node positive patients.

For the proposed use of the test to identify patients who are AJCC stage I/II who should receive enhanced surveillance, there is also a lack of evidence that imaging surveillance or increased frequency of surveillance improves outcomes in stage I//I patients. The National Comprehensive Cancer Network guidelines state that imaging surveillance is not recommended for stage I-IIA and can be ‘considered’ for IIB-IV but that there is an absence of meaningful data on the association of rigorous routine surveillance imaging with improved long-term outcome for stage IIB-IIC and the recommendations regarding consideration of imaging surveillance remain controversial. While earlier detection of recurrence is thought to be beneficial because lower tumor burden and younger age are associated with improved treatment response and survival, this has not been proven and RCTs are needed to assess whether enhanced surveillance improves survival. The optimal frequency and duration of follow-up surveillance are not standardized and how the surveillance would be altered for DecisionDx class 2 patients has not be defined.

No evidence was identified that demonstrated that adjuvant therapy or increased surveillance improves net health outcomes in AJCC stage I or II patients who are DecisionDx class 2.

Section Summary: Clinically Useful

There is no direct evidence of clinical utility. A chain of evidence for clinical utility cannot be created due to lack of robust evidence of clinical validity and lack of evidence-based management pathway.

Summary of Evidence

For individuals with suspicious pigmented lesions (based on ABCDE and/or ugly duckling criteria) being considered for biopsy who receive gene expression profiling (GEP) with the DermTech Pigmented Lesion Assay to determine which lesions should proceed to biopsy, the evidence includes observational studies. Relevant outcomes are overall survival, disease-specific survival, validity, and resource utilization. The Pigmented Lesion Assay has one clinical validity study with many methodologic and reporting limitations. Therefore, performance characteristics are not well-characterized. Also, the test has not been compared with dermoscopy, another tool frequently used to make biopsy decisions. No direct evidence of clinical utility was identified. Given that the evidence is insufficient to demonstrate test performance, no inferences can be made about clinical utility through a chain of evidence. The evidence is insufficient to determine the effects of the technology on health outcomes.

For individuals who have melanocytic lesions with indeterminate histopathologic features who receive GEP with the myPath Melanoma test added to histopathology to aid in the diagnosis of melanoma, the evidence includes observational studies. Relevant outcomes are overall survival, disease-specific survival, test validity, change in disease status, treatment-related morbidity. The myPath test has one clinical validity study, which includes long-term follow-up for metastasis as the reference standard. However, it is not clear if the study population included lesions that were indeterminate following histopathology and the study had other methodologic and reporting limitations. Therefore, performance characteristics are not well-characterized. No direct evidence of clinical utility was identified. Given that the evidence is insufficient to demonstrate test performance, no inferences can be made about clinical utility through a chain of evidence. The evidence is insufficient to determine the effects of the technology on health outcomes.

For individuals with American Joint Committee on Cancer (AJCC) stage I or II cutaneous melanoma who receive GEP with the DecisionDx-Melanoma test to inform management decisions regarding enhanced surveillance, the evidence includes retrospective and prospective observational studies. Relevant outcomes are overall survival, disease-specific survival, test validity, change in disease status, resource utilization and treatment-related morbidity. The DecisionDx-Melanoma test has three independent clinical validity studies that have reported five-year recurrence-free survival (RFS) in AJCC stage I or II patients. Gerami et al (2015) reported RFS rates of 37% for DecisionDx class 2 (high-risk) in patients in AJCC stage I and II patients combined. Zager et al (2018) reported RFS rates of 85% (95% confidence interval [CI], 74% to 97%) for DecisionDx class 2 patients in AJCC stage 1 and 55% (95% CI, 44% to 69%) for DecisionDx class 2 in AJCC stage II disease. RFS does not appear to be well-characterized as evidenced by the variation in estimates across studies. This indication is to 'rule-in' patients for enhanced surveillance; therefore, specificity and positive predictive value (PPV) are key performance characteristics. Zager et al (2018) and Greenhaw et al (2018) the specificities were 71% and 87% respectively while the PPV were 48% and 24%, respectively. The PPV suggests that the majority of patients identified as high-risk by the DecisionDx test would not develop metastasis and would be unnecessarily subjected to additional surveillance. Greenhaw et al (2018) also reported that in 219 AJCC stage I patients, 201 had DecisionDx class 1 (low-risk) scores and 18 had DecisionDx class 2 (high-risk) scores. The only metastasis in stage I patients occurred in a patient with a DecisionDx class 1 score. Therefore none of their stage 1 patients benefited from DecisionDx testing but 18 (8%) were incorrectly identified as high-risk for metastasis. and could have received unnecessary surveillance. Five-year RFS data are not available for the subgroup of patients for whom a 'rule-out' test would be relevant (class IIB through III). There is no evidence that changes to the frequency and methods for surveillance improve outcomes. Given that the evidence is insufficient to demonstrate test performance and there is no evidence that changes in surveillance improve outcomes, no inferences can be made about clinical utility through a chain of evidence. The evidence is insufficient to determine the effects of the technology on health outcomes.

For individuals with AJCC stage I or II cutaneous melanoma who receive GEP with the DecisionDx-Melanoma test to inform management decisions regarding adjuvant therapy, the evidence includes retrospective and prospective observational studies. Relevant outcomes are overall survival, disease-specific survival, test validity, change in disease status, resource utilization and treatment-related morbidity. The DecisionDx-Melanoma test has three independent clinical validity studies that have reported five-year RFS in AJCC stage I or II patients. Gerami et al (2015) reported RFS rates of 37% for DecisionDx class 2 (high-risk) in patients in AJCC stage I and II patients combined. Zager et al (2018) reported RFS rates of 85% (95% CI, 74% to 97%) for DecisionDx class 2 patients in AJCC stage 1 and 55% (95% CI, 44% to 69%) for DecisionDx class 2 in AJCC stage II disease. RFS does not appear to be well-characterized as evidenced by the variation in estimates across studies. This indication is to 'rule-in' patients for adjuvant therapy; therefore, specificity and PPV are key performance characteristics. Zager et al (2018) and Greenhaw et al (2018) the specificities were 71% and 87% respectively while the PPV were 48% and 24%, respectively. The PPV suggests that the majority of patients identified as high-risk by the DecisionDx test would not develop metastasis and would be unnecessarily subjected to additional treatment. Greenhaw et al (2018) also reported that in 219 AJCC stage I patients, 201 had DecisionDx class 1 (low-risk) scores and 18 had DecisionDx class 2 (high-risk) scores. The only metastasis in stage I patients occurred in a patient with a DecisionDx class 1 score. Therefore none of their stage 1 patients benefited from DecisionDx testing but 18 (8%) were incorrectly identified as high-risk for metastasis. and could have received unnecessary treatment..There is no evidence that adjuvant therapy improves outcomes in these patients. Given that the evidence is insufficient to demonstrate test performance and there is no evidence that adjuvant therapy improves outcomes, no inferences can be made about clinical utility through a chain of evidence. The evidence is insufficient to determine the effects of the technology on health outcomes.

For individuals with cutaneous melanoma with clinically negative sentinel node basins who are being considered for sentinel lymph node biopsy (SLNB) who receive GEP with the DecisionDx-Melanoma test to determine whether to perform SLNB), the evidence includes retrospective observational studies. Relevant outcomes are overall survival, disease-specific survival, test validity, change in disease status, resource utilization and treatment-related morbidity. The DecisionDx-Melanoma test has three independent clinical validity studies that have reported five-year RFS in AJCC stage I or II patients. Gerami et al (2015) reported RFS rates of 98% in DecisionDx class 1 (low-risk) without CIs, in AJCC stage I or II patients. Zager et al (2017) reported RFS rates of 96% (95% CI, 94% to 99%) for DecisionDx class 1 in patients with AJCC stage I disease; they also reported RFS rates of 74% (95% CI, 60% to 91%) for DecisionDx class 1 in patients with AJCC stage II disease. Although CIs were not available for the first study, RFS does not appear to be well-characterized as evidenced by the variation in estimates across studies. Zager et al (2017) also reported that in 56 patients who were DecisionDx class 1 (low-risk) but SLNB-positive, 22 recurrences (39%) occurred over 5 years. If the DecisionDx test were used as a triage for SLNB, these patients would not undergo SLNB and would likely not receive adjuvant therapy, which has shown to be effective at prolonging time to recurrence in node-positive patients. Data on five-year RFS is not available for the target population (Class 1A patients ≤55 years old who have tumors less than 2 mm deep [T1-T2]) outside of the retrospective cohort that was used to identify the target population. No direct evidence of clinical utility was identified. Given that the evidence is insufficient to demonstrate test performance, no inferences can be made about clinical utility through a chain of evidence. The evidence is insufficient to determine the effects of the technology on health outcomes.

SUPPLEMENTAL INFORMATION
Practice Guidelines and Position Statements
National Comprehensive Cancer Network

The National Comprehensive Cancer Network guidelines (v.2.2020) for melanoma made the following statements on use of gene expression profiling^63,

The guidelines state the following regarding diagnostic testing for indeterminate melanocytic neoplasms following histopathology:"Melanocytic neoplasms of uncertain biologic potential present a unique challenge to pathologists and treating clinicians. Ancillary methods to aid in benign versus malignant differentiation include molecular cytogenetics (eg, comparative genomic hybridization [CGH]), fluorescence in situ hybridization [FISH]), gene expression profiling (GEP), next generation sequencing (NGS), and immunohistochemistry (IHC), among others. While limited report on the intermediate category of melanocytic neoplasia show evolutionary pathogenic genetic alteration during melanoma progression, there are insufficient data from histologically ambiguous melanocytic neoplasms."

The guidelines state the following regarding prognostic testing:

• ‘Prognostic gene expression profiling (GEP) to differentiate melanomas at low versus high risk for metastasis may provide information on individual risk of recurrence, as an adjunct to standard AJCC staging. However, the currently available prognostic molecular techniques should not replace pathologic staging procedures and the use of GEP testing, according to specific melanoma stage (before or after sentinel lymph node biopsy [SLNB]) requires further prospective investigation in large, contemporary data sets of unselected patients.’
• Commercially available GEP tests are marketed as being able to classify cutaneous melanoma into separate categories based on risk of metastasis. However it remains unclear whether these tests provide clinically actionable prognostic information when used in addition to or in comparison with known clinicopathologic factors or multivariable nomograms that incorporate patient sex, age, tumor location and thickness, …. Furthermore, the impact of these tests on treatment outcomes or follow-up schedules has not been established.’
• ‘Various (mostly retrospective) studies of prognostic GEP testing suggest its role as an independent predictor of worse outcomes, though not superior to Breslow thickness or SLN status. It remains unclear whether this GEP profile is reliably predictive of outcome… Prospective validation studies (as have been performed in breast cancer) are required to more accurately define the clinical utility of molecular testing prior to widespread implementation of GEP for prognostication of cutaneous melanoma and in particular to determine its role in guiding surveillance imaging, SLNB, and adjuvant treatment decisions.’

In 2019, the American Academy of Dermatology published guidelines of care for the management of primary cutaneous melanoma.^64, The guidelines state the following regarding GEP tests:

• Regarding diagnostic GEP tests:
  • "Diagnostic molecular techniques are still largely investigative and may be appropriate as ancillary tests in equivocal melanocytic neoplasms, but they are not recommended for routine diagnostic use in CM. These include comparative genomic hybridization, fluorescence in situ hybridization, gene expression profiling (GEP), and (potentially) next-generation sequencing."
  • "Ancillary diagnostic molecular techniques (eg, CGH, FISH, GEP) may be used for equivocal melanocytic neoplasms."
• Regarding prognostic GEP tests:
  • "...there is also insufficient evidence of benefit to recommend routine use of currently available prognostic molecular tests, including GEP, to provide more accurate prognosis beyond currently known clinicopathologic factors" (Strength of evidence: C, Level of evidence II/III)
  • "Going forward, GEP assays should be tested against all known histopathologic prognostic factors and contemporary eighth edition of AJCC CM staging to assess their additive value in prognostication."
  • "Routine molecular testing, including GEP, for prognostication is discouraged until better use criteria are defined. The application of molecular information for clinical management (eg, sentinel lymph node eligibility, follow-up, and/or therapeutic choice) is not recommended outside of a clinical study or trial."

In 2019, the National Society for Cutaneous Medicine published appropriate use criteria for the integration of diagnostic and prognostic gene expression profile assays for management of cutaneous melanoma. ^65, The criteria were developed with "unrestricted educational grants from related companies involved with these technologies". The majority of the panel members were consultants or advisors for Castle BioSciences or Myriad. The criteria were consensus-based using a modified Delphi approach. Numerous recommendations were made for each of the tests reviewed here. Some of the recommendations are as follows:

• Using PLA test for patients with atypical lesions requiring assessment beyond visual inspection to help in selection for biopsy (B = Inconsistent or limited quality patient-oriented evidence)
• Using myPath for differentiation of a nevus from melanoma in an adult patient when the morphologic findings are ambiguous by light microscopic parameters (A = Consistent, good-quality patient-oriented evidence)
• Using DecisionDx by integrating results into the decision to adjust follow up regimens or to assess need for imaging (B = Inconsistent or limited quality patient-oriented evidence)
• Using DecisionDx by integrating results into subsequent management of patients:

  - Who are sentinel node negative (A = Consistent, good-quality patient-oriented evidence)

  - Who are in AJCC “low risk” categories: (Thin (<1 mm), Stage I-IIA, SLNBx-) (B= Inconsistent or limited quality patient-oriented evidence)

• Using DecisionDx by integrating 31GEP results as a criteria for inclusion in a chemotherapy regimen (C = Consensus, disease-oriented evidence, usual practice, expert opinion, or case series)

Not applicable.

Ongoing and Unpublished Clinical Trials

Some currently unpublished trials that might influence this policy are listed in Table 16.

Table 16. Summary of Key Trials

NCT No.	Trial Name	Planned Enrollment	Completion Date
Ongoing
NCT02355574a	An Ongoing, 5-year Post Market Study to Track Clinical Application of DecisionDx-Melanoma Gene Expression Profile (GEP) Assay Results and the Impact on Patient Outcomes and Health Economics	1672	Jun 2024
NCT02355587a	An Open, 5-year Registry Study to Track Clinical Application of DecisionDx-Melanoma Gene Expression Profile Assay Results and Associated Patient Outcomes	5000	Feb 2024

NCT: national clinical trial.
^a Denotes industry-sponsored or cosponsored trial.]
________________________________________________________________________________________

Horizon BCBSNJ Medical Policy Development Process:

This Horizon BCBSNJ Medical Policy (the “Medical Policy”) has been developed by Horizon BCBSNJ’s Medical Policy Committee (the “Committee”) consistent with generally accepted standards of medical practice, and reflects Horizon BCBSNJ’s view of the subject health care services, supplies or procedures, and in what circumstances they are deemed to be medically necessary or experimental/ investigational in nature. This Medical Policy also considers whether and to what degree the subject health care services, supplies or procedures are clinically appropriate, in terms of type, frequency, extent, site and duration and if they are considered effective for the illnesses, injuries or diseases discussed. Where relevant, this Medical Policy considers whether the subject health care services, supplies or procedures are being requested primarily for the convenience of the covered person or the health care provider. It may also consider whether the services, supplies or procedures are more costly than an alternative service or sequence of services, supplies or procedures that are at least as likely to produce equivalent therapeutic or diagnostic results as to the diagnosis or treatment of the relevant illness, injury or disease. In reaching its conclusion regarding what it considers to be the generally accepted standards of medical practice, the Committee reviews and considers the following: all credible scientific evidence published in peer-reviewed medical literature generally recognized by the relevant medical community, physician and health care provider specialty society recommendations, the views of physicians and health care providers practicing in relevant clinical areas (including, but not limited to, the prevailing opinion within the appropriate specialty) and any other relevant factor as determined by applicable State and Federal laws and regulations.

___________________________________________________________________________________________________________________________

Index:
Gene Expression Profiling for Cutaneous Melanoma
Cutaneous Melanoma, Gene Expression Profiling for
Melanoma, Cutaneous, Gene Expression Profiling for
DermTech Pigmented Lesion Assay
Pigmented Lesion Assay
myPath Melanoma Test
DecisionDx-Melanoma Test

References:
1. Chang AE, Karnell LH, Menck HR. The National Cancer Data Base report on cutaneous and noncutaneous melanoma: a summary of 84,836 cases from the past decade. The American College of Surgeons Commission on Cancer and the American Cancer Society. Cancer. Oct 15 1998; 83(8): 1664-78. PMID 9781962

2. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018. CA Cancer J Clin. Jan 2018; 68(1): 7-30. PMID 29313949

3. Gilchrest BA, Eller MS, Geller AC, et al. The pathogenesis of melanoma induced by ultraviolet radiation. N Engl J Med. Apr 29 1999; 340(17): 1341-8. PMID 10219070

4. Gandini S, Sera F, Cattaruzza MS, et al. Meta-analysis of risk factors for cutaneous melanoma: III. Family history, actinic damage and phenotypic factors. Eur J Cancer. Sep 2005; 41(14): 2040-59. PMID 16125929

5. Caini S, Gandini S, Sera F, et al. Meta-analysis of risk factors for cutaneous melanoma according to anatomical site and clinico-pathological variant. Eur J Cancer. Nov 2009; 45(17): 3054-63. PMID 19545997

6. Goldstein AM, Chan M, Harland M, et al. Features associated with germline CDKN2A mutations: a GenoMEL study of melanoma-prone families from three continents. J Med Genet. Feb 2007; 44(2): 99-106. PMID 16905682

7. Wendt J, Rauscher S, Burgstaller-Muehlbacher S, et al. Human Determinants and the Role of Melanocortin-1 Receptor Variants in Melanoma Risk Independent of UV Radiation Exposure. JAMA Dermatol. Jul 01 2016; 152(7): 776-82. PMID 27050141

8. Wiesner T, Obenauf AC, Murali R, et al. Germline mutations in BAP1 predispose to melanocytic tumors. Nat Genet. Aug 28 2011; 43(10): 1018-21. PMID 21874003

9. Chen T, Fallah M, Forsti A, et al. Risk of Next Melanoma in Patients With Familial and Sporadic Melanoma by Number of Previous Melanomas. JAMA Dermatol. Jun 2015; 151(6): 607-15. PMID 25671687

10. Jiang AJ, Rambhatla PV, Eide MJ. Socioeconomic and lifestyle factors and melanoma: a systematic review. Br J Dermatol. Apr 2015; 172(4): 885-915. PMID 25354495

11. Abbasi NR, Shaw HM, Rigel DS, et al. Early diagnosis of cutaneous melanoma: revisiting the ABCD criteria. JAMA. Dec 08 2004; 292(22): 2771-6. PMID 15585738

12. Grob JJ, Bonerandi JJ. The 'ugly duckling' sign: identification of the common characteristics of nevi in an individual as a basis for melanoma screening. Arch Dermatol. Jan 1998; 134(1): 103-4. PMID 9449921

13. Wilson RL, Yentzer BA, Isom SP, et al. How good are US dermatologists at discriminating skin cancers? A number-needed-to-treat analysis. J Dermatolog Treat. Feb 2012; 23(1): 65-9. PMID 21756146

14. National Center for Biotechnology Information. PRAME preferentially expressed antigen in melanoma. 2018; https://www.ncbi.nlm.nih.gov/gene/23532. Accessed March 30, 2018.

15. DermTech. Pigmented Lesion Assay: Non-invasive gene expression analysis of pigmented skin lesions. Performance and Development Notes. 2015; http://dermtech.com/wp-content/uploads/2015/10/White-Paper- DermTech-Melanoma-Assay-.pdf. Accessed March 16, 2018.

16. Wachsman W, Morhenn V, Palmer T, et al. Noninvasive genomic detection of melanoma. Br J Dermatol. Apr 2011; 164(4): 797-806. PMID 21294715

17. Gerami P, Alsobrook JP, Palmer TJ, et al. Development of a novel noninvasive adhesive patch test for the evaluation of pigmented lesions of the skin. J Am Acad Dermatol. Aug 2014; 71(2): 237-44. PMID 24906614

18. Gerami P, Yao Z, Polsky D, et al. Development and validation of a noninvasive 2-gene molecular assay for cutaneous melanoma. J Am Acad Dermatol. Jan 2017; 76(1): 114-120.e2. PMID 27707590

19. Vestergaard ME, Macaskill P, Holt PE, et al. Dermoscopy compared with naked eye examination for the diagnosis of primary melanoma: a meta-analysis of studies performed in a clinical setting. Br J Dermatol. Sep 2008; 159(3): 669-76. PMID 18616769

20. Murzaku EC, Hayan S, Rao BK. Methods and rates of dermoscopy usage: a cross-sectional survey of US dermatologists stratified by years in practice. J Am Acad Dermatol. Aug 2014; 71(2): 393-5. PMID 25037790

21. Engasser HC, Warshaw EM. Dermatoscopy use by US dermatologists: a cross-sectional survey. J Am Acad Dermatol. Sep 2010; 63(3): 412-9, 419.e1-2. PMID 20619490

22. Bossuyt PM, Irwig L, Craig J, et al. Comparative accuracy: assessing new tests against existing diagnostic pathways. BMJ. May 06 2006; 332(7549): 1089-92. PMID 16675820

23. Ferris LK, Gerami P, Skelsey MK, et al. Real-world performance and utility of a noninvasive gene expression assay to evaluate melanoma risk in pigmented lesions. Melanoma Res. Oct 2018; 28(5): 478-482. PMID 30004988

24. Ferris LK, Jansen B, Ho J, et al. Utility of a Noninvasive 2-Gene Molecular Assay for Cutaneous Melanoma and Effect on the Decision to Biopsy. JAMA Dermatol. Jul 01 2017; 153(7): 675-680. PMID 28445578

25. Myriad. n.d. Understanding the myPath Melanoma Results; https://mypathmelanoma.com/about-mypath- melanoma/understanding-the-mypath-melanoma-results/. Accessed March 30, 2018.

26. Clarke LE, Warf MB, Flake DD, et al. Clinical validation of a gene expression signature that differentiates benign nevi from malignant melanoma. J Cutan Pathol. Apr 2015; 42(4): 244-52. PMID 25727210

27. Clarke LE, Flake DD, Busam K, et al. An independent validation of a gene expression signature to differentiate malignant melanoma from benign melanocytic nevi. Cancer. Feb 15 2017; 123(4): 617-628. PMID 27768230

28. Reimann JDR, Salim S, Velazquez EF, et al. Comparison of melanoma gene expression score with histopathology, fluorescence in situ hybridization, and SNP array for the classification of melanocytic neoplasms. Mod Pathol. Nov 2018; 31(11): 1733-1743. PMID 29955141

29. Gaiser T, Kutzner H, Palmedo G, et al. Classifying ambiguous melanocytic lesions with FISH and correlation with clinical long-term follow up. Mod Pathol. Mar 2010; 23(3): 413-9. PMID 20081813

30. Vergier B, Prochazkova-Carlotti M, de la Fouchardiere A, et al. Fluorescence in situ hybridization, a diagnostic aid in ambiguous melanocytic tumors: European study of 113 cases. Mod Pathol. May 2011; 24(5): 613-23. PMID 21151100

31. Ko JS, Clarke LE, Minca EC, et al. Correlation of melanoma gene expression score with clinical outcomes on a series of melanocytic lesions. Hum Pathol. Apr 2019; 86: 213-221. PMID 30566894

32. Clarke LE, Pimentel JD, Zalaznick H, et al. Gene expression signature as an ancillary method in the diagnosis of desmoplastic melanoma. Hum Pathol. Dec 2017; 70: 113-120. PMID 29079183

33. Minca EC, Al-Rohil RN, Wang M, et al. Comparison between melanoma gene expression score and fluorescence in situ hybridization for the classification of melanocytic lesions. Mod Pathol. Aug 2016; 29(8): 832-43. PMID 27174586

34. Ko JS, Matharoo-Ball B, Billings SD, et al. Diagnostic Distinction of Malignant Melanoma and Benign Nevi by a Gene Expression Signature and Correlation to Clinical Outcomes. Cancer Epidemiol Biomarkers Prev. Jul 2017; 26(7): 1107-1113. PMID 28377414

35. Cockerell C, Tschen J, Billings SD, et al. The influence of a gene-expression signature on the treatment of diagnostically challenging melanocytic lesions. Per Med. Mar 2017; 14(2): 123-130. PMID 28757886

36. Cockerell CJ, Tschen J, Evans B, et al. The influence of a gene expression signature on the diagnosis and recommended treatment of melanocytic tumors by dermatopathologists. Medicine (Baltimore). Oct 2016; 95(40): e4887. PMID 27749545

37. Gershenwald JES, R.A.; Hess, K.R.; et al. Melanoma of the Skin. Chicago, IL: American Joint Committee on Cancer; 2017.

38. Eggermont AM, Chiarion-Sileni V, Grob JJ, et al. Prolonged Survival in Stage III Melanoma with Ipilimumab Adjuvant Therapy. N Engl J Med. Nov 10 2016; 375(19): 1845-1855. PMID 27717298

39. Weber J, Mandala M, Del Vecchio M, et al. Adjuvant Nivolumab versus Ipilimumab in Resected Stage III or IV Melanoma. N Engl J Med. Nov 09 2017; 377(19): 1824-1835. PMID 28891423

40. Long GV, Hauschild A, Santinami M, et al. Adjuvant Dabrafenib plus Trametinib in Stage III BRAF-Mutated Melanoma. N Engl J Med. Nov 09 2017; 377(19): 1813-1823. PMID 28891408

41. 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 01 2015; 21(1): 175-83. PMID 25564571

42. Zager JS, Gastman BR, Leachman S, et al. Performance of a prognostic 31-gene expression profile in an independent cohort of 523 cutaneous melanoma patients. BMC Cancer. Feb 05 2018; 18(1): 130. PMID 29402264

43. Wrightson WR, Wong SL, Edwards MJ, et al. Complications associated with sentinel lymph node biopsy for melanoma. Ann Surg Oncol. Jul 2003; 10(6): 676-80. PMID 12839853

44. Soong SJ, Ding S, Coit DG, et al. AJCC: Individualized melanoma patient outcome prediction tools. n.d.; http://www.melanomaprognosis.net/. Accessed March 21, 2018.

45. Callender GG, Gershenwald JE, Egger ME, et al. A novel and accurate computer model of melanoma prognosis for patients staged by sentinel lymph node biopsy: comparison with the American Joint Committee on Cancer model. J Am Coll Surg. Apr 2012; 214(4): 608-17; discussion 617-9. PMID 22342785

46. Dicker TJ, Kavanagh GM, Herd RM, et al. A rational approach to melanoma follow-up in patients with primary cutaneous melanoma. Scottish Melanoma Group. Br J Dermatol. Feb 1999; 140(2): 249-54. PMID 10233217

47. Garbe C, Paul A, Kohler-Spath H, et al. Prospective evaluation of a follow-up schedule in cutaneous melanoma patients: recommendations for an effective follow-up strategy. J Clin Oncol. Feb 01 2003; 21(3): 520-9. PMID 12560444

48. Faries MB, Steen S, Ye X, et al. Late recurrence in melanoma: clinical implications of lost dormancy. J Am Coll Surg. Jul 2013; 217(1): 27-34; discussion 34-6. PMID 23643694

49. Hsueh EC, DeBloom JR, Lee J, et al. Interim analysis of survival in a prospective, multi-center registry cohort of cutaneous melanoma tested with a prognostic 31-gene expression profile test. J Hematol Oncol. Aug 29 2017; 10(1): 152. PMID 28851416

50. Podlipnik S, Carrera C, Boada A, et al. Early outcome of a 31-gene expression profile test in 86 AJCC stage IB-II melanoma patients. A prospective multicentre cohort study. J Eur Acad Dermatol Venereol. May 2019; 33(5): 857-862. PMID 30702163

51. Gastman BR, Gerami P, Kurley SJ, et al. Identification of patients at risk of metastasis using a prognostic 31-gene expression profile in subpopulations of melanoma patients with favorable outcomes by standard criteria. J Am Acad Dermatol. Jan 2019; 80(1): 149-157.e4. PMID 30081113

52. Gastman BR, Zager JS, Messina JL, et al. Performance of a 31-gene expression profile test in cutaneous melanomas of the head and neck. Head Neck. Apr 2019; 41(4): 871-879. PMID 30694001

53. Vetto JT, Hsueh EC, Gastman BR, et al. Guidance of sentinel lymph node biopsy decisions in patients with T1-T2 melanoma using gene expression profiling. Future Oncol. Apr 2019; 15(11): 1207-1217. PMID 30691297

54. Marks, Etan et al. Establishing an evidence-based decision point for clinical use of the 31-gene expression profile test in cutaneous melanoma. SKIN The Journal of Cutaneous Medicine, [S.l.], v. 3, n. 4, p. 239-249, july 2019. ISSN 2574-1624. Available at: <https://jofskin.org/index.php/skin/article/view/628>. Date accessed: 25 mar. 2020. doi:https://doi.org/10.25251/skin.3.4.2.

55. Greenhaw BN, Zitelli JA, Brodland DG. Estimation of Prognosis in Invasive Cutaneous Melanoma: An Independent Study of the Accuracy of a Gene Expression Profile Test. Dermatol Surg. Dec 2018; 44(12): 1494-1500. PMID 29994951

56. Keller J, Schwartz TL, Lizalek JM, et al. Prospective validation of the prognostic 31-gene expression profiling test in primary cutaneous melanoma. Cancer Med. May 2019; 8(5): 2205-2212. PMID 30950242

57. 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): 780-5.e3. PMID 25748297

58. Ferris LK, Farberg AS, Middlebrook B, et al. Identification of high-risk cutaneous melanoma tumors is improved when combining the online American Joint Committee on Cancer Individualized Melanoma Patient Outcome Prediction Tool with a 31-gene expression profile-based classification. J Am Acad Dermatol. May 2017; 76(5): 818-825.e3. PMID 28110997

59. 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. Sep 2016; 32(9): 1599-604. PMID 27210115

60. Farberg AS, Glazer AM, White R, et al. Impact of a 31-gene Expression Profiling Test for Cutaneous Melanoma on Dermatologists' Clinical Management Decisions. J Drugs Dermatol. May 01 2017; 16(5): 428-431. PMID 28628677

61. Schuitevoerder D, Heath M, Cook RW, et al. Impact of Gene Expression Profiling on Decision-Making in Clinically Node Negative Melanoma Patients after Surgical Staging. J Drugs Dermatol. Feb 01 2018; 17(2): 196-199. PMID 29462228

62. Dillon LD, Gadzia JE, Davidson RS, et al. Prospective, multicenter clinical impact evaluation of a 31-gene expression profile test for management of melanoma patients. Skin. 2018;2(2):111-121.

63. National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines in Oncology. Cutaneous Melanoma. Version 2.2020. Accessed April 2020.

64. Swetter SM, Tsao H, Bichakjian CK, et al. Guidelines of care for the management of primary cutaneous melanoma. J Am Acad Dermatol. Jan 2019; 80(1): 208-250. PMID 30392755

65. Berman, et.al. Appropriate Use Criteria for the Integration of Diagnostic and Prognostic Gene Expression Profile Assays into the Management of Cutaneous Malignant Melanoma: An Expert Panel Consensus-Based Modified Delphi Process Assessment. SKIN. 2019; 3(5):291-298.

0089U
81479
81599
84999