Economic Analysis of a Noninvasive Molecular Pathologic Assay for Pigmented Skin Lesions

This economic analysis assesses the costs of use of the pigmented lesion assay vs visual and histopathologic assessment in model inputs of patients with primary pigmented lesions suggestive of melanoma.

Key Points

Question

What are the economic implications of using a noninvasive gene expression test (the pigmented lesion assay) to rule out melanoma and the need for surgical biopsy of atypical pigmented lesions suggestive of melanoma?

Findings

In this cost-savings analysis of the pigmented lesion assay using model inputs for patients with pigmented lesions suggestive of melanoma, a $447 (47%) cost reduction per assessed pigmented lesion vs the current histopathologic standard of care was achievable if the assay was priced at $500. Savings were driven by reduction in initial biopsies and excisions and reduced stage-related treatment costs from missing fewer melanomas.

Meaning

These findings suggest that the pigmented lesion assay reduces cost and may improve the care of patients with primary pigmented skin lesions suggestive of melanoma.

Abstract

Importance

A recently described noninvasive gene expression test (the pigmented lesion assay [PLA]) with adhesive patch–based sampling has the potential to rule out melanoma and the need for surgical biopsy of pigmented lesions suggestive of melanoma with a negative predictive value of 99% compared with 83% for the histopathologic standard of care. The cost implications of using this molecular test vs visual assessment followed by biopsy and histopathologic assessment (VAH) have not been evaluated.

Objective

To determine potential cost savings of PLA use vs the VAH pathway.

Design, Setting, and Participants

This health economic analysis performed from a US payer perspective was based on consensus treatment guidelines and fee schedules from the Centers for Medicare & Medicaid Services. Data for model input were derived from routine use of the test in US dermatology practices and literature. Participants included patients with primary cutaneous pigmented lesions suggestive of melanoma. Data were analyzed from February 8 to December 1, 2017.

Main Outcomes and Measures

The primary analysis consisted of the relative reduction in costs of diagnostic surgical procedures for PLA vs VAH management. Additional analyses included stage-related treatment costs associated with delays in diagnosis.

Results

In the cost analysis for this economic model, the relative reduction in surgical procedure costs (biopsy and subsequent excision), assuming $0 for the PLA to facilitate multiple comparison scenarios, was −$395 compared with VAH. The relative reduction in stage-related treatment costs associated with the PLA was −$433 compared with VAH, primarily associated with avoidance of delays due to false-negative diagnoses. Surveillance costs were reduced by −$119 with the PLA. The total cost of fully adjudicating a lesion suggestive of melanoma by VAH was $947. At a mean selling price reference point for PLA of $500, cost savings of $447 (47%) per lesion tested could be realized.

Conclusions and Relevance

The results of this analysis suggest that the PLA reduces cost and may improve the care of patients with primary pigmented skin lesions suggestive of melanoma.

Introduction

Management of atypical pigmented lesions involves ruling out melanoma via visual assessment, followed by surgical biopsy and histopathologic assessment (VAH).^1,2,3 The goal of this assessment is to identify melanomas at their earliest stages (in situ or stage I) when a high cure rate is possible by wide excision.⁴ Although the purpose of the VAH pathway is to rule out melanoma, the poor performance metrics of this diagnostic pathway lead to a low negative predictive value (NPV) for early-stage disease. The low specificity of visual assessment (3.7%-32.0%) results in a high number of lesions with false-positive biopsy results.^{5,6,7,8,9,10,11} Therefore, the primary and difficult role of histopathologic assessment in this setting is to identify the small number of true-positive lesions from a large pool, including a large number of false-positive lesions. However, significant overlap in the histopathologic diagnostic criteria exists between atypical nevi and early-stage melanoma, invariably leading to false-negative diagnoses and a relatively low sensitivity of histopathologic assessment (81%-84%).^12,13,14 With the prevalence of early-stage melanoma in biopsy specimens at approximately 6% and ranging from 2% to 10% in many settings,^1,2,14,15,16 the NPV of the surgical biopsy plus histopathologic diagnostic paradigm is unexpectedly low in most settings. In a study by Malvehy et al,¹⁴ 206 cases of melanoma in situ and stage IA invasive melanoma (thickness <0.75 mm) were diagnosed with a sensitivity of 81%, a specificity of 10%, and an NPV of 83%.

This low NPV for the current standard of care pathway is accompanied by a high number of unnecessary surgical procedures driven by the poor specificity of visual assessment.⁸ The mean number of surgical biopsies needed to identify 1 melanoma (number needed to biopsy [NNB]) is approximately 20 and ranges from 8 to 30 depending on the setting.^{9,10,11,12,13,14} Conservative management of biopsied lesions with atypia and positive margins leads to a high number of subsequent unnecessary excisions with margins.^{13,14,15,16,17} Approximately 5.2 excisions with margins are performed per melanoma identified.^{13,17,18,19,20,21,22} Less than 1.0% of lesions with atypia and positive margins that undergo excision are diagnostically upgraded to melanoma.^12,13 This notion that the current pathway has a significant number of unnecessary surgical procedures is also supported by other investigators,^{13,14,15,16,17,21,23} who recently found that more than 90% of skin biopsies to rule out melanoma were attributed to benign and low-risk lesions. Approximately 3.0 million surgical biopsies and 780 000 excisions are performed in the United States annually to find approximately 150 000 in situ and invasive melanomas as part of the current diagnostic pathway for atypical pigmented lesions.^8,9,23,24,25 These findings stress the need for cost-effective tools that may improve management of pigmented lesions.

One such tool, the pigmented lesion assay (PLA), is a gene expression test that helps rule out melanoma and the need for a surgical biopsy of atypical pigmented lesions. The PLA is based on a new platform technology for noninvasive genomic testing of the skin, which allows the analysis of samples collected from adhesive patches.^{11,22,24,25,26,27} In contrast to the current VAH pathway, the PLA has a very high NPV (>99%) that reduces the probability of missing melanomas from about 17% using the current standard of care (NPV of 83%) to less than 1%.^26,27 In addition, the noninvasive sampling leads to dramatic reduction in surgical biopsies and subsequent excisions. The aim of this study was to demonstrate the economic implications of the noninvasive PLA compared with the current standard of care, the VAH pathway.

Methods

The Figure depicts the decision paradigm used in the model. The PLA and the VAH rely on visual assessment to identify lesions suggestive of melanoma with 1 or more ABCDE (asymmetry, border, color, diameter, and evolution) criteria. The PLA provides a dichotomous result (positive or negative), with positive test results entering the surgical biopsy pathway. For the VAH, all (100%) lesions suggestive of melanoma enter the biopsy pathway. Management after biopsy is based on histopathologic diagnosis. All studies used for the economic analysis were approved by the Western-Copernicus Group independent review board and were conducted in accordance with the Declaration of Helsinki principles. Informed consent was obtained for all the underlying studies required to conduct this analysis.

Figure. Comparison of the Visual and Histopathologic Assessment (VAH) and the Pigmented Lesion Assay (PLA) Pathways.

The diagram provides an overview on modalities physicians can select to rule out melanoma in pigmented lesions suggestive of melanoma with 1 or more ABCDE (asymmetry, border, color, diameter, and evolution) risk factors when the VAH and the PLA are available. NPV indicates negative predictive value.

^aNegative results are found for 87% of PLA tests, dramatically reducing the number of surgical biopsies and follow-up excisions. The high NPV reduces the probability of missing melanomas to less than 1% compared with 17% for the VAH.

Data were analyzed from February 8 to December 1, 2017. An economic model was constructed from the US payer perspective²⁸ aimed at assessing the initial surgical costs associated with the VAH. For this purpose, data were presented with (1) a theoretical zero cost ($0) relative reference point for the PLA to facilitate assessment of various cost-saving scenarios and (2) pricing of the PLA at $500, close to the expected mean selling price. Direct medical costs for initial surgical management included office visits, biopsy procedures, histopathologic evaluation and special stains, and management of repairs and complications. Unit costs were based on the Centers for Medicare & Medicaid Services reimbursement rates for the corresponding Current Procedural Terminology codes.²⁸ All costs were adjusted to 2017 US dollars. Primary cost outcomes included were initial surgical procedures (biopsy and excisions), stage-related melanoma treatment costs, indirect costs, and costs related to patient experience (quality of life).^{20,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45}

Key diagnostic test input variables to compare costs for initial surgical procedures include the sensitivity and specificity of the modalities, the type of initial biopsy procedure, and the excision rate for histopathologically atypical and/or ambiguous lesions with positive margins. For the PLA, the sensitivity and specificity were derived from a retrospective medical record review of real-world use of 280 PLA-negative and 39 PLA-positive cases at 2 dermatology sites to maximize clinical relevance of the data obtained (Table 1). For the VAH, a specificity of 32% was used based on a utility study by Ferris et al¹¹ that compared the PLA with the VAH in 60 pigmented lesions. The sensitivity of the VAH is well published in the literature.^{1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,46} However, to minimize bias and to facilitate comparisons, the concordance of histopathologic diagnoses at the primary site against consensus diagnoses obtained by a panel of 3 dermatopathologists in 128 cases of in situ and stage I invasive melanoma of our own archival cases was evaluated in this study. This review demonstrated a sensitivity of 84% for the VAH for representative early-stage lesions. A tangential biopsy (shave or scoop), as performed most often in the United States, was assumed at a rate of 70%, with the remaining 30% assumed to be excisional or punch procedures.³¹ The analysis includes the incidence of procedure-related complications, such as infection and bleeding.^31,32,33 A literature review indicated that a rate of excision after initial biopsy of atypia with positive tissue margins of 20% was suitable.^20,34,35,36

Table 1. Real-world PLA Results at 2 US Dermatology Sites^a.

PLA Result	Melanoma Cases, No.	Nonmelanoma Cases, No.
Positive (n = 39)	14 (8 in situ and 6 invasive)	25 (14 [56%] atypical nevi; 2 [8%] nonmelanocytic lesions)
Negative (n = 280)	1b	279 (2 follow-up biopsies with histopathologic findings of nonmelanoma)c

Abbreviation: PLA, pigmented lesion assay.

^aData were obtained from review of medical records, including pathology reports of positive test results and 3- to 6-month follow-up procedures for negative test results. All 39 (100%) of PLA-positive cases underwent surgical biopsy to establish a histopathologic diagnosis. Two hundred seventy-seven (99%) of PLA-negative cases underwent clinical monitoring.

^bOne PLA-negative case underwent surgical biopsy on the same day the PLA was performed.

^cAssumes for this analysis that tests with no biopsy performed at the 3- to 6-month follow-up visits are nonmelanoma.

To assess stage-related treatment costs, the false-negative rate of the PLA and VAH were used to establish the proportion of patients with delayed diagnosis. A delay of at least 3 months in diagnosing melanoma was assumed based on guidelines for interval surveillance recommended by the National Comprehensive Cancer Network.³⁰ The effects on clinical stage of delayed diagnoses were derived from Losina et al³⁷ in a model that suggests that the rate of progression of an undiagnosed lesion to a more advanced stage is 10% per year.⁴⁷ Distribution and overall survival of melanoma by stage is based on population statistics reported by the American Cancer Society.⁴⁵

Stage-related costs attributable to melanoma after initial surgical management varied from a low of $6041 for management of stage 0 disease (primarily surveillance of new lesions) to more than $200 000 for management of stage IV disease (systemic therapy, including recently approved options and end-of-life care).^39,40 Indirect costs were estimated based on assumptions made on the time to attend a clinic visit, including commute time (0.75-2.00 hours), with a unit cost of $26 per hour.⁴⁰ King et al⁴⁴ reported disutilities for tangential biopsy procedures at −0.001, excision procedures at −0.004, and subsequent repair procedures at −0.021. The model multiplied mean durations of overall survival by the utility for each stage, weighted by the proportion of patients diagnosed in each stage to derive mean melanoma-related quality-adjusted life-years. A univariate sensitivity analysis was performed on key variables based on published estimates with 95% CIs of −33% to 33%.²⁸

Results

Economic Model Base Case

The Figure compares how use of the PLA and the VAH compare at the decision diagram level. Key economic model input variables and base-case results are summarized in Table 2 and Table 3. For the base-case model, we assumed the pretest probability of melanoma was 6% (range, 2%-10%) and that 56% of lesions evaluated would be considered clinically suggestive enough to recommend biopsy.^1,2,13,22

Table 2. Input Variables of Cost and Test Characteristics and Data Sources.

Variable	Base-Case Analysis	Source
Pretest probability of melanoma, %	6	Friedman et al,1 2009; Schäfer et al,2 2006; Malvehy et al,14 2014; Lott et al,15 2018; Hodis et al,16 2012
Sensitivity of VAH, %	84	128 cases in present study
Specificity of VAH, %	32	Ferris et al,11 2017
Fold change in sensitivity of PLA to detect melanoma	1.11	Table 1 in present study
Fold change in specificity of PLA to detect melanoma	2.87	Table 1 in present study
Surgical procedures, %
Initial surgical biopsy	56	Tong et al,33 2016
Proportion of initial biopsy tangential (shave or scoop)	70	Tong et al,33 2016
Excision with atypia and positive margins	20	Strazzula et al,20 2014; Grelck et al,34 2013; Fleming et al,35 2016
Annual rate of progression to next stage if undetected, %	10	Losina et al,37 2007
Probability of residual tumor requiring change in therapy, %	6	Hieken et al,38 2013
Costs, $
PLA test	0/500	Present study
Office visits
For diagnosis	167	CMS28
For surveillance	138	CMS28
Special stain for pathologic reading	115	CMS28
Additional treatment for delayed diagnosis	5040	Losina et al,37 2007; Hieken et al,38 2013
Melanoma care after surgery (true-positive result)	39 719	Styperek and Kimball,39 2012; Alexandrescu,40 2009
Wage per hour	26	US Bureau of Labor Statistics41
Disutilities
Benign lesion	1.000	King et al,44 2011
Melanoma
Stage I	0.809	King et al,44 2011
Stage II	0.802	King et al,44 2011
Stage III	0.703	King et al,44 2011
Stage IV	0.796	King et al,44 2011
Surgical biopsy	−0.001	Seidler et al,43 2009
Excision and repair	−0.025	Seidler et al,43 2009

Abbreviations: CMS, Centers for Medicare & Medicaid Services; PLA, pigmented lesion assay; VAH, visual and histopathologic assessment.

Table 3. Base-Case Results.

Variable	PLA	VAH	Difference
Rate of procedures, %
Initial biopsy	13.3	69.0	−55.7
Excision	7.5	18.8	−11.2
Any procedure	13.7	69.9	−56.2
Efficiency measures
NNB	2.7	15.7	−13.0
Excisions per melanoma	1.18	2.49	−1.31
Proportion of patients with immediate diagnosis, %	75.5	68.0	7.5
Direct medical costs, $
Initial biopsy	76	396	−320
Subsequent excisions	50	125	−75
Direct cost of initial surgical procedures, $	126	521	−395
Office visits
For diagnosis	167	167	0
For surveillance	19	138	−119
Treatment costs	1616	2049	−433
Indirect costs, $	54	101	−47
Total, $	1982	2976	−994
QALYs	16.743	16.707	0.036

Abbreviations: NNB, number needed to biopsy; PLA, pigmented lesion assay; QALY, quality-adjusted life-year; VAH, visual and histopathologic assessment.

The proportion of patients undergoing initial surgical biopsy of any type was 69.0% with the VAH vs 13.3% with the PLA. The improved accuracy of the PLA reduces the number of patients undergoing subsequent excision for melanoma from 18.8% with the VAH to 7.5% with the PLA. The NNB dropped from 15.7 with VAH to 2.7 with the PLA. This result aligns well with the NNB of 2.8 found at the 2 dermatology clinic sites assessed. With the VAH, 68.0% of patients with melanoma had an immediate and accurate diagnosis of melanoma compared with 75.5% of patients undergoing PLA assessment, such that 7.5% of patients would avoid a delayed diagnosis and progress to a more advanced stage.

Cost Differential PLA vs VAH

Assuming a theoretical $0 PLA cost reference point to facilitate comparisons of multiple scenarios, costs associated with initial surgical biopsies were lower for the PLA ($76) compared with the VAH ($396) (relative difference, −$320) (Table 3). The cost of excisional procedures was also lower with the PLA ($50) vs the VAH ($125), demonstrating that the clinical performance of the PLA reduced overall costs of surgical procedures associated with the assessment of pigmented lesions (PLA vs VAH, −$395). No difference existed in patient visits for initial diagnostic evaluation, but the PLA had a lower cost for subsequent office visits for surveillance (PLA vs VAH, −$119). Because of the high treatment costs for stage-related management of melanoma, particularly with newer targeted therapies, avoidance of diagnostic delays significantly affects treatment costs. Lower false-negative diagnoses with the PLA reduces these treatment costs by $433 (PLA, $1616 per melanoma; VAH, $2049 per melanoma).³⁹ Indirect costs from lost employee wages and productivity were $101 for VAH and $54 for the PLA. Applying the utility estimates by melanoma stage from King et al,⁴⁴ the difference in quality-adjusted life-years was 0.016 (VAH, 16.707; PLA, 16.743). Use of the PLA at an assumed mean selling price reference point of $500 leads to $447 in direct cost reductions relative to the VAH for each evaluated primary pigmented skin lesion clinically suggestive of melanoma.

Sensitivity Analysis

The PLA pathway was dominant compared with the VAH pathway in all sensitivity analyses (Table 4), meaning it reduced total costs and improved the patient experience. The most influential variables on costs were the estimate of increased specificity for PLA relative to VAH (range, −$1014 to −$457) and melanoma treatment costs. The probabilistic sensitivity analysis showed that the incremental costs and quality-adjusted survival points always fell in the bottom right quadrant of the cost-effectiveness plane. The PLA was a less costly and more effective strategy in all simulations regardless of which utilities for melanoma stage were applied.

Table 4. One-way Deterministic Sensitivity Analysis.

Variable	Base-Case Finding	Range	PLA vs VAH Cost-effectiveness Plane, $
			Left Quadrant	Right Quadrant	Difference
Fold increase in specificity of PLA over VAH (%)	2.87 (91)	1.92 (73) to 3.0 (100)	−457	−1014	557
Cost of melanoma treatment for true-positive cases, $	39 179	27 009 to 52 827	−992	−1428	436
Rate of excision after initial biopsy, %	20	13.4 to 26.6	−776	−1118	342
Cost of initial surgical biopsy, $	373	252 to 552	−1159	−1276	117
Fold increase in sensitivity of PLA over VAH (%)	1.11 (93)	1.0 (84) to 1.19 (100)	−964	−935	29
Cost of additional treatment for delayed diagnosis, $	5040	3405 to 7459	−918	−923	5

Abbreviations: PLA, pigmented lesion assay; VAH, visual and histopathologic assessment.

Discussion

The PLA is used as a rule-out test to aid the decision by physicians to perform a surgical biopsy and not as a screening test on pigmented lesions without clinical risk factors. Positive PLA test results are followed up with a surgical biopsy and histopathologic assessment, whereas the negative test results (a much larger group) are followed up with surveillance per standard of care. This economic model demonstrates that incorporating the noninvasive PLA into the diagnostic paradigm for pigmented skin lesions suggestive of melanoma can yield substantial cost savings relative to the VAH standard of care.

The economic benefits of the PLA are driven by high specificity that substantially reduces the NNB to find melanoma by 5-fold (from 15.7 to approximately 2.7). Most surgical biopsies have negative findings and can be considered unnecessary. In a study by Lott et al¹⁵ of 80 368 surgical biopsies, 83% were considered class I (benign or mildly atypical) and 8% were considered class II (moderately to severely atypical). In a review of more than 15 000 PLA test results, 88% of tests had negative results, eliminating unnecessary surgical procedures (Laura Ferris, MD, PhD; written communication; February 1, 2018).

As many as 30% of biopsied lesions have cellular atypia on histopathologic assessment with positive tissue margins.²⁰ Conservative management of these results leads to a high number of subsequent excisions with margins. Only 0.2% to less than 1.0% of lesions with atypia and positive margins that undergo excision are diagnostically upgraded to melanoma in situ, suggesting that most of these excisions are similarly unnecessary.^14,16 Thus, as the model demonstrates, reducing the number of unnecessary surgical biopsies will also reduce unnecessary excisional procedures, thereby driving even more savings.

In this study, a relatively generous specificity of 32% was given to the VAH pathway. Work by Monheit et al⁵ and Argenziano et al⁶ describe visual assessment specificity of 3.7% to 6.0%, and the VAH specificity is likely less than 10%. In the real world, this lower specificity would be associated with even more surgical procedures, and an even larger economic benefit of the noninvasive PLA may be realized.

In contrast to the current pathway, the PLA also has a very high NPV (>99%),²⁶ which reduces the likelihood that a melanoma will be missed and subject to a delayed diagnosis. Melanomas with a delayed diagnosis have higher stage-related treatment costs. This finding may be more relevant for early-stage disease. Conic and colleagues⁴⁷ recently described a delay of surgery of more than 29 days as having a negative effect on overall survival for stage I melanoma while not providing information on melanoma-specific survival. Their study leaves unclear whether delays in treatment resulted in patients dying of melanoma, but the data can be seen as supporting the notion that minimizing delays in treatment may benefit patients with melanoma. The PLA test has a 72-hour turnaround time; although patients with positive test results must return for a biopsy, that return visit is unlikely to lead to a significant delay.

Limitations

Limitations of the PLA test include that it does not work on palms of hands and soles of feet owing to insufficient RNA in tissue samples obtained from these locations via adhesive patches. Adhesive patches are also not suited to harvest of tissue from mucous membranes. The potentially biggest limitation of this economic model stems from the assumption that small delays in diagnosis and stage progression owing to a missed melanoma combined with the significant cost of late-stage therapy create meaningful opportunities for cost savings. We attempted to base this notion on evidence from the literature most prominently supported by the work by Conic et al,⁴⁷ who found that overall survival decreased in patients waiting longer than 90 days for surgical treatment regardless of stage and that delay of surgery beyond the first 29 days had a negative effect on overall survival of stage I melanoma. Other limitations are based on assumptions that attempted to generalize direct medical costs for office visits (which vary by state and are influenced by a patient’s insurance type and coverage), for biopsy procedures and practices (which also vary among health care professionals and dermatology offices and centers), for histopathologic practices (including the frequency of use of special stains), and for the differential management of repairs and complications. Other variables we attempted to capture within the confines of assumptions include conditions for reexcisions, margin selection, and stage-related treatment costs. With generalized assumptions, we were also unable to fully capture the rapidly changing cost structure and landscape of checkpoint inhibitors, BRAF antagonists, and emerging combination strategies for late-stage melanoma and the future use of these high-cost treatment options and adjuvant and neoadjuvant settings.

Conclusions

The PLA provides clinical utility in the assessment of pigmented lesions suggestive of melanoma by its ability to transform the current clinical pathway from one that is subjective (hence variable in implementation), invasive, and with a relatively low NPV to one that is objective (hence more predictable), noninvasive, and with a high NPV.^{5,6,13,14,20,26} Pricing of the PLA at $500 would lead to $447 or 47% in health care cost savings vs the VAH standard of care pathway per assessed primary cutaneous lesion suggestive of melanoma. Use of the PLA reduces surgical procedures, missed melanomas, and cost.