Efficiency of Detecting New Primary Melanoma Among Individuals Treated in a High-risk Clinic for Skin Surveillance | Cancer Screening, Prevention, Control | JAMA Dermatology | JAMA Network
[Skip to Navigation]
Sign In
Table 1.  Description of Each Melanoma HRC Study Center and Surveillance Technology Used
Description of Each Melanoma HRC Study Center and Surveillance Technology Used
Table 2.  Characteristics of High-risk Clinic Participants From All Centers Combined, by Subgroupa
Characteristics of High-risk Clinic Participants From All Centers Combined, by Subgroupa
Table 3.  Incidence of Melanoma, NMSC, Benign Melanocytic, and Nonmelanocytic Lesions Excised During Follow-up Surveillance in High-risk Clinics, by Subgroupa,b
Incidence of Melanoma, NMSC, Benign Melanocytic, and Nonmelanocytic Lesions Excised During Follow-up Surveillance in High-risk Clinics, by Subgroupa,b
Table 4.  Incidence of New Primary Melanoma Diagnosed During Follow-up Surveillance, by Center
Incidence of New Primary Melanoma Diagnosed During Follow-up Surveillance, by Center
Table 5.  Lesion Characteristics of Incident Primary Melanoma Diagnosed During Follow-up Surveillance in High-risk Clinics, by Centera
Lesion Characteristics of Incident Primary Melanoma Diagnosed During Follow-up Surveillance in High-risk Clinics, by Centera
1.
Olsen  CM, Green  AC, Pandeya  N, Whiteman  DC.  Trends in melanoma incidence rates in eight susceptible populations through 2015.   J Invest Dermatol. 2019;139(6):1392-1395. doi:10.1016/j.jid.2018.12.006 PubMedGoogle ScholarCrossref
2.
Australian Institute of Health and Welfare. Skin cancer in Australia. Published July 13, 2016. Accessed August 3, 2020. https://www.aihw.gov.au/reports/cancer/skin-cancer-in-australia/contents/table-of-contents
3.
Doran  CM, Ling  R, Byrnes  J,  et al.  Estimating the economic costs of skin cancer in New South Wales, Australia.   BMC Public Health. 2015;15:952. doi:10.1186/s12889-015-2267-3 PubMedGoogle ScholarCrossref
4.
Australian Institute of Health and Welfare. Cancer in Australia 2019. Published 2019. Accessed August 3, 2020. https://www.aihw.gov.au/getmedia/8c9fcf52-0055-41a0-96d9-f81b0feb98cf/aihw-can-123.pdf.aspx?inline=true
5.
Youlden  DR, Youl  PH, Soyer  HP, Aitken  JF, Baade  PD.  Distribution of subsequent primary invasive melanomas following a first primary invasive or in situ melanoma in Queensland, Australia, 1982-2010.   JAMA Dermatol. 2014;150(5):526-534. doi:10.1001/jamadermatol.2013.9852 PubMedGoogle ScholarCrossref
6.
Cust  AE, Badcock  C, Smith  J,  et al.  A risk prediction model for the development of subsequent primary melanoma in a population-based cohort.   Br J Dermatol. 2020;182(5):1148-1157. doi:10.1111/bjd.18524 PubMedGoogle ScholarCrossref
7.
Müller  C, Wendt  J, Rauscher  S,  et al.  Risk factors of subsequent primary melanomas in Austria.   JAMA Dermatol. 2019;155(2):188-195. doi:10.1001/jamadermatol.2018.4645PubMedGoogle ScholarCrossref
8.
Gershenwald  JE, Scolyer  RA, Hess  KR,  et al; for members of the American Joint Committee on Cancer Melanoma Expert Panel and the International Melanoma Database and Discovery Platform.  Melanoma staging: evidence-based changes in the American Joint Committee on Cancer eighth edition cancer staging manual.   CA Cancer J Clin. 2017;67(6):472-492. doi:10.3322/caac.21409PubMedGoogle ScholarCrossref
9.
Ruiz  ES, Morgan  FC, Zigler  CM, Besaw  RJ, Schmults  CD.  Analysis of national skin cancer expenditures in the United States Medicare population, 2013.   J Am Acad Dermatol. 2019;80(1):275-278. doi:10.1016/j.jaad.2018.04.035 PubMedGoogle ScholarCrossref
10.
Buja  A, Rivera  M, De Polo  A,  et al.  Real-world data for direct stage-specific costs of melanoma healthcare.   Br J Dermatol. 2020;183(1):171-172. doi:10.1111/bjd.18896 PubMedGoogle ScholarCrossref
11.
Elliott  TM, Whiteman  DC, Olsen  CM, Gordon  LG.  Estimated healthcare costs of melanoma in Australia over 3 years post-diagnosis.   Appl Health Econ Health Policy. 2017;15(6):805-816. doi:10.1007/s40258-017-0341-y PubMedGoogle ScholarCrossref
12.
Lyth  J, Carstensen  J, Synnerstad  I, Lindholm  C.  Stage-specific direct health care costs in patients with cutaneous malignant melanoma.   J Eur Acad Dermatol Venereol. 2016;30(5):789-793. doi:10.1111/jdv.13110 PubMedGoogle ScholarCrossref
13.
Cancer Council Australia. Clinical practice guidelines for the diagnosis and management of melanoma. Accessed August 3, 2020. https://wiki.cancer.org.au/australia/Guidelines:Melanoma
14.
Moloney  FJ, Guitera  P, Coates  E,  et al.  Detection of primary melanoma in individuals at extreme high risk: a prospective 5-year follow-up study.   JAMA Dermatol. 2014;150(8):819-827. doi:10.1001/jamadermatol.2014.514 PubMedGoogle ScholarCrossref
15.
Watts  CG, Cust  AE, Menzies  SW, Mann  GJ, Morton  RL.  Cost-effectiveness of skin surveillance through a specialized clinic for patients at high risk of melanoma.   J Clin Oncol. 2017;35(1):63-71. doi:10.1200/JCO.2016.68.4308 PubMedGoogle ScholarCrossref
16.
Adler  NR, Kelly  JW, Guitera  P,  et al.  Methods of melanoma detection and of skin monitoring for individuals at high risk of melanoma: new Australian clinical practice.   Med J Aust. 2019;210(1):41-47. doi:10.5694/mja2.12033 PubMedGoogle ScholarCrossref
17.
Salerni  G, Carrera  C, Lovatto  L,  et al.  Benefits of total body photography and digital dermatoscopy (“two-step method of digital follow-up”) in the early diagnosis of melanoma in patients at high risk for melanoma.   J Am Acad Dermatol. 2012;67(1):e17-e27. doi:10.1016/j.jaad.2011.04.008 PubMedGoogle ScholarCrossref
18.
Watts  CG, Madronio  CM, Morton  RL,  et al.  Diagnosis and clinical management of melanoma patients at higher risk of a new primary melanoma: a population-based study in New South Wales, Australia.   Australas J Dermatol. 2017;58(4):278-285. doi:10.1111/ajd.12530 PubMedGoogle ScholarCrossref
19.
Carli  P, De Giorgi  V, Crocetti  E,  et al.  Improvement of malignant/benign ratio in excised melanocytic lesions in the “dermoscopy era”: a retrospective study 1997-2001.   Br J Dermatol. 2004;150(4):687-692. doi:10.1111/j.0007-0963.2004.05860.x PubMedGoogle ScholarCrossref
20.
Menzies  SW, Emery  J, Staples  M,  et al.  Impact of dermoscopy and short-term sequential digital dermoscopy imaging for the management of pigmented lesions in primary care: a sequential intervention trial.   Br J Dermatol. 2009;161(6):1270-1277. doi:10.1111/j.1365-2133.2009.09374.x PubMedGoogle ScholarCrossref
21.
Petty  AJ, Ackerson  B, Garza  R,  et al.  Meta-analysis of number needed to treat for diagnosis of melanoma by clinical setting.   J Am Acad Dermatol. 2020;82(5):1158-1165. doi:10.1016/j.jaad.2019.12.063 PubMedGoogle ScholarCrossref
22.
Green  AC, Baade  P, Coory  M, Aitken  JF, Smithers  M.  Population-based 20-year survival among people diagnosed with thin melanomas in Queensland, Australia.   J Clin Oncol. 2012;30(13):1462-1467. doi:10.1200/JCO.2011.38.8561 PubMedGoogle ScholarCrossref
23.
Aguilera  P, Malvehy  J, Carrera  C,  et al.  Clinical and histopathological characteristics between familial and sporadic melanoma in Barcelona, Spain.   J Clin Exp Dermatol Res. 2014;5(5):231. doi:10.4172/2155-9554.1000231PubMedGoogle Scholar
24.
Gassenmaier  M, Stec  T, Keim  U,  et al.  Incidence and characteristics of thick second primary melanomas: a study of the German Central Malignant Melanoma Registry.   J Eur Acad Dermatol Venereol. 2019;33(1):63-70. doi:10.1111/jdv.15194 PubMedGoogle ScholarCrossref
25.
Glasziou  PP, Jones  MA, Pathirana  T, Barratt  AL, Bell  KJ.  Estimating the magnitude of cancer overdiagnosis in Australia.   Med J Aust. 2020;212(4):163-168. doi:10.5694/mja2.50455 PubMedGoogle ScholarCrossref
26.
McLoone  JK, Watts  KJ, Menzies  SW, Barlow-Stewart  K, Mann  GJ, Kasparian  NA.  Melanoma survivors at high risk of developing new primary disease: a qualitative examination of the factors that contribute to patient satisfaction with clinical care.   Psychooncology. 2013;22(9):1994-2000. doi:10.1002/pon.3243 PubMedGoogle ScholarCrossref
27.
van der Leest  RJ, Flohil  SC, Arends  LR, de Vries  E, Nijsten  T.  Risk of subsequent cutaneous malignancy in patients with prior melanoma: a systematic review and meta-analysis.   J Eur Acad Dermatol Venereol. 2015;29(6):1053-1062. doi:10.1111/jdv.12887 PubMedGoogle ScholarCrossref
28.
Read  RL, Madronio  CM, Cust  AE,  et al.  Follow-up recommendations after diagnosis of primary cutaneous melanoma: a population-based study in New South Wales, Australia.   Ann Surg Oncol. 2018;25(3):617-625. doi:10.1245/s10434-017-6319-z PubMedGoogle ScholarCrossref
29.
Vuong  K, Armstrong  BK, Weiderpass  E,  et al; Australian Melanoma Family Study Investigators.  Development and external validation of a melanoma risk prediction model based on self-assessed risk factors.   JAMA Dermatol. 2016;152(8):889-896. doi:10.1001/jamadermatol.2016.0939 PubMedGoogle ScholarCrossref
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    2 Comments for this article
    EXPAND ALL
    An additional study which failed to look at the most important outcome to measure in melanoma
    Robert Swerlick, MD | Department of Dermatology, Emory University School of Medicine
    In 2019 Johansson and co-authors published a Cochrane review entitled:
    "Screening for reducing morbidity and mortality in malignant melanoma".
    They reviewed two studies with almost 65,000 participants. The summary of what outcomes were found was the following (page 4 of the review):

    1. Total mortality - Not measured
    2. Overdiagnosis of malignant melanoma - Not measured
    3. Quality of life/psychosocial consequences - Not measured
    4. Mortality specific to malignant melanoma - Not measured
    5. False positive rates - Not measured
    6. False negative rates - Not measured

    The next time
    they update the review, they may be able to include another study but the summary will look the same.

    (Citation: Johansson M, Brodersen J, Gøtzsche PC, Jørgensen KJ. Screening for reducing morbidity and mortality in malignant melanoma. Cochrane Database of Systematic Reviews 2019, Issue 6. Art. No.: CD012352. DOI: 10.1002/14651858.CD012352.pub2.)
    CONFLICT OF INTEREST: None Reported
    READ MORE
    Melanoma surveillance aided by TBP/SDDI needs further investigation before benefits over traditional surveillance can be claimed
    Hilary Brown, B.Med FRACGP | South East Dermatology Brisbane Australia
    Guitera et al. document intensive surveillance of high-risk Australian melanoma patients aided by total body photography (TBP) and serial digital dermoscopic imaging (SDDI) . They conclude this technique has “consistent and sustainable benefits” which “may be implemented on a large scale”. Without a control arm or comparison to other studies of melanoma diagnosis is this claim justified?

    To document efficacy researchers should compare their intervention to current practice using a control group. Surveillance with TBP/SDDI would be compared to the widespread current practice of regular skin examinations with dermoscopy and excision of suspect lesions.

    Australia has
    large numbers of high-risk melanoma patients undergoing usual surveillance by a similar range of medical practitioners as those in Guitera et al. A control arm was readily available. Absence of a control arm means relative efficacy cannot be documented.

    A proxy measure of early melanoma detection is the ratio of in-situ to invasive melanomas and the percentage greater than 0.8mm thick. Two Australian papers report outcomes of traditional surveillance without TBP/SDDI. They show more favourable in-situ to invasive melanoma ratios (2.9:12 and 4:13) and a lower percentage of melanomas over 0.8mm thick (4.2%) compared to Guitera et al (2.2:1 and 10% respectively).

    SDDI risks imaging rather than excising melanoma. Guitera et al. found more melanomas in the third and fourth year of surveillance than the first two. Combined with the higher percentage of invasive and thick lesions these results may suggest that TBP/SDDI results in delayed melanoma diagnosis compared to traditional practice. A control arm would have resolved this question.

    Guitera et al report a loss-to-follow up rate of 14.4%. Without a control arm we cannot know the figures for traditional surveillance. However, by imaging suspect lesions rather than immediately excising them, non-compliance with SDDI creates an additional risk.

    Without a control arm, this paper cannot document benefit over current practice. Comparison of TBP/SDDI to traditional surveillance suggests TBP/SDDI may result in inferior outcomes. There are risks with monitoring, delayed excision and compliance that await investigation. Promotion of TBP/SDDI outside of properly controlled trials is unjustified.

    We await the outcomes of trials comparing TBP/SDDI and traditional surveillance.
    CONFLICT OF INTEREST: None Reported
    READ MORE
    Original Investigation
    March 17, 2021

    Efficiency of Detecting New Primary Melanoma Among Individuals Treated in a High-risk Clinic for Skin Surveillance

    Author Affiliations
    • 1Melanoma Institute Australia, The University of Sydney, Sydney, Australia
    • 2Faculty of Medicine and Health, Sydney Medical School, The University of Sydney, Sydney, Australia
    • 3Sydney Melanoma Diagnostic Centre, Royal Prince Alfred Hospital, Sydney, Australia
    • 4Newcastle Skin Check, Newcastle, Australia
    • 5School of Medicine, The University of Queensland, Brisbane, Australia
    • 6Department of Dermatology, Westmead Clinical School, The University of Sydney, Sydney, Australia
    • 7Sydney School of Public Health, The University of Sydney, Sydney, Australia
    • 8Westmead Institute for Medical Research, The University of Sydney, Sydney, Australia
    • 9The John Curtin School of Medical Research, ANU College of Health and Medicine, The Australian National University, Canberra, Australia
    JAMA Dermatol. 2021;157(5):521-530. doi:10.1001/jamadermatol.2020.5651
    Key Points

    Question  Are the favorable excision rates and melanoma early detection outcomes from a previously implemented structured surveillance program for people at high risk of melanoma sustained in the longer term and replicated in other centers, including a primary care skin cancer clinic?

    Findings  Of 171 new melanomas detected among 593 participants in this cohort study, 96% had a Breslow thickness of 1 mm or less, and 67% of melanomas were found with the assistance of total-body photography or sequential digital dermoscopy imaging. The overall benign to malignant excision ratio was 0.8:1.0, and the benign melanocytic to melanoma excision ratio was 2.4:1.0, both of which were similar across centers.

    Meaning  The findings of this cohort study suggest that the structured surveillance program may be implemented on a larger scale, including at primary care skin cancer clinics, with consistent and sustainable benefits observed.

    Abstract

    Importance  A previous single-center study observed fewer excisions, lower health care costs, thinner melanomas, and better quality of life when surveillance of high-risk patients was conducted in a melanoma dermatology clinic with a structured surveillance protocol involving full-body examinations every 6 months aided by total-body photography (TBP) and sequential digital dermoscopy imaging (SDDI).

    Objective  To examine longer-term sustainability and expansion of the surveillance program to numerous practices, including a primary care skin cancer clinic setting.

    Design, Setting, and Participants  This prospective cohort study recruited 593 participants assessed from 2012 to 2018 as having very high risk of melanoma, with a median of 2.9 years of follow-up (interquartile range, 1.9-3.3 years), from 4 melanoma high-risk clinics (3 dermatology clinics and 1 primary care skin cancer clinic) in New South Wales, Australia. Data analyses were conducted from February to September 2020.

    Exposures  Six-month full-body examination with the aid of TBP and SDDI. For equivocal lesions, the clinician performed SDDI at 3 or 6 months.

    Main Outcomes and Measures  All suspect monitored or excised lesions were recorded, and pathology reports obtained. Outcomes included the incidence and characteristics of new lesions and the association of diagnostic aids with rates of new melanoma detection.

    Results  Among 593 participants, 340 (57.3%) were men, and the median age at baseline was 58 years (interquartile range, 47-66 years). There were 1513 lesions excised during follow-up, including 171 primary melanomas. The overall benign to malignant excision ratio, including keratinocyte carcinomas, was 0.8:1.0; the benign melanocytic to melanoma excision ratio was 2.4:1.0; and the melanoma in situ to invasive melanoma ratio was 2.2:1.0. The excision ratios were similar across the 4 centers. The risk of developing a new melanoma was 9.0% annually in the first 2 years and increased with time, particularly for those with multiple primary melanomas. The thicker melanomas (>1-mm Breslow thickness; 7 of 171 melanomas [4.1%]) were mostly desmoplastic or nodular (4 of 7), self-detected (2 of 7), or clinician detected without the aid of TBP (3 of 7). Overall, new melanomas were most likely to be detected by a clinician with the aid of TBP (54 of 171 [31.6%]) followed by digital dermoscopy monitoring (50 of 171 [29.2%]).

    Conclusions and Relevance  The structured surveillance program for high-risk patients may be implemented at a larger scale given the present cohort study findings suggesting the sustainability and replication of results in numerous settings, including a primary care skin cancer clinic.

    Introduction

    The incidence of melanoma has been increasing in many countries.1 Australia has the highest incidence rates, and together with other nonmelanoma (keratinocyte) skin cancers, melanoma represents the most expensive cancer for the health care system.2-4 Those who develop an in situ or invasive primary melanoma are at much greater risk of developing subsequent melanoma compared with the general population,5 especially for those with additional risk factors, such as multiple primary melanomas, dysplastic nevi, family history, or a melanoma-predisposing gene variant.6,7 Early detection is associated with better survival,8 less morbidity from treatment, and fewer health system costs.9-12

    Australian clinical practice guidelines recommend that individuals at very high risk of melanoma be checked regularly by a clinician, with full skin examinations every 6 months supported by dermoscopy and using the aids of sequential digital dermoscopy imaging (SDDI) and total-body photography (TBP).13 This recommendation was based partly on a previous Australian study that showed a structured surveillance program was less expensive and associated with more quality-adjusted life-years of survival than standard care.14,15 The use of surveillance technologies, including dermoscopy, short-term and long-term SDDI, and TBP,16 facilitated the detection of changing lesions and minimized the excision of benign lesions that add considerably to health system costs.11,15 A Spanish study17 also showed a favorable Breslow thickness distribution among a high-risk cohort followed up with a structured surveillance protocol, albeit with a higher ratio of excised benign to malignant melanocytic lesions (10.7:1) than the Australian study (4.4:1).14,15

    As highlighted in clinical practice guidelines for the diagnosis and management of melanoma,13 there are some key gaps and limitations that hinder the wider implementation of these recommendations in routine clinical practice, such as the single-center design of the previous studies, limited replication, and uncertainty around the extent to which reduced rates of excisions for benign lesions can be sustained in the longer term and achieved in other dermatology clinics and in primary care settings, such as primary care skin cancer clinics, that diagnose approximately 17% of early-stage melanoma cases in Australia.18 The present study addresses these gaps.

    Methods
    Study Design and Population

    We used a cohort study design. Melanoma high-risk clinics (HRCs) were established at 4 centers in New South Wales, Australia (Table 1). The Sydney Melanoma Diagnostic Centre (SMDC) at Royal Prince Alfred Hospital, a tertiary referral center, had a dedicated trained resident medical officer or general practice physician (S.W.M., E.C.) under specialist supervision. Dermatologists (P.F.-P., R.L.) led the HRCs at Westmead Hospital, an outpatient clinic in a major teaching hospital, and at Melanoma Institute Australia (P.G., H.C.), a major tertiary referral center. Newcastle Skin Check is a primary care skin cancer clinic (A.A., A.L.). Skin cancer clinics are staffed by general practice physicians; however, they focus solely on skin cancer. The study was approved by the human research ethics committees of the Sydney Local Health District at Royal Prince Alfred Hospital. All participants provided written informed consent that was obtained in a manner consistent with the Australian National Statement on Ethical Conduct in Human Research. No one received compensation or was offered any incentive for participating in this study.

    Patients 18 years of age or older were invited to attend the HRC for their skin surveillance as part of this research study if they were considered to be at very high risk of cutaneous melanoma, assessed as meeting at least 1 of 4 eligibility criteria (eMethods in the Supplement).14 The HRC study initially started as a single-center study with recruitment of participants from 2006 to 2009 at the Sydney Melanoma Diagnostic Centre; follow-up data collected at that center from 2006 to 2011 as part of the initial study have been previously reported.14 Thus, only data from 2012 onward are included in this analysis from that center. This study reports the outcomes from the expansion of the HRC study, which involved continuation of the existing cohort and recruitment of new participants at the Sydney Melanoma Diagnostic Centre and at the 3 other centers.

    The median follow-up time for participants in this expanded cohort was 2.9 years (interquartile range [IQR], 1.9-3.3 years), and there was a median of 2.7 (IQR, 2.3-3.2) clinic visits per year per participant (Table 1). The length of follow-up was based on available research funding.

    HRC Skin Surveillance Protocol and Data Collection

    The HRC skin surveillance protocol has been previously described.14 A clinician conducted a full-body skin examination every 6 months with the aid of a handheld dermoscopy device and a comparison of the skin with baseline TBP. In addition, SDDI of some lesions, either short term (>3 months) or long term (≥6 months), was scheduled as required. Table 1 shows the surveillance equipment used. More details are in the eMethods in the Supplement.

    Statistical Analysis

    In total, 593 participants from 4 centers were included in the analysis (Table 1) after excluding 100 of 693 individuals (14.4%) lost to follow-up (eMethods in the Supplement). Participants’ follow-up times were censored at the end of the specified follow-up time (Table 1), or earlier if there were more than 12 months between HRC visits, or at the time of death (n = 11).

    The patient’s first HRC visit during the specified follow-up dates was considered the baseline visit, except for the initial cohort at the Sydney Melanoma Diagnostic Centre and patients from the Melanoma Institute Australia, who were already following a photography surveillance protocol similar to the HRC and were thus considered not to have a baseline visit for this analysis.

    Descriptive data are shown as frequencies and percentages for categorical variables and as mean (SD) values or median values with IQRs for normally and nonnormally distributed continuous variables, respectively. The P values for differences across centers or subgroups were calculated using the χ2 test or the Mantel-Haenszel test for trend. A 2-sided value of P < .05 was considered statistically significant. Poisson regression models using generalized estimating equations were used to estimate incidence rate ratios and 95% CIs for excisions and new primary melanomas after 2 years compared with within the first 2 years of follow-up. The analysis was conducted using SAS software, version 9.4 (SAS Institute Inc). Data analyses were performed from February to September 2020.

    Results
    Characteristics of Participants

    The characteristics of the participants are shown in Table 2 for the cohort overall and in eTable 1A, B, C, and D in the Supplement for each of the 4 centers. The majority of 593 participants met the eligibility criteria for multiple primary melanoma (n = 546), followed by dysplastic nevus syndrome (DNS) and previous melanoma (n = 332), with fewer participants having a strong family history and previous melanoma (n = 83) or a CDKN2A gene variant (no requirement of previous melanoma was made; n = 16). The overlap of participants in the 4 eligibility criteria subgroups, overall and by center, is shown in eFigure 1 in the Supplement. The distribution of participants in the different subgroups differed across centers. The median age at first visit to the HRC was 58 years (IQR, 47-66 years). The sample of 593 participants was predominately men (340 [57.3%]), but this differed by subgroup, with more women (50 of 83 [60.2%]) in the strong family history subgroup (P = .001 compared with no strong family history) and the CDKN2A subgroup (10 of 16 [62.5%] were women; P = .17). Participants with a strong family history had a higher proportion of high-risk phenotypic characteristics (red hair, many freckles).

    Lesions Excised at Baseline

    Among 261 participants who were not already under close photography surveillance, there were 78 lesions excised at the baseline HRC visit (or within 8 weeks), of which 10 (12.8%) were melanomas, 34 (43.6%) were nonmelanoma skin cancer, 17 (21.8%) were benign melanocytic lesions, and 17 (21.8%) were benign nonmelanocytic lesions (eTable 2 in the Supplement).

    Lesions Excised and Excision Ratios During Follow-up Surveillance

    There were 1513 lesions excised during follow-up surveillance in the HRCs (Table 3). Of the excised lesions, 171 (11.3%) were melanomas, 690 (45.6%) were nonmelanoma skin cancer, 410 (27.1%) were benign melanocytic lesions, and 234 (15.5%) were benign nonmelanocytic lesions. The probability of an excised lesion being melanoma (positive predictive value) was 11.3% overall and 29.4% for melanocytic lesions. Of 593 participants, 114 (19.2%) had 1 or more primary melanomas excised during the median 2.9-year follow-up surveillance period (IQR, 1.9-3.3 years), 214 (36.1%) had 1 or more nonmelanoma skin cancer lesions, 213 (35.9%) had 1 or more benign melanocytic lesions, and 142 (23.9%) had 1 or more benign nonmelanocytic lesions. The occurrence of a new primary melanoma was less common in the DNS subgroup (14.8%, P = .03).

    The mean excision rate for all lesions was 0.9 (95% CI, 0.8-1.0) excisions per person-year of follow-up in the first 2 years of HRC surveillance, and the mean excision rate was 1.2 (95% CI, 1.0-1.4) excisions per person-year of follow-up in years 2 to 4 (eFigure 2 in the Supplement). Thus, the excision rate was 1.3 times as high (95% CI, 1.1-1.5; P = .002) in years 2 to 4 compared with years 0 to 2. When compared across centers, the mean excision rates in years 0 to 2 ranged from 0.6 to 1.5, and the incidence excision rate ratios ranged from 1.0 to 1.8.

    For all centers combined, the total benign to malignant excision ratio was 0.8:1.0, and the benign melanocytic to melanoma ratio was 2.4:1.0 (Table 3). The excision ratios were higher for the DNS subgroup and lower for the strong family history subgroup. The total excision ratios were similar across the 4 centers (Table 3). The lesions excised for each center are shown in eTable 3 in the Supplement. The melanoma in situ to invasive melanoma ratio was 2.2:1.0.

    Melanomas Detected During Follow-up Surveillance

    The 171 melanomas detected during follow-up surveillance occurred among 114 participants. Although most participants (479 of 593 [80.8%]) experienced no melanomas during follow-up, some participants had multiple (up to 6) primary melanomas detected (Table 4). Male participants were more likely than female participants to develop another melanoma (82 of 340 [24.1%] vs 32 of 253 [12.6%]; P < .001).

    The mean melanoma incidence rate was 0.09 (95% CI, 0.08-0.12) per person-year of follow-up in the first 2 years of HRC surveillance (equivalent to a 9.0% annual risk of melanoma in each of the first 2 years) and 0.15 (95% CI, 0.11-0.20) per person-year of follow-up in years 2 to 4 (equivalent to a 15.0% annual risk of melanoma in years 2-4) (Table 4). Thus, the incidence rate of new primary melanomas was 1.6 times as high (95% CI, 1.2-2.2; P = .004) in years 2 to 4 compared with years 0 to 2. The increase over time was more pronounced for the Melanoma Institute Australia center than for the other centers (Table 4) and for the subgroup with multiple primary melanomas than for the other subgroups (eFigure 3 in the Supplement).

    The lesion characteristics of the incident primary melanomas diagnosed during follow-up surveillance in the HRCs are shown in Table 5. Melanomas occurred most frequently on the trunk (56 of 171 [32.7%]) and the upper limb and shoulder (56 of 171 [32.7%]). The median Breslow thickness was in situ (IQR, in situ to 0.40 mm); of 171 melanomas, 117 (68.4%) were in situ, 37 (21.6%) had a Breslow thickness of 0.1 to less than 0.8 mm, 10 (5.8%) had a Breslow thickness of 0.8 to 1.0 mm, and 7 (4.1%) had a Breslow thickness of more than 1.0 mm. There were some differences by center, such as a higher proportion of lentigo maligna diagnoses at Westmead Hospital (8 of 24 melanomas [33.3%]). Only 9 of 171 melanomas (5.3%) were of a nodular or desmoplastic subtype, but they represented 4 of the 7 thicker (>1 mm) melanomas detected (eTable 4 in the Supplement). Two of the thicker melanomas were self-detected, and 5 were detected by a clinician, of which 2 were detected with the aid of TBP. Overall, new primary melanomas were most likely to be detected by a clinician with the aid of TBP (54 of 171 [31.6%]) followed by SDDI (50 of 171 [29.2%], of which 32 [18.7%] were short term and 18 [10.5%] were long term), but there were differences across centers (Table 5).

    Discussion

    Using a structured surveillance protocol aided by TBP and SDDI for optimizing the detection of new primary melanoma among individuals at very high risk, we observed favorable long-term early detection and excision results sustained for more than 10 years at the original center (SMDC) and replicated at 3 other centers. The clinicians following the surveillance protocol included dermatology specialists, trained dedicated residents, and primary care physicians in hospital outpatient clinics and in a primary care skin cancer clinic.

    The sustained long-term results at SMDC are reassuring because they indicate that thick melanomas were unlikely to be missed despite a low benign to malignant excision ratio. The overall benign to malignant excision ratio of 0.8:1.0 and the overall benign melanocytic to melanoma ratio of 2.4:1.0 in this cohort were better than the commonly accepted benign to malignant excision ratios of 5:1 for dermatology specialists and 20:1 for generalists.19 Australia often has lower excision ratios than other countries because skin cancer is common.20 A recent international meta-analysis on the number needed to excise or biopsy to diagnose melanoma concluded that pigmented lesion specialists have the lowest number (5.9), followed by dermatologists (9.6) and primary care doctors (22.6).21 Nevertheless, our results showed similar outcomes across centers, indicating that the diagnostic tools and structured surveillance protocol were more important than the clinical specialty. The low number needed to excise or biopsy was associated with the use of photography surveillance affecting the threshold for biopsy and would also be expected to be lower for clinicians experienced in skin cancer detection and for regions with higher incidence of melanoma, as was the case in the present study.

    Based on the original study,14 Watts et al15 reported the cost-effectiveness of this structured surveillance protocol (mean savings per patient of A$6828 [approximately $5205] during 10 years), showing that the main factor associated with the savings was the detection of melanoma at an earlier stage, resulting in less extensive treatment (mean quality-adjusted life-year gain of 0.31) and a low annual mean excision rate (0.81 vs 2.55 in standard care). In the present expanded cohort, the mean excision rate was 0.9 per person-year of follow-up in the first 2 years and 1.2 per person-year in years 2 to 4. However, this rate also corresponded with an overall 1.6 times as high melanoma incidence rate in years 2 to 4 vs the first 2 years. This result is a distinct difference from the initial study in which the melanoma incidence rate decreased with time.14 The melanoma incidence rate was also higher in the present expanded cohort (9.0% annual risk in each of the first 2 years and 15.0% annual risk thereafter) compared with 12.7% cumulative 2-year risk in the original study. This higher and increasing incidence rate may be because 46.9% of participants had multiple primary melanomas at baseline in the first cohort, whereas 92.1% did so in the expanded cohort, and multiple primary melanomas are a strong determinant of subsequent melanoma risk (mean 5-year risk of 8% after 1 melanoma and 47% after 2 melanomas6). Another possible explanation is that the protocol relied heavily on photographic change to detect melanoma, shifting the diagnosis to later time points. Yet this shift did not appear to be meaningfully associated with the stage at diagnosis, probably because measuring the change allows detection of incipient melanoma that may not yet have developed dermoscopic features.13 The increased incidence rate was more pronounced for the Melanoma Institute Australia than for the other centers; this finding may be partly due to a higher proportion of male patients at the Melanoma Institute Australia because male patients were twice as likely as female patients to develop new primary melanoma during the surveillance period. Male patients also have a higher risk of melanoma mortality.22

    The Breslow thickness distribution was even more favorable in the expanded cohort; in the initial cohort, the median Breslow thickness was in situ (IQR, in situ to 0.60 mm), and there were 4 of 61 lesions (6.6%) that had a Breslow thickness of more than 1 mm, compared with the expanded cohort in which the median was in situ (IQR, in situ to 0.40 mm), and 7 of 171 lesions (4.1%) had a Breslow thickness of more than 1 mm. Thus, these results, together with substantially increasing health system costs for melanoma treatment,11 suggest that cost-effectiveness may be higher than previously estimated and provide impetus to scale up the program. This was, however, an observational study; more definitive evidence on mortality reduction or rates of thicker melanoma requires a randomized clinical trial. In addition, second primary melanomas and familial melanomas tend to be thinner than sporadic melanomas.23,24

    One concern of increased surveillance is for overdiagnosis and overtreatment, particularly for in situ melanoma, such that some melanomas left undetected (and untreated) would never transform into invasive disease causing symptoms or harm.25 We diagnosed 2.2 in situ melanomas to each invasive melanoma in this study, compared with a ratio of 1.6:1.0 in the general population.4 The harm associated with overdiagnosis in the HRCs is likely lower relative to the benefits obtained from attending the clinic because previous work by members of our group have shown fewer excisions compared with usual care,15 and the psychological impact of a new diagnosis may be less given that most people have already had a previous melanoma. Patients are also reassured by expert care,26 and there was minimal loss to follow-up (14.0%). In addition, two-thirds (66.7%) of melanomas in our cohort were found because of changes on photography (SDDI or TBP) indicating biological activity.

    Participants with DNS had a higher benign to malignant excision ratio but fewer new primary melanomas developed compared with the other subgroups, and this highlights the difficulty in correctly distinguishing melanomas from dysplastic nevi masquerading as melanomas. Nodular and desmoplastic melanomas are particularly difficult to diagnose, and these were overrepresented in the thicker (>1 mm) melanomas detected. Only 2 of the 7 thicker melanomas detected were identified using photography, and 2 were self-detected, showing that educating patients and physicians to recognize these difficult lesions remains a priority.

    Limitations

    Our data were limited to the Australian population and may not be generalizable to regions with lower melanoma incidence or to all Australian centers. Melanomas in Australia are routinely detected in dermatology settings and primary care (general practice and skin cancer clinics).18 Most primary care physicians in Australia use dermoscopy, whereas TBP and particularly SDDI are more commonly used in primary care skin cancer clinics than in generalist primary care (Victoria Mar, PhD, Director of the Victorian Melanoma Service, written personal communication, September 2020). In the primary care skin cancer clinic in our study, TBP and SDDI were routinely used prior to the study. Hence, significant uncertainty exists when generalizing the findings to all primary care clinics. However, nearly everyone in this cohort had a previous melanoma, which is consistently a strong predictor for developing a subsequent melanoma in other countries,27 and close surveillance is recommended.28 People at low or average risk may benefit less from this surveillance program. Use of risk prediction algorithms may help to accurately select people at high risk and tailor surveillance intervals according to personal risk.6,29 Advanced diagnostic photographic tools and high-quality, low-cost dermatoscopes provide an opportunity for primary care physicians and even patients to equip themselves with this technology. Incorporating artificial intelligence to enhance melanoma diagnosis may further change this paradigm of skin surveillance.

    Conclusions

    The structured surveillance program for individuals at high risk of new primary melanoma may be implemented on a larger scale, including primary care skin cancer clinics, given the study findings suggesting consistent and sustainable benefits.

    Back to top
    Article Information

    Accepted for Publication: January 19, 2021.

    Published Online: March 17, 2021. doi:10.1001/jamadermatol.2020.5651

    Corresponding Author: Anne E. Cust, MPH(Hons), PhD, Sydney School of Public Health, Building A27, The University of Sydney, New South Wales 2006, Australia (anne.cust@sydney.edu.au).

    Author Contributions: Dr Cust and Ms Badcock had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: Guitera, Menzies, Coates, Mann, Cust.

    Acquisition, analysis, or interpretation of data: All authors.

    Drafting of the manuscript: Guitera, Coates, Mann, Cust.

    Critical revision of the manuscript for important intellectual content: Menzies, Coates, Azzi, Fernandez-Penas, Lilleyman, Badcock, Schmid, Watts, Collgros, Liu, van Kemenade, Mann, Cust.

    Statistical analysis: Badcock, Liu, Cust.

    Obtained funding: Menzies, Coates, Mann, Cust.

    Administrative, technical, or material support: Menzies, Coates, Fernandez-Penas, Schmid, Watts, Collgros, Liu, van Kemenade, Mann.

    Supervision: Guitera, Menzies, Coates, Fernandez-Penas, Mann, Cust.

    Conflict of Interest Disclosures: Dr Menzies reported receiving personal fees from SciBase AB. Dr Fernandez-Penas reported receiving personal fees from AbbVie, Amgen, Boehringer Ingelheim, Bristol Myers Squibb, Eli Lilly and Company, Janssen, Leo Pharma, Merck Sharp & Dohme, Novartis, Pfizer, Roche, Sanofi, Sun Pharmaceuticals, and UCB outside the submitted work; and conducting clinical trials for AbbVie, Akaal Pharma, Amgen, Arena Pharmaceuticals, Boehringer Ingelheim, Bristol Myers Squibb, CSL Behring, Pfizer, Eli Lilly and Company, Eisai, Galderma, GlaxoSmithKline, Jiangsu Hengrui, Kyowa Hakko Kirin, mRage, Novartis, OncoSec Medical Incorporated, Regeneron, Roche, Sun Pharmaceuticals, UCB, and Xoma. No other disclosures were reported.

    Funding/Support: Financial support was provided by the Centre of Research Excellence in Melanoma grant 1135285 from the NHMRC to Drs Guitera, Fernandez-Penas, Mann, and Cust; Career Development Fellowship 1147843 from the NHMRC to Dr Cust; and Translational Program Grant 10 TPG 1-02 from the Cancer Institute NSW to Drs Menzies and Mann.

    Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Additional Contributions: Assistance with data collection was provided by Marina Ali, PhD, Giuliana Carlos, MBBS, Linda Chan, MBBS, Deepal Deshpande, MBBS, Shelley Ji Eun Hwang, MBBS, Rupal Patel, MSc, Melissa Peera, MBBS, and Cathy Zhao, MBBS, all with the Department of Dermatology, Westmead Clinical School, The University of Sydney. Management of the database development was provided by Anthea Azzi, BNutrDiet(Hons), Newcastle Skin Check; Ritta Khoury, BMedSci, and Michelle Menzies, BSc, Sydney Melanoma Diagnostic Centre, Royal Prince Alfred Hospital; Phoebe Star, MBBS, MPhil, BSc, BA, Melanoma Institute Australia, The University of Sydney; and Leo Raudonikis, Westmead Institute for Medical Research, The University of Sydney.

    References
    1.
    Olsen  CM, Green  AC, Pandeya  N, Whiteman  DC.  Trends in melanoma incidence rates in eight susceptible populations through 2015.   J Invest Dermatol. 2019;139(6):1392-1395. doi:10.1016/j.jid.2018.12.006 PubMedGoogle ScholarCrossref
    2.
    Australian Institute of Health and Welfare. Skin cancer in Australia. Published July 13, 2016. Accessed August 3, 2020. https://www.aihw.gov.au/reports/cancer/skin-cancer-in-australia/contents/table-of-contents
    3.
    Doran  CM, Ling  R, Byrnes  J,  et al.  Estimating the economic costs of skin cancer in New South Wales, Australia.   BMC Public Health. 2015;15:952. doi:10.1186/s12889-015-2267-3 PubMedGoogle ScholarCrossref
    4.
    Australian Institute of Health and Welfare. Cancer in Australia 2019. Published 2019. Accessed August 3, 2020. https://www.aihw.gov.au/getmedia/8c9fcf52-0055-41a0-96d9-f81b0feb98cf/aihw-can-123.pdf.aspx?inline=true
    5.
    Youlden  DR, Youl  PH, Soyer  HP, Aitken  JF, Baade  PD.  Distribution of subsequent primary invasive melanomas following a first primary invasive or in situ melanoma in Queensland, Australia, 1982-2010.   JAMA Dermatol. 2014;150(5):526-534. doi:10.1001/jamadermatol.2013.9852 PubMedGoogle ScholarCrossref
    6.
    Cust  AE, Badcock  C, Smith  J,  et al.  A risk prediction model for the development of subsequent primary melanoma in a population-based cohort.   Br J Dermatol. 2020;182(5):1148-1157. doi:10.1111/bjd.18524 PubMedGoogle ScholarCrossref
    7.
    Müller  C, Wendt  J, Rauscher  S,  et al.  Risk factors of subsequent primary melanomas in Austria.   JAMA Dermatol. 2019;155(2):188-195. doi:10.1001/jamadermatol.2018.4645PubMedGoogle ScholarCrossref
    8.
    Gershenwald  JE, Scolyer  RA, Hess  KR,  et al; for members of the American Joint Committee on Cancer Melanoma Expert Panel and the International Melanoma Database and Discovery Platform.  Melanoma staging: evidence-based changes in the American Joint Committee on Cancer eighth edition cancer staging manual.   CA Cancer J Clin. 2017;67(6):472-492. doi:10.3322/caac.21409PubMedGoogle ScholarCrossref
    9.
    Ruiz  ES, Morgan  FC, Zigler  CM, Besaw  RJ, Schmults  CD.  Analysis of national skin cancer expenditures in the United States Medicare population, 2013.   J Am Acad Dermatol. 2019;80(1):275-278. doi:10.1016/j.jaad.2018.04.035 PubMedGoogle ScholarCrossref
    10.
    Buja  A, Rivera  M, De Polo  A,  et al.  Real-world data for direct stage-specific costs of melanoma healthcare.   Br J Dermatol. 2020;183(1):171-172. doi:10.1111/bjd.18896 PubMedGoogle ScholarCrossref
    11.
    Elliott  TM, Whiteman  DC, Olsen  CM, Gordon  LG.  Estimated healthcare costs of melanoma in Australia over 3 years post-diagnosis.   Appl Health Econ Health Policy. 2017;15(6):805-816. doi:10.1007/s40258-017-0341-y PubMedGoogle ScholarCrossref
    12.
    Lyth  J, Carstensen  J, Synnerstad  I, Lindholm  C.  Stage-specific direct health care costs in patients with cutaneous malignant melanoma.   J Eur Acad Dermatol Venereol. 2016;30(5):789-793. doi:10.1111/jdv.13110 PubMedGoogle ScholarCrossref
    13.
    Cancer Council Australia. Clinical practice guidelines for the diagnosis and management of melanoma. Accessed August 3, 2020. https://wiki.cancer.org.au/australia/Guidelines:Melanoma
    14.
    Moloney  FJ, Guitera  P, Coates  E,  et al.  Detection of primary melanoma in individuals at extreme high risk: a prospective 5-year follow-up study.   JAMA Dermatol. 2014;150(8):819-827. doi:10.1001/jamadermatol.2014.514 PubMedGoogle ScholarCrossref
    15.
    Watts  CG, Cust  AE, Menzies  SW, Mann  GJ, Morton  RL.  Cost-effectiveness of skin surveillance through a specialized clinic for patients at high risk of melanoma.   J Clin Oncol. 2017;35(1):63-71. doi:10.1200/JCO.2016.68.4308 PubMedGoogle ScholarCrossref
    16.
    Adler  NR, Kelly  JW, Guitera  P,  et al.  Methods of melanoma detection and of skin monitoring for individuals at high risk of melanoma: new Australian clinical practice.   Med J Aust. 2019;210(1):41-47. doi:10.5694/mja2.12033 PubMedGoogle ScholarCrossref
    17.
    Salerni  G, Carrera  C, Lovatto  L,  et al.  Benefits of total body photography and digital dermatoscopy (“two-step method of digital follow-up”) in the early diagnosis of melanoma in patients at high risk for melanoma.   J Am Acad Dermatol. 2012;67(1):e17-e27. doi:10.1016/j.jaad.2011.04.008 PubMedGoogle ScholarCrossref
    18.
    Watts  CG, Madronio  CM, Morton  RL,  et al.  Diagnosis and clinical management of melanoma patients at higher risk of a new primary melanoma: a population-based study in New South Wales, Australia.   Australas J Dermatol. 2017;58(4):278-285. doi:10.1111/ajd.12530 PubMedGoogle ScholarCrossref
    19.
    Carli  P, De Giorgi  V, Crocetti  E,  et al.  Improvement of malignant/benign ratio in excised melanocytic lesions in the “dermoscopy era”: a retrospective study 1997-2001.   Br J Dermatol. 2004;150(4):687-692. doi:10.1111/j.0007-0963.2004.05860.x PubMedGoogle ScholarCrossref
    20.
    Menzies  SW, Emery  J, Staples  M,  et al.  Impact of dermoscopy and short-term sequential digital dermoscopy imaging for the management of pigmented lesions in primary care: a sequential intervention trial.   Br J Dermatol. 2009;161(6):1270-1277. doi:10.1111/j.1365-2133.2009.09374.x PubMedGoogle ScholarCrossref
    21.
    Petty  AJ, Ackerson  B, Garza  R,  et al.  Meta-analysis of number needed to treat for diagnosis of melanoma by clinical setting.   J Am Acad Dermatol. 2020;82(5):1158-1165. doi:10.1016/j.jaad.2019.12.063 PubMedGoogle ScholarCrossref
    22.
    Green  AC, Baade  P, Coory  M, Aitken  JF, Smithers  M.  Population-based 20-year survival among people diagnosed with thin melanomas in Queensland, Australia.   J Clin Oncol. 2012;30(13):1462-1467. doi:10.1200/JCO.2011.38.8561 PubMedGoogle ScholarCrossref
    23.
    Aguilera  P, Malvehy  J, Carrera  C,  et al.  Clinical and histopathological characteristics between familial and sporadic melanoma in Barcelona, Spain.   J Clin Exp Dermatol Res. 2014;5(5):231. doi:10.4172/2155-9554.1000231PubMedGoogle Scholar
    24.
    Gassenmaier  M, Stec  T, Keim  U,  et al.  Incidence and characteristics of thick second primary melanomas: a study of the German Central Malignant Melanoma Registry.   J Eur Acad Dermatol Venereol. 2019;33(1):63-70. doi:10.1111/jdv.15194 PubMedGoogle ScholarCrossref
    25.
    Glasziou  PP, Jones  MA, Pathirana  T, Barratt  AL, Bell  KJ.  Estimating the magnitude of cancer overdiagnosis in Australia.   Med J Aust. 2020;212(4):163-168. doi:10.5694/mja2.50455 PubMedGoogle ScholarCrossref
    26.
    McLoone  JK, Watts  KJ, Menzies  SW, Barlow-Stewart  K, Mann  GJ, Kasparian  NA.  Melanoma survivors at high risk of developing new primary disease: a qualitative examination of the factors that contribute to patient satisfaction with clinical care.   Psychooncology. 2013;22(9):1994-2000. doi:10.1002/pon.3243 PubMedGoogle ScholarCrossref
    27.
    van der Leest  RJ, Flohil  SC, Arends  LR, de Vries  E, Nijsten  T.  Risk of subsequent cutaneous malignancy in patients with prior melanoma: a systematic review and meta-analysis.   J Eur Acad Dermatol Venereol. 2015;29(6):1053-1062. doi:10.1111/jdv.12887 PubMedGoogle ScholarCrossref
    28.
    Read  RL, Madronio  CM, Cust  AE,  et al.  Follow-up recommendations after diagnosis of primary cutaneous melanoma: a population-based study in New South Wales, Australia.   Ann Surg Oncol. 2018;25(3):617-625. doi:10.1245/s10434-017-6319-z PubMedGoogle ScholarCrossref
    29.
    Vuong  K, Armstrong  BK, Weiderpass  E,  et al; Australian Melanoma Family Study Investigators.  Development and external validation of a melanoma risk prediction model based on self-assessed risk factors.   JAMA Dermatol. 2016;152(8):889-896. doi:10.1001/jamadermatol.2016.0939 PubMedGoogle ScholarCrossref
    ×