Screening and novel diagnostic technologies for melanoma: an update

Abstract

Melanoma incidence has increased over recent decades, yet mortality has been relatively stable. This pattern has raised concern that many newly diagnosed melanomas, particularly melanoma in situ, may reflect overdiagnosis rather than a true increase in disease burden. Screening can detect melanoma earlier but is likely associated with overdiagnosis and overdetection, which may lead to excess morbidity with little survival benefit. This review examines global trends in melanoma incidence and mortality, the effects of screening programs, and the consequences of overdiagnosis. We evaluate both population-based and risk-directed screening strategies and assess diagnostic tools such as dermoscopy, total body photography, and artificial intelligence devices. Further research is needed to determine how these adjunctive technologies can be effectively integrated into screening strategies to improve clinical outcomes.

ARTICLE HIGHLIGHTS

1. The rise in melanoma incidence is primarily attributed to the increased detection of in situ and thin invasive melanomas with low mortality risk.

2. Melanoma-specific mortality has been relatively stable even as incidence has increased, indicating likely melanoma overdiagnosis.

3. Population-wide skin cancer screening has not demonstrated a sustained mortality benefit and may delay care for high-risk patients.

4. Increased overdiagnosis and overdetection of melanoma can lead to excess morbidity from unnecessary biopsies along with higher healthcare costs and patient distress.

5. Rare and aggressive melanoma subtypes like acral and amelanotic melanomas are often diagnosed late and are inadequately addressed by current screening approaches.

6. Public and physician education should include not only the ABCDE rule and ugly duckling sign but also mnemonics like CUBED, ABCDEF, the 3 Rs, and EFG to improve awareness of acral, amelanotic, and nodular melanoma.

7. Risk-directed and lesion-directed screening strategies improve diagnostic yield and may reduce overdiagnosis and overtreatment.

8. Adjunctive screening technologies may support physicians in screening but raise concerns about diagnostic accuracy and algorithmic bias.

9. Prioritizing population-level education that reaches patients often missed by current screening may reduce melanoma mortality more effectively and affordably than improving screening technology or increasing screening frequency.

1. Introduction

Cutaneous melanoma (CM) is an aggressive form of skin cancer that originates from melanocytes [1]. These melanin-producing cells originate from neural crest cells and are located in the basal layer of the epidermis [2]. When melanocytes transform into melanoma, they can metastasize and develop immune evasion mechanisms [3]. Even though it makes up approximately 1% of all skin cancers diagnosed, CM accounts for 66–83% of skin cancer-related deaths within countries that report high skin cancer mortality [4,5]. This review examines global trends in melanoma incidence and mortality, the balance between screening and overdiagnosis, and the need for more targeted, risk-based screening strategies.

1.1. Epidemiology and incidence

Melanoma is the fifth most common cancer in the United States (US). It accounts for approximately 5% of all newly diagnosed cancers each year. The American Cancer Society estimates that 104,960 new melanomas will be diagnosed and 8,430 people will die from melanoma in 2025 [4]. In 2022 alone, an estimated 330,000 new cases of melanoma were diagnosed, and 60,000 people died from melanoma worldwide [6]. By 2040, the number of cases is projected to increase by 50% worldwide [7]. Among countries reporting data to the World Health Organization (WHO), the US, Australia, and Europe have the highest incidence of melanomas (Figure 1) [8]. In 2018, the age-standardized rates of invasive melanoma were 21 per 100,000 in Scotland, 32 per 100,000 in the US White population, and 73 per 100,000 in Queensland, Australia, demonstrating geographical differences in incidence rates [9]. The highest melanoma incidence rates are reported in Australia in both males and females, with melanoma as the most common cancer among Australians 15–29 years old [10,11]. Skin cancer incurs the highest costs among all cancers within Australian healthcare [12]. Over 75% of all new CM cases are induced by ultraviolet radiation (UVR), 89% of which arise in highly developed countries [13]. The increased incidence has been associated with an increased number of total body skin exams (TBSEs) in certain countries like Australia, the US, and the United Kingdom (UK) [12,14–18].

Figure 1.

The worldwide age-standardized incidence rate for melanoma of skin by GLOBOCAN.

1.2. Melanoma outcome trends

Patients diagnosed with localized melanoma generally have an excellent prognosis with a 5-year relative survival (RS) greater than 99%. However, the 5-year RS declines to 74.8% and 35.0% with regional lymph node and distant metastatic disease involvement, respectively [19]. While most melanoma deaths have occurred in patients who initially presented with localized disease (53%), most deaths were in those with melanomas thicker than 1 mm rather than thin melanomas (≤ 1 mm). Among patients with localized thin melanomas, the cumulative incidence of melanoma-specific mortality at 10 years was only 2.6%, while it was 45.9% for melanomas > 4 mm [20].

Multiple population datasets show that melanoma mortality has declined in most countries [10]. The highest age-standardized rates of melanoma-specific mortality worldwide are in the US, Europe, Australia, South Africa, and Namibia (Figure 2) [21]. Survival outcomes have recently improved in the US, likely due to advancements in melanoma treatment options since 2011 [4,22]. After increasing by 7.5% from 1986 to 2013, melanoma mortality in the US Caucasian population declined by 17.9% between 2013 and 2016 [23]. This decline closely coincided with the introduction of immune checkpoint inhibitors (ICIs), beginning with the Food and Drug Administration (FDA) approval of ipilimumab in 2011, as well as subsequent approvals of targeted therapies and additional immunotherapies not only for metastatic disease but in the adjuvant setting for regional and locally advanced disease [24]. From 2017 to 2020, mortality continued to decline, though at a slower rate of −1.6% annually. Globally, similar trends have been observed. Based on 2019 Global Burden of Disease data, annual incidence rates of melanoma have increased since 1990 across Europe, the US, and Australia, while joinpoint analyses revealed melanoma-specific mortality has declined in all countries except the United Kingdom [25]. At least some of these changes likely reflect the mortality benefits of newer therapies as ICIs and targeted therapies became standard practice [26]. Many more types of therapies have been approved that help continue to decrease the melanoma mortality rate since 2013 [4], including dual immunotherapy in 2015, adjuvant PD-1 therapy in 2017, triple combination therapy in 2020, combined relatlimab with nivolumab in 2022 (first-line treatment option for metastatic and unresectable melanoma based on the National Comprehensive Cancer Network 2025 guidelines), and tumor-infiltrating lymphocyte (TIL) therapies like lifileucel in 2024 [26–31].

Figure 2.

The worldwide age-standardized mortality rate for melanoma of skin by GLOBOCAN.

While melanoma mortality has either stabilized or declined with the advent of ICIs and targeted therapy, keratinocyte carcinoma (KC) mortality continues to rise. With these shifting trends, KC-related mortality is projected to surpass melanoma mortality in Scotland by 2028, in the US by 2031, and in Australia by 2033 [5].

1.3. Classification and screening implications

There are four main histologic subtypes of cutaneous malignant melanoma: superficial spreading melanoma (SSM), lentigo maligna melanoma (LMM), acral lentiginous melanoma (ALM), and nodular melanoma (NM). SSM is the most common (70–79% of CMs), followed by NM (15–20%), LMM (5–8%), and ALM (2–3%) [32,33]. The distribution of melanoma subtypes widely varies between racial and ethnic populations. Even though ALM is the rarest CM subtype in non-Hispanic White (NHW) populations, it makes up the largest proportion of melanomas in East Asian (50–58%), Hispanic/Latin-American (49–61%), and Black/African-American populations (39–75%) [34–37].

The 2018 WHO classification of melanoma introduced a more detailed framework, categorizing melanomas into nine distinct evolutionary “pathways” based on clinicopathologic, epidemiologic, and genetic characteristics: (1) low-cumulative solar damage melanoma/SSM), (2) high-cumulative solar damage melanoma/LMM, (3) desmoplastic melanoma (DM), (4) Spitz melanoma (SM), (5) acral melanoma, (6) mucosal melanoma (MM), (7) melanoma arising in congenital nevi, (8) melanoma arising in blue nevi, and (9) uveal melanoma. NM does not belong to a single pathway, as it can develop through multiple mechanisms [38,39].

Most skin cancer screening programs focus on detecting melanomas in chronic or intermittently sun-exposed areas, including the trunk, head, neck, and non-acral extremities, where SSM and LMM subtypes predominate. These subtypes are more frequently diagnosed in NHW populations, whereas ALM, which often occurs on the palms, soles, and nail unit, occurs in individuals of all skin tones [40]. Standard skin cancer screening approaches may be less effective at identifying non-UV-induced melanomas, such as acral melanoma, since they typically present in anatomical sites that receive less attention during routine skin exams [41]. The US Preventive Services Task Force (USPSTF) 2023 Recommendation Statement also notes the paucity of data on screening for ALM [42].

1.4. Melanoma genesis

CM arises through multiple tumorigenic pathways. Over two-thirds of cases develop de novo rather than from preexisting nevi, which represent only 21–28% of CM [43–45]. Nevus-associated melanomas (NAM) tend to develop more frequently in intermittently sun-exposed anatomic locations like the trunk and shoulders, whereas de novo melanomas arise more often in chronically sun-exposed areas like the scalp, head, and neck [46]. NAMs are more common in younger individuals, with the likelihood of melanoma originating from a preexisting nevus decreasing with age. One study found that NAMs account for 64% of melanomas in patients under 20 years old but only 11% in those over 90 years old [44]. The transition from benign nevus to melanoma in NAM also follows an age-related pattern. Another study found that the average age of patients diagnosed with mild or moderate dysplastic nevi (DN) was 35 years, compared to 42 years for severe DN and 47 years for invasive melanoma [47].

The main risk factors for melanoma include prolonged UVR exposure, lighter skin phototypes (Fitzpatrick skin type I-II), family history of melanoma, increased number of melanocytic nevi, and inheritance of certain genetic risk factors [48–50]. BRAF mutations are the most commonly linked to CM, representing 41–50% of all cases [51]. NRAS mutations are the second most common mutation, representing about 15–20% of all CMs [52]. Most cutaneous melanomas are directly related to UVR exposure (60–90%), specifically UVA (315–400 nm) and UVB (280–315 nm), both of which cause cumulative DNA damage [48,53]. UVC (200–290 nm) is mostly absorbed by the ozone layer due to its shorter wavelength, so it has limited clinical relevance [49].

Even though UVR exposure is estimated to be responsible for most CMs, some genetic factors confer an intrinsic susceptibility to melanoma independent of UVR [53,54]. The melanocortin-1-receptor (MC1R) gene regulates pigmentation via cyclic adenosine monophosphate (cAMP) signaling. Loss-of-function MC1R variants, common in Fitzpatrick skin types I-II and individuals with red hair, predispose carriers to melanoma irrespective of UV exposure. These pathogenic variants decrease the synthesis of eumelanin, a photoprotective pigment, while increasing pheomelanin, a genotoxic pigment that provides inadequate UV protection and generates reactive oxygen species (ROS) in response to UVA [55,56].

Similarly, CDKN2A mutations increase melanoma risk by disrupting tumor suppression rather than through UV-induced mutagenesis. These mutations interfere with the p16INK4a and p14ARF pathways, which leads to uncontrolled cell growth and melanoma development regardless of UVR exposure [57]. CDKN2A mutations are found in approximately 20–40% of melanoma families [58] and are implicated in familial atypical multiple mole melanoma (FAMMM) syndrome [59,60].

Unlike CM, rare melanoma subtypes like mucosal and acral melanoma are unrelated to UVR exposure. Even though their pathogenesis has not been fully elucidated, it is well-established that the mucosal and acral melanoma genesis pathways are different from UV-induced CM [54]. Mucosal melanoma exhibits low tumor mutational burden with SF3B1, KIT, and NF1 as key drivers of melanoma [61]. Acral melanoma, found on palms, soles, and subungual sites, is also associated with a lower tumor mutational burden and is more often associated with copy number amplifications in CDK4, CCND1, KIT, and TERT rather than point mutations [62].

2. Detection of melanoma

2.1. Public awareness campaigns

Public awareness campaigns have been central to melanoma prevention and early detection efforts. Nationwide initiatives like “Slip! Slop! Slap!” and SunSmart in Australia have likely contributed to a decline in melanoma incidence among younger populations, with rates for 15–24 year-olds decreasing since the mid-1990s [12,63]. By 2009, the SunSmart Primary School and Early Childhood Programs were integrated into all Australian states and territories [63]. From 1988 to 2003, SunSmart is estimated to have prevented 9,000 melanomas and reduced skin cancer-related deaths in Victoria by 1,000 [64].

Educating the public about sun exposure and melanoma risk is important since approximately half of CMs are detected by patients themselves [65,66]. Teaching individuals to recognize early warning signs may improve early diagnosis and survival outcomes. Many of these campaigns now incorporate the ABCDE mnemonic in their education, including the American Academy of Dermatology (AAD) [67], but there is mixed evidence on its benefit for use by the general population. While the ABCDE criterion has been shown to improve sensitivity in identifying concerning lesions, it also lowers specificity and increases the false-positive rate in the general population [68]. The “ugly duckling sign,” in which a suspicious melanocytic lesion looks different from the patient’s other nevi, is more specific than and similarly sensitive to the ABCDE mnemonic for patients [69].

Public education campaigns for rare melanoma subtypes, such as acral or amelanotic melanoma, are still lacking. Unlike CM, acral melanoma is not strongly associated with UV radiation, and its delayed recognition often results in late-stage diagnoses [62]. Amelanotic melanoma, which lacks pigmentation and does not fit the typical visual cues of melanoma, is frequently misdiagnosed or identified at a more advanced stage [70]. Late-stage diagnosis of melanoma carries a significantly worse prognosis. The diagnostic mnemonics CUBED and ABCDEF were developed to aid in identifying acral and subungual melanoma, respectively [71]. The 3 Rs (Red, Raised, with Recent change) mnemonic has been proposed to help detect more amelanotic melanomas [72]. The EFG rule (Elevation, Firmness to touch, Growth) has been developed to identify NM subtypes [73]. Compared to the ABCDE mnemonic, these rarer melanoma subtype memory aids are not well-known. Both primary care physicians (PCPs) and patients are often unaware of how to spot most rare melanoma subtypes [71,72]. ALMs have a much higher misdiagnosis rate (34%) than melanomas on other anatomic sites and have been mistaken for many other conditions, including warts, diabetic foot ulcers, and onychomycosis [74,75]. More education on how to identify rare melanoma subtypes for both the general population and PCPs might help reduce delayed diagnoses and improve diagnostic accuracy [76].

2.2. Nationwide or population-based skin cancer screening efforts

Detecting melanoma at an early stage while it is still thin has been viewed as a core component for improving outcomes and reducing mortality from melanoma [20,77]. The rationale for melanoma screening was that early excision of thin lesions can be curative, whereas later-diagnosed melanomas are more advanced with a much worse prognosis [20]. Formal skin cancer screening usually refers to a TBSE by a dermatologist or a PCP in dermatologist-scarce regions in asymptomatic individuals. Unlike many screening programs for other cancers, such as colorectal or breast cancer, there is no substantial evidence that population-based skin cancer screening reduces melanoma mortality [42,78,79].

More studies are needed to assess the risks and benefits of population-based melanoma screening more conclusively. As of June 2025, there have been no randomized control trials of melanoma screening, primarily due to a lack of feasibility [80]. Available evidence regarding universal or nationwide skin cancer screening is limited to studies from individual countries and single-institution pilot programs (Table 1). While more melanomas are identified because of increased screening and societal awareness through public awareness campaigns [12,14], most associated melanomas caught from screening are thin [78,83]. A recent prospective cohort study of skin cancer screening in primary care clinics found that screened patients had a higher incidence of melanoma in situ (MIS) or thin invasive melanomas. They also reported a nearly significant trend toward reduced odds of thick melanomas among those aged 65 and older who were screened compared to those not screened [78]. The German skin cancer pilot program, called Skin Cancer Research to Provide Evidence for Effectiveness of Screening in Northern Germany (SCREEN), showed a 48% reduction in melanoma mortality 5 years after starting the initiative, which prompted a nationwide skin cancer screening initiative in 2008. However, this benefit was not sustained, and subsequent analyses at the 10-year mark found no significant long-term impact on melanoma-related mortality [42,84,85].

Table 1.

Universal or nationwide skin cancer screening programs.

Author (year)	Years	Location	Name of screening program (if applicable)	Who was eligible? (if applicable)	# Patients screened (TOTAL)	No. Confirmed melanoma In situ (%)	No. Confirmed melanoma (%)	No. Melanomas ≤ 1 mm (%)	No. Melanomas > 1 mm (%)	Positive predictive value (%)	Effect on OS	Effect on MSS
Koh et al. 1996 [81]	1992–1994	United States	National Skin Cancer Early Detection and Screening Program	Any individual	282,555	151 (0.05)	364 (0.13)	NA	NA	17	NA	NA
van der Leest et al. 2011 [15]	2009–2010	Europe	Euromelanoma	Any individual	260,000	NA	136 (0.35)	NA	NA	13	NA	NA
Boniol et al. 2015 [77]	2003–2012	Schleswig-Holstein, Germany	Skin Cancer Research to Provide Evidence for Effectiveness of Screening in Northern Germany (SCREEN)	Individuals aged ≥ 20 years	NA	NA	NA	NA	NA	NA	NA	Initial reduction in MSS by 50% in 5 years following intervention, change not sustained
Okhovat et al. 2018 [14]	1992–2010	United States	SPOTme	Patients aged 15–90 years	1,299,765	NA	14908 (1.15)	NA	NA	NA	NA	NA
Matsumoto et al. 2022 [72]	2014–2018	Pittsburgh, United States	NA	Primary care patients aged ≥ 35 years	144,851	172 (0.12)	246 (0.17)	132 (0.09)	52 (0.04)	NA	NA	NA
Whiteman et al. 2022 [82]	2010	Queensland, Australia	NA	Patients aged 40–69 years	28,155	556 (0.02)	967 (0.03)	NA	NA	NA	NA	NA

Abbreviations: OS, overall survival; MSS, melanoma-specific survival; HR, hazard ratio; MIS, melanoma in situ.

Even though universal screening for all adults has not been shown to reduce melanoma mortality in a significant way, there may be consequences to eliminating dermatologist access to skin cancer diagnosis and screening. Between 2019 and 2020, when the COVID-19 pandemic started, in-person TBSEs were reduced by an estimated 34%, and melanoma diagnoses decreased by 16% in the US [86,87]. Multi-institutional and cancer registry studies from the US, Chile, France, Belgium, Spain, and the UK have found that the pandemic resulted in reduced melanoma incidence [7,86, 88–91]. Delays in routine dermatologic care led to a measurable shift in melanoma presentation in the US, with a decrease in early-stage diagnoses [86]. The Global Coalition for Melanoma Patient Advocacy estimates that 34% of skin checks were missed, and 21% of melanomas went undiagnosed during the pandemic [87]. The long-term consequences of these pandemic-era changes in melanoma diagnoses have yet to be elucidated and will require future studies during post-pandemic follow-up.

2.3. Targeted and risk-directed screening initiatives

A recent USPSTF 2023 report found insufficient evidence for recommending nationwide routine skin cancer screening in asymptomatic adolescents or adults to reduce melanoma-related mortality. Their recommendation is consistent with the prior 2016 USPSTF recommendation [42,92]. Targeted, risk-directed screening is a much more feasible alternative and has been shown to have higher sensitivity and specificity [93]. Many risk stratification tools and models have been developed to best determine who is at higher risk of melanoma, including the Melanoma Institute Australia (MIA) First Primary Melanoma Risk Prediction Tool, the Self-Assessment of Melanoma risk score (SAMScore), and the Mackie risk stratification tool [94–99]. These models incorporate different sets of risk factors, so their predictive abilities can vary depending on the clinical setting and patient population. None have been fully validated for widespread use, and their clinical applicability is mainly limited to the countries where they were developed [100].

Some overlapping risk factors between these risk stratification models include those with fairer skin tone, a high total nevus count, the presence of atypical nevi, freckling, lighter hair color (red or blond), a family history of melanoma, a history of severe sunburns, and residence in regions with high UVR exposure [95,97, 101–104]. Certain populations with a higher risk of melanoma likely should continue to receive regular TBSEs. From 2009 to 2013, NHW men over 50 years old made up 56% of all melanoma-related deaths [105]. This subgroup should be considered in any risk-directed screening. A cost-effectiveness analysis by Adamson et al. reported that screening NHW men every two years, starting at age 50 or 60 and continuing until age 75, may be cost-effective. Screening starting at age 50 in this group was estimated to reduce melanoma mortality by 30% with an incremental cost-utility of $67,970 USD per quality-adjusted life year (QALY) while starting at age 60 had an estimated reduction in melanoma mortality by 20% with an incremental cost-utility of $26,503 USD per QALY [106]. Patients with CDKN2A pathogenic variants have an estimated 30–70% lifetime risk of melanoma and are recommended to have biannual TBSEs beginning at 10 years old [59].

Lesion-directed screening (LDS) is a targeted skin cancer screening approach proposed as an alternative to TBSE. Because it limits skin cancer screening to only patients who present with concerning lesions based on predefined criteria (e.g., ABCD rule, ugly duckling sign), LDS can help alleviate strain on dermatology resources and time. A study comparing LDS to universal TBSEs in patients older than 18 found that LDS took nearly six times less time while maintaining a similar skin cancer detection rate (3.2% vs. 2.3%) [107]. An observational study of LDS that included only patients who passed through a phone question screen regarding the lesion found a 13.2% skin cancer detection rate (4.1% for melanoma), several times higher than population-based screening detection rates [108].

A few other examples of targeted skin cancer screening programs (Table 2). A targeted screening program at the Veterans Affairs (VA) Palo Alto Health Care System (VAPAHCS) focused on screening veterans at higher risk of skin cancer detected skin cancer or precancerous lesions in 54% of participants [109]. An Australian targeted skin cancer screening initiative found that routine skin screenings were associated with significantly lower all-cause mortality but not melanoma-specific mortality [65]. An Austrian pilot skin cancer targeted screening initiative, however, was surprisingly associated with increased melanoma mortality compared to the general population. This higher mortality may reflect the screened population’s elevated baseline risk rather than an effect of screening itself. Without a direct comparison to an unscreened high-risk group, since the general population was the comparator, the impact of targeted screening efforts on melanoma-specific mortality remains unclear [79]. More studies are needed to clarify the effectiveness of targeted screening strategies in improving melanoma outcomes for patients at higher risk of developing melanoma.

Table 2.

Targeted and risk-directed skin cancer screening initiatives.

Author (year)	Years	Location	Program name	Eligibility	# Patients screened	No. Confirmed melanoma (%)	Thin melanomas (≤1 mm), n (%)	Thick melanomas (>1 mm), n (%)	Positive predictive value (%)	OS HR (95% CI)	MSS HR (95% CI)
Swetter et al. (2003) [102]	1999–2002	Northern California, USA	NA	veterans with increased risk of skin cancer	374	1 (.003%), screening focused on both keratinocyte carcinoma and melanoma. Among 203 veterans with suspicious lesions, only 101 followed up and only 36 underwent a biopsy.	NA	NA	12.5%	NA	NA
Hoorens et al. (2016) [100]	2014	East Flanders, Belgium		Individuals with a lesion meeting ≥ 1 criterion: ABCD rule, ugly duckling sign, new lesion >4 weeks, or red nonhealing lesions	314	8 (2.5%)	NA	NA	56.6%	NA	NA
Watts et al. (2021) [59]	2006–2018	New South Wales, Australia	NA	patients diagnosed with a histologically confirmed primary MIS or invasive cutaneous melanoma	2452	858 (35%)	709 (83%)	146 (17%)	NA	0.75 (95% CI, 0.63–0.90; P = .006)	0.68 (95% CI, 0.44–1.03; P = .13)
Rat et al. (2021)	2011–2015	Loire Atlantique and Vendee, France	NA	patients at high risk for developing melanoma based on the SAMScore	3917	3169 (81%)	2425 (77%), MIS and stage 1	716 (23%)	NA	NA	NA
Mylle et al. (2021) [21]	2017–2019	Ghent, Belgium		Patients with 1–2 concerning lesions meeting ≥1 criterion (ugly duckling, non-healing lesion, new/changed nevus) and/or referred by a non-dermatologist contacted dermatology via telephone for screening.	342	14 (4.1%)	NA	NA	NA	NA	NA

Abbreviations: OS, overall survival; MSS, melanoma-specific survival; HR, hazard ratio; MIS, melanoma in situ.

3. Overdiagnosis

Overdiagnosis occurs when a cancer is detected that would not have caused symptoms or led to death [110]. This has been a concern with melanoma screening, as well as with thyroid, prostate, breast, and lung cancer [82,111–113]. Population and registry studies have shown an increased incidence of melanoma over the last 25–50 years [114–116].

Increased incidence alone does not establish overdiagnosis. A more reliable indicator is stable or declining mortality in the context of unchanged treatment practices, as seen in melanoma cases diagnosed in the US prior to the introduction of systemic therapies in 2011 [83]. Data from the Surveillance, Epidemiology, and End Results (SEER) program and the National Vital Statistics System indicate that melanoma incidence continued to increase while mortality remained stable until 2014 [83]. Between 1975 and 2017, overall melanoma incidence increased by 41.6 cases per 100,000 (+459%) and by 20.1 per 100,000 (+236%) for invasive melanoma, whereas mortality rose by only 0.8 per 100,000 from 1975 to 2010 and decreased by 0.6 per 100,000 between 2010 and 2017 [117]. The very recent decline in mortality may be attributable to the FDA approval of novel immune- and targeted therapies since 2011. From 1975 to 2017, the most significant increase in incidence was observed in MIS, which rose by 25.4 per 100,000 (+4,675%) [117]. Subgroup analyses of the same data suggest that overdiagnosis is likely more pronounced in younger and middle-aged females, whereas older males experience a combination of overdiagnosis and a true increase in disease burden [117]. From 2010 to 2019, SEER also documented an increase in T4 melanoma incidence and a decline in T1–T3 cases within metropolitan areas [118]. These shifts may be related to differences among subpopulations that warrant further investigation.

MIS likely plays a significant role in overdiagnosis. Although patients with MIS are at increased risk of developing subsequent melanomas, which may indicate some clinical benefit to earlier diagnosis, recent epidemiologic data raise concerns about overdiagnosis. A 2022 study reported that the US’s MIS to invasive melanoma incidence ratio rose from 0.14 in 1982 to 0.93 in 2018, with the steepest increases observed among older adults [9]. Another ecological analysis estimated that 85–89% of MIS diagnoses in 2018 were likely attributable to overdiagnosis [119]. A US population-based cohort study found that patients with a first-and-only MIS had a 15-year melanoma-specific survival of 98.4%, an RS of 112.4%, a melanoma-specific standardized mortality ratio (SMR) of 1.89, and an all-cause SMR of 0.68. These outcomes are consistent with low-risk melanoma detected in health-seeking individuals who, on average, are healthier and live longer than the general population [120].

Findings suggestive of overdiagnosis have also been detected in other countries. A cancer registry study from Denmark examining melanoma diagnosed between 1999 and 2019 detected an increase in melanoma incidence, with the highest percent increase being a 476% increase in MIS in males. In 2021, rates of overdiagnosis in Australia were estimated at 71–76% for MIS and 28–34% for thin melanomas [121]. An Australian registry study also found higher melanoma and biopsy rates in screened versus unscreened patients but no significant difference in invasive melanoma incidence after adjusting for covariates [122]. This pattern points to a disconnect between the rising detection of early-stage melanomas and relatively stable rates of thick melanomas. The premise of effective cancer screening is that earlier detection should ultimately lead to a reduction in advanced cases. The increasing incidence of thin melanomas without a corresponding decrease in thick tumors may be due to either overdiagnosis or a true rise in melanoma occurrence. However, if incidence is truly increasing, it would be atypical for that trend to selectively involve only thin melanomas without a parallel increase in thicker melanomas.

Multiple theories have been proposed to explain the rise in melanoma diagnoses. One theory is that ultraviolet radiation (UVR) exposure trends may have changed over time. UVR is a well-known carcinogen, and although the number of sunburns appears to increase the risk of melanoma, long-term low-dose or moderate exposures do not increase the risk of melanoma, independent of hair and skin color [123]. Based on estimates using historical data, decreases in ozone layer density between 1970 and 2000 likely resulted in a 7% increase in sunburn-causing UVR in winter and spring at mid-latitudes and 4% in summer and fall [124], which could potentially account for a subsequent increase in melanoma incidence many years later. A population-based US study of the SEER registry found no correlation between UVR exposure and melanoma or MIS incidence, although it should be noted that UVR exposure was inferred with regional UVR data at the county level and did not account for lifestyle confounders [125]. Additionally, indoor tanning bed use rates have likely not risen fast enough to account for the increase in melanoma incidence [126,127]. Among EU residents, summer travel to foreign countries increased from 2012 to 2019 [128], but further investigation is needed to substantiate the connection to melanoma incidence since melanoma diagnoses would likely lag behind increases in sun exposure.

Another possible explanation for the higher rates of melanoma detection could be increased surveillance in patients with elevated genetic risk. Hereditary cancer syndromes, such as familial atypical multiple mole melanoma, MC1R polymorphisms, xeroderma pigmentosum, and BAP1 tumor predisposition syndrome, can predispose patients to develop melanoma [129–132]. In France, relaxed rules for genetic testing for pathogenic mutations in genes associated with melanoma susceptibility led to the detection of mutations in up to 6.5% of patients, leading the authors to suggest that genetic testing should be performed in patients with two or more melanomas or cancers in a patient or family, called the “rule of 2” [133]. More recent studies have estimated germline pathogenic variants in 10.6–15.8% of patients with melanoma [134]. Using polygenic risk scores (PGS) for melanoma, one study found that melanoma incidence increased slightly with a higher PGS and was mildly correlated with higher screening rates, but there were no statistically significant differences by PGS tertile [135]. These findings suggest that increased screening in high-risk individuals with a genetic predisposition is unlikely to contribute to higher rates of melanoma detection.

One factor that may be contributing to an increase in the diagnosis of melanoma and makes comparisons across decades difficult is an improvement in screening technologies, such as dermoscopy. Dermoscopy is more accurate than visual inspection alone, with a 2018 Cochrane review and meta-analysis showing increased sensitivity and specificity for melanoma detection [136]. A single-center retrospective study found that 40% of melanomas detected with dermoscopy did not meet the diameter criterion (size > 6 mm) of the ABCD rule, indicating that more melanomas may be detected with increased use of dermoscopy [137]. Reported use of dermoscopy among PCPs varies widely, from only 8% in France and the Netherlands to 40% in Australia and 81–87% in the US and Europe [138–142]. Experience and training with dermoscopy improve diagnostic accuracy, as shown in a study where family medicine residents had significantly higher diagnostic scores after a dermoscopy course [143]. Dermoscopy is increasingly used in primary care in some countries like Australia, but clinically distinguishing benign from malignant lesions can be challenging for dermatologists and even more so for PCPs, advanced practice providers, and less experienced dermatologists [144,145]. Data on interobserver reliability with dermoscopic diagnosis is limited. One study investigating the reliability of dermoscopic criteria found poor to fair interrater agreement for many individual dermoscopic criteria, but high diagnostic accuracy for melanoma with many of these criteria. Greater standardization of these criteria is warranted. However, this analysis did not stratify by level of training, and only 56% of participants were dermatologists. Reliability may be higher among more experienced providers who are more familiar with dermoscopy [146]. In the context of diagnostic uncertainty, clinicians may have a lower threshold for biopsy to avoid missing a malignant lesion. This approach may increase the likelihood of diagnosing lesions that would not have progressed, thereby contributing to overdiagnosis [147].

Physician and screening-mediated effects are likely major contributing factors to melanoma overdiagnosis. Patients may undergo more frequent skin examinations, or physicians may have a lower threshold for biopsy. Dermatopathology overcalling of lesions also likely plays a significant role in the rise of melanoma diagnoses. In a study that asked dermatopathologists to reevaluate melanocytic specimens 20 years later, 14% of cases initially identified as dysplastic nevus with severe atypia were upgraded to melanoma [148]. Subsequent intra- and interobserver pathologic agreement studies have similarly identified low reproducibility for severely dysplastic nevi and MIS [149–151].

Furthermore, teledermatology is increasingly used for melanoma screening. Teledermatology programs have been associated with a low number needed to excise for melanoma and a greater probability of thin melanoma at diagnosis with a more favorable prognosis, but implications for overdiagnosis remain unknown [152,153].

3.1. Economic considerations in melanoma screening and prevention

The financial impact of melanoma screening should be considered alongside the consequences of both overdiagnosis and delayed detection. Diagnosing clinically insignificant lesions increases procedural costs and burdens dermatologic services more. The annual increased costs from melanoma overdiagnosis are up to AUD $21.4 million in Australia [121].

Delayed diagnoses can lead to more advanced melanomas, resulting in increased treatment costs and a large financial burden. In Europe, COVID-19-related diagnostic delays for melanoma were estimated to cost an additional $7.65 billion and 111,464 years of life [154]. According to the Irish National Cancer Registry, healthcare costs associated with stage IV melanoma are more than 25 times than those for stage IA melanoma [155]. Following the advent of immunotherapy and targeted therapy for advanced melanoma in the past decade, costs have multiplied. Between 2018 and 2019, the average treatment costs for stage III and stage IV melanoma increased to $67,108 and $117,450, respectively, compared to $46,511 and $47,739 during 2007–2012 [156].

Another concern is the cost of skin cancer screening for patients [157]. It can significantly vary based on the area, but the proportion of dermatologists that accept Medicaid insurance is only 11–34% [158–160]. Under US law, Medicare is only required to cover screening services that receive a grade A or B from the USPSTF (42 U.S.C. § 1395x(ddd)(1)) [81]. Therefore, skin cancer screening in asymptomatic adults, which was assigned an “I” statement by the USPSTF in 2023 [42], is not covered by Medicare. Still, as discussed in Section 2.3, a study comparing systematic melanoma screening to no screening found that screening for melanoma every two years starting at age 60 reduced melanoma mortality by 20% and was deemed cost-effective [106]. While the AAD SPOT Skin Cancer Screening Program provides a platform for dermatologist AAD members to share with patients about free skin cancer screenings, 54% (27/50) of states do not have a single location listed. These gaps in access may contribute to well-established differences in stage at diagnosis and prognosis among patients with reduced access, such as those living in rural communities [161].

Public education campaigns have proven to be very cost-effective primary prevention strategies in reducing healthcare costs associated with melanoma. It is estimated that Australia gained an estimated AUD $2.30 for every dollar spent on its SunSmart skin cancer prevention program and saved 28,000 disability-adjusted life-years in the first two decades of its inception [64]. A national skin cancer prevention program in the US is estimated to save USD$2.7 billion from melanoma treatment between 2020 and 2030 [162].

The cost of increasing skin cancer screening should be weighed against primary preventative measures such as public education that would be cheaper to implement at scale. Thicker melanomas are more often found in patients with limited knowledge of melanoma [163]. Improving public understanding of early melanoma signs may have a greater impact on reducing the burden of advanced disease than increasing screening frequency or incorporating expensive screening technologies that have limited demonstrated benefits [164].

3.2. Psychological impacts of skin cancer screening

The effects of skin cancer screening are often assessed through incidence, mortality, and cost-effectiveness. The psychological impacts, however, are more difficult to quantify. A US population-based study found an increased incidence of suicide in patients diagnosed with melanoma [165]. Though it is unknown whether the melanoma diagnosis or an unmeasured confounder was driving suicide rates, concerns have been raised that expanded screening could lead to unnecessary psychological distress due to false positives [83,166]. However, data from the University of Pittsburgh Medical Center Internet Curriculum for Melanoma Early Detection (INFORMED) program indicate that screening conducted by PCPs who completed the INFORMED training does not increase anxiety or depression, and patients undergoing biopsies for suspicious lesions often report better emotional outcomes, as measured by the Psychological Consequences of Screening Questionnaire (PCQ). These results suggest that screening and biopsy may not necessarily heighten distress and, in some cases, could provide reassurance [167,168]. On the other hand, another study found that many patients diagnosed with localized melanoma experience persistent fear and anxiety of recurrence, even if their prognosis is favorable. Overdiagnosis and overdetection can cause lasting psychological distress and should be factored into discussions on the potential harms of identifying indolent disease [169].

3.3. New modalities for screening

Advancements in technology have led to more options for screening. Total body photography (TBP), for example, improves documentation of lesion progression and may facilitate earlier melanoma detection in high-risk patients [170–173]. TBP allows lesions to be compared over time, resulting in fewer unnecessary biopsies in cases of uncertainty about the stability of nevi. In support of this potential benefit, a 2016 retrospective study demonstrated a reduction in biopsies of nevi from 1.62 per year to 0.34 per year following the implementation of TBP in a pigmented lesion clinic [174]. A meta-analysis of cohorts using TBP in high-risk individuals, including 10 studies of 41,703 patients and 6,203 biopsies through 2020, reported a mean number-needed-to-biopsy (NNB) of 8.6 to diagnose one melanoma. A prior meta-analysis through 2018 found an NNB of 7.5 using standard clinical examination by dermatologists, although differences in study design limit direct comparison [147,175]. The TBP meta-analysis lacked a comparator group, and no studies have evaluated whether adding TBP to routine dermatologic care improves outcomes over standard care alone. Three-dimensional (3D) TBP systems incorporate automated assessment of lesion size, color, and symmetry to improve diagnostic accuracy and reduce unnecessary biopsies [176]. At the time of writing, only one randomized controlled trial has evaluated 3D TBP, which compared 3D TBP through teledermatology to usual care and found higher excision rates and MIS but fewer invasive melanomas detected in the intervention group [177]. However, clinicians performing excisions in the intervention arm only saw standard clinical images, not longitudinal 3D TBP images, which precluded the assessment of the potential of 3D TBP to reduce excisions by documenting lesion changes over time.

MelaFind is a multispectral digital dermoscopy device combined with computer vision that has demonstrated high sensitivity in clinical trials [178,179]. It received FDA approval in 2011 for use by dermatologists but was discontinued in 2017 due to low specificity and concerns regarding additional skin biopsies [180]. Nevisense, a medical device using electrical impedance spectroscopy for melanoma detection, was approved in 2017 following a randomized, blinded clinical trial that showed a sensitivity of 96.6% [181,182].

Artificial intelligence (AI) has prompted both research and commercial efforts to develop clinical tools that assist with melanoma screening. The International Skin Imaging Collaboration AI working group published a checklist to guide the development of training datasets and the performance evaluations of AI models in dermatology [183]. Although most current technologies that incorporate AI into clinical workflows for screening were evaluated years before these guidelines were released, regulatory requirements tend to be stringent on the use of AI for diagnostic purposes [184]. In Europe, an AI algorithm for skin cancer detection with dermoscopic images, Deep Ensemble for the Recognition of Malignancy (DERM), was recently approved for autonomous use and deployed in the UK after a prospective masked study demonstrated a negative predictive value of 99.8% for detection of melanoma in a prospective masked study [185], but its approval and deployment still require human oversight in cases deemed to be positive [186]. These requirements align with the perceptions of both patients and clinicians. Studies report that patients felt safer when AI screening was combined with dermatologist evaluation than with either alone, and most dermatologists believed improving AI would increase diagnostic accuracy [187,188].

Elastic scattering spectroscopy (ESS), a tool that measures changes in scattering and absorption of near-infrared light to detect melanomas, is a novel tool for melanoma screening by non-dermatologists. Although chromophores like hemoglobin and melanin absorb light at shorter wavelengths, malignant pigmented lesions demonstrate differences in scattering at both short and long wavelengths compared to benign pigmented lesions, allowing their differentiation with ESS [189]. One ESS device, DermaSensor, applies a machine learning algorithm to aid with classification and has been evaluated in prospective clinical trials, which have demonstrated sensitivity of 85–95% and specificity of 26–40% [190–192]. Notably, minimal training is required to use the device; when used in combination with clinical judgment in a cohort of PCPs, the area under the receiver operator curve (AUROC) for diagnosis of malignancy increased from 0.619 to 0.683 with the addition of ESS [191]. In a multicenter investigator-blinded prospective trial with dermatologists in specialty pigmented lesion clinics, there was no difference in AUROC between ESS device and dermatologist-investigator predictions [192]. In 2024, DermaSensor was approved by the FDA for non-dermatologist physicians to determine whether to refer a patient to a dermatologist [184]. These clinical trials support the device’s adjunctive use for assessing both pigmented and non-pigmented lesions in primary care settings. However, all studies excluded lesions in non-accessible sites, limiting available performance data in these areas. The specificity of all currently available devices also remains low, increasing the risk of inappropriate referrals, unnecessary biopsies, and overdiagnosis.

Smartphone applications using AI are now publicly available for patients to assess their own skin lesions. However, evaluations of these applications have been inconsistent since they often exclude low-quality images, use images taken by researchers instead of patients, or include both premalignant and malignant lesions in performance analyses [193]. There are currently no approved smartphone applications for skin cancer screening in the US or Canada. A recent study of 25 applications claiming to identify melanoma reported an average sensitivity of 28%, specificity of 81%, and overall accuracy of 59% [194]. Most evaluations of these applications have taken place in controlled settings. In the few studies performed in real-world settings, diagnostic accuracy was lower than that of PCPs and dermatologists, especially when multiple lesions were present [195]. These findings suggest a significant risk of missed diagnoses due to high false-negative rates [196]. More refinement and prospective validation are needed before these mobile applications can be incorporated into routine screening. However, even aside from the aforementioned technologies, both medical devices and smartphone applications represent a rapidly evolving area, with a range of lesion-targeted and total body devices currently under development.

4. Conclusion

Skin cancer screening strategies face an ongoing challenge of balancing the admirable goal of earlier detection with the risk of overdiagnosis of melanoma. While increased awareness and screening efforts have led to higher detection rates, much of the rise encompasses MIS and thin melanoma. These ecological trends have raised concerns that increased screening may predominantly detect lesions that would not have caused harm if left untreated. At the same time, failure to identify late-stage melanoma can result in much worse patient outcomes and high treatment costs.

Nationwide screening programs have not demonstrated consistent reductions in melanoma-specific mortality, and no randomized trials have been conducted. Routine screening of asymptomatic individuals is not supported by current evidence and is unlikely to be sustainable due to primary care and dermatology workforce constraints. Risk-stratified screening may be more feasible, but its effectiveness has not been adequately studied. Devices like TBP, AI, and polygenic risk scores may help in lesion assessment and risk prediction, but their integration into screening protocols requires further validation. Prospective studies are needed to determine whether these risk-directed screening approaches or adjunctive technologies improve melanoma detection and patient outcomes.

5. Future perspective

Melanoma incidence and mortality have been studied across many countries, but several rarer subtypes are still poorly characterized. Mucosal and acral melanomas have a high mortality rate and are frequently missed by routine screening due to their anatomic location. Amelanotic melanoma is often diagnosed at later stages, yet national data temporal trends in detection are limited [70].

AI tools have been introduced as adjuncts to skin cancer screening, but their performance in clinical settings is not well established. Most models have only been tested in controlled environments, and few studies have assessed whether they improve diagnostic accuracy or patient outcomes. Lack of transparency in developing these AI tools is another concern. Most training datasets are composed of images from patients with lighter skin tones, with minimal representation of skin of color [197]. This imbalance can compromise diagnostic performance in underrepresented populations. In one evaluation of benign pigmented acral lesions in Black patients, the AI algorithm that won in the 2020 International Skin Imaging Collaboration Challenge misclassified 95.7% of volar, 98.6% of dorsal, and 100% of nail lesions as melanoma [198]. More research should be performed on heterogeneous patient populations, and future guidelines should prioritize representative datasets, standardized methodology, and prospective evaluation in clinical settings.

Funding Statement

Dr. Hartman is supported by the Department of Defense under award number W81XWH2110820 and the Department of Veterans Affairs under award number VA CSR&D IK2 CX-002531.

Author contributions

Conceptualization: J.C.H., B.L.P., and R.I.H.; Data Curation: J.C.H. and B.L.P.; Formal Analysis: J.C.H. and B.L.P.; Funding Acquisition: R.I.H.; Investigation (Literature Search): J.C.H. and B.L.P.; Methodology (Search Strategy): J.C.H. and B.L.P.; Project Administration: R.I.H.; Resources (Access to Databases and Articles): J.C.H., B.L.P., and R.I.H.; Supervision: R.I.H.; Writing – Original Draft: J.C.H., B.L.P.; Writing – Review & Editing: J.C.H., B.L.P., and R.I.H.; Concept and Design: J.C.H., B.L.P., and R.I.H.

Disclosure statement

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Writing assistance

No writing assistance was utilized in the production of this manuscript.