Skip to content

Melanoma histopathology and prognostic factors

TL;DR — Breslow thickness and ulceration are the two histological variables that carry the AJCC 8th-edition T category; mitotic rate was removed as a T1 criterion in that revision (Gershenwald 2017, PMID 29028110). The reproducibility of the underlying diagnosis is poor in the middle of the spectrum — intraobserver agreement 35.2% for moderate atypia, 59.5% for severe atypia/in situ and 63.2% for pT1a, with accuracy against expert consensus of 25%, 40% and 43% respectively (Elmore 2017, PMID 28659278) — and second opinions reduce but do not eliminate it (Piepkorn 2019, PMID 31603483). Several reported prognostic factors are directionally counter-intuitive: histological regression is protective (relative risk of death 0.77, 95% CI 0.61–0.97, across 10 studies and 8,557 patients) (Gualano 2018, PMID 28386936), and brisk tumour-infiltrating lymphocytes associate with better 5-year overall survival (HR 0.62, 0.44–0.88) in meta-analysis but showed no independent association with disease-specific survival in a contemporary 1,017-patient cohort (Sun 2020, PMID 32023619; Straker 2022, PMID 35301610). Histotype matters after adjustment for thickness only for acral lentiginous melanoma (adjusted HR for disease-free survival 1.25, 1.02–1.52), not for nodular melanoma (Mandalà 2022, PMID 35421840).

The reproducibility floor

Every prognostic factor below is measured on a slide whose diagnosis is itself uncertain in the diagnostic middle.

Interpretation class Intraobserver reproducibility Accuracy vs expert consensus
Nevus / mild atypia 76.7% 92% (90–94)
Moderate atypia 35.2% 25% (22–28)
Severe atypia / melanoma in situ 59.5% 40% (37–44)
pT1a early invasive melanoma 63.2% 43% (39–46)
≥pT1b invasive melanoma 82.6% 72% (69–75)

Source: 187 pathologists, 240 cases, 8,976 interpretations (Elmore 2017, PMID 28659278). Population-level estimates: 82.8% (81.0–84.5) of melanocytic biopsy diagnoses would be confirmed by an expert panel, 8.0% overinterpreted, 9.2% underinterpreted (PMID 28659278). Universal second opinions by subspecialty dermatopathologists reduce misclassification from 52.8% to 36.7% but in situ and thin invasive melanoma remain "relatively intractable to all examined strategies" (PMID 31603483). Molecular adjuncts are being tested against exactly this problem: combined 611-gene sequencing plus 850k methylation arrays separated Spitz nevi from both nevi and melanoma, with multiple copy-number alterations and TERT promoter mutations found only in melanomas (Zaremba 2022, PMID 35737508).

The staged prognostic variables

Breslow thickness

Thickness is the dominant continuous predictor and the backbone of the T category (PMID 29028110). The 8th edition made two measurement changes with practical consequences: thickness is recorded to the nearest 0.1 mm rather than 0.01 mm, and T1a/T1b were redefined as <0.8 mm without ulceration versus 0.8–1.0 mm with or without ulceration or <0.8 mm with ulceration (PMID 29028110). Because partial biopsies underestimate depth in a minority of cases — positive deep margins in 33% of shave and 23% of punch biopsies, with residual melanoma at definitive excision in 40.6% and 60.4% respectively — thickness from a partial biopsy is a lower bound (Jones 2023, PMID 36739830). Across 145 melanomas, 88% of non-excisional biopsies nonetheless returned a Breslow depth ≥ the excision depth, with excisional biopsy the most accurate method (Ng 2003, PMID 12637923).

Ulceration

Ulceration is the second AJCC T-category variable, and its prognostic significance was established decades ago (Balch 1980, PMID 7388745). Two refinements matter:

  • Extent, not just presence. Ulceration width has been examined alongside lymphovascular invasion, microscopic satellitosis, perineural invasion and mitotic rate in 1,898 patients undergoing sentinel node biopsy at a single comprehensive cancer centre (median thickness 1.25 mm, median follow-up 6.7 years); the strongest risk factor for sentinel-node positivity was thickness, followed by lymphovascular invasion (Namikawa 2018, PMID 29464914). Extent of ulceration was separately shown to carry prognostic information (Grande Sarpa 2006, PMID 17063079).
  • Incipient ulceration is a distinct category. In a matched case-control study from the Melanoma Institute Australia database (2005–2015), 40 incipiently ulcerated melanomas were matched 1:2 against non-ulcerated and ulcerated controls. Median Breslow thickness was 2.8 mm (IQR 1.7–4.1) for incipient cases versus 1.0 mm (0.6–2.1) non-ulcerated and 5.3 mm (3.5–8.0) ulcerated; median mitotic rate 5.0/mm² versus 1.0/mm² in non-ulcerated controls (Paver 2023, PMID 37910123). Incipient ulceration sits biologically between the two AJCC categories that currently have no space for it.

Mitotic rate — removed from staging, retained in practice

Mitotic rate was dropped as a T1 criterion in AJCC 8 (PMID 29028110), and independent validation found that removing it decreased stage I prognostic value, though not significantly (RFS concordance index falling from 0.63, 0.56–0.69 to 0.56, 0.49–0.63; P = .07) (Bajaj 2020, PMID 31977051). Two measurement problems explain the ambivalence:

  • It is section-dependent. Among 82 T1 melanomas sectioned 5–10 times, 44 (54%) were classified T1b under AJCC 7; the number of sections showing a mitotic figure ranged from 1 of 5 (11.4% of T1b cases) to all 5 (45.5%). The authors recommend reviewing 3–5 sections to satisfy AJCC recommendations (Knezevich 2014, PMID 25239732).
  • It is inconsistently reported. In 107,134 NCDB patients, mitotic-rate reporting rose from 64.3% to 80.9% between 2010 and 2013, with reporting predicted by facility type, facility volume, patient income, education and metropolitan county — i.e. by where the patient was treated (Lorimer 2017, PMID 28335082).

Mitotic rate nonetheless continues to inform sentinel-node decisions in thin melanoma; a validated nomogram for T1 sentinel-node metastasis incorporates it (Maurichi 2020, PMID 32167862).

Factors that behave counter-intuitively

Factor Direction Evidence
Histological regression Protective Relative risk of death 0.77 (0.61–0.97) across 10 studies, 8,557 patients; no funnel-plot evidence of publication bias (Gualano 2018, PMID 28386936)
Brisk TILs, meta-analysis Protective 13 studies, 7,633 patients: 5-year OS HR 0.62 (0.44–0.88, I²=0), 5-year DSS HR 0.53 (0.30–0.96), endpoint DSS HR 0.51 (0.30–0.87) (Sun 2020, PMID 32023619)
TILs, contemporary single-centre cohort No independent DSS association 1,017 patients ≥1 mm undergoing wide excision and SLNB, 2006–2019: any TILs associated with more regression (OR 1.86, P = .016), less acral lentiginous histology (OR 0.22, P < .001) and lower SLN positivity (OR 0.64, P = .042), but multivariable analysis found no association between brisk TILs and disease-specific survival (Straker 2022, PMID 35301610)
Nodular histotype No independent effect after adjustment Adjusted HR vs superficial spreading: DFS 1.04 (0.93–1.15, P = .513), OS 0.96 (0.86–1.08, P = .548) in 6,734 stage I–II patients (Mandalà 2022, PMID 35421840)
Acral lentiginous histotype Independently worse Adjusted DFS HR 1.25 (1.02–1.52, P = .028) vs superficial spreading, after adjustment for age, sex, Breslow thickness, ulceration and sentinel-node status (PMID 35421840)

Nodular melanoma's apparent aggressiveness is largely explained by thickness and mitotic rate; acral lentiginous melanoma's is not. That asymmetry is the histological form of the argument in acral and mucosal melanoma.

Thin nodular melanoma is nonetheless a distinct entity. Across 20,132 melanomas from 17 centres (5,062 nodular, 15,070 superficial spreading), T1 nodular melanoma was less likely than T1 superficial spreading melanoma to show regression (OR 0.46, 0.29–0.72) or nevus remnants (OR 0.60, 0.42–0.85), and more likely to have mitoses (OR 1.97, 1.33–2.93), regional metastasis (OR 1.77, 1.02–3.05) and a higher mitotic rate (adjusted geometric mean 2.2, 1.9–2.5 versus 1.6, 1.5–…) (Dessinioti 2019, PMID 30863861).

Desmoplastic melanoma

Desmoplastic melanoma has distinct biology and pathology and is subclassified into pure and mixed histological subtypes that correlate with outcome (Nicolson 2019, PMID 30481377). The subtype distinction changes surgical decision-making: meta-analysis of 18 studies and 1,671 patients found a pooled sentinel-node positivity rate of 9% (95% CI 7–12), with 6% (4–8) for pure and 15% (10–20) for mixed desmoplastic melanoma, and no significant association between positivity and Breslow depth (β = 0.058, P = .749) or ulceration (β = 0.763, P = .677) (Yee 2026, PMID 41870859). A histological subtype that decouples nodal risk from thickness and ulceration is an explicit exception to the AJCC logic — see sentinel node and nodal management. In the WHO 2018 scheme, desmoplastic melanoma sits in the high-cumulative-solar-damage pathway (Elder 2020, PMID 32057276).

Nodal pathology beyond presence or absence

Sentinel-node tumour burden carries prognostic information the AJCC N category does not. In 736 sentinel-node-positive patients with mean follow-up 64.4 months, melanoma-specific survival was significantly better in stage IIIA, IIIB and IIIC patients with lower burden at cut-offs of ≤0.5 mm and ≤1 mm maximum deposit diameter; in multivariable Cox analysis the maximum deposit diameter was an independent predictor, with 0.5 mm giving a marginally higher AUC (0.617) than 1 mm (0.599) (Satzger 2019, PMID 31677550). Lymph node ratio adds information beyond AJCC N stage even when surgical quality is standardised (Spillane 2011, PMID 21119509).

Emerging measurement

  • AI-quantified TILs. In 1,202 patients with advanced melanoma treated with first-line anti-PD-1 ± anti-CTLA-4 across 11 Dutch centres, TIL percentage in pretreatment metastases was determined by a model trained on 161,835 pathologist-verified annotated cells, with objective response rate as the primary outcome and PFS/OS secondary (Schuiveling 2025, PMID 41100131). This moves TIL scoring from a categorical pathologist judgement to a continuous automated measure — the same transition that has repeatedly changed which prognostic factors survive multivariable analysis.
  • Multigene RNA assays on archival tissue. An 11-gene RNA assay has been evaluated for prognosis in stage I–III archival tissue (Gambichler 2021, PMID 33278769), alongside the 31-gene expression profile described in molecular subtypes and genomics (Zager 2018, PMID 29402264).

Immunohistochemistry

Immunohistochemistry does not resolve the reproducibility problem but it narrows specific differentials.

Marker Role
SOX10 Considered the optimal general melanocytic marker — similar sensitivity to S100 with higher specificity (Saleem 2022, PMID 35016807)
HMB-45 Less sensitive than S100 but useful in confirming deceptively banal small-cell and nevoid melanoma variants where deep nests of melanocytes are highlighted (PMID 35016807)
PRAME 93.5% (72/77) of lentigo maligna cases were PRAME-positive and 94.7% (18/19) of cases with no residual lentigo maligna were PRAME-negative in a 96-case series, with histological margin measurement by PRAME comparable to conventional assessment (Gradecki 2021, PMID 33280156). It has been evaluated for staged-excision margin assessment alongside Melan-A and SOX10 (de Wet 2023, PMID 36669074) and in cytology specimens for metastatic melanoma (PMID 38558495)
PRAME — the caveat Systematic immunohistochemical study of >5,800 tumours found PRAME variably expressed across many non-melanocytic malignancies, including >50% positivity among endometrial (82%), uterine serous (82%) and ovarian clear-cell (90%) and serous (63%) carcinomas, and in normal testis and proliferative endometrium (Kaczorowski 2022, PMID 35973038). PRAME positivity is not melanoma-specific.
Practice Use of immunohistochemical markers in the histopathological diagnosis of melanocytic lesions varies by pathologist and setting (PMID 32383301); broader reviews cover the current panel (PMID 40507250; PMID 41790207)

Intraoperative immunohistochemistry during Mohs or staged excision lowers local recurrence for invasive melanoma from 1.8% (0.8–2.8) to 0.3% (0–0.6), P < .001, without changing nodal recurrence, distant recurrence or disease-specific mortality (O'Hern 2024, PMID 38530980) — see surgical management and margins.

Interpretation rules for this page

  • A prognostic factor measured on an unreliable diagnosis inherits that unreliability. The diagnostic middle has 25–43% accuracy against expert consensus (PMID 28659278).
  • Distinguish univariable from adjusted effects. Nodular histotype is strongly adverse univariably (DFS HR 2.29) and null after adjustment (1.04); acral is adverse in both (PMID 35421840).
  • Ask how many sections were examined before accepting a mitotic rate, and whether the report includes one at all (PMID 25239732; PMID 28335082).
  • Thickness from a partial biopsy is a floor (PMID 36739830; PMID 12637923).
  • Regression and lymphocytic infiltrate are favourable, not adverse, contrary to a persistent clinical intuition (PMID 28386936; PMID 32023619) — although the TIL effect did not survive multivariable analysis in the most recent large cohort (PMID 35301610).
  • Desmoplastic melanoma breaks the thickness–nodal-risk relationship and must be staged with its subtype named (PMID 41870859).

Open questions

  • Can a molecular assay resolve the moderate-to-severe-atypia band that morphology cannot (PMID 28659278; PMID 35737508)?
  • Should incipient ulceration be a distinct AJCC category, given its intermediate thickness and high mitotic rate (PMID 37910123)?
  • Should mitotic rate return to staging with a specified number of sections, or be replaced by an automated proliferation measure (PMID 25239732; PMID 28335082)?
  • Should sentinel-node tumour burden be incorporated into the N category, and at what cut-off (PMID 31677550)?
  • Why does the TIL survival association hold in meta-analysis but not in a contemporary multivariable cohort — era, adjustment set, or scoring method (PMID 32023619; PMID 35301610)?
  • Does automated TIL quantification predict immunotherapy response better than pathologist grading (PMID 41100131)?
  • Why is acral lentiginous histotype independently adverse after full adjustment while nodular is not (PMID 35421840)?

References

  1. Gershenwald JE, et al. Melanoma staging: Evidence-based changes in the American Joint Committee on Cancer eighth edition cancer staging manual. CA: a cancer journal for clinicians. 2017;67:472-492. PMID 29028110
  2. Elmore JG, et al. Pathologists' diagnosis of invasive melanoma and melanocytic proliferations: observer accuracy and reproducibility study. BMJ (Clinical research ed.). 2017;357:j2813. PMID 28659278
  3. Piepkorn MW, et al. Assessment of Second-Opinion Strategies for Diagnoses of Cutaneous Melanocytic Lesions. JAMA network open. 2019;2:e1912597. PMID 31603483
  4. Gualano MR, et al. Prognostic role of histological regression in primary cutaneous melanoma: a systematic review and meta-analysis. The British journal of dermatology. 2018;178:357-362. PMID 28386936
  5. Sun Q, et al. Prognostic Significance of Tumor-Infiltrating Lymphocyte Grade in Melanoma: A Meta-Analysis. Dermatology (Basel, Switzerland). 2020;236:481-492. PMID 32023619
  6. Straker RJ 3rd, et al. Prognostic Significance of Primary Tumor-Infiltrating Lymphocytes in a Contemporary Melanoma Cohort. Annals of surgical oncology. 2022;29:5207-5216. PMID 35301610
  7. Mandalà M, et al. Acral lentiginous melanoma histotype predicts outcome in clinical stage I-II melanoma patients: an International multicenter study. ESMO open. 2022;7:100469. PMID 35421840
  8. Zaremba A, et al. Genetic and methylation profiles distinguish benign, malignant and spitzoid melanocytic tumors. International journal of cancer. 2022;151:1542-1554. PMID 35737508
  9. Jones S, et al. Clinical Impact and Accuracy of Shave Biopsy for Initial Diagnosis of Cutaneous Melanoma. The Journal of surgical research. 2023;286:35-40. PMID 36739830
  10. Ng PC, et al. Evaluating invasive cutaneous melanoma: is the initial biopsy representative of the final depth?. Journal of the American Academy of Dermatology. 2003;48:420-4. PMID 12637923
  11. Balch CM, et al. The prognostic significance of ulceration of cutaneous melanoma. Cancer. 1980;45:3012-7. PMID 7388745
  12. Namikawa K, et al. Clinical impact of ulceration width, lymphovascular invasion, microscopic satellitosis, perineural invasion, and mitotic rate in patients undergoing sentinel lymph node biopsy for cutaneous melanoma: a retrospective observational study at a comprehensive cancer center. Cancer medicine. 2018;7:583-593. PMID 29464914
  13. Grande Sarpa H, et al. Prognostic significance of extent of ulceration in primary cutaneous melanoma. The American journal of surgical pathology. 2006;30:1396-400. PMID 17063079
  14. Paver EC, et al. Prognostic Significance of Incipient Ulceration in Primary Cutaneous Melanoma. JAMA dermatology. 2023;159:1359-1367. PMID 37910123
  15. Bajaj S, et al. Melanoma Prognosis: Accuracy of the American Joint Committee on Cancer Staging Manual Eighth Edition. Journal of the National Cancer Institute. 2020;112:921-928. PMID 31977051
  16. Knezevich SR, et al. Variability in mitotic figures in serial sections of thin melanomas. Journal of the American Academy of Dermatology. 2014;71:1204-11. PMID 25239732
  17. Lorimer PD, et al. Reporting of mitotic rate in cutaneous melanoma: A study using the national cancer data base. Journal of surgical oncology. 2017;115:281-286. PMID 28335082
  18. Maurichi A, et al. Factors Affecting Sentinel Node Metastasis in Thin (T1) Cutaneous Melanomas: Development and External Validation of a Predictive Nomogram. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2020;38:1591-1601. PMID 32167862
  19. Dessinioti C, et al. Distinct Clinicopathological and Prognostic Features of Thin Nodular Primary Melanomas: An International Study from 17 Centers. Journal of the National Cancer Institute. 2019;111:1314-1322. PMID 30863861
  20. Nicolson NG, et al. Desmoplastic melanoma. Journal of surgical oncology. 2019;119:208-215. PMID 30481377
  21. Yee P, et al. Sentinel Lymph Node Biopsy for Desmoplastic Melanoma: A Systematic Review and Meta-analysis. Annals of surgical oncology. 2026;33:6719-6727. PMID 41870859
  22. Elder DE, et al. The 2018 World Health Organization Classification of Cutaneous, Mucosal, and Uveal Melanoma: Detailed Analysis of 9 Distinct Subtypes Defined by Their Evolutionary Pathway. Archives of pathology & laboratory medicine. 2020;144:500-522. PMID 32057276
  23. Satzger I, et al. Melanoma-specific survival in patients with positive sentinel lymph nodes: Relevance of sentinel tumor burden. European journal of cancer (Oxford, England : 1990). 2019;123:83-91. PMID 31677550
  24. Spillane AJ, et al. Lymph node ratio provides prognostic information in addition to american joint committee on cancer N stage in patients with melanoma, even if quality of surgery is standardized. Annals of surgery. 2011;253:109-15. PMID 21119509
  25. Schuiveling M, et al. Artificial Intelligence-Detected Tumor-Infiltrating Lymphocytes and Outcomes in Anti-PD-1-Based Treated Melanoma. JAMA oncology. 2025;11:1470-1478. PMID 41100131
  26. Gambichler T, et al. Prognostic significance of an 11-gene RNA assay in archival tissue of cutaneous melanoma stage I-III patients. European journal of cancer (Oxford, England : 1990). 2021;143:11-18. PMID 33278769
  27. Zager JS, et al. Performance of a prognostic 31-gene expression profile in an independent cohort of 523 cutaneous melanoma patients. BMC cancer. 2018;18:130. PMID 29402264
  28. Saleem A, et al. Immunohistochemistry in melanocytic lesions: Updates with a practical review for pathologists. Seminars in diagnostic pathology. 2022;39:239-247. PMID 35016807
  29. Gradecki SE, et al. PRAME immunohistochemistry as an adjunct for diagnosis and histological margin assessment in lentigo maligna. Histopathology. 2021;78:1000-1008. PMID 33280156
  30. de Wet J, et al. Staged Excision of Lentigo Maligna of the Head and Neck: Assessing Surgical Excision Margins With Melan A, SOX10, and PRAME Immunohistochemistry. The American Journal of dermatopathology. 2023;45:107-112. PMID 36669074
  31. Saad M, et al. Performance of preferentially expressed antigen in melanoma (PRAME) immunohistochemistry for metastatic melanoma in cytology specimens. Diagnostic cytopathology. 2024;52:362-368. PMID 38558495
  32. Kaczorowski M, et al. PRAME Expression in Cancer. A Systematic Immunohistochemical Study of >5800 Epithelial and Nonepithelial Tumors. The American journal of surgical pathology. 2022;46:1467-1476. PMID 35973038
  33. May CJ, et al. Factors associated with use of immunohistochemical markers in the histopathological diagnosis of cutaneous melanocytic lesions. Journal of cutaneous pathology. 2020;47:896-902. PMID 32383301
  34. Voiculescu VM, et al. Immunohistochemistry for Skin Cancers: New Insights into Diagnosis and Treatment of Melanoma. Cancers. 2025;17. PMID 40507250
  35. Mentzel T. [Immunohistochemistry of melanocytic tumours]. Pathologie (Heidelberg, Germany). 2026;. PMID 41790207
  36. O'Hern K, et al. Intraoperative Immunohistochemistry During Mohs Micrographic Surgery and Staged Excision Decreases Local Recurrence Rates for Invasive Cutaneous Melanoma: A Systematic Review and Meta-Analysis. Dermatologic surgery : official publication for American Society for Dermatologic Surgery [et al.]. 2024;50:601-610. PMID 38530980