Diagnostic and Prognostic Values of PRAME Immunohistochemistry in Acral Lentiginous Melanoma

Abstract

Background:

PReferentially Expressed Antigen in MElanoma (PRAME) was introduced as an ancillary tool for confirming the diagnosis of cutaneous melanoma, including the acral lentiginous melanoma (ALM). However, its standardized cut-off values and prognostic performance remain inconclusive. This study assessed the diagnostic performance of PRAME immunohistochemistry (IHC) in ALM across different thresholds and explored its prognostic associations.

Methods:

This retrospective cohort study reviewed 90 acral melanocytic lesions, including 60 cases of acral benign nevi and 30 cases of primary ALM. PRAME immunohistochemistry was performed on the biopsy specimens. Sensitivity and specificity were calculated at varying cut-offs. The optimal threshold was applied to link PRAME expression with the outcome and features of ALM.

Results:

A PRAME staining threshold of 50% (score ≥3) provided the optimum cut-off for differentiating ALM, with a high specificity of 98.3% and sensitivity of 53.3%. Positive PRAME was significantly associated with higher TNM stages (ordinal odds ratio [95% confidence interval], 13.11 [2.55–67.27]) and higher rates of recurrence (hazard ratio, 6.62 [1.42–30.94]) but not with any unfavorable histopathological features or mortality.

Conclusion:

PRAME can be used as an adjunctive diagnostic tool to differentiate malignant from benign acral melanocytic lesions. It may serve as a prognostic marker for poor ALM outcomes.

INTRODUCTION

PReferentially Expressed Antigen in MElanoma (PRAME), a tumor-associated antigen that belongs to the cancer-testis antigen family, is expressed in various malignancies including melanoma, and has emerged as a promising ancillary marker for melanoma diagnosis.¹ While physiologic expression is limited to the testis, ovary, adrenal glands, and placenta, PRAME typically demonstrates nuclear staining in various melanoma subtypes.^2,3 Beyond cutaneous melanoma, PRAME has been investigated as both a diagnostic and prognostic biomarker across several cancers, such as uveal melanoma, mucosal melanoma, breast carcinoma, lung cancer, and sarcomas.^4,5 The nuclei of malignant melanocytes, but not of benign nevi, express PRAME. Nevertheless, recent studies underscored that PRAME expression varies across histologic subtypes and anatomical sites, warranting interpretation within an integrated clinicopathologic context.⁶

Acral lentiginous melanoma (ALM) represents the most common subtype of melanoma among Asian populations, accounting for approximately 50%–70% of reported cases.^7,8 As ALM typically occurs on sun-protected sites such as the soles and nail apparatus, the diagnosis is often delayed. Thus, wide local excision or amputation is often required, resulting in functional impairment and disfigurement. Even so, the prognosis of such ALM cases is grim. Early and accurate diagnosis of ALM is therefore critical for improving clinical outcomes.

Published studies have delineated a plethora of definitions for PRAME positivity, including diffuse nuclear staining (≥75%), ≥60% nuclear staining, ≥50% nuclear staining, and composite scoring systems that combine the extent and intensity of expression.^9–13 However, data on PRAME expression and its prognostic significance in ALM, particularly in Asian populations, remain scarce, and no universal consensus has been established regarding the optimal threshold for PRAME interpretation.

This study aimed to evaluate the diagnostic performance of PRAME immunohistochemistry (IHC) at different cut-offs in differentiating ALM from acral melanocytic nevus (AMN), and to explore the association between PRAME expression, tumor stage, and clinical outcomes in ALM.

METHOD

Study Design

This retrospective case–control study reviewed archival data of patients diagnosed with AMN and ALM at the Division of Dermatology, Department of Medicine, Ramathibodi Hospital, Bangkok, Thailand, over a 7-year period (January 2017–December 2023). This study was approved by the Institutional Review Board of Human Rights, Ramathibodi Hospital, Mahidol University (Protocol Number MURA 2024/152) in accordance with the Declaration of Helsinki.

Case Selection

Eligible cases were required to have complete histopathological data and available formalin-fixed, paraffin-embedded tissue blocks for PRAME immunohistochemical staining. Inclusion criteria comprised adults (18 years and older) with histologically confirmed diagnoses of ALM or AMN. Metastatic or recurrent melanoma specimens were excluded. The number of AMN included was at least equal to that of ALM cases to ensure balanced group comparison. A total of 190 acral melanocytic lesions were initially reviewed, comprising 143 AMN and 47 ALM. After applying the eligibility criteria, 30 ALM and 72 AMN cases remained. According to the case–control study design, 60 AMN were randomly selected using a computer-generated randomization sequence in STATA/SE version 18 (StataCorp LP, College Station, TX) to maintain a 2:1 ratio of benign to malignant lesions for subsequent analysis. The 2 groups were not age- or sex-matched, reflecting the natural epidemiologic tendency for melanoma to occur more frequently in older individuals.

Data Collection

Clinical and histopathological data were retrieved from electronic medical records and pathological reports. Recorded parameters included the patient's age, sex, underlying diseases, family history of melanoma, lesion location, biopsy type, histological subtype, and treatment. Additional variables for ALM included disease onset, Breslow thickness, Clark level, lymphovascular invasion, dermal mitotic rate, ulceration, tumor-infiltrating lymphocytes, neurotropism, surgical margin status, serum lactate dehydrogenase level, sentinel lymph node biopsy result, tumor staging according to AJCC eighth edition criteria,¹⁴ metastasis, recurrence, morbidity (eg, amputation), and mortality events.

IHC and PRAME Evaluation

PRAME IHC was performed on formalin-fixed, paraffin-embedded tissue sections using a commercially available anti-PRAME antibody (Cell Marque, clone EP461, dilution 1:50) on an automated immunostainer (Ventana BenchMark Ultra, Tucson, AZ) following standard protocols. All tissue sections chosen for immunostaining were those that did not require any additional modification such as decalcification or softening.

The extent of expression was evaluated as the percentage of tumor cell nuclei showing positivity and categorized into 5 grades: zero, 0%; 1+, 1%–25%; 2+, 26%–50%; 3+, 51%–75%; 4+ (diffused), 76%–100%. Staining intensity was assessed as strength of nuclear expression (0, absent; 1, weak; 2, moderate; 3, strong).¹⁵ The greatest intensity was recorded in cases where the staining intensity was inhomogeneous. A combination of expression and intensity (E + I) score was calculated by summing both parameters, yielding a total range of 0–7.

All slides were independently evaluated by 1 dermatopathologist (S.R.) and 1 pathologist (B.P.), who were blinded to both clinical and histopathological data. In cases of discordant interpretation, the slides were jointly reviewed and a consensus diagnosis was recorded as the final result.

Statistical Analysis

Categorical data were described as numbers and percentages, and continuous data as the mean and SD or median and interquartile range. Continuous variables were compared using the unpaired t test, and categorical variables using the χ² test or Fisher exact test, as appropriate. Diagnostic performance of PRAME expression was calculated using sensitivity, specificity, positive and negative likelihood ratios (LR+/LR−), positive and negative predictive values, and the area under the receiver-operating-characteristic curve (AUC). The optimum cut-off will then be used as a diagnostic criterion for calculating associative factors and prognostic factors. Associations between PRAME positivity and clinicopathological features were analyzed using logistic regression models. Prognostic impact on recurrence-free and overall survival was assessed using Cox proportional-hazards regression. Two-tailed P < 0.05 was considered statistically significant. All analyses were performed using STATA/SE version 18 (StataCorp LP, College Station, TX).

RESULTS

Demographics and Pathological Characteristics

A total of 90 melanocytic lesions were included in this case–control study, including 60 AMN and 30 primary ALM. Within the ALM group, 24 cases (80%) were invasive and 6 (20%) were melanoma in situ (MIS). The plantar surface of the foot was the most common site for both groups, accounting for 29 (48.3%) of AMN and 18 (60%) of ALM. Junctional nevus was the most common clinicopathological subtype of AMN (55%) (Table 1). Regarding pathological features of ALM, the mean Breslow thickness was 5.2 mm, around 95% of cases' margins were beyond the reticular dermis (Clark level IV and V), ulceration was present in 58%, and mitotic activity ≥1/mm² was found in 75% of cases, consistent with the advanced-stage disease profile of tertiary referral populations (Table 2).

TABLE 1.

Demographic Characteristics of the Patients and Their Tumors

Characteristics	AMN (n = 60)	ALM (n = 30)
Age, mean ± SD (yr)	44.55 ± 15.98	58.68 ± 16.34
Sex, n (%), female	42 (70.0)	18 (60.0)
Family history of melanoma, n (%)	0 (0.0)	0 (0.0)
Tumor site, n (%)
Hand
Palm	16 (26.7)	0 (0.0)
Dorsum of hand	7 (11.7)	0 (0.0)
Finger	2 (3.3)	2 (6.7)
Finger nail	1 (1.7)	4 (13.3)
Feet
Plantar	29 (48.3)	18 (60.0)
Dorsum of foot	3 (5.0)	1 (3.3)
Toe	2 (3.3)	4 (13.3)
Toe nail	0 (0.0)	1 (3.3)
Clinicopathological subtypes of ALM, n (%)
Invasive melanoma	—	24 (80.0)
Melanoma in situ (MIS)	—	6 (20.0)
Clinicopathological subtypes of AMN, n (%)
Junctional nevus	33 (55.0)	—
Compound nevus	21 (35.0)	—
Intradermal nevus	2 (3.3)	—
Blue nevus	2 (3.3)	—
Congenital nevus	1 (1.7)	—
Compound dysplastic nevus	1 (1.7)	—

AMN, acral melanocytic nevus; ALM, acral lentiginous melanoma.

— indicates not evaluated.

TABLE 2.

Pathological Features of Invasive Acral Lentiginous Melanoma

Pathological Features	n = 24, n (%)
Breslow thickness
Mean ± SD	5.23 ± 3.92
<0.8 mm	2 (8.3)
≥0.8 mm	22 (91.7)
Clark level
I	0 (0.0)
II	1 (4.2)
III	0 (0.0)
IV	18 (75.0)
V	5 (20.8)
Lymphovascular invasion (LVI)
Present	3 (16.7)
Absent	15 (83.3)
Mitotic rate (/mm2)
<1	6 (25.0)
≥1	18 (75.0)
Ulceration
Present	14 (58.3)
Absent	10 (41.7)
Tumor-infiltrating lymphocyte
Present	11 (45.8)
Absent	13 (54.2)
Neurotropism
Present	2 (8.3)
Absent	22 (91.7)

PRAME Extent and Intensity of Expression

PRAME staining results are summarized in Table 3. Nearly all AMN lesions (90%) were completely negative for PRAME; the only AMN showing extent of expression greater than 2+ (≥26%) were junctional nevi subtype. Among ALM, more than half (53.3%) exhibited extent of expression more than 3+ (≥51%), whereas half of the MIS cases was negative for staining, and the remaining half demonstrated only minimal staining (1+, 1%–25%). Representative staining patterns of PRAME 3+ and 4+ cases are shown in Figure 1.

TABLE 3.

Extent of PRAME Expression Among Subtypes of Acral Melanocytic Nevus and Acral Lentiginous Melanoma According to Different Cut-off Criteria

PRAME Expression	AMN (n = 60)							ALM (n = 30)			P
	AMN	Junctional (n = 33)	Compound (n = 21)	Intradermal (n = 2)	Blue (n = 2)	Congenital (n = 1)	Dysplastic (n = 1)	ALM	MIS (n = 6)	Invasive ALM (n = 24)
4+ (76%–100%), n (%)	1/60 (1.7%)	1	0	0	0	0	0	14/30 (46.7%)	0	14	<0.05
3+ (51%–75%), n (%)	0 (0.0%)	0	0	0	0	0	0	2/30 (6.7%)	0	2	0.11
2+ (26%–50%), n (%)	1/60 (1.7%)	1	0	0	0	0	0	2/30 (6.7%)	0	2	0.26
1+ (1%–25%), n (%)	4/60 (6.7%)	0	2	1	0	0	1	5/30 (16.7%)	3	2	0.14
0 (0%), n (%)	54/60 (90.0%)	31	19	1	2	1	0	7/30 (23.3%)	3	4	<0.05

AMN, acral melanocytic nevus; ALM, acral lentiginous melanoma; PRAME, PReferentially expressed Antigen in MElanoma.

FIGURE 1.

Representative of PRAME immunohistochemical staining in 2 cases of invasive acral lentiginous melanoma. Diffuse nuclear extent of expression (100%, 4+) with strong staining intensity (3) (A: H&E, original magnification X400; B: PRAME, original magnification ×400). Partial nuclear extent of expression (60%, 3+) with moderate staining intensity (2) (C: H&E, original magnification ×400; D: PRAME, original magnification ×400).

Both the extent and intensity of expression were recorded in detail (see Supplementary Table 1, Supplemental Digital Content 1, http://links.lww.com/AJDP/A183). The combined extent and intensity of expression (E + I) score showed a similar distribution pattern, with greater positivity proportion among ALM compared with AMN. With combined PRAME cutoff of ≥4, 98.3% of AMN is negative (59/60) while 60% (18/30) of ALM is positive. Two cases of AMN, 1 junctional nevus and 1 dysplastic nevus, exhibited PRAME positivity (4 + 3 and 2 + 1, respectively) while 7 cases of ALM were completely PRAME negative.

The diagnostic value of PRAME expression is summarized in Table 4. The extent of expression cut-off of ≥3 yielded the highest specificity (98.33%; 95% CI, 91.06–99.96) and a positive predictive value of 94.11%, with an area under the ROC curve (AUC) of 0.76 (95% CI, 0.67–0.85). Although a combined extent and intensity of expression (E + I) score of ≥4 demonstrated a slightly higher AUC (0.79; 95% CI, 0.70–0.88) (see Supplementary Table 2, Supplemental Digital Content 1, http://links.lww.com/AJDP/A183), the improvement was not statistically significant (P = 0.15). The expression criteria were therefore selected as the optimal diagnostic threshold owing to its high specificity and greater clinical practicality, as staining intensity assessment is more subjective and less routinely recorded in practice. The extent of expression ≥3 is later selected as a criterion for calculating prognostic values.

TABLE 4.

Diagnostic Performance of the Extent of PRAME Expression for Differentiating ALM from AMN at Various Cut-off Thresholds

	Sensitivity (95% CI)	Specificity (95% CI)	LR+ (95% CI)	LR- (95% CI)	Positive Predictive Value (95% CI)	Negative Predictive Value (95% CI)	AUC (95% CI)
PRAME expression cut-off of 4+	46.67 (28.34–65.67)	98.33 (91.06–99.96)	28.00 (3.86–202.97)	0.54 (0.39–0.76)	93.33 (68.05–99.83)	78.67 (67.68–87.29)	0.73 (0.63–0.82)
PRAME expression cut-off of 3+	53.33 (34.33–71.66)	98.33 (91.06–99.96)	32.00 (4.45–229.96)	0.47 (0.32–0.7)	94.11 (71.31–99.85)	80.8 (69.9–89.1)	0.76 (0.67–0.85)
PRAME expression cut-off of 2+	60.00 (40.60–77.34)	96.67 (88.47–99.59)	18.00 (4.47–72.53)	0.41 (0.27–0.64)	90.00 (68.30–98.77)	82.86 (71.97–90.82)	0.78 (0.69–0.88)
PRAME expression cut-off of 1+	76.67 (57.72–90.07)	90.00 (79.49–96.24)	7.67 (3.50–16.8)	0.26 (0.13–0.50)	79.31 (60.28–92.01)	88.52 (77.78–95.26)	0.83 (0.75–0.92)

Sensitivity, specificity, likelihood ratios, predictive values, and area under the curve (AUC) are shown with 95% confidence intervals. The extent of PRAME expression ≥3+ demonstrates the highest specificity and was selected as the optimal diagnostic threshold.

AUC, area under the curve; CI, confidence interval; LR−, negative likelihood ratio; LR+, positive likelihood ratio; PRAME, PReferentially expressed Antigen in MElanoma.

Clinicopathological correlations of the extent of expression ≥3 in ALM are summarized in Table 5. The PRAME positivity was significantly associated with a higher TNM stage (OR = 13.11, 95% CI, 2.55–67.27; P < 0.05). However, our study revealed that no significant associations were observed between PRAME status and poor prognostic histopathological variables such as Breslow thickness, ulceration, lymphovascular invasion, mitotic rate, or neurotropism.

TABLE 5.

Clinicopathological Associations of the Extent of PRAME Expression ≥3+ in Acral Lentiginous Melanoma

	OR (95% CI)	P
Invasive ALM (n = 24)
Breslow thickness (>0.8 mm)	5.37 (0.4-∞)	0.10
Clark 5	0.69 (0.09–5.29)	0.73
LVI	4.26 (0.49–∞)	0.13
Mitotic rate (>1)	2.60 (0.39–17.45)	0.32
Ulceration	3.67 (0.62–21.73)	0.14
Tumor-infiltrating lymphocyte	0.78 (0.14–4.27)	0.77
Neurotropism	0.47 (0.03–8.60)	0.61
LDH (>220) (n = 18)	1.33 (0.10–18.19)	0.83
SNLB positive (n = 20)	2.67 (0.24–30.07)	0.40
ALM (n = 30)
Staging	13.11 (2.55–67.27)	<0.05*
Stage 4	2.21 (0.17–∞)	0.49

Odds ratios (OR) with 95% confidence intervals and corresponding P-values are demonstrated. Staging according to AJCC 8th guideline. The number of cases with LDH and SNLB was evaluated only in cases with available data.

^*Statistical significance with P < 0.05.

ALM, acral lentiginous melanoma; LDH, lactate dehydrogenase; LVI, lymphovascular invasion; SNLB, sentinel lymph node biopsy.

Kaplan–Meier survival analysis (Fig. 2) showed that PRAME-positive ALM (expression ≥3) had significantly shorter recurrence-free survival compared with PRAME-negative cases (HR = 6.62, 95% CI, 1.42–30.94, P < 0.05). The association between PRAME expression and overall survival could not be estimated, as all 4 deaths occurred in the PRAME-positive group.

FIGURE 2.

Kaplan–Meier survival curve comparing recurrence-free survival between PRAME-positive and PRAME-negative acral lentiginous melanoma (ALM) with an extent of expression cut-off of ≥3. ALM with PRAME positivity was associated with a higher risk of recurrence (HR = 6.62; 95% CI, 1.42–30.94; P < 0.05).

DISCUSSION

In this study, we evaluated the diagnostic utility of PRAME IHC in differentiating ALM from benign AMN and its prognostic applicability in ALM cases. PRAME immunostaining demonstrated a consistent trend of higher expression in ALM compared with benign acral lesions. Using a semiquantitative grading system, we identified a cut-off of the extent of expression ≥3 as the optimal diagnostic threshold, providing the highest specificity and AUC for distinguishing malignant from benign acral melanocytic lesions. Moreover, PRAME positivity correlated with advanced TNM stage and increased risk of recurrence, supporting its potential prognostic significance in acral melanoma.

In this study, PRAME was expressed in 76.7% of ALM and 10.0% of AMN (P < 0.001), whereas 90% of nevi were entirely negative. A threshold of the extent of expression ≥3 (≥51% nuclear positivity) produced the best diagnostic performance (sensitivity 53.33%, specificity 98.33%, AUC, 0.76). Our findings parallel those reported in a previous study on PRAME expression in nail melanocytic lesions,^16,17 which also demonstrated that a threshold of more than 50% and 75% of nuclear positivity yielded high specificity for melanoma. In our series, a threshold ≥3 (50% of tumor nuclei) achieved comparable diagnostic accuracy while maintaining strong specificity. This finding is consistent with emerging evidence suggesting that a 3 + cut-off yields better balance between sensitivity and specificity, especially in morphologically heterogeneous tumors.^11,18 Although the combination of the extent and intensity of expression (E + I score) produced a marginally higher AUC, intensity assessment remains subjective and less reproducible in daily practice. For this reason, expression-based thresholds alone may be more practical for routine dermatopathological evaluation, particularly in laboratories without digital quantification tools.

Acral melanomas are biologically and genetically distinct from nonacral subtypes. In our series, half of the acral MIS cases were PRAME-negative, contrasting with the >90% diffuse nuclear positivity reported in previous studies.^1,19 Similar discordance was observed by Santandrea et al,¹⁶ who found that acral and nail melanomas frequently showed only partial PRAME staining despite overt histologic atypia. Gassenmaier et al²⁰ further reported diffuse (>75%) expression in only 59% of thin melanomas (≤1 mm), while Miao et al¹⁸ demonstrated that a lower (≥3+) threshold improved detection of acral MIS by capturing lesions with patchy or uneven staining. This discrepancy may result from the difference in study population and the use of different antibody clones and different immunostaining techniques. A recent meta-analysis by Kunc et al²¹ evaluated more than 2915 melanocytic lesions from multiple studies. They reported the high diagnostic accuracy of PRAME IHC for cutaneous melanoma over benign nevi with a pooled sensitivity of 0.74 and specificity of 0.92. The authors identified a ≥3 (≥50%) nuclear positivity threshold as the optimal cut-off, providing the best balance between sensitivity and specificity across melanoma subtypes. Although the analysis included limited ALM cases, its findings support the broader applicability of a semiquantitative 3+ criterion for distinguishing melanoma from benign melanocytic proliferations. Together, these findings indicate that reduced PRAME expression in early or cell-poor acral lesions likely reflects inherent biological and architectural factors rather than technical artifact. Accordingly, PRAME negativity should not exclude a diagnosis of early ALM, and adopting a lower (≥3+) or semiquantitative interpretive threshold may optimize detection while maintaining high specificity. Larger multicenter studies are warranted to validate site-specific criteria and standardize interpretation in acral melanoma.

Beyond its diagnostic role, PRAME expression was associated with higher TNM stage (OR, 13.11; 95% CI, 2.55–67.27) (P < 0.05) and recurrence (HR, 6.62, 95% CI, 1.42–30.94) (P < 0.05) in our cohort. These observations coincide with those of Asato et al,²² who found that diffuse PRAME expression (>75%) predicted shorter overall survival but not melanoma-specific, and Parra et al,²³ who reported that PRAME correlated with increased mitotic activity but lacked independent prognostic significance after adjustment for Breslow thickness. These findings emphasize PRAME's main utility as a diagnostic adjunctive rather than a reliable indicator of melanoma-specific outcomes. However, the hazard ratio for mortality, in our study, could not be estimated owing to the finding that all 4 deaths occurred exclusively in PRAME-positive group. Despite having been associated with higher TNM staging and recurrence in our study, PRAME has not been shown to be associated with any of the histopathologic features that reflects poorer outcome such as Breslow thickness or ulceration. These unparallelled findings between PRAME and the histopathologic features have also been observed by a few other studies of both cutaneous and uveal melanoma.^22–24 This may be because PRAME exerts its oncogenic effects primarily through molecular and immunological pathways – namely, repression of RAS signaling, promotion of epithelial-to-mesenchymal transition, and creation of an immunologically “cold” tumor microenvironment through downregulation of antigen-presenting molecules such as Beta-2 microglobulin.⁴ These mechanisms operate at a cellular and immunological level and are found to be associated with a higher rate of metastasis but may not be reflected the conventional histopathological measurements.^22,24 Another possible explanation is the ceiling effect from the uniformly advanced-stage profile of our referral center population.

Previous studies have demonstrated strong concordance between PRAME IHC and cytogenetic assays, namely fluorescent in situ hybridization and single nucleotide polymorphism array, as well as consistent positivity in nodal metastases, supporting its effectiveness as a reliable and practical tool for melanoma diagnosis and staging.^9,25 Although our sample size limits definitive prognostic interpretation, the trend observed supports PRAME's potential value as a marker of poor clinical outcome in acral melanoma, and promote for a prompt management of the malignancy.

The main limitation in this study was being a single-center, retrospective study with a limited sample size. Therefore, the prognostic analysis was constrained by small numbers, and mortality outcomes may have been confounded by treatment variability, which could not be statistically adjusted. Larger, multicenter studies with standardized sampling and follow-up are warranted to validate our findings and clarify whether PRAME expression demonstrates stronger prognostic associations in ALM.

In conclusion, PRAME IHC is a practical adjunct for distinguishing ALM from benign acral melanocytic nevi. A cut-off of the extent of expression ≥3 provides optimal diagnostic performance while remaining applicable in routine practice. However, with the current information and understanding, it is tantamount that PRAME IHC should still be used in conjunction with clinical, histomorphology, and, when needed, other ancillary studies in the diagnosis of melanoma. PRAME positivity correlated with higher tumor stage and recurrence, supporting its potential prognostic relevance. The low staining in acral MIS suggests site-specific variation, suggesting further validation.

Supplementary Material

ajd-48-684-s001.docx ^{(24.8KB, docx)}