Epidemiological data

Samples from 115 patients were examined for expression of NY-ESO-1, tyrosinase, MAGE-A3, and TPTE. Pediatric melanomas (n = 25, 12:13 male:female, ages 1–17 years, median tumor thickness 3.25 mm) were compared with three control cohorts: melanomas in young adults (n = 31, 13:18 male:female, ages 20–30 years, median tumor thickness 0.75 mm), adult melanomas (n = 29, 13:16 male:female, ages 33–95 years, median tumor thickness 2.5 mm), and benign melanocytic nevi of childhood (n = 30, 15:15 male:female, ages 0–17 years). Pediatric melanomas were predominantly stage II (48%) at diagnosis according to AJCC 2017, while melanomas in young adults were predominantly stage I (77%), and adult melanomas were evenly distributed between stages I, II, and III (each 31%). The most common subtype of pediatric melanomas was spitzoid melanoma (48%), whereas superficial spreading melanoma was predominant in young adult (71%) and adult melanomas (66%) (Table 1).

Table 1 Epidemiological dataExpression profile of TAAs in pediatric melanoma, melanoma in young adults, adult melanoma, and benign melanocytic nevi of childhood

TAA expression in primary tumor and metastasis samples was analyzed on the protein level by immunohistochemical staining and quantified automatically using QuPath (Fig. 1). Cutoff for positivity for all markers was defined as ≥ 1% positively stained tumor cells. The expression pattern of the four markers showed great variability in terms of homogeneity within the same study group as well as between different study groups with a tendency towards a more homogeneous staining for tyrosinase (Supplementary Fig. 4). Staining intensity was not assessed in this study because of non-standardized preanalytical conditions, such as cold ischemia time, fixation time, and fixative. In general, tyrosinase was frequently expressed in all four cohorts (86.2–100.0%), while NY-ESO-1, MAGE-A3, and TPTE were less frequently expressed overall with marked differences between cohorts (Fig. 2a–d). Supplementary Tables 2 to 5 list TAA expression for each cohort in relation to clinical data, follow-up, and outcome.

Fig. 1

Immunohistochemical staining of TAA protein expressions in pediatric melanomas. a Positive NY-ESO-1 staining of patient K5, b negative NY-ESO-1 staining of patient K1, c positive tyrosinase staining of patient K22, d negative tyrosinase staining of adult patient A20 as negative staining of tyrosinase is absent in our cohort of pediatric melanoma, e positive MAGE-A3 staining of patient K1, f negative MAGE-A3 staining of patient K5, g positive TPTE staining of patient K1, h negative TPTE staining of patient K27. Scale bars are equal to 50 µm

Fig. 2

Expression levels of NY-ESO-1, tyrosinase, MAGE-A3, and TPTE a–d Protein expression by immunohistochemistry. a Pediatric melanoma, b melanoma in young adults, c adult melanoma, d benign nevi of childhood. e–h Transcript expression by RT-qPCR. e pediatric melanoma, f melanoma in young adults, g adult melanoma, h benign nevi of childhood. Samples with late or no Cqs for targets were classified as negative. i Concordance of IHC and RT-qPCR over all samples (n = 114). OPA overall percent agreement, PPA positive percent agreement, NPA negative percent agreement. j–m Protein expression of TAAs in different age strata and subtypes of pediatric melanoma. j Childhood melanoma defined as patients under 12 years, k adolescent melanoma defined as patients from 12 to 18 years, l Spitzoid melanoma, m non-spitzoid melanoma. CM conventional melanoma (includes SSM, NM, and ALM), CNM congenital nevus melanoma, MET metastasis, OT other type of melanoma

Tyrosinase was expressed in all pediatric melanoma samples, followed in frequency by TPTE (44.0%, n = 11), MAGE-A3 (12.0%, n = 3), and NY-ESO-1 (8.0%, n = 2) (Fig. 2a, Supplementary Table 2). Tyrosinase was the only TAA expressed in a majority (56.0%, n = 14) of pediatric melanoma samples. Some pediatric melanomas expressed two (24.0%, n = 6) or three (20.0%, n = 5) TAAs, and none expressed all four.

Almost all young adult melanomas expressed tyrosinase (96.8%, n = 30) and expressed the other three TAAs in notably different frequencies than pediatric melanomas (Fig. 2b, Supplementary Table 3). TPTE was much less frequently expressed in young adult melanomas (3.2%, n = 1), whereas NY-ESO-1 and MAGE-A3 were more frequently expressed (both 19.4%, n = 6). Most young adult melanomas expressed one TAA (61.3%, n = 19), almost a third expressed two TAAs (29.0%, n = 9), while few expressed three TAAs (6.5%, n = 2). Simultaneous expression of all four TAAs was not detected in young adult melanomas.

Adult melanomas expressed NY-ESO-1 (48.3%, n = 14) and MAGE-A3 (75.9%, n = 22) much more frequently than either young adult or pediatric melanomas (Fig. 2c, Supplementary Table 4). Adult melanomas expressed TPTE (48.3%, n = 14) as frequently as pediatric melanomas. Again, adult melanomas frequently expressed tyrosinase, though slightly less so than the other melanoma cohorts (86.2%, n = 25). In contrast to the younger melanoma cohorts, a minority of adult melanomas expressed one TAA (20.7%, n = 6), with most expressing either two (13.8%, n = 4), three (37.9%, n = 11), or four (24.1%, n = 7) TAAs. One sample expressed none of the TAAs (3.4%).

Almost all benign melanocytic nevi of childhood controls except for two expressed tyrosinase (93.3%, n = 28), but none of the other showed detectable TAA expression (Fig. 2d, Supplementary Table 5).

We also investigated transcript expression of the TAAs in our cohorts using RT-qPCR (Fig. 2e–h). RT-qPCR and IHC results agreed well for all four TAAs over all four cohorts, with OPA ranging from 96.5 to 99.1% (Fig. 2i). Ten samples (n[NY-ESO-1] = 3, n[tyrosinase] = 3, n[MAGE-A3] = 4, n[TPTE] = 1) showed inconsistent transcript and protein expression data. One of these samples resulted in two inconsistent cases; the other cases were distributed to different samples. In eight of eleven inconsistent cases the transcript result was negative with late or undetected targets, only for one case the dCq result was located near its RT-qPCR cutoff. These samples predominantly represented junctional tumors, thin tumors, had few tumor cells, or were archived for extended times.

To further explore differences in TAA expression frequency within the cohort of pediatric melanomas, TAA expression was analyzed separately for children (n = 17, ages 0 < 12 years) and adolescents (n = 8, ages 12 < 18 years) (Fig. 2j, k). The only marked difference between these subgroups was found in MAGE-A3 expression which should not be overinterpreted due to the limited sample number.

To investigate whether TAA expression differed between melanoma subtypes with different prognostic outcomes, we compared the expression patterns of spitzoid melanomas, which have a favorable prognosis, to the combined expression patterns of conventional melanoma (CM; including SSM, NM, and ALM), congenital nevus melanoma (CNM), other subtypes (OT), and metastatic melanoma (MET), which all have a poor prognosis (Fig. 2l, m). All NY-ESO-1-positive samples fell within the CM/CNM/OT/MET group. Additionally, MAGE-A3 and TPTE expression frequency was marginally lower in the CM/CNM/OT/MET group.

Expression of TAAs in relation to patient age

To assess whether the differences of TAA expression patterns in different age strata were associated with patient age, we plotted TAA expression frequencies versus age (20-year bins) (Fig. 3a, c, e, g) and performed logistic regression models (Fig. 3b, d, f, h). The analysis was performed over all melanoma samples including metastatic samples. The first group (0–20 years) contains all pediatric (n = 25) and young adult samples until 20 years (n = 2, ntotal = 27). The age groups 41–60 and 80 + are underrepresented with seven and five samples, respectively.

Fig. 3

Expression of TAAs (IHC) in relation to patient age Samples were combined in age groups of 20 years: 0–20 (n = 27), 21–40 (n = 32), 41–60 (n = 7), 61–80 (n = 14), 81 + (n = 5). a, b NY-ESO-1, c, d tyrosinase, e, f MAGE-A3, g, h TPTE, a, c, e, g relative expression frequency per age group in pediatric, young adult, and primary adult melanomas, b, d, f, h logistic regression models over pediatric, young adult, and adult melanoma samples. p Values were determined with likelihood ratio G2 test

Regression analysis showed that as patient age increased, the prevalence of tyrosinase expression decreased significantly while NY-ESO-1 and MAGE-A3 expression prevalence increased significantly. The slight decrease in NY-ESO-1 expression for patients older than 80 years is negligible because of the small patient number in that age group (Fig. 3a, b). Prevalence of TPTE expression did not significantly correlate with patient age (Fig. 3g, h).

Expression of TAAs in relation to patient survival

To investigate whether expression of individual TAAs may have a prognostic value for pediatric melanoma patients, we performed Kaplan–Meier analyses of event-free survival in a cohort of 22 patients with available follow-up data, four of whom died during the analyzed period (Fig. 4). Statistical analysis was not performed for tyrosinase expression because this TAA was detected in all pediatric melanoma samples (Fig. 4c). No prognostic significance was found for the other three TAAs (NY-ESO-1, p = 0.13; MAGE-A3, p = 0.19; TPTE, p = 0.1; Fig. 4a, e, g). These results are also evident in deceased pediatric melanoma patients. Three of the four deceased children expressed only tyrosinase. Only one deceased patient expressed three TAAs (NY-ESO-1, tyrosinase, and TPTE) (see Supplementary Table 2). In contrast, a comparison of the NGS analyses shows that two of the four deceased patients had a BRAF V600 mutation and one patient had an NRAS mutation (see Supplementary Table 2).

Fig. 4

Event-free survival of melanoma patients stratified by TAA expression a, c, e, g Pediatric melanoma. As all pediatric melanomas were positive for tyrosine (c), an analysis was not possible. b, d, f, h Non-pediatric melanoma. a, b NY-ESO-1 expression; c, d tyrosinase expression; e, f MAGE-A3 expression; g, h TPTE expression

For comparison, we performed the same survival analysis on a combined cohort of 58 young adults and adults with melanoma with available follow-up data, seven of whom died during the analyzed time period. Both MAGE-A3 (p < 0.001; Fig. 4f) and TPTE (p = 0.0045; Fig. 4h) expression were significantly associated with melanoma-related events, whereas NY-ESO-1 (p = 0.77; Fig. 4b) and tyrosinase (p = 0.25; Fig. 4d) expression were not significantly associated with prognosis. Six of the seven adults in the cohort who died of melanoma showed expression of more than one TAA. Notably, expression of MAGE-A3 was detected in all seven deceased patients in this cohort (Supplementary Table 3 and 4).