TERT Promoter Mutations Are Predictive of Aggressive Clinical Behavior in Patients with Spitzoid Melanocytic Neoplasms

Spitzoid neoplasms constitute a morphologically distinct category of melanocytic tumors, encompassing Spitz nevus (benign), atypical Spitz tumor (intermediate malignant potential), and spitzoid melanoma (fully malignant). Currently, no reliable histopathological criteria or molecular marker is known to distinguish borderline from overtly malignant neoplasms. Because TERT promoter (TERT-p) mutations are common in inherently aggressive cutaneous conventional melanoma, we sought to evaluate their prognostic significance in spitzoid neoplasms. We analyzed tumors labeled as atypical Spitz tumor or spitzoid melanoma from 56 patients with available follow-up data for the association of TERT-p mutations, biallelic CDKN2A deletion, biallelic PTEN deletion, kinase fusions, BRAF/NRAS mutations, nodal status, and histopathological parameters with risk of hematogenous metastasis. Four patients died of disseminated disease and 52 patients were alive and disease free without extranodal metastasis (median follow-up, 32.5 months). We found TERT-p mutations in samples from the 4 patients who developed hematogenous metastasis but in none of tumors from patients who had favorable outcomes. Presence of TERT-p mutations was the most significant predictor of haematogenous dissemination (P < 0.0001) among variables analyzed. We conclude that TERT-p mutations identify a clinically high-risk subset of patients with spitzoid tumors. Application of TERT-p mutational assays for risk stratification in the clinic requires large-scale validation.

diagnosed as AST are reclassified as melanoma once distant metastasis develops. To date, no single histopathological criterion 7 or molecular marker is known to predict with certainty the risk of subsequent aggressive disease with these tumors.
Telomerase activity is crucial for tumorigenesis and cancer progression 11,12 . The activity of telomerase, the enzyme responsible for maintaining telomeric DNA during replication, is regulated by the telomerase reverse transcriptase (TERT) gene 13 . Next-generation sequencing studies have identified somatic mutations in the core promoter region of TERT that by generating Ets/TCF transcription binding motifs increase the transcriptional activity of the gene 14,15 . TERT promoter (TERT-p) mutations have been found in 22%-71% of cutaneous melanoma in adult series [14][15][16][17][18][19] and in the majority of conventional pediatric melanoma in a study from our group 20 , suggesting that they contribute to TERT regulation in melanoma. Interestingly, in our original series of pediatric melanoma, we found a hot-spot TERT-p mutation in the single patient with SM who died of disease but in no other patients with spitzoid tumors who had favorable outcomes 20 . We postulated that the molecular mechanism for maintaining telomere may be similar in SM and conventional melanoma. Herein, we investigated the presence of TERT-p mutations in 56 patients with histopathologically well-characterized atypical spitzoid neoplasms for whom follow-up information was available.

Results
Clinical Features. The clinical characteristics of patients are provided in Table S1. Tumors occurred in 33 female and 23 male patients aged 2-61 years (median, 9; mean, 14.6). They arose in skin of the lower extremity (n = 26), upper extremity (n = 9), face (n = 7), trunk (n = 6), ear (n = 5), and scalp (n = 3). Of the 42 patients who underwent sentinel lymph node evaluation, 21 (50%) had at least 1 positive lymph node, of which 9 had extensive nodal metastasis. Fifty-two patients were alive with no evidence of disease at last follow-up (mean, 32.5 months). Four patients developed hematogenous metastasis and died of widespread disease.
Histologic Features. The histologic features of the 56 tumors are provided in Table S2. The Breslow tumor thickness ranged from 0.3 to 13.3 mm (median, 2.85 mm) in the 52 tumors with a favorable behavior and 1.3 to 8 mm (median, 5.25 mm) in the 4 tumors with an unfavorable behavior. The lesional diameter ranged from 1.5 to 17 mm (median, 6.5 mm) in tumors with a favorable behavior and 4 to 12 mm (median, 11 mm) for those with an unfavorable behavior. Ulceration was present in 10 of 52 (19%) tumors with a favorable and in 3 of 4 (75%) tumors with an unfavorable behavior. A high mitotic rate (>5/mm 2 ) was seen in 5 of 52 (10%) tumors with a favorable behavior and in 3 of 4 (75%) tumors with an unfavorable behavior.
BRAF and NRAS Mutations. Of the 56 tumors, 3 (5%) carried a BRAF mutation (2 V600 E in patients with a favorable and 1 V600 K in a patient with an unfavorable clinical course). None of the tumors had an NRAS mutation (Fig. 1). CDKN2A (p16). FISH identified biallelic CDKN2A deletion in 12 of 49 (24%) successfully tested samples from patients with a favorable clinical course and in 2 of 4 (50%) samples from patients with an unfavorable clinical course (Fig. 1). In some patients, the status of p16 was different between their primary and metastatic samples (see Discussion). As expected, biallelic CDKN2A deletion was predictive of loss of p16 by immunohistochemistry (Table S3). In 2 samples, immunohistochemical analysis showed loss of p16 expression without evidence of biallelic gene deletion by FISH (Table S3), suggesting that CDKN2A might be inactivated by mechanisms other than large deletion.
PTEN. Biallelic deletion of PTEN was found in 4 of 40 (10%) samples from patients with a favorable clinical course and in none of the 3 successfully tested samples from patients with an unfavorable clinical course (Table 1).
Scientific RepoRts | 5:11200 | DOi: 10.1038/srep11200 Association Analysis. The presence of TERT-p mutations was significantly associated with the risk of extranodal metastatic disease or death (P < 0.0001) ( Table 1). In addition, age ≥10 years, mitotic rate >5/mm 2 , and ulceration were each associated with the risk of extranodal metastasis or death (P <0.05). In contrast, no statistically significant association was found between the presence of biallelic loss of CDKN2A and extranodal metastasis or death (P = 0.56) ( Table 1). Gender, nodal metastasis, primary tumor thickness, and tumor infiltrating lymphocytes were also not associated with extranodal metastasis. Lesional diameter was marginally associated with extranodal metastasis (P = 0.054). The presence of TERT-p mutations was correlated with age ≥10 years at diagnosis (P = 0.034), as well as mitotic rate >5/mm 2 and ulceration. A multiple-regression analysis to adjust the age effect was not feasible due to the small number of events (metastasis/death) in the cohort.

Discussion
ASTs account for the majority of so-called melanomas encountered in children. Patients with atypical spitzoid neoplasms have frequent sentinel lymph node involvement, but their outcomes are much better than for patients with similar-staged conventional melanoma 21,22 . Lallas et al. conducted a systemic review of the literature and found that having positive sentinel lymph nodes did not predict a worse outcome in patients with AST 9 . In our study, 21 of 42 (50%) patients who underwent sentinel nodal sampling had positive nodes and 2 (5%) developed extranodal disease. Consistent with previous studies 9,22-26 , we find no correlation between regional nodal metastasis and subsequent development of aggressive disease in our patients (P = 0.61).
The difficulty in predicting the risk of hematogenous metastasis in patients with atypical spitzoid lesions has prompted extensive investigations. Sptaz et al. proposed a histopathologic grading scheme for risk stratification of pediatric AST 27 . We found a similar set of variables (age ≥10 years, ulceration, and mitotic activity >5/mm 2 ) associated with a later development of extranodal metastasis, although no single factor by itself was predictive of outcomes. Similar to the study by Heidenreich et al. 19 , the presence of TERT-p mutations in our cohort correlated with histopathologic parameters of poor prognosis, such as mitotic activity and ulceration. It is known that spitzoid tumors developing in the first decade of life (prepubertal ages), irrespective of having histologic attributes of malignancy, do not progress to distant metastasis, except in the rarest instances 21,28 . The absence of TERT-p mutations in tumors from younger patients (<10 years of age) in our study is consistent with this observation, although the underlying biological mechanism for the phenomenon remains to be elucidated.
Activated kinase signaling pathways via chromosomal translocations are responsible for tumorigenesis in spitzoid melanocytic tumors 29,30 . Wiesner et al. found kinase fusions of NTRK1, ROS1, ALK, BRAF, or RET in 55% of Spitz nevi, 56% of ASTs, and 39% of SMs 30 . Similarly, kinase fusions were found across the entire biological spectrum of spitzoid neoplasms in our cohort, indicating that they cannot be applied as a means to predict the biological behavior of spitzoid lesions. The discovery of recurrent DNA copy number gains and losses in melanoma, and the absence of such changes in nevi, prompted the development of diagnostic assays to assist proper classification of histologically challenging melanocytic lesions 31,32 . The multiprobe melanoma FISH assay targets the common regions of alterations in melanoma: 6p25 (targeting RREB1), 6q23 (MYB), 11q13 (CCND1), and 9p21 (CDKN2A) [33][34][35][36] ; whereas comparative genomic hybridization (CGH) studies provide a genome wide view of copy number changes 37,38 . Although both assays are valuable tools to discriminate nevi from borderline or malignant lesions, their ability in risk stratification for spitzoid tumors is uncertain. In our series, samples from 23 patients showed multiple copy number changes by CGH studies or an abnormal result on multicolor melanoma FISH assay, which supported an SM diagnosis, but only in 1 patient a clinically malignant phenotype became apparent (Table S3).
Biallelic 9p21 deletion is proposed as a marker for spitzoid tumors that are at high risk for aggressive behavior. In a study by Gerami et al., of 37 pediatric patients with AST, 9 developed extensive locoregional disease and 2 developed distant metastasis. Tumors in 2 patients with distant metastasis and 7 of 9 patients with advanced locoregional disease harbored homozygous loss of 9p21, suggesting that this is a marker for disease progression 39 . We too observed the acquisition of biallelic deletion of CDKN2A (a reflection of tumor progression) in the nodal metastasis of at least 2 patients in whom the primary tumors retained both copies of the gene (Table S3), but neither patient developed extranodal metastasis in follow-up. As a whole in our series, the presence of biallelic CDKN2A deletion did not have a statistically significant association with the future risk of extranodal metastasis. Of the 4 patients who

Kinase fusion
NRAS mutation 0 (0%) 0 (0%) Not found 23 (49%) 2 (50%) NA 5 0 Table 1. Association of molecular markers and nodal status at diagnosis with outcome in 56 patients with atypical spitzoid melanocytic neoplasms (52 with favorable and 4 with unfavorable clinical outcome). * P-values for association between each factor and risk of hematogenous metastasis/death developed disseminated disease, 2 did not have biallellic CDKN2A loss in their primary or metastatic tumor and 2 had biallelic deletion in a subset of melanocytes that were not present at distant metastatic sites. Moreover, biallelic CDKN2A deletion was relatively common in the entire cohort, including in 23% of patients with a favorable clinical course. Overall, the findings from our study do not support that bialleclic 9p21 deletion is a reliable marker predictive of extranodal metastasis in spitzoid neoplasms. Replicative immortality, a required element for overt malignant transformation in neoplastic cells, is commonly achieved by telomerase activation [11][12][13]40 . Several mechanisms are implicated in the control and reactivation of TERT in cancer cells, such as TERT gene copy gain or epigenetic modulation through TERT-p methylation [41][42][43] . In addition, mutations in the core promoter region of TERT have been shown to increase the transcriptional activity of telomerase and to be an independent marker of poor prognosis in various cancers, such as glioblastoma, thyroid carcinoma [44][45][46][47][48] , and cutaneous conventional melanoma 16,17 . The finding of TERT-p mutations in tumors from patients with fatal outcome in our series, and its absence in those with favorable behavior, strongly suggest that TERT-p mutations are a predictive marker of aggressive clinical behavior in patients with spitzoid lesions.
The negative predictive value of TERT-p mutations, on the other hand, requires further validation. TERT-p mutations may be acquired late during multistep melanoma development 49 . Conceivably, subclones of immortalized melanocytes with TERT-p mutations might be present in a lesion but be missed if below the detection level of standard sequencing assays. Moreover, the diverse mechanisms by which cancer cells maintain telomere length suggest that alternative mechanisms may be enacted alone or with TERT-p mutations to restore telomere length in melanocytes 41,42 . In addition, the follow-up time in our cohort was not extended enough to document the ultimate long-term outcomes of patients with wild-type TERT-p. One of our patients presented with disseminated disease 10 years after an initial diagnosis of AST (see Supplementary material). To prove the reliability and reproducibility of TERT-p mutations as a screening tool in the clinic, the assay needs to be evaluated on primary-site spitzoid neoplasms in large-scale studies with long term follow-up. We are currently collaborating with the Pediatric Melanoma Registry, an international multiinstitutional registry of pediatric melanoma based at the University of Pittsburgh, to work toward this aim.
The unpredictable clinical course of patients with spitzoid neoplasms has made the optimal management of these patients debatable, often leading to overtreatment of patients at minimal risk for disease progression. The use of TERT-p mutations as a marker of aggressive disease can help stratify a subgroup of patients with spitzoid tumors that likely behave in a malignant fashion and for whom more intensive therapies are warranted.

Materials and methods
Tissue Specimens. The study was approved by the institutional review board at St. Jude Children's Research Hospital and the methods were carried in accordance with the approved guidelines. Written informed consent was not required under a HIPAA waiver IRB approval.
Tissue specimens from patients diagnosed with AST or SM were obtained from the surgical pathology archives. Inclusion criteria were as follows: (1) lesions showing some histologic features of Spitz nevus and meeting previously described criteria for AST and SM 4,20,21,50 ; (2) availability of sufficient tissue for sequencing assays; and (3) availability of demographic and follow-up information. A comprehensive histologic evaluation by study investigators (RLB, RD, and AB) identified specimens from 56 patients with atypical spitzoid neoplasms (labeled as SM in 33 and AST in 23), including specimens from 5 previously reported patients 20 . The study material included formalin-fixed paraffin-embedded (FFPE) tissue specimens consisting of primary tumors (n = 49), paired primary and metastatic tumors (n = 5), and metastatic tumors (n = 2) from 56 patients.
Mutational Analysis of BRAF, NRAS, and TERT Promoter. FFPE tumor sections were manually microdissected guided by H&E slides to obtain at least 50% tumor purity in the material used for DNA extraction. Genomic DNA was extracted according to the manufacturer's protocol using Maxwell ® 16 FFPE Plus LEV DNA Purification Kit (Promega). Mutational hotspots for BRAF (exon 15), NRAS (exons 1 and 2), and a portion of TERT-p (HG19 coordinates, chr5: 1295151-1295347) were screened in genomic tumor DNA of the 56 tumors. PCRs were performed, using GoTaq ® Long PCR Master Mix (Promega, Madison, WI) or AmpliTaq Gold ® 360 Master Mix (Applied Biosystems, Foster City, CA) using amplification primers as previously described 20 . Direct sequencing of PCR products was performed using BigDye version 3.1 and a 3730XL DNA analyzer (Applied Biosystems, Foster City, CA). Results were screened using CLC Main Workbench sequence analysis software version 6.0.2 (CLC bio, Cambridge, MA). Dual-color FISH was applied on 4-μ m FFPE sections as previously described 20 . FISH was successfully performed for a panel of kinase fusions (51 samples), copy number CDKN2A (58 primary/metastatic samples), and copy number PTEN (43 samples).
Immunohistochemical Analysis. FFPE tumor sections were processed for immunohistochemical analysis for p16 (JC8; Santa Cruz) as previously described 20 .
Statistical Analysis. Contingency tables were generated to study associations between outcome and risk factors. Pearson's chi-square exact tests were used to test the associations.