Molecular Profiling of Multiple Primary Merkel Cell Carcinoma to Distinguish Genetically Distinct Tumors From Clonally Related Metastases

Key Points

Question  Do clinically designated second primary Merkel cell carcinomas (MCC) actually represent metastases?

Finding  In 2 of 4 cases of clinically designated multiple primary MCC, the second MCC tumor was genetically unrelated to the first tumor, consistent with a new primary. In the other 2 cases, tumors were genetically related, consistent with metastases.

Meanings  Given that clinicopathologic criteria are imperfect for distinguishing second primary MCC from metastases, molecular analysis may assist in the determination of clonality in clinically challenging cases.

Importance  Merkel cell carcinoma (MCC) is an aggressive cutaneous neuroendocrine carcinoma. In rare cases, the development of an additional cutaneous MCC tumor is clinically consistent with a second primary MCC tumor rather than a cutaneous metastasis, which has important treatment and prognostic implications.

Objective  To evaluate genetic relatedness in 4 cases with the clinical diagnosis of multiple primary MCCs.

Design, Setting, and Participants  In this case series, 7 cases of clinically designated multiple primary MCC were identified; 4 cases met inclusion criteria for next-generation sequencing (NGS) analysis. Mutations, copy number alterations, and Merkel cell polyomavirus (MCPyV) sequence were analyzed and compared between clinically designated multiple primary tumors to characterize genetic relatedness and hence assess clonality. Patients with clinically designated multiple primary MCC were identified from the multidisciplinary MCC Program at the University of Michigan, a tertiary care center.

Main Outcomes and Measures  Four cases of clinically designated multiple primary MCC were characterized by tumor sequencing and targeted MCPyV sequencing to distinguish independent primary tumors from related metastases.

Results  Overall, 4 patients in their 70s or 80s were included and analyzed. Cases 1 and 4 were verified as genetically distinct primary tumors and did not harbor similar copy number alterations or demonstrate significant mutational overlap. Cases 2 and 3 were designated as clonally related based on overlapping copy number alterations. In clonally related tumors, chromosomal copy number changes were more reliable than mutations for demonstrating clonality. Regardless of clonality, we found that MCPyV status was concordant for all tumor pairs and MCPyV positive tumors harbored predominatly subclonal mutations.

Conclusions and Relevance  Our findings suggest that patients with MCC may develop a second genetically distinct primary tumor; in this case, the subsequent tumor is likely to develop through similar mechanisms of pathogenesis, either MCPyV-mediated or ultraviolet light–mediated. Next–generation sequencing analysis of chromosomal copy number changes and mutations is useful in distinguishing multiple primary MCCs from progression of MCC clinically resembling multiple primaries, allowing appropriate staging of the patient.

Merkel cell carcinoma (MCC) is a rare cutaneous neuroendocrine neoplasm and presents as a red-to-violaceous nodule, typically on sun-exposed skin of older individuals.¹ At the time of diagnosis of the primary tumor, there is at least a 15% to 20% risk of clinically occult metastasis to the regional nodal basin.^2-4 The most common site of metastasis is the regional lymph node basin, followed by skin, lung, liver, bone, and other sites of distant metastasis. Evidence suggests that MCC may arise via 2 pathways: a virus-associated pathway mediated by the integration of the oncogenic Merkel cell polyomavirus (MCPyV) or an ultraviolet light–damage pathway associated with a high mutation burden, ultraviolet light–signature mutations, and inactivation of the tumor suppressors RB1 and TP5.^5-7

The phenomenon of multiple primary tumors has been observed in melanoma, where distinct cutaneous primary tumors are clonally unrelated.⁸ Importantly, the designation of an additional distinct primary melanoma impacts management, as the lesion is treated as a primary melanoma with excision and possibly additional staging with sentinel lymph node biopsy, rather than a distant cutaneous metastasis. Following a diagnosis of MCC, the development of additional cutaneous tumors, whether adjacent to the treated primary site, within the draining lymphatics or on distant skin, is usually thought to represent a local, in-transit or distant cutaneous recurrence of the original tumor. However, in rare cases, patients may present with a second cutaneous MCC that is spatially and/or temporally separated such that the lesion is clinically designated an independent primary MCC rather than a cutaneous metastasis.^9-16 Given the rarity of multiple primary MCCs, genetic relatedness has been evaluated only in 3 cases that we know of.^11,15,16 One case demonstrated genetic unrelatedness by analyzing sequences of the integrated MCPyV,¹¹ and the other cases demonstrated clonality by comparative genomic hybridization analysis of chromosomal copy number changes.^15,16 The distinction between 2 primary tumors and a primary-metastasis pair has significant impact on treatment and prognosis.

Next-generation sequencing (NGS), which provides a broad profile of mutations and chromosomal copy number alterations (CNAs) within a tumor, is ideally suited for clonality analysis.¹⁷ In this study, we evaluate clonal relatedness in 4 patients with clinically designated multiple primary MCCs using NGS.

All studies were conducted according to protocols previously approved by the institutional review board of the University of Michigan; archival formalin-fixed, paraffin-embedded tissues collected for diagnostic purposes were used according to waiver of consent protocol approved by the institutional review board. Seven MCC cases (14 tumors) designated as multiple primary tumors (distant metastases not suspected) were identified from a database of 473 cases at the Multidisciplinary MCC Program at the University of Michigan from January 2006 through April 2016 (eMethods in the Supplement). Inclusion criteria were availability of paraffin blocks, adequate tumor for sequencing, and adequate quality DNA for NGS analysis; 4 cases (8 tumors) met these criteria. For case 4, a matched regional lymph node metastasis was also sequenced. Metastases in other cases did not yield adequate tumor purity or DNA quantity and/or quality for inclusion. Clinicopathologic features are summarized in Table 1. Two primary-metastasis pairs previously sequenced by the same methods were included in analysis.¹⁸

Targeted NGS assessing the complete coding sequence of 409 cancer related genes on archived formalin-fixed, paraffin-embedded (FFPE) material to identify somatic mutations and CNAs was performed using the Ion Ampliseq Comprehensive Cancer Panel (CCP) (Thermo Fisher Scientific) as described previously^18,19 and in eMethods in the Supplement.

Next-generation sequencing data analysis was performed as described^18,19 to exclude probable germline mutations and nominate high-confidence somatic single nucleotide variants and insertions and/or deletions (indels). Further details are in eMethods in the Supplement.

Copy number alterations were identified as described^18-20 and further detailed in eMethods in the Supplement. Library quality parameters are in eTable 1 in the Supplement.

The fraction of shared somatic mutations and CNAs between samples was determined by similarity index calculation.^21,22 Further details are in eMethods in the Supplement.

Detection of MCPyV sequences in tumor DNA was performed by quantitative polymerase chain reaction (qPCR) and PCR-Sanger as described^23-26 using novel or described primers (eMethods, eTable 2 in the Supplement).

Seven cases were identified in which a second MCC tumor was clinically designated a second primary tumor rather than a recurrence or distant metastasis. Four cases had sufficient quantity and quality of DNA for clonality assessment. The clinical histories of the 4 analyzed cases are summarized in Table 1.

A man in his 70s was diagnosed with a primary MCC on the left third finger. He underwent amputation and sentinel lymph node biopsy (SLNB) revealing clear margins at the primary site and 2 negative sentinel lymph nodes (SLNs) from the left axilla. He did not undergo adjuvant radiation therapy. Six months later, biopsy of a lesion on his right (contralateral) first finger showed MCC. Restaging imaging was negative for metastatic disease. He underwent excision and SLNB revealing clear margins at the primary site and 3 negative SLNs from the right axilla. Adjuvant radiation therapy was not indicated. He has been free of disease since treatment of the presumed second primary MCC almost 6 years ago.

A man in his 80s was diagnosed with a primary MCC on the left elbow. He underwent excision with clear margins and left axillary SLNB interpreted as negative at an outside institution. Nineteen months later, biopsy of a cutaneous lesion on his left thigh revealed MCC, at which time he presented to the University of Michigan for further evaluation. Upon review, the initial left axillary SLNB was found to be positive for microscopic MCC in 1 of 2 lymph nodes. Therefore, at the time, the newly diagnosed lesion on the thigh was presumed to be a distant cutaneous metastasis. Restaging imaging was negative for metastatic disease. The lesion on the thigh was excised, but SLNB or adjuvant therapy was not recommended based on the presumption of stage IV disease. Four months later, he developed a nodal MCC metastasis in the left groin. Inguinal lymph node dissection revealed 2 of 26 lymph nodes positive for MCC. He did not undergo adjuvant radiation therapy. At this point, based on patterns of metastasis, the tumors were considered to represent 2 primary MCCs, each with regional nodal metastases. The patient has been free of disease for nearly 4 years since the left inguinal lymph node dissection.

A woman in her 70s was diagnosed with 2 concurrent MCCs on the left and right nasal alae. Staging imaging did not demonstrate metastatic disease. The lesions were designated as 2 primary tumors, and she underwent excision and SLNB for both sites. Pathology revealed clear margins at both primary sites and 2 of 4 positive SLNs in the left neck, and 3 negative SLNs in the right neck. Based on the uncertain clinical presentation, the patient underwent bilateral neck dissection revealing additional 23 and 41 negative lymph nodes in the left and right neck, respectively. She did not undergo adjuvant radiation and has been free of disease for 1.3 years after surgery.

A man in his 70s was diagnosed with a large primary MCC on the left cheek with a concurrent nodal metastasis in the left parotid gland. He underwent excision, parotidectomy, and left neck dissection, which revealed a positive deep margin at the primary site and 9 of 20 lymph nodes positive for metastatic MCC. He underwent adjuvant radiation therapy to the left cheek and neck. He had no evidence of recurrence until 42 months later, when he developed a second MCC on the right cheek. Restaging imaging was negative for metastatic disease. Under the presumption that the lesion represented a second primary MCC, he underwent excision and SLNB, which revealed clear margins at the primary site and 2 of 2 right parotid SLNs positive for MCC. He underwent right parotidectomy and neck dissection with 22 additional negative lymph nodes in the right neck. He declined adjuvant radiation therapy due to significant xerostomia from his prior contralateral treatment. He had progression of disease with recurrence in the right neck and liver.

Overall, 6 of 8 (75%) of the primary tumors in cases 1 through 4 harbored MCPyV large and small T antigen sequences (Figure 1; Table 1) (eFigure 1 in the Supplement). Both tumors from a given patient had similar MCPyV copy number (Table 1). One patient (case 4) lacked detectable MCPyV in the 2 primary tumors (Figure 1; Table 1) (eFigure 1 in the Supplement). In 5 tumors, adequate DNA remained for partial sequencing of the MCPyV large T antigen exon 2, revealing identical sequences across all cases, without tumor-specific mutations in the analyzed region (eTable 2 in the Supplement).

We assessed somatic, high-confidence mutations and copy number alterations in 409 cancer-related genes across the primary tumors from the 4 cases, as well as a regional lymph node metastasis in case 4. Results are summarized in Table 1 and Table 2, with additional details in eTables 3 and 4 in the Supplement. Recurrent events that suggest clonality were identified in a subset of paired tumors, as described in clonality analysis below (Figure 2) (eTable 4 in the Supplement).

Clonality was evaluated in the clinically designated multiple primary MCC tumors by quantitating the fraction of shared genetic events in a pair using the similarity index calculation.²² To determine the expected range of overlap for clonally related MCC tumors, we first assessed genetic similarity in the primary MCC tumor and matched regional metastasis from case 4, as well as 2 primary-metastasis pairs (MCC9-MCC14 and MCC10-MCC16; cases 5 and 6) (eTable 5 in the Supplement) previously sequenced using an identical approach.¹⁸ Primary MCC tumor-metastasis pairs (n = 3) displayed similarity indices, ranging from 0.21 to 1.0 for CNA and 0.09 to 0.0.91 for mutational analysis (Figure 3). Thus, CNA and mutational similarity indices greater than 0.21 and 0.09, respectively, were used as the minimum scores representing likely clonal relatedness. Analysis of random pairings (primary tumors from different patients, representing 24 total pairings of cases 1-4), served as a negative control for clonality analysis and displayed no similarities (similarity index = 0 for both CNA and mutational analyses). For case 1, similarity indices were 0 (not clonal) and 0.014 (not clonal), for CNA and mutational analyses, respectively. For case 2, similarity indices were 0.8 (clonal) and 0 (not clonal) for CNA and mutational analyses, respectively. For case 3, similarity indices were 0.67 (clonal) and 0 (not clonal) for CNA and mutational analyses, respectively. For case 4, similarity indices were 0 (not clonal) for both CNA and mutational analyses. In summary, cases 1 and 4 were confirmed as independent primary tumors without genetic overlap, while cases 2 and 3 demonstrated genetic relatedness in chromosomal copy number changes and were designated as clonally related (Figures 2 and 3).

To evaluate the discrepancy in the clonality results between CNA and mutational analyses, we evaluated the allele frequency of mutations. Mutations in MCPyV-negative tumors were observed at an average allele frequency of 49.2%, consistent with heterozygous mutations present in most or all tumor cells. In contrast, mutations in MCPyV-positive MCC display significantly lower allele frequencies (mean, 14.3%), suggesting mutations affecting only a minority of tumor cells (eFigure 2A and B and eTable 6 in the Supplement). This pattern was consistent in an independent cohort of previously sequenced MCC tumors (eFigure 2C in the Supplement).⁵ Mutations affecting a minority of cells in MCPyV-positive tumors might not be present in tumor cells that metastasize, and/or might arise after metastasis (eFigure 2D in the Supplement). Hence, lack of shared mutations is not informative regarding clonality in this context. Taken together, our results support CNAs as earlier events in MCPyV-positive MCCs, with the vast majority of somatic mutations being subclonal (thus passenger mutations).

Prioritized somatic driving mutations were also evaluated in the multiple primary tumor cases determined to be clonally unrelated tumors. Prioritized driver mutations were not identified in case 1, which displayed unrelated primary MCC tumors on the left and right fingers. Prioritized driver TP53 and RB1 mutations were identified in case 4, which displayed unrelated primary MCC tumors on the left and right cheek and a matched nodal metastasis from the right cheek. The primary tumor on the right cheek and the matched nodal metastasis harbored the same TP53 (Q317X and P32L) and RB1 (Q637X) mutations. The primary tumor on the left cheek harbored distinct TP53 (H179Y) and RB1 (R552X) mutations. Together these results support the assessment of clonally independent tumors based on separate driving mutations.

The clinical diagnosis of multiple primary MCC tumors is rare but important for management and prognosis. Treatment of a second primary MCC includes excision and staging with SLNB, and prognosis depends upon sentinel lymph node status. Conversely, the diagnosis of a distant cutaneous metastasis is managed as stage IV disease. Here, we used NGS to analyze clonality in 4 cases of clinically designated multiple primary MCC tumors. For maximal sensitivity, we compared tumor mutations, chromosomal copy number changes, and MCPyV sequence.

Multiple primary MCCs were verified in case 1 (second primary MCC arising on contralateral hand) and case 4 (second primary MCC arising on contralateral cheek) as the tumor pairs did not harbor similar copy number alterations or significant mutational overlap (Table 1). In case 4, tumors displayed distinct TP53 and RB1 mutations, consistent with the proposed role for inactivation of these tumor suppressor genes as driving, early events in MCPyV-negative MCC.^5,27 Of note, in case 1, which was MCPyV-positive, a limited number of discordant CNAs drove the determination of multiple primary MCC status, as identified mutations in each sample were nonprioritized and subclonal based on variant allele frequency assessment. Clonal relatedness was identified in case 2 (MCC on left elbow and subsequently on left thigh) and case 3 (MCCs on bilateral nasal alae). In case 2, both tumors shared multiple copy number changes, that are predicted to be early events, compared with mutations arising later. Although speculative, we hypothesize that the second MCC tumor on the left thigh arose from hematogenous spread from the primary tumor or the untreated left axillary nodal metastasis and that the nodal recurrence in the left groin was from lymphatic drainage from the left thigh metastasis.

In case 3, both tumors displayed significant overlap in CNAs. Given the synchronous occurrence, these tumors likely represent a primary and in-transit metastasis pair. A less likely scenario is that both tumors represent in-transit metastases from a regressed midline primary tumor.

Rare cases of multiple primary MCCs have been reported.^9-16 The designation of multiple primary MCC tumors should be assigned carefully. In our multidisciplinary MCC experience, some cases reported as multiple primary MCC tumors are more consistent with the presentation of in-transit metastatic disease, such as cases of multiple MCC tumors on the scalp⁹ and the ankle.¹⁰ Importantly, genetic relatedness has been evaluated by alternative technologies in 3 cases with findings similar to our current study.^11,15,16 Sharma et al¹¹ confirmed the diagnosis of independent primary MCCs by sequence analysis of the MCPyV genomes in a patient with a MCC on the left upper arm who developed a second MCC on the right elbow 6 years later. Ahronowitz et al¹⁵ demonstrated clonality by array comparative genomic hybridization analysis in a patient with a primary tumor on the right cheek and the second tumor on the left leg 2 months later. Nagy et al¹⁶ used comparative genomic hybridization analysis to demonstrate shared and distinct molecular patterns in a primary MCC tumor on the lip and a second tumor on the palatine tonsil 7 years later. Nagy et al¹⁶ conclude that the second tumor was an independent primary MCC with a field effect from the first tumor. Given the genetic overlap, we and others interpret these tumors to be clonally related.²⁸

Merkel cell polyomavirus is found to be integrated in up to 80% of MCCs.²⁹ Interestingly, we found that tumors from the same patient were consistently MCPyV negative or MCPyV positive, regardless of clonal relatedness. Although our cohort is small, this observation suggests that a given patient may be predisposed to development of either MCPyV-negative or MCPyV-positive MCC. Furthermore, MCPyV status alone is not a reliable indicator of clonal relatedness between MCC tumors.

This study has several limitations. The occurrence of clinically designated multiple primary MCCs is exceedingly rare, which limits appropriate cases with tissue suitable for analysis. However, that we know of, this series represents the largest to date, the first to use NGS for clonality analysis, and the only study to analyze clonality in multiple primary MCCs by multiple parameters in parallel: mutations, CNAs, and MCPyV sequence. Of these, we found CNA similarity to be a more consistent indicator of clonality than mutations in genetically related tumors, thus mutation analysis is less informative in this context. In particular, MCPyV-positive MCC has low mutational burden, with the vast majority of somatic mutations being subclonal, nondriving mutations, which may explain why examination of tumor mutations was not reliable for demonstrating clonality in these tumors. Unlike a previous study, we did not find MCPyV sequence analysis to be useful in evaluating clonal relatedness, as all MCPyV-positive tumors displayed high similarity in viral sequence. However, coverage of the MCPyV sequence was limited in many cases due to low remaining material; therefore, we cannot exclude the possibility that more extensive analysis of viral sequences might be informative. Likewise, we only assessed a portion of the tumor genome, and hence, are unable to assess fine subclonal structure in the related tumors. However, the purpose of our study was to assess clonal relatedness between 2 apparent primary tumors, not detailed intratumoral or intertumoral heterogeneity.

Our findings show that patients with MCC may develop a second genetically distinct primary tumor, which is likely to develop through similar mechanisms of pathogenesis. Our study also supports clonality, and hence metastasis, in 2 cases of presumed multiple primary tumors. These findings underscore the challenge to correctly distinguish a second primary MCC from an isolated distant cutaneous metastasis, which has critically important prognostic and therapeutic implications. Our findings also support copy number analysis as more effective than mutational analysis for determining clonality in MCC, including MCPyV-positive tumors with low mutational burden. Given that clinicopathologic criteria may be imperfect, as seen in multiple primary lung carcinoma,³⁰ copy number analysis by array comparative genomic hybridization or NGS may assist in the determination of clonality in clinically challenging cases.

Corresponding Author: Paul W. Harms, MD, PhD, University of Michigan Medical School, 3261 Medical Science I, 1301 Catherine, Ann Arbor, MI 48109-5602 (paulharm@med.umich.edu).

Accepted for Publication: December 8, 2016

Published Online: April 12, 2017. doi:10.1001/jamadermatol.2017.0507

Author Contributions: Drs K. L. Harms and P. W. Harms, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: K. L. Harms, Johnson, Tomlins, P. W. Harms.

Acquisition, analysis, or interpretation of data: K. L. Harms, Lazo de la Vega, Hovelson, Rahrig, Cani, Liu, Fullen, Wang, Andea, Bichakjian, Tomlins, P. W. Harms.

Drafting of the manuscript: K. L. Harms, Lazo de la Vega, Rahrig, P. W. Harms.

Critical revision of the manuscript for important intellectual content: K. L. Harms, Lazo de la Vega, Hovelson, Cani, Liu, Fullen, Wang, Andea, Bichakjian, Johnson, Tomlins, P. W. Harms.

Statistical analysis: Lazo de la Vega, Hovelson, P. W. Harms.

Obtained funding: P. W. Harms.

Administrative, technical, or material support: Liu, Wang, Andea, Tomlins, P. W. Harms.

Supervision: K. L. Harms, Cani, Johnson, Tomlins, P. W. Harms.

Conflict of Interest Disclosures: Dr Tomlins has received travel and had a separate sponsored research agreement with ThermoFisher Scientific and is a cofounder, equity holder, and consultant for Strata Oncology. No other conflicts are reported.

Funding/Support: This study was supported in part by the A. Alfred Taubman Medical Research Institute (award to Dr Tomlins as the A. Alfred Taubman Emerging Scholar). This study was also supported by the Dermatology Foundation Dermatopathology Research Career Development Award (P. W. Harms), as well as the Anatomic Pathology Project Fund of the University of Michigan Department of Pathology.

Role of the Funder/Sponsor: The funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: We are indebted to Nisha Meireles, BSc, for maintaining the Merkel cell carcinoma clinical database.