A, H-E stain reveals a discrete nest of atypical epithelioid melanocytes in the superficial dermis (arrowhead). These cells contain ample cytoplasm, enlarged nuclei, and central nucleoli, cytologically identical to the overlying in situ component and therefore consistent with microinvasive melanoma (H-E score, 2) (original magnification ×400). B, Melanoma antigen (melan-A) immunostain highlights these melanoma cells (melan-A score, 2) (original magnification ×400).
A, H-E stain shows large nests within the epidermis (top left) consistent with in situ melanoma. Rare, similarly appearing cells are seen in the dermis (arrowheads). Although suspicious for invasion, these cells are obscured by admixed lymphocytes and histiocytes that preclude definitive diagnosis (H-E score, 1) (original magnification ×400). B, Double-staining with melan-A and SOX10 immunostains highlight these cells (arrowheads) with cytoplasmic staining for melan-A (brown) and nuclear staining for SOX10 (red), confirming the presence of microinvasion (melan-A score, 2) (original magnification ×400).
A, Hematoxylin-eosin (H-E) stain reveals numerous atypical melanocytes within the epidermis, consistent with in situ melanoma. No dermal invasion was identified (H-E score, 0) (original magnification ×200). B, Melan-A immunostain highlights the in situ melanoma as well as a few atypical cells with similar cytomorphology in the superficial dermis (arrowhead) (melan-A score, 2). Double melan-A/SOX10 immunostains shown in the inset confirm the melanocytic origin of these cells by demonstrating both cytoplasmic staining for melan-A (brown) and nuclear staining for SOX10 (red). These findings support the diagnosis of microinvasive melanoma (original magnification ×200; inset, ×400).
A, Hematoxylin-eosin (H-E) stain reveals an increased number of atypical epithelioid melanocytes disposed singly along the dermoepidermal junction with foci of pagetoid spread, consistent with in situ melanoma. No dermal melanocytes were identified (H-E score, 0) (original magnification ×200). B, Melanoma antigen (melan-A) immunostain highlights the in situ melanoma and also rare single cells in the dermis (melan-A score, 1). These dermal cells (black arrowhead) are significantly smaller and blander than the in situ melanoma cells (white arrowhead) and are best interpreted as occult dermal nevus cells (original magnification ×200).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Bax MJ, Johnson TM, Harms PW, et al. Detection of Occult Invasion in Melanoma In Situ. JAMA Dermatol. 2016;152(11):1201–1208. doi:10.1001/jamadermatol.2016.2668
Why do patients with in situ melanoma develop metastasis?
Occult invasive melanoma was detected in one-third of 34 unequivocally diagnosed in situ melanomas; approximately half were identified only by immunohistochemistry and the other half were noted by cutting deeper into the tissue block. This evidence provides a plausible explanation for the rare occurrence of metastasis from in situ melanoma.
Patients with in situ melanoma are at risk for occult invasion; thus, history and physical examination, education, and surveillance recommendations should be similar between patients with in situ and thin invasive melanoma.
It is unclear why some patients with in situ melanoma develop metastases. Few reports demonstrate occult invasion with immunohistochemistry staining, which were discordant with reports interpreting such staining as false-positive.
To investigate the occurrence of occult invasive disease within in situ melanoma by using methods to circumvent potential limitations in prior study designs.
Design, Setting, and Participants
Unequivocal in situ melanoma without associated nevi or regression was identified using a consecutive sample of 33 cases plus 1 index case in an academic medical center. After cutting deeper into the most representative tissue block, 3 sequential slides were stained with hematoxylin-eosin (H-E), melanoma antigen (melan-A), and again with H-E. Melan-A–stained slides showing definitive invasion were double-stained with Sry-related HMg-Box gene 10 (SOX10) to confirm the melanocytic nature of the cells of interest. The study evaluated the possibilities of occult invasion detected by immunohistochemistry, sectioning deeper into the tissue block, or both. Slides were independently scored by 3 dermatopathologists with interrater reliability assessed. The study was conducted from January 1, 2012, to July 31, 2014.
Main Outcomes and Measures
Assessment of the occurrence of occult invasion, diagnosis of invasion by immunohistochemistry alone vs cutting deeper into the tissue block, and occurrence of false-positive results using immunohistochemistry alone.
Occult invasive melanoma was detected in 11 of 33 consecutive cases (33%) of previously diagnosed unequivocal in situ melanoma. Six of 11 melanomas (55%) were diagnosable only by immunohistochemistry. The remaining 5 tumors (45%) were diagnosable by both melan-A and H-E staining, likely as a result of simply cutting deeper into the tissue block. Four cases (12%) showed a few melan-A–positive cells in the dermis, which was insufficient for a diagnosis of invasive melanoma and most consistent on a cytomorphologic basis with occult nevi.
Conclusions and Relevance
Although rare, in situ melanoma may metastasize. Occult microinvasion was demonstrated in up to one-third of the specimens in the present study, which provides a plausible explanation for this adverse event. Thus, history and physical examination including regional lymph nodes, education, and surveillance recommendations should be based on a very low, but not zero, risk of metastasis.
In 2016 an estimated 144 860 new cases of melanoma will be diagnosed, including 68 480 in situ melanomas.1 With the inclusion of in situ tumors, melanoma is the third most common cancer type in men and the fourth most common in women. An estimated 10 130 deaths due to melanoma will occur in the United States in 2016. An increased risk of metastasis and death correlates with a deeper Breslow depth of invasion. Before invasion, most melanomas develop from melanocytic precursors in the epidermis. By definition, in situ or stage 0 (TisN0M0) melanoma is confined to the epidermis with theoretically no risk of metastasis. Yet, although rare, in situ melanoma metastasizes and patients die of the disease. Clinicians and patients may not fully appreciate this occurrence, which can lead to suboptimal surveillance, counseling, and recommendations.
Based on an index case of metastatic in situ melanoma, our objective was to investigate the possibility of occult invasive disease within in situ melanoma. The presence of occult invasion posits a plausible explanation for development of metastasis with in situ melanoma and provides a rationale to modify clinical practice. We also sought to critically examine and compare our results with those of prior conflicting reports regarding the frequency of occult invasion in cases with an original pathologic diagnosis of in situ melanoma by using a method to circumvent potential pitfalls and limitations in prior study designs.
We queried the melanoma database on patients with in situ melanoma diagnosed and treated at the University of Michigan Comprehensive Cancer Center from January 1, 2012, to July 31, 2014. The index case (a patient who developed regional lymph node metastasis 7 years after the initial diagnosis in 2007) was also included. The electronic medical record for each case was reviewed by one of us (T.M.J.) to confirm an unequivocal diagnosis of primary in situ melanoma. Only cases with all tissue blocks from diagnosis to excision internally available and present within the University of Michigan Comprehensive Cancer Center were included. Histopathologic material was rereviewed by a dermatopathologist with melanoma expertise (M.P.C.) for inclusion for further study. Exclusion criteria included locally recurrent lesions, presence of regression, associated nevus clinically or histopathologically, equivocal diagnosis (eg, concerning for early or evolving in situ melanoma), or any suspicion for microinvasion on rereview of slides. Histopathologic factors recorded included melanoma subtype (lentigo maligna, superficial spreading, or acral lentiginous), appendageal involvement (present or absent), and inflammation (present or absent). Clinical features included age at diagnosis, sex, and location. The study was approved by the University of Michigan Medical School Institutional Review Board for Human Subject Research with waiver of informed consent.
The histologic tissue block most representative of the initial diagnosis of in situ melanoma was selected. The block was “faced” (shaved deeper into the block) before 3 consecutive 4-µm sections were obtained. The first section was stained with hematoxylin-eosin (H-E), the second with melanoma antigen (melan-A) immunohistochemistry, and the third again with H-E. For melan-A, sections were deparaffinized and heat-induced epitope retrieval was performed (PT Link with a proprietary Tris buffer pH 9, FLEX Target Retrieval solution with high pH; Dako). After blocking endogenous peroxidase activity, slides were incubated for 60 minutes at room temperature with a mouse monoclonal melan-A antibody (catalog No. 281A-86; Cell Marque) and subsequently detected (EnVision FLEX + mouse 3,3′-diaminobenzidine detection system; Dako) using brown chromogen.
The H-E– and melan-A–stained slides were independently reviewed by 3 dermatopathologists (P.W.H., D.R.F., and M.P.C.). A score of 0 was given for no invasive melanoma, 1 for dermal cells equivocal or suspicious for invasion (cannot exclude nevus cells or histiocytes), and 2 for definitive invasion. For cases with discordant scores rendered by the 3 dermatopathologists, the majority concordant score between 2 raters was chosen. If 3 different scores (0, 1, and 2) were rendered, the case was rereviewed collaboratively by all 3 dermatopathologists to achieve a consensus final score. The Breslow depth was measured by conventional methods in millimeters if invasion was present. Definitive invasion was defined as the presence of atypical dermal cells morphologically similar to the in situ melanoma cells, characterized by large size, ample amount of cytoplasm, and/or visible nucleoli. One melan-A and 1 H-E score (combining findings of both H-E slides) were rendered for each case.
Melan-A–stained slides showing definitive invasion were subsequently double-stained with Sry-related HMg-Box gene 10 (SOX10) to confirm melanocytic origin of the invasive cells. Briefly, the melan-A–stained slides were soaked in xylene to remove coverslips and rehydrated with distilled water. After buffer rinse, the slides were immersed in FLEX Retrieval solution, high pH, at 70°C for 10 minutes to remove unconjugated detection reagents from the previous melan-A staining. The slides were incubated for 2 hours at room temperature with polyclonal SOX10 (catalog No. 383A-75; Cell Marque) and subsequently detected with the Biocare MACH 4 alkaline phosphatase system (catalog No. M4U536; Biocare Medical) using Warp Red (catalog No. WR806; Biocare Medical) as the chromogen. The slides were counterstained with hematoxylin, dehydrated, and coverslipped.
Fisher exact tests or χ2 tests were used to evaluate the association of categorical variables (sex, location, melanoma subtype, appendageal involvement, and inflammation) with the presence of invasion. A Wilcoxon rank sum test was used to assess for age differences between groups with and without invasion. The Kendall coefficient (0, no agreement, to 1, complete agreement) was calculated to assess concordance among the 3 dermatopathologists. For all statistical tests, P < .05 was considered significant. All analyses were conducted using SAS, version 9.4 (SAS Institute Inc).
Thirty-four in situ melanomas were identified for analysis; 33 in a consecutive cohort plus 1 nonconsecutive index case. Clinical and histopathologic results are listed in the Table. The median age of the patients was 68 years (range, 24-95 years) with an equal distribution of 17 men and 17 women. Eighteen of 34 melanomas (53%) occurred on the head and neck, 3 (9%) on the trunk, 7 (21%) on the upper extremity, and 6 (18%) on the lower extremity. Twenty (59%) were lentigo maligna subtype, 13 (38%) were superficial spreading, and 1 (3%) was acral lentiginous. Appendageal involvement was present in 18 (53%) and inflammation was present in 15 melanomas (44%).
Quiz Ref IDOccult invasive melanoma was detected in 12 of 34 melanomas (35%) overall, with 11 of 33 tumors (33%) in the consecutive case cohort. Definitive invasion was evident in melan-A– and H-E–stained sections in 6 of 34 cases (18%; Breslow depth, 0.12-0.45 mm) (Figure 1). Definitive invasion was present in the melan-A–stained section and suspicious for invasion in H-E–stained sections in 3 of 34 melanomas (9%; Breslow depth, 0.21-0.49 mm) (Figure 2). Definitive invasion was present in the melan-A–stained section and absent in H-E–stained sections in 3 of 34 melanomas (9%; Breslow depth, 0.12-0.45 mm) (Figure 3). In all 12 cases with occult invasive disease, tumor burden was low with small aggregates or rare single cells present in the papillary dermis. Four cases (12%) showed a few small melan-A–positive cells in the dermis that were not identifiable in the flanking H-E sections; these sections were insufficient for a diagnosis of invasive melanoma and were most consistent on a cytomorphologic basis with occult nevi (Figure 4).
Concordance was high among the 3 dermatopathologists (P.W.H., D.R.F., and M.P.C.). The Kendall coefficient of concordance was 0.81 for the H-E score and 0.82 for the melan-A score. Most of the discrepant scores differed by no more than 1 point. Only 2 cases (6%) showed discordant H-E scores ranging from 0 to 2. These cases were rereviewed to reach consensus on the final scores. Of all variables evaluated (age, sex, location, subtype, appendage involvement, and inflammation), Quiz Ref IDonly inflammation showed marginal association with invasion.
Occult invasion was detected in 33% of our consecutive in situ melanoma case cohort using immunohistochemistry and deeper sectioning into the tissue block. We applied unique methods to circumvent potential limitations of prior reports, including the absence of preexisting nevus to minimize false-positive immunostaining and regression with the associated potential for regressed microinvasion. The immunostained slide was flanked by H-E slides to examine the yield of occult invasion detected by immunohistochemistry alone and not simply by cutting deeper into the block. Furthermore, we applied a double-staining method using a cytoplasmic melan-A and nuclear SOX10 overlay to confirm the melanocytic nature of the cells of interest.2-4 Finally, cases were interpreted independently by 3 dermatopathologists with expertise in melanoma and assessed for concordance.
Our results are consistent with those of prior reports. Undetected invasive disease with an original pathologic diagnosis of in situ melanoma was reported in a study5 of 104 patients in 2002 using an immunohistochemistry design. Specimens were stained in sequential sections with H-E and at least 2 sections were stained with melan-A, also termed MART-1 (melanoma antigen recognized by T cells 1). Thirty (29%) of 104 patients with in situ melanoma had an invasive component noted by melan-A staining of cells in the dermis. The Breslow depth of invasion ranged from 0.25 or less to 1.00 mm. One patient with a Breslow depth of 0.43 mm developed regional nodal recurrence and died of melanoma. Others have also reported occult invasive disease within lesions of in situ melanoma primarily based on immunohistochemistry alone. One study6 using a sequential section of the tissue block of in situ melanoma detected invasion in 9 of 27 cases (33%) using melan-A with Breslow depths from 0.15 to 0.35 mm. The melan-A–positive dermal cells also stained positive with S-100 in an additional section. Another report7 demonstrated invasive disease in lentigo maligna using S-100 staining of sequential sections in 14 of 91 cases (15%). Quiz Ref IDOur study added a nuclear SOX10 overlay stain to confirm the melanocytic nature of the melan-A–positive dermal cells to avoid the potential pitfall of nonspecific melan-A staining of melanosomal debris in histiocytes.8 In addition, by analyzing the melan-A slides and flanking H-E slides, we found that 6 of the 11 occult invasive melanomas (55%) were diagnosable only by immunohistochemistry. The remaining 5 tumors (45%) were diagnosable by both melan-A– and H-E–stained sections, likely as a result of having deeper sections into the tissue block when it was faced.
The interpretation of malignant vs benign cells using immunohistochemistry is fraught with the potential for false-positive interpretation. This hazard was highlighted in a study9 of 188 excisions of melanoma in situ, atypical junctional melanocytic hyperplasia, and other nonmelanocytic neoplasms stained with melan-A and SOX10. The 188 excisions evaluated included cases of melanoma in situ (90 [48%] cases), atypical junctional melanocytic hyperplasia (34 [18%]), invasive squamous cell carcinoma (18 [9%]), squamous cell carcinoma in situ (11 [6%]), and basal cell carcinoma (35 [19%]). Thirty-two of the 188 cases examined (17%) exhibited dermal cells that were positive for both melan-A and SOX10 by immunohistochemistry. Twelve of 112 cases (10%) from the melanocytic group and 20 of 64 cases (31%) from the nonmelanocytic group exhibited dermal cells that were positive for both melan-A and SOX10. These cells lacked cytologic atypia and were interpreted as benign and reactive dermal melanocytes. Simply interpreting positive staining as an indicator of malignancy would have resulted in misdiagnosis as invasive melanoma in the melanocytic subgroup. The correct interpretation depends on the cytomorphologic features, including cell size, amount of cytoplasm, degree of nuclear atypia, and comparison with the overlying in situ component. Essentially identical to those described above, in our 22 cases without definite invasion on H-E or melan-A, we identified rare dermal melan-A–positive cells in 4 cases (18%). These cells were significantly smaller and blander than the in situ melanoma cells, either small round (as in type B nevus cells) or spindle (as in type C nevus cells) in shape, and were interpreted as benign. In contrast, the cells in invasive melanoma were significantly larger and epithelioid, with a greater amount of cytoplasm, larger nuclei, and cytology similar to that of the overlying in situ component. In addition, to avoid misinterpretation of tangentially cut intraepidermal nests as invasion, at least some of these cells should be present below the level of most of the adjacent rete ridges and should not be associated with keratinocytes in any of the 3 planes of sections examined. The presence of multiple scattered single cells, as opposed to a solitary nest, further supports true invasion. The overall high level of concordance between our dermatopathologists (Kendall coefficient of concordance >0.80) supports the diagnostic accuracy of our results.
Serial sectioning deeper into the tissue block without immunohistochemistry may also result in a greater Breslow depth in melanoma. In a previous study,10 the difference in Breslow depth was compared between standard bread loaf serial sectioning at 3-mm intervals and complete exhaustive serial sectioning at 5-µm intervals through 2 tissue blocks immediately adjacent to and containing the deepest Breslow depth on the original diagnostic slides. Nineteen cases were studied with Breslow depths from 0.26 to 0.76 mm (mean, 0.52 mm). The number of sections from exhaustive serial sectioning at 5-µm intervals in 2 blocks ranged from 147 to 859 (mean, 451). All cases showed an increase in maximum tumor thickness when exhaustively serially sectioned, ranging from 0.03 to 0.44 mm (mean, 0.13 mm). Consistent with these findings, we detected occult invasive disease simply based on deeper sectioning into the block without the need for immunohistochemistry in 6 of 34 cases (18%). Only inflammation was marginally associated with invasion. Detecting microinvasive melanoma may be more difficult when tumor cells are obscured by inflammatory cells. Age, sex, location, and melanoma subtype had no association with microinvasion. Appendageal involvement also had no effect on the rate of microinvasion detection inferring that misinterpretation of tangentially cut intraepithelial tumor nests as invasive melanoma was unlikely to be a confounding factor.
Theoretically, stage 0 (TisN0M0) in situ melanoma should have a metastatic rate of 0%. However, reports11-13 of metastatic disease from presumed in situ melanoma exist, including our index case. Reclassification from stage 0 (TisN0M0) to stage IA (T1aN0M0) with Breslow depths ranging from 0.12 to 0.49 mm occurred in 11 of 33 cases (33%) in our consecutive cohort. Even ultrathin invasive melanoma has metastatic potential clearly noted in several series. A multi-institutional series of 43 patients with metastatic melanoma from thin primary tumors (Breslow depth range, 0-0.95 mm; mean, 0.59 mm) with 12 with a Breslow depth between 0.10 and 0.50 mm and 2 in situ melanomas was published.11 One of the 2 in situ melanomas had extensive regression; the other had neither regression nor other concerning features. Both patients developed regional lymph node and systemic metastases. Another small series12 reported melanoma-specific death of 4 patients with melanoma of less than a 0.50-mm Breslow depth. Survival data stratified for ultrathin tumors were published from the New South Wales central cancer registry and the Sydney Melanoma Unit.13 The New South Wales 10- and 15-year melanoma-specific survival for 817 patients with melanomas having less than a 0.20-mm Breslow depth were 97.7% and 95.4%; for 5310 patients with 0.21- to 0.40-mm Breslow depth, 97.9% and 97.5%; and for 5693 patients with 0.41- to 0.60-mm Breslow depth, 97.3% and 95.5%. The Sydney Melanoma Unit 10-year melanoma-specific survival for melanomas with less than a 0.75-mm Breslow depth was 96.9%.
Based on our data, which are supported by cumulative data demonstrating occult microinvasive disease in up to one-third of in situ melanoma lesions, surveillance recommendations should be based on a very low, but not zero, risk of metastasis. Dermal tumor burden identified within in situ melanoma with occult invasion is predictably less than that of nonoccult invasive melanoma of the same depth. Quiz Ref IDWhat is important is that, although the risk of occult invasion is quantitatively less than the the risk of its nonoccult invasive counterpart, it is not zero. Late recurrence after 10 years may be as common as before 10 years for ultrathin melanoma; therefore, surveillance should be lifelong.13 This surveillance includes routine examination of the entire skin surface and mucous membranes for detection of additional primary melanomas as well as skin and subcutaneous metastases. Examination of the lymph node system with attention to the regional draining nodal basin is indicated. Patients should be educated on the importance of monthly self-skin and regional lymph node examination based on an improved outcome for early vs late clinically detected primary skin and regional lymph node disease.14 Performing a melanoma-focused review of systems with symptom-directed workup during surveillance visits should also be considered.
The implications for modification of current surgical margin recommendations for in situ melanoma are uncertain but probably unwarranted. To our knowledge, no randomized clinical trials have been conducted to determine optimal margins for in situ melanoma. In addition, the lentigo maligna subtype occurring in chronically sun-damaged skin is known to be associated with greater subclinical extension. Current practice guidelines, such as those of the National Comprehensive Cancer Network,15 provide relatively uniform recommendations of 0.5- to 1.0-cm clinical margins for in situ melanoma. Numerous studies16-19 involving large numbers of patients with adequate follow-up confirm that conventional excision with 0.5-cm–measured clinical margins for in situ melanoma, nonlentigo maligna subtype, is associated with a high rate of negative histologic margins and a low rate of local recurrence. Our data and those of similar studies should not be used to justify or prompt dermatopathologists to routinely perform immunostaining or exhaustive serial sectioning in cases of in situ melanoma. The magnitude of all melanoma prognosis data is based on routine serial section sampling Breslow depth rather than exhaustive section absolute Breslow depth.20Quiz Ref ID The use of immunohistochemistry or extensive serial sectioning in every in situ melanoma case is impractical, costly without associated high benefit, and unrealistic in time, efficiency, and scope for pathologists and laboratory processing. Rather, the patient with in situ melanoma should be monitored and educated based on a very low risk of metastasis, which, if detected early in the regional lymph nodes, is associated with improved outcomes.
Melanoma in situ may rarely metastasize. History and physical examination, including regional lymph nodes, education, and surveillance recommendations should be based on a very low, but not zero, risk of melanoma metastasis.
Corresponding Author: Timothy M. Johnson, MD, Department of Dermatology, University of Michigan Medical School and Comprehensive Cancer Center, 1500 E Medical Center Dr, 1910 A. Alfred Taubman Center, SPC 5314, Ann Arbor, MI 48109 (email@example.com).
Accepted for Publication: June 14, 2016.
Published Online: August 10, 2016. doi:10.1001/jamadermatol.2016.2668
Author Contributions: Drs Bax and Johnson had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Bax, Johnson, Schwartz, Fullen, Chan.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Bax, Johnson, Zhao, Chan.
Critical revision of the manuscript for important intellectual content: Bax, Johnson, Harms, Schwartz, Fullen, Chan.
Statistical analysis: Zhao.
Administrative, technical, or material support: Harms.
Study supervision: Johnson, Chan.
Conflict of Interest Disclosures: None reported.
Funding/Support: Funds from the Steven and Jacklynn Tracey Endowment were used for costs related to additional tissue processing and stains.
Role of the Funder/Sponsor: The funding organization 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.
Create a personal account or sign in to: