The estimated local recurrence rates were 1.4% at 5 years, 1.8% at 7.5 years, and 2.2% at 10 years. Solid line indicates the cumulative incidence; dashed lines, 95% CI.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Moyer JS, Rudy S, Boonstra PS, et al. Efficacy of Staged Excision With Permanent Section Margin Control for Cutaneous Head and Neck Melanoma. JAMA Dermatol. 2017;153(3):282–288. doi:10.1001/jamadermatol.2016.4603
Does staged excision with comprehensive hematoxylin-eosin–stained permanent section margin control in melanoma arising in chronically photodamaged skin of the head and neck have low recurrence rates?
In a cohort study of 806 patients with a median follow-up of 9.3 years, the estimated local recurrence rates of melanoma were 1.4% at 5 years, 1.8% at 7.5 years, and 2.2% at 10 years. For each 50-mm2 increase in the size of the clinical lesion, there was a 9% increase in the rate of local recurrence.
Staged excision with comprehensive permanent section margin control of melanomas arising in chronically sun-damaged skin on the head and neck has favorable recurrence rates when melanoma margins are difficult to assess, and recurrence rates are higher with traditional techniques.
Melanoma arising in chronically photodamaged skin, especially on the head and neck, is often characterized by poorly defined clinical margins and unpredictable occult extension. Staged excision techniques have been described to treat these challenging melanomas.
To investigate the local recurrence rates and margin to clearance end points using staged excision with comprehensive hematoxylin-eosin–stained permanent section margin control.
Design, Setting, and Participants
In this observational cohort study performed from October 8, 1997, to December 31, 2006, with a median follow-up of 9.3 years, 806 patients with melanoma on the head and neck, where clinical occult extension is common, were studied at an academic medical center.
Staged excision with comprehensive hematoxylin-eosin–stained permanent section margin control commonly known as the square technique.
Main Outcomes and Measures
Local recurrence rates and margin to clearance end points.
A total of 806 patients (276 women [34.2%]; 805 white [99.9%]) with a median age at the time of first staged excision procedure of 65 years (range, 20-94 years) participated in the study. The estimated local recurrence rates were 1.4% at 5 years, 1.8% at 7.5 years, and 2.2% at 10 years. For each 50-mm2 increase in the size of the clinical lesion, there was a 9% increase in the rate of local recurrence (hazard ratio, 1.09; 95% CI, 1.02-1.15; P = .02). The mean (SD) margin from lesion to clearance for melanoma in situ was 9.3 (5.1) mm compared with 13.7 (5.9) mm for invasive melanoma. For melanoma in situ, margins were clear after 5 mm or less in 232 excisions (41.1%) and after 10 mm or less in 420 excisions (74.5%). For invasive melanoma, margins were clear after 5 mm or less in 8 excisions (3.0%) and after 10 mm or less in 141 excisions (52.2%).
Conclusions and Relevance
Staged excision with comprehensive permanent section margin control of melanomas arising in chronically sun-damaged skin on the head and neck has favorable recurrence rates when melanoma margins are difficult to assess, and recurrence rates are high with traditional techniques.
Most melanomas on the trunk and extremities can be excised with standard margins and tissue processing techniques with high rates of local control. Melanoma arising in chronically photodamaged skin of the head and neck, however, is often characterized by poorly defined clinical margins and unpredictable occult extension, possibly because of higher melanocytic density compared with non–head and neck locations. Excision with standard margins is likely inadequate in this subset of patients and is associated with a higher risk of local recurrence.1-6 Staged excision techniques using paraffin-embedded permanent section tissue processing with more comprehensive margin assessment may achieve the highest local control rates with improved tissue preservation in critical areas.1-3,5,7,8 Several variations in staged excisions, with subtle differences in techniques, have been described that all aim to optimize comprehensive permanent section margin control.1,5,6,8-22 However, relatively small numbers and short follow-up significantly limit the assessment of these techniques. This study investigates the local recurrence rates and margin to clearance end points using staged excision with comprehensive hematoxylin-eosin–stained permanent section margin control in a large population of patients with melanoma on the head and neck, where clinical occult extension is common.
The University of Michigan Melanoma Database, Ann Arbor, was queried for all patients with a melanoma diagnosis treated with a staged excision technique with formalin-fixed permanent section margin control from October 8, 1997, to December 31, 2006. The cutoff date was selected at the inception of study design in 2011 to ensure a minimum 5-year follow-up time. All histopathologic material was interpreted by dermatopathologists participating in the University of Michigan Multidisciplinary Melanoma Program. The electronic medical records (EMRs), including photographs, notes, and pathology reports, were reviewed. After data extraction from the EMR, a maximum of 3 telephone calls ending May 2014 were made to patients and/or their physician(s) with specific questions about long-term scar satisfaction and an update of pertinent research record data. A death index report was queried when status was unknown. This study was approved by the institutional review board of the University of Michigan Medical School, and no informed consent was required.
The following elements were recorded: date of diagnosis, date of last follow-up, anatomical location, age at diagnosis, race/ethnicity, sex, and immunosuppression (long-term immunosuppression therapy or chronic lymphocytic leukemia). Lesions were classified as melanoma in situ or invasive melanoma and as primary (never treated), recurrent, or incompletely excised (positive margin after definitive excision intent). Clinical lesion size in 2 dimensions, number of staged excisions performed, any change in invasion depth after staged excision, and total greatest margin to clearance were recorded. For incompletely excised lesions, the original surgical margin was added to the staged excision margin(s) for total greatest margin distance to clearance. Lesions were grouped into 5 histologic patterns or subtypes as follows: desmoplastic, lentigo maligna, superficial spreading, other (spindle cell, nodular, Spitzoid, and unclassified), and no pattern reported. A total of 111 of 115 patients (96.5%) with no pattern reported were the result of dermatopathologist clinical practice during part of the study period to not report a subtype specifically for any in situ melanoma, whereas 4 (3.5%) had no subtype because the specimen was fragmented (n = 1) or only focally microinvasive with tumor burden insufficient for definitive subtype classification (n = 3).
The date of any local recurrence was recorded. Local recurrence status from the EMR required examination by a dermatologist or a physician with melanoma knowledge, with specific note documentation of a history of melanoma within the visit or specific examination of the primary melanoma site. In addition, for patients not routinely followed up at our institution or for whom recurrence data were not already known, a maximum of 3 telephone calls were made with specific questions about follow-up with a local dermatologist or primary care physician for melanoma, any additional treatment or recurrence at the site, or any pigment in the area. For persistent cold cases, one call to the referring dermatologist or primary care physician was made. The last follow-up date was based on the last clinic visit in dermatology or primary care for melanoma evaluation noted in the EMR or confirmed with the follow-up physician. Local recurrence was defined as a new clinical lesion developing within 5 mm of the surgical scar, with the biopsy revealing invasive melanoma, melanoma in situ, or atypical junctional melanocytic hyperplasia consistent with early or evolving melanoma in situ. Equivocal clinical or histopathologic findings were recorded as local recurrence. Dates of death were recorded. Satisfaction with reconstruction was assessed by telephone or follow-up visit and was recorded as very satisfied, somewhat satisfied, somewhat dissatisfied, or very dissatisfied.
Incisional or excisional biopsy was performed in all patients to obtain a tissue diagnosis of melanoma and the Breslow depth for invasive lesions. If a patient had a residual lesion suggestive of invasive melanoma after a partial biopsy, the clinically apparent residual lesion was reexcised with a 1-mm margin for histopathologic microstaging to better determine the optimal surgical margin of the initial staged excision procedure. Candidates for sentinel lymph node biopsy (Breslow depth ≥1.00 mm or 0.75-0.99 mm with other adverse features) underwent standard wide local excision with sentinel lymph node biopsy. If peripheral margins were positive for melanoma (incomplete excision), the patient was considered for a subsequent staged excision.
Patients underwent staged excision with local anesthesia in a treatment room setting with methods previously described (the square procedure).23-25 Surgical margins of 0.5 cm for melanoma in situ and 1.0 cm for invasive melanoma were typically used for the first stage. Tissue was processed for formalin-fixed permanent section comprehensive margin control, theoretically containing 100% of the peripheral margin. Immunohistochemistry was not used. After dermatopathologic interpretation, any areas of positivity were excised as a subsequent stage, most often with an additional 0.5-cm margin around the areas of positivity. Margins less than 0.5 cm were used in cosmetically sensitive areas that needed optimal tissue sparing.
Peripheral margins were assessed histopathologically and defined as positive if melanoma in situ or trailing edge of melanoma in situ (atypical junctional melanocytic hyperplasia) was present. Melanoma in situ was defined as confluent or near-confluent proliferation of cytologically atypical nested and single-unit melanocytes at the dermoepidermal junction with variable dyscohesion, pagetoid scatter, and/or adnexal epithelial extension. Atypical junctional melanocytic hyperplasia (trailing edge of melanoma in situ) was defined as an increase in atypical single-unit melanocytes, often with nuclear hyperchromasia and nucleoli, beyond what would be expected for actinically damaged skin.
In total, 834 staged excision procedures were performed on 806 patients. End points recorded included local recurrence, death without local recurrence, or unavailable for follow-up while recurrence free. The follow-up period was defined as the time after staged excision until the date of local recurrence, the date of death, or the date of last follow-up, whichever occurred first. All patients known to be alive and recurrence free as of December 10, 2015 (the date of the final record review), were administratively censored. We used the likelihood ratio to assess statistical significance and quantify uncertainty. All analyses were performed with R statistical software.26
The primary aim was to quantify the overall local recurrence rate. Follow-up time was quantified via reverse Kaplan-Meier method.27 The cumulative incidence of local recurrence, treating death as a competing outcome, was calculated.28 To quantify associations between local recurrence and patient and lesion characteristics, multivariable proportional hazards regression was used29; patients who died while free of local recurrence were censored at their date of death. Backward selection on a prospectively identified set of patient and lesion variables was used to identify a subset of important associations. Beginning with a multivariable model of 10 variables listed in Table 1 (race/ethnicity, Breslow depth, and reconstruction service were a priori excluded), variables were removed sequentially in the order of smallest decrease of the partial log-likelihood and continuing until the Bayesian information criterion stopped decreasing, that is, when removing the next variable would decrease the partial log-likelihood by more than d × loge(17)/2, where d is the number of free parameters, or degrees of freedom, used by a variable and 17 is the number of local recurrences observed.30 Quantitative variables were assumed to multiplicatively scale the baseline hazard in a continuous fashion, and categorical variables were used as multivariate dummy variables, with 1 level serving as a reference category. Statistical significance was assessed via the likelihood ratio. To investigate possible sensitivity to correlation between outcomes attributable to more than 1 distinct site treated with staged excision in some patients, the analysis was repeated after including a patient-level frailty in the proportional hazards model,31 fixing the variance component such that approximately 1 df was used by the frailty term. Standard diagnostic techniques were used to assess the quality of our model fit and identify potentially influential data points.
A secondary aim was to determine whether the size of the margin to clearance was associated with any of the patient and lesion variables. We fit a linear model using the logarithm of the margins to clearance as the outcome. Analogous to the primary aim, backward selection procedure using Bayesian information criterion was implemented. Except for the number of staged excision procedures performed, all the same variables in our primary aim were considered for inclusion in this model, with lesion size transformed to the logarithmic scale, based on the observed association between size of margins to clearance and lesion size.
A total of 806 patients (276 women [34.2%]; 805 white [99.9%]) with a median age at the time of first staged excision procedure of 65 years (range, 20-94 years) participated in the study (Table 1). Based on the reverse Kaplan-Meier method, the median potential follow-up time was 9.3 years, and the observed median time of last follow-up was 8.4 years. Of 834 unique staged excision procedures, a total of 17 local recurrences were observed. The estimated local recurrence rates were 1.4% at 5 years, 1.8% at 7.5 years, and 2.2% at 10 years (Figure 1).
Variables affecting the overall local recurrence rate were analyzed using multivariable proportional hazards regression (Table 2). Three variables were ultimately retained in the model: lesion size, cutaneous tumor site (4 sites compared with a fifth reference site), and immunosuppression. For each 50-mm2 increase in the size of the primary lesion, there was a 9% increase in rate of local recurrence (hazard ratio, 1.09; 95% CI, 1.02-1.15; P = .02). In addition, when the cutaneous tumor site was periocular, the rate for local recurrence was 12.5 times that of lesions excised from the scalp, cheek, forehead, or neck (hazard ratio, 12.48; 95% CI, 4.08-39.54; P < .001). For the remaining sites, the estimated hazard ratios were 14.20 for lip (95% CI, 0.74-89.05; P = .07), 0.91 for nose (95% CI, 0.05-5.64; P = .93), and 2.76 for ear (95% CI, 0.39-12.97; P = .27) compared with the scalp, cheek, forehead, or neck. No immunosuppressed patients developed recurrences (hazard ratio, 0; 95% CI, 0-0.79; P = .04). Accounting for within-patient correlation by including patient-level frailty did not change statistical significance.
We also examined the association between the lesion size and the distance from the lesion needed to obtain pathologically tumor-free margins (Table 3). The following clinical and pathologic characteristics were positively associated with larger margins to tumor clearance: the size of the lesion, a diagnosis of invasive melanoma vs melanoma in situ, and excising a previously incompletely excised lesion. Greater margins were needed for tumor clearance for invasive melanoma compared with melanoma in situ (Table 1). The mean margin to clearance for melanoma in situ was 9.3 (5.1) mm compared with 13.7 (5.9) mm for invasive melanoma. Furthermore, for melanoma in situ, margins were clear after 5 mm or less in 232 excisions (41.1%) and after 10 mm or less in 420 excisions (74.5%). For invasive melanoma, margins were clear after 5 mm or less in 8 excisions (3.0%) and after 10 mm or less in 141 excisions (52.2%). Figure 2 depicts the predicted margins required to excise a melanoma in situ or invasive melanoma based on primary lesion size.
We also prospectively investigated the subjective reconstructive outcomes in the patient population using the following Likert scale: very satisfied, somewhat satisfied, somewhat dissatisfied, and very dissatisfied. Of the 806 patients who were enrolled in the study, 162 (20.1%) were deceased, and these data could not be collected. Of the remaining 644 patients, 320 patients (49.7%) were able to complete the outcome assessment survey. Among this selected group of patients for whom data were collected, satisfaction was high, with 301 (94.1%) of the 320 contacted patients reporting being very (271 [84.7%]) or somewhat (30 [9.4%]) satisfied with their reconstruction results. No association was found between subjective reconstructive outcomes and the size of the area reconstructed.
Melanoma occurring in chronically sun-damaged skin on the head and neck may be associated with asymmetric, deceptively extensive occult extension, resulting in higher local recurrence rates with the current recommended margins and tissue processing techniques.3,4,7,17,32 As a result, numerous staged excision techniques with the commonality of incorporating comprehensive permanent section peripheral margin assessment without the need for immunohistochemistry have evolved. To our knowledge, we report the largest series to date, involving 834 lesions in 806 patients with a median follow-up of 9.3 years. The local recurrence rates with staged excision and permanent margin control were at 1.4% at 5 years, 1.8% at 7.5 years, and 2.2% at 10 years.
Publications5,6,9-16,18-22,33 on comparable staged excision techniques report local recurrence rates ranging from 0% to 10%. Collectively, these studies are limited by small numbers and short follow-up but suggest a higher local recurrence rate with longer follow-up. In our cohort, approximately 36% of recurrences developed after 5 years, clearly indicating that a longer time horizon is necessary before the true absence of a local recurrence can be assessed.
Although there is no standard definition for local recurrence, we included a new lesion within 5 mm of the original scar and histologically equivocal cases. Factors strongly associated with local recurrence included increasing lesion size and periocular location, with a weak association with lip location. Immunosuppression appeared protective in the regression model, which is interesting and deserves further study. No statistically significant association was found with recurrence status, histologic pattern, age, or sex. Although 55 of 62 periocular lesions (88.7%) did not recur, 7 of the 17 local recurrences (41.2%) were in the periocular region. Our current technique now incorporates cryotherapy of the conjunctival margin to improve disease control in this high-risk region.25
Our data contribute valuable information to help the clinician provide counseling and guidance with respect to anticipated margin to clearance in patients with melanoma in situ or invasive melanoma on the head and neck (Figure 2). Our results also explain, in part, the higher local recurrence rate and the challenge in obtaining clear margins using standard techniques by demonstrating that the mean margin to histologic clearance was 9.3 mm for melanoma in situ and 13.7 mm for invasive melanoma. This finding contrasts the standard margin recommendation in current practice guidelines of 5 to 10 mm for melanoma in situ and 10 mm for invasive melanoma with a Breslow depth of 1.0 mm or less.2 For melanoma in situ, margins were clear after 5 mm or less in 232 excisions (41.1%) and after 10 mm or less in 420 excisions (74.5%). For invasive melanoma, margins were clear after 5 mm or less in 8 excisions (3.0%) and after 10 mm or less in 141 excisions (52.2%). Factors associated with a greater margin to clearance included increasing lesion size, invasive vs in situ disease, and incompletely excised vs primary or recurrent lesions. Incompletely excised lesions undergoing a staged excision represent a bias subset. These lesions had more extensive occult junctional melanocytic disease on standard excision. Subsequent treatment in this group with staged excision resulted in low recurrence rates similar to all other lesions.
Clinically apparent residual lesions suggestive of invasive disease were excised before initiation of our staged excision. Remarkably, 32 of 834 lesions (3.8%) still had unsuspected, deeper residual disease. Sixteen of these 32 lesions were originally diagnosed as melanoma in situ and had occult invasive disease, with Breslow depths ranging from 0.2 to 1.12 mm (mean, 0.48 mm) on the staged excision. The other 16 lesions, those with invasive melanoma at initial diagnosis, had deeper invasive disease from a preprocedure mean Breslow depth of 0.47 mm (range, 0.2-0.65 mm) to postprocedure mean Breslow depth of 0.85 mm (range, 0.32-2.47 mm). Although this increase in Breslow depth was not selected for inclusion in our final model, 4 of these 16 patients were upstaged by American Joint Committee on Cancer criteria after staged excision. Rates of invasive disease noted on staged excision specimens range widely from approximately 5% to 57% in prior reports.1,8,10-12,14,15,18,20 These data, combined with the 3.8% occult rate in the current study, reinforce the fact that unsuspected invasive disease may be present in this patient population that could affect prognosis and may be relevant in treatment planning.
Despite relatively large facial defects, patient satisfaction with the final reconstructive result was high, with 301 (94.1%) of the 320 contacted patients reporting being very satisfied (271 [84.7%]) or somewhat satisfied (30 [9.4%]). No association was found between satisfaction and the area of skin excised. This observation is noteworthy because there has been some concern that a more aggressive local excision approach to melanoma on the head and neck might result in unacceptably high morbidity. This concern did not appear to be valid in this study population.
The study is limited by the retrospective nature of the data collection and the necessary reliance on data extraction from EMR reviews. In addition, the study is from a single center, which may not be representative of the results that might be obtained in other centers performing this technique.
The 2016 National Comprehensive Cancer Network melanoma guidelines note that techniques for more exhaustive histologic assessment of margins should be considered for the lentigo maligna pattern, which is found most commonly on the head and neck. Our large data set with long-term follow-up provides evidence of efficacy using a staged excision method with comprehensive formalin-fixed permanent section total peripheral margin control. The low local recurrence rates and patient satisfaction obtained with this technique are the result of the collaborative efforts of a multidisciplinary approach with experts in excision technique, histopathologic analysis, and reconstruction of each lesion. Our results should be interpreted as one method to achieve high rates of local control and may serve as a reference point for future practice guidelines.
Corresponding Author: Jeffrey S. Moyer, MD, Department of Otolaryngology–Head and Neck Surgery, University of Michigan Medical School, 1500 E Medical Center Dr, TC 1904, Ann Arbor, MI 48199 (firstname.lastname@example.org).
Accepted for Publication: August 1, 2016.
Published Online: December 21, 2016. doi:10.1001/jamadermatol.2016.4603
Author Contributions: Drs Moyer 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.
Study concept and design: Moyer, Boonstra, Chinn, Schwartz, Johnson.
Acquisition, analysis, or interpretation of data: Moyer, Rudy, Boonstra, Kraft, Chinn, Baker, Bichakjian, Fullen, Durham, Lowe, Johnson.
Drafting of the manuscript: Moyer, Rudy, Boonstra, Kraft, Chinn, Johnson.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Rudy, Boonstra, Kraft, Chinn.
Administrative, technical, or material support: Moyer, Kraft, Chinn, Lowe, Johnson.
Conflict of Interest Disclosures: None reported.