Introduction:
Extramammary Paget disease (EMPD) is a rare cutaneous adenocarcinoma originating from apocrine-rich areas with potential for locoregional invasion and distant metastasis.1,2 EMPD poses a management challenge due to difficulty in margin assessment secondary to subclinical extension and frequent multifocal growth pattern. This translates into high recurrence rates after treatment.3–7 Blind scouting biopsies to assess subclinical extension provide only focal information and can have false-negative results.8,9Moreover, margins needed to clear tumor have not been well-defined. Some studies shown a 1cm surgical margins might suffice to clear EMPD;5,9 however, studies using Mohs micrographic surgery (MMS) have shown that up to 5-cm margins are needed to clear 97% of cases.10,11
Recently, non-invasive imaging modalities including reflectance confocal microscopy (RCM) have been used to evaluate EMPD.12 Preliminary RCM data from small cohorts have shown its utility in the diagnosis and management of EMPD.13–18 Most RCM studies, however, included Asian cohorts17 with few studies evaluating RCM role in Caucasians.13,14 We use RCM as an adjunctive presurgical tool to define subclinical extension, delineate surgical margins, assess response to non-surgical treatments, and monitor for recurrences. In this study, we describe an EMPD cohort to elucidate: surgical margins needed to clear tumors using staged excision, utilization of non-surgical treatment options, and to evaluate the bedside use of RCM as a complementary diagnostic aid.
Patients and methods:
This was an IRB approved, prospective study conducted at a tertiary cancer center between July 2017 and July 2019 on consecutive patients with diagnosis of EMPD.
Inclusion and exclusion criteria:
Adult patients (≥18 years old), who presented to dermatology, with biopsy-proven diagnosis of primary or recurrent EMPD were included. We excluded patients in which a diagnosis of EMPD was not confirmed.
Clinical, tumor, and systemic workup characteristics:
Patient age, gender, pre-surgical size (cm), and location were recorded. All biopsy slides were reviewed at our institution by dermatopathology. Diagnosis was confirmed based on standard stains (hematoxylin and eosin), periodic-acid Schiff stain (PAS), cytokeratin 7 (CK7), and additional immunostains as deemed necessary by the dermatopathologist (e.g. CK20, CDX2, ER, PR). Tumor characteristics, depth, and immunohistochemistry panel were recorded. All patients underwent regionally specific work-up to rule out internal malignancy (and diagnosis of secondary EMPD) including: colonoscopy, gynecologic evaluation, ultrasound or computed tomography scan of chest/abdomen/pelvis, cystoscopy of bladder, prostate exam, and breast imaging. The case was labelled as primary EMPD if systemic workup was negative, and secondary if an underlying malignancy was found.
Treatment modalities:
Definitive treatment included surgical and non-surgical approaches, Surgical techniques included staged excision with radial section margin assessment (SE) (performed by dermatologic surgery), or wide local excision (performed by urology or gynecology). SE with radial margin assessment was performed using formalin-fixed, paraffin-embedded tissue, as described by Hazan et al.19 The initial margins were 1.0 cm. SE was performed in consecutive days with rushed permanent histology and delayed reconstruction until margins were clear. If margins were positive on histopathology, additional margins were excised until clear. Final defect size and repair type were recorded.
In selected cases after discussing the risks and benefits of treatment, anatomical extent of disease (rectal/anal extension), inoperability, and patient’s preferences, non-surgical treatment modalities were used. These included imiquimod-5% (5 times a week for 12 weeks; 60 applications), red light aminolevulinic-acid photodynamic therapy (PDT), or radiation therapy (RT).
Reflectance confocal microscopy study:
Patients were assessed in the initial visit clinically and with Wood’s lamp. Then, handheld RCM (Vivascope 3000, Caliber ID, Rochester, NY) was used to evaluate areas suspicious for invasion or subclinical extension outside of the clinical/Wood’s lamp margin. We used 6-mm paper rings to guide RCM imaging of areas of clinical interest selected by the dermatologic surgeon (Figure 1A, inset).20 Single images, stacks, and videos were obtained at different skin levels from stratum corneum to superficial dermis (depth 200μm).
We evaluated previously defined RCM EMPD features (Table 1).13,14,17 Imaged sites were labeled ‘positive’ or ‘negative’ for EMPD under RCM if any ‘EMPD-specific features’ were seen (dark holes and glandular nests). We also evaluated for ‘dermoepidermal junction (DEJ) disarrangement’ as a feature related to possible invasion. RCM evaluations were recorded by consensus, in real-time, at bedside by 3 experienced confocalists (C.N-D, S.A., and M.C.), blinded to the final histopathologic results. Biopsies (4 mm punch biopsies) were performed in RCM-examined areas; RCM findings were correlated with the histopathology results on those areas (i.e. 1-to-1 correlation). Some patients had more than 1 biopsy, depending on surgeon’s clinical concern or RCM suspicion for invasion on a case-by-case basis. All biopsies were correlated with histopathology.
Statistical analysis:
Descriptive statistics including means, range, standard deviation (SD) and relative frequencies were used to describe study participants, tumor and histopathologic characteristics, treatments, and the RCM features. Correlation was calculated with kappa Cohen test. Fisher’s square test was used for categorical variables. Independent-samples student’s t test was performed for comparisons of continuous variables. Alpha-level was 5% for all evaluations, and all tests were two-sided. Analyses were performed using SPSS 23.0 (SPSS, Armonk, NY, USA).
Results:
Thirty-four patients were evaluated in the study period. One patient was excluded due to inguinoscrotal disease suspicious for EMPD, but was ultimately diagnosed with metastatic colon cancer to the skin. Final analysis included 36 EMPD lesions in 33 patients. Mean age at diagnosis was 71.7 years (SD 8.7; range 54 – 96 years); 23 males (69.7%). Mean duration of disease prior to diagnosis (available for 16 patients) was 31.6 months (SD 23.9; range 6 – 96 months). Twenty-eight patients (84.8%) identified as Caucasian, 4 (12.1%) as Asian, and 1 (3.0%) as Hispanic/Latino.
History of previous treatment:
Seventeen out of 36 lesions (47.2%) had prior treatment and were categorized as ‘recurrent / persistent’ EMPD on our initial evaluation. Prior treatments included: WLE (n=4; 23.5%), vulvectomy (n=4; 23.5%), imiquimod (n=3; 17.6%), WLE twice (n=2; 11.8%), RT (n=1; 5.9%), ED&C (n=1; 5.9%), imiquimod and PDT (n=1; 5.9%), and WLE and RT (n=1; 5.9%).
Tumor and histopathologic characteristics:
Mean lesional size was 4.3 cm (SD 2.9; range 0.5 – 10.0 cm). Majority were located on the inguino-genital area (n=27 patients; 75.0%), followed by pubis (n=4; 11.1%), perianal (n=4; 11.1%), and axilla (n=1; 2.7%). Two patients (6.1%) had bilateral synchronous, discontiguous involvement. One patient had metachronous disease (inguino-genital, then pubis). Three out of 33 patients (9.1%) were labelled as secondary EMPD (2 colorectal adenocarcinomas, 1 lung carcinoma).
In all, 28 out of 36 (77.8%) EMPD lesions had no invasion on histopathology. Invasive cases (n= 8; 22.2%) showed a mean depth of 1.1 mm (SD 1.06; range 0.2 – 3.0 mm). Among invasive cases, there was a linear correlation between clinical size and invasion (R2= 0.95). Immunostains revealed CK7 positivity in 32 out of 33 lesions tested (96.9%); CK20 positivity in 5 out of 17 cases tested (29.4%); and CDX2 in 4 out of 8 cases tested (50%) (Table 1). There was no statistical correlation between depth and CK7, CK20 or CDX2 positivity (p=1.0, 1.0, and 1.0, respectively). CK20 and CDX2 were positive in 100% of perianal cases tested (n=4; p=0.02 and 0.018, respectively). CK20 and CDX2 positivity were not associated with primary vs secondary disease (26.7% vs 50.0%; p=0.5 and 42.8% vs 50%; p=1.0, respectively).
Treatment modalities:
Twenty-one EMPD lesions were treated surgically (58.3%); of these, 5 (23.8%) had prior treatment before presentation. Fifteen EMPD lesions (41.6% total; 71.4% of surgically-treated patients) underwent SE in dermatology under local anesthesia. Of these 15 SE cases, mean presurgical size was 5.0 cm (SD 3.5; range 0.5 – 10 cm). Mean number of surgical stages needed to clear margins was 1.9 (SD 0.9; 1.0 – 3.0 stages). Mean margin needed to clear tumor was 1.8 cm (SD 1.1; range 1.0 – 5.0 cm). Fourteen out of 15 (93.3%) cases were cleared with margins of 3.0 cm or less. There was no correlation between clinical size and margins needed to clear (R2=0.04). Mean final defect size was 11.1 cm (SD 4.7; range 1.5 – 17.0 cm). Nine lesions (60%) were repaired with a flap, 1 lesion with a combined flap and graft (6.6%), and 5 lesions (33.3%) with primary closure. After a mean follow-up time of 18.3 months (SD 13.5 range 3 – 47 months) no patients have recurred clinically (Table 2, Figure 1–2). Six EMPD lesions (16.6% of total; 28.5% of surgically treated) were treated in the operating room under general anesthesia by urology (n=2) or gynecology (n=4). Of these, 1 male patient (50%) had positive margins and 2 female patients (50%) treated with vulvectomy had positive margins.
Twelve EMPD lesions were treated with non-surgical modalities; of these, 6 (50%) had prior treatment before presentation to our group. We employed RT (n=5; 41.6%), imiquimod (n=5; 41.6%), and ALA-PDT (n=1; 8.3%); 25% recurred (1 imiquimod, 2 RT) after a mean follow-up of 25.3 months (SD 18.8; range 9 - 70 months). One additional patient (8.3%) that was initially treated with ED&C (outside institution) and imiquimod had invasive disease metastatic to lymph nodes and bone. The patient was subsequently treated with RT and systemic chemotherapy, but passed away from metastatic disease (Figure S1 - Supplemental Digital Content).
Three patients with recurrent disease declined any additional treatment following our initial evaluation (Figure S2 - Supplemental Digital Content).
Reflectance confocal microscopy characteristics and histopathological correlation:
RCM was performed on 33 EMPD lesions. Forty-one sites were selected for biopsy by the dermatologic surgeon. Biopsies were positive for EMPD in 31 (75.6%) of the 41 biopsy sites. RCM was positive in 30 out of the 41 (73.1%) biopsy sites. The kappa Cohen correlation between RCM and punch biopsies was 0.93 (p<0.001). Main RCM features in patients with active EMPD sites were ‘dark holes in the epidermis’ (90.3%) and ‘glandular nests’ (67.7%) (Table 1 and Figures 1–2, Figures S1–S2 – SDC). The presence of ‘disruption of the DEJ’ (83.3% vs 25.9%) was more commonly associated with invasive EMPD cases on histopathology (p=0.01). Yet, 16.7% EMPD biopsies were invasive and lacked disruption of DEJ on RCM (1 out of 6 invasive cases evaluated with RCM). The visualization of ‘glandular nests’ did not show significant correlation with invasion (83.3% invasive vs 59.2% non-invasive; p=0.4). A normal ring pattern was more common in biopsy-proven negative EMPD cases (100% vs 0%; p<0.001).
We performed histopathological correlation with RCM findings. ‘Dark holes in the epidermis’ corresponded to single-to-few-cell aggregates of pagetoid cells with abundant cytoplasm in the epidermis. ‘Glandular nests’ corresponded to large aggregates of mucin-rich cells at the basal layer or DEJ.
Discussion:
In this prospective study of 36 EMPD sites in 33 patients, almost 50% of patients presented in the initial evaluation with persistent or recurrent disease, highlighting the complexity of EMPD. In all, 41.6% of cases were treated with SE by dermatology, 16.6% with WLE by urology or gynecology, and 33.3% of cases were treated non-surgically; interestingly, 8.3% of patients declined treatment. Of those treated with SE, clear margins were achieved in all patients, with an average margin of 1.8 cm and mean number of 1.9 stages. The final defect size after SE was 11.1 cm, which is relevant since most EMPDs arise in highly functional and sensitive areas such as the genitalia, perineum, or perianal area. Prior studies have shown that margins of 5 cm are needed to clear 97% of EMPDs.10 In our series, 92.8% of EMPD cases were cleared with margins of 3 cm or less. We also confirm by using comprehensive margin evaluation that 1 cm margins seems insufficient to clear most EMPDs.5,9
Techniques that provide circumferential margin assessment, such as MMS and SE, have been associated with a recurrence rate of 12.2% and 5-year tumor-free rate of 83.6%.21 This is in contrast to traditional surgical techniques which are associated with recurrence rates as high as 70%.22–24 In our study, 50% of EMPD lesions treated with WLE without circumferential margin assessment had persistent disease/positive margins. None of our EMPD patients treated with SE have recurred after a mean follow-up time of 18.3 months. Subclinical extension of EMPD is challenging to estimate pre-surgically despite the physician’s clinical acumen. Prior studies found patients typically require multiple stages when treated with MMS or SE.5 Hendi et al. showed the mean number of stages to clear EMPD using MMS was 3.1 (range 1 – 9) and mean margins were 2.5 cm (range 0.6 - 11 cm).10 In our study, initial pre-surgical margins were estimated based on erythematous areas or darkening under Wood’s lamp examination. We showed an average of 1.9 stages to clear margins which is lower than previously reported (Figure 1–2).15–18,25–28
In our study, RCM was used to evaluate EMPD lesions in their initial visit. A novel finding of our study was that the presence of ‘disruption of the DEJ’ on RCM was associated with the presence of invasive disease on histopathology (83.3% vs 20.0%; p=0.005). Therefore, if ‘disruption of the DEJ’ is seen under RCM, suspicion for invasion should be raised. However, disruption of the DEJ was absent in one out of six cases (16.7%) of invasive EMPD examined with RCM: The absence of this finding does not completely rule-out invasion. Interestingly, ‘glandular nests’ were not associated with invasive disease. Correctly identifying invasion in EMPD can potentially help triage patients to surgical versus non-surgical treatments. This study also confirmed prior reported RCM features on a larger cohort. Prior studies mainly described RCM findings in Asian cohorts. Of 41 EMPD sites evaluated with RCM and biopsied, we found a kappa Cohen correlation of 0.92 to histopathology. The main RCM findings were ‘dark holes’ in the epidermis and ‘glandular nests,’ seen in 90.3% and 67.7% of EMPD lesions, respectively.13 Future, larger studies evaluating additional RCM features suggestive of invasive disease and interrater reliability are needed to validate our results.
Despite surgery being the gold standard treatment for EMPD, it might not always be feasible since EMPD involves functional and cosmetically sensitive areas. In our series, 33.3% of EMPD lesions were treated with either imiquimod, RT, or PDT. One third of them failed non-surgical treatment (3 failures, 1 death). One patient with perianal disease, treated non-surgically, then with RT and cytotoxic chemotherapy, died of metastatic disease. A recent systematic review highlighted the role non-surgical EMPD treatment options.29The main concern when treating EMPD with non-surgical modalities is the possible risk of untreated invasive disease and lack of histopathological confirmation of clearance. RCM may emerge as a potential tool to overcome these limitations. Whenever ‘disarranged DEJ’ is seen on RCM, invasive EMPD should be considered, and therefore the use of surgery or scouting biopsies to rule out invasion may be warranted. RCM can potentially help with both treatment monitoring and recurrence detection as well.12,13,22,30 In this study, one patient treated 17 years prior with imiquimod presented with non-specific scrotal erythema. RCM showed EMPD-specific features, prompting a biopsy that confirmed recurrent EMPD (Figure S2 -SDC).
Limitations:
Relatively small sample size; however, EMPD is a rare neoplasm and few studies have included prospective data collection. Nearly half (47.2%) had prior treatment and were categorized as ‘recurrent / persistent’ EMPD; there were few invasive or metastatic EMPD cases. Follow-up time was limited and hence no definitive conclusions on recurrence can be made from this study. No interrater correlation was calculated given RCM evaluations were performed by consensus. Additional RCM limitations preventing its wider adoption include cost associated with the device, time required by the treating physician and the patient, technician, and learning curve.
Conclusions:
EMPD is a challenging and complex disease. Pretreatment evaluation, including lesion demarcation and evaluating for invasive disease, is of the utmost importance. When surgery is feasible, obtaining clear margins before reconstruction may help reduce disease persistence/recurrence. The mean margins needed to clear EMPD using SE with radial sectioning were 1.8 cm, less than previously reported. Given potential surgical morbidity, non-surgical modalities including RT, imiquimod or PDT can be considered in selected cases. For all, systemic evaluation and close follow-up are necessary. Further studies are needed to evaluate the RCM cost-effectiveness but our experience suggests RCM is a valuable non-invasive imaging modality for the management, triage, and follow-up of EMPD treated both surgically and non-surgically.
Acknowledgments:
This research is funded in part by a grant from the National Cancer Institute / National Institutes of Health (P30-CA008748) made to the Memorial Sloan Kettering Cancer Center.
Dr. Rossi has received separate grant funding from The Skin Cancer Foundation, Society of MSKCC, Regen Pharmaceuticals, LeoPharma, and the A.Ward Ford Memorial Grant. He also served on advisory board, as a consultant, or given educational presentations: for Allergan, Inc; Galderma Inc; Evolus Inc; Elekta; Biofrontera, Quantia; Merz Inc; Dynamed; Skinuvia, Perf-Action, and LAM therapeutics.
Authors:
Cristian Navarrete-Dechent, Saud Aleissa, Miguel Cordova, Brian P. Hibler, Andres M. Erlendsson, Max Polansky, Frank Cordova, Erica H. Lee, Klaus J. Busam, Travis Hollmann, Cecilia Lezcano, Andrea Moy, Melissa Pulitzer, Mario M Leitao, Jr., and Anthony M. Rossi.