A Territory-Wide Follow-Up of Primary and Secondary Extramammary Paget Disease of 2 Decades: Effects of Local Disease on Survival

Extramammary Paget disease (EMPD) is an uncommon malignant cutaneous neoplasm. Primary EMPD is believed to be of cutaneous origin, whereas secondary EMPD is that that is associated with genitourinary (GU) and/or gastrointestinal (GI) carcinomas [1]. These two entities are not easily distinguishable from each other and often rely on detailed history taking and extensive investigations to exclude a diagnosis of secondary EMPD. It is of clinical significance to recognize secondary from primary EMPD. In this multicenter study, histologically proven cases of EMPD were reviewed for clinicopathological parameters and outcomes. In particular, subgroup analysis was conducted for secondary EMPD, which represents the rarer form of EMPD and has limited published data with quantitative analysis reported.

Computerized searches of the pathology department archives involved institutions were performed for all surgical specimens with a diagnosis of “EMPD” (primary, secondary, or without specifying primary or secondary), from the year 1996 to 2022, including biopsy (incision and excisional) and resection specimens. Pathology reports of the retrieved cases were reviewed and cases with a confirmed diagnosis of EMPD were included. Repeated cases identified by identity number were removed. Demographical, clinicopathological, and follow-up information, including baseline demographics, history of malignancy, resection margins clearance, recurrence, and survival status, was retrieved by reviewing pathology reports and electronic case notes, which includes all attendances in public hospitals and affiliated clinics of Hong Kong, and outcome data from the deaths registry that records hospital and nonhospital deaths. Cases of secondary EMPD were identified by a medical history of or subsequent diagnosis of GI or GU carcinomas, confirmed by hospital diagnostic code, imaging findings, and/or tissue biopsy. Cases were considered as primary EMPD if there were clinical and radiological investigations performed to exclude secondary EMPD. Cases without sufficient correlating information were excluded (online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000541394).

Statistical analysis was performed with the statistical software SPSS (Microsoft Windows, version 26). The Mann-Whitney U test, chi-squared test, and Fischer exact test were used to compare clinicopathological features of primary and secondary EMPD as continuous and categorical variables. Survival analyses for overall survival (OS), disease-free survival (DFS), and disease-specific survival (DSS) were performed using Kaplan-Meier analysis, hazard ratio, and Cox regression (with the backward-Wald method) for univariate and multivariate analysis. A p value of <0.05 was considered significant.

A total of 109 cases from 64 male and 45 females, with a median age of 70 years, were retrieved. Follow-up data were available for 107 cases, with both cases without follow-up data being primary EMPD, and the median duration of follow-up was 63 months. The cases without follow-up data were excluded from outcome analysis. There were 86 cases with local excision performed and margin status retrievable. The majority of anatomical sites involved were of the genitoperineal region (n = 97/109, 89.0%); the remaining were from the abdomen (n = 7/109, 6.4%), axilla (n = 3/109, 2.8%), and face (n = 2/109, 1.8%). There were 19 cases of secondary EMPD, most commonly associated with colorectal (n = 6/19, 31.6%), anal (n = 5/19, 26.3%), and prostatic carcinomas (n = 3/19, 15.8%). In 4 cases, malignancies of other primaries (breast, liver, and lung) were documented (Table 1). Lymph node and distant metastases were documented in 8 and 6 cases, respectively, which were all future recurrences except for 1 case of liver metastasis. The median and median OS and DSS were shorter comparing secondary EMPD to primary EMPD.

Comparing primary and secondary EMPD, there was a significant difference observed between older age and secondary EMPD (77 vs. 68.5 years, p = 0.016). There were no statistically significant differences observed between primary and secondary EMPD in terms of patient sex and anatomical site (Table 2). Kaplan-Meier survival analysis was performed comparing sex, age (over 60 years old), margin involvement, site of lesion, primary versus secondary EMPD, presence of invasion, and any history of non-GI/GU malignancy. Male sex (p = 0.018), age over 60 years (p = 0.004), and involvement of margins (p = 0.018) were associated with shorter OS (Fig. 1). As for DSS, involvement of margins (p = 0.009) was an adverse predictor. Secondary EMPD showed a trend in worse OS (p = 0.077) and statistically worse DSS than primary EMPD (p = 0.005) (Fig. 2). DFS did not demonstrate any statistically significant correlations.

Cox regression demonstrated independent association between age (HR = 4.221, 1.266–14.076, p = 0.019), margin involvement (HR = 2.361, 1.169–4.469, p = 0.017), and secondary (vs. primary) EMPD (HR = 2.185, 1.001–4.770, p = 0.050) and shorter OS. Shorter DSS was independently associated with secondary EMPD (HR = 3.445, 1.317–9.012, p = 0.012) and margin involvement (HR = 3.157, 1.219–8.179, p = 0.018) (Table 3). Similarly, no statistically significant associations were seen with DFS.

Subgroup analysis was performed on the 19 cases of secondary EMPD. On Kaplan-Meier analysis, margin involvement was associated with shorter OS (p = 0.007) and DSS (p = 0.003). Results from Cox regression incorporating other variables showed that margin involvement was an independent predictor of shorter OS (HR = 6.845, 1.359–34.466, p = 0.020) with a similar trend for DSS (HR = 8.523, 0.869–83.624, p = 0.066) (Table 3).

EMPD is an uncommon cutaneous neoplasm that poses as a diagnostic and therapeutic dilemma. EMPD is categorized into primary EMPD and secondary EMPD. EMPD is considered to be secondary when there is an underlying and/or contiguous carcinoma, most classically of GU and GI origin [1]. Primary EMPDs are believed to originate from cutaneous adnexal structures, with evidence suggesting apocrine glands as a possibility [2]. Although EMPD is considered as an in situ malignancy, dermal invasion can be seen occasionally [3]. There is high heterogeneity among all disease entities under the umbrella of “EMPD” encompassing primary cutaneous EMPD, secondary EMPD associated with various visceral malignancies with invasive and noninvasive forms [4]. In this cohort, archives over 2 decades from three regional institutions serving a population of over one million were reviewed [5] to compare the clinicopathological features and outcome of histologically proven EMPDs.

EMPD most commonly presents as a crusted, erythematous ill-defined lesion [2, 3, 6]. Its appearance overlaps with that of other dermatoses such as eczema, fungal infection, and psoriasis [7, 8] and other cutaneous neoplasms including melanocytic and squamous lesions [9]. Despite primary and secondary EMPD having significantly different underlying etiologies, in the absence of clinical, radiological, or pathological evidence of an attributable GI or GU malignancy, it is not possible to differentiate primary from secondary EMPD. Secondary EMPD is rare, constituting less than 20% of all EMPDs [4]. As such, many authors have opted to perform analysis on all EMPDs without subgrouping [10, 11], or only review cases of primary EMPD [12].

Smaller cohorts making descriptive comparisons of primary and secondary EMPD have suggested secondary EMPDs being of smaller size and associated with a younger age [1]. Comparison of age and sex in the current cohort showed a trend favoring older age in patients with secondary EMPD. The site of lesion did not show significant correlation (Table 2); as even though the GI and GU tract is in continuation with the genitalia and perineum, primary EMPD also frequently develops in the genitoperineal region and secondary EMPD is sometimes seen outside of the perineum [13]. Despite the clinical resemblance of primary and secondary EMPD, the OS and DSS in patients with secondary EMPD are significantly worse upon multivariable analysis (Table 3).

Univariable survival analysis showed old age, history of non-GI/GU malignancy, and presence of invasion being associated with shorter OS, but only involvement of margins was correlated with both shorter OS and DSS. These findings illustrate that comorbidities such as age and other concurrent malignancies do not affect disease-specific outcome. The significance in prognostic value of margin involvement is further evidenced by multivariable Cox regression (Table 3).

The positive correlation between survival and clear margins is supported by some authors [14, 15] with a SEER review of invasive EMPD suggesting that patients who did not receive surgery suffered from increased mortality [16]. However, conflicting results from proponents of functional preservation and minimal surgery have presented cohorts without apparent association between margin and survival [10, 17]. By design, survival analyses are retrospective, and it would not be possible to assign a control arm to forgo complete excision when it is surgically feasible. Similarly, involved margins may be due to surgical and anatomical limitations that cannot be overcome. Hence, it remains unclear whether excision with clear margins by itself confers survival benefit, or that the improved outcomes reflect a lesser extent of involvement or better fitness in patients with complete resection, in other words resectability. Recommendations on re-excision also cannot be made. It may be prudent to infer that resectability is a strong and independent predictor of longer OS and DSS in the cohort and subgroup analysis of secondary EMPD only (Table 3). The lack of associations between DFS and clinicopathological prognostic factors is in line with the high risk of local, but treatable, recurrence in EMPD [18], for which radiotherapy [19, 20], repeat surgical excision [18], and topical chemotherapy [21] are available options in disease management.

The current study is limited by a modest number of cases. Data collection is difficult with other published clinical reviews reporting less than 100 cases even with extended data collection periods [8, 10]. Although just over 100 cases were collected in the current cohort, there were less than 20 cases of secondary EMPD, and the weight of individual cases may be amplified. However, all cases in the current cohort were reviewed case by case with electronical case notes, clinical and outcome data on a region-wide electronic patient system, as opposed to database analysis suffering from the inability of assessing individual patient data [22] and thus difficulty to distinguish between primary and secondary EMPD. Immunohistochemistry is not uncommonly used in identifying secondary EMPD by demonstrating positivity to GI and urothelial markers (or vice versa, negativity to apocrine and cutaneous markers) [23, 24]. Immunohistochemistry was not reviewed for the current cohort, as the reference standard of diagnosing secondary EMPD is the presence of an attributable carcinoma, of which a follow-up period averaging more than 6 years should be considered sufficient for identifying these cases.

Primary and secondary EMPD cannot be distinguished reliably by patient demographics nor clinical features, albeit secondary EMPD demonstrating a trend toward older age at presentation. Older age is independently associated with shorter OS, while margin involvement and secondary EMPD (vs. primary EMPD) are predictors of poorer OS and DSS. In subgroup analysis of secondary EMPD, margin involvement remained as a strong and independent indicator of shorter OS and DSS. Regardless of the association of visceral carcinomas in secondary EMPD, resectability is a strong predictor of favorable outcome. Excision with clear margins may confer survival benefits and should be attempted when surgically feasible.

The authors would like to thank Ms. Stella Leung for her assistance in graphical editing of the tables and figures and Ms. Nikki Kit Ying Hon, Jasmine Hei Nga Law, and Mr. Ka Wun See for arranging the tables and data.

The study was approved by the Joint Chinese University of Hong Kong-New Territories East Cluster Clinical Research Ethics Committee (Approval reference No. 2020.320) with wavier of the requirement of written consent.

The authors declare that there is no conflict of interest regarding the publication of this paper.

The authors have no funding to declare.

J.K.M.N.: conceptualization, methodology, investigation, and writing – original draft. A.W.S.C.: formal analysis and validation. C.M.T.C.: resources and validation. P.C.L.C.: resources and supervision. E.C.C.I. and W.H.L.L.: investigation and resources. J.S.H.L.: validation and resources. J.J.X.L.: conceptualization, data curation, formal analysis, methodology, investigation, and writing – review and editing.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from the corresponding author [J.J.X.L.] upon reasonable request.