Epidemiology of primary cutaneous γδ T‐cell lymphoma and subcutaneous panniculitis‐like T‐cell lymphoma in the U.S.A. from 2006 to 2015: a Surveillance, Epidemiology, and End Results‐18 analysis

Dear Editor, Primary cutaneous gamma/delta T‐cell lymphoma (pcGDTCL) is a rare, aggressive T‐cell lymphoma that presents with ulcerated nodules, extranodal dissemination, constitutional symptoms and often, haemophagocytic lymphohistiocytosis.1 pcGDTCL is so named because of the gamma/delta (γδ) phenotype of the T‐cell receptor (TCR). pcGDTCL is a histological mimic of subcutaneous panniculitis‐like T‐cell lymphoma (SPTCL), a lymphoma with an α/β TCR phenotype. SPTCL has an indolent course and presents with recurrent, painless, indurated, erythematous subcutaneous nodules that may ulcerate.1 Prior to 2005, cases of pcGDTCL and SPTCL were grouped as SPTCL.2 In this study, we sought to delineate further the demographics of patients with pcGDTCL and patients with SPTCL in the Surveillance, Epidemiology, and End Results (SEER)‐18 database during the period 2006–2015. We identified all patients with SPTCL and pcGDTCL diagnosed between 2006 and 2015 in the 18 SEER cancer registries. Cumulative incidences were calculated using the SEER*Stat software package, version 8·3·4 (Surveillance Research Program, National Cancer Institute, Bethesda, MD, U.S.A.). Routine methods of categorical (χ²‐test) and continuous (Student's t‐test) data analysis were applied with statistical significance defined for P < 0·05. Cox proportional hazard modelling was performed using Stata (StataCorp 2017, Stata Statistical Software: Release 15; StataCorp LLC, College Station, TX, U.S.A). This study was exempt from institutional review board approval. Search of the SEER‐18 registries identified 37 cases of pcGDTCL and 132 of SPTCL. Cumulative incidences of pcGDTCL and SPTCL were 0·40 and 1·51 per 10 million [95% confidence interval (CI) 0·28–0·56 and 1·26–1·80], respectively. Patients with pcGDTCL were significantly older than those with SPTCL [median 63 years, interquartile range (IQR) 49·5–72 vs. median 45 years, IQR 23–62; P < 0·001]. Patients with pcGDTCL were more likely to be male than were those with SPTCL (P < 0·013). There was no significant difference in terms of race/ethnicity, stage at diagnosis, cause of death, tumour location or presence of B symptoms. The majority of patients with either pcGDTCL or SPTCL who died did so of their lymphoma (83% and 86% of cases, respectively). There was no significant difference in median time to death between the two groups (pcGDTCL 6 months, IQR 1·5–10·25; SPTCL 10·5 months, IQR 2–40; P > 0·05). Median follow‐up for pcGDTCL was significantly shorter than that for SPTCL (13 months vs. 23 months, P < 0·001). All deaths in patients with pcGDTCL occurred within the first 24 months of diagnosis, yielding a 2‐year and 5‐year disease‐specific survival (DSS) of 73% and a 2‐year and 5‐year overall survival (OS) of 67% (Fig. 1a), compared with a 2‐year DSS of 95%, 5‐year DSS of 91%, 2‐year OS of 90% and 5‐year OS of 86% for SPTCL. (a) Kaplan–Meier plot describing the survival distribution function for 37 cases of primary cutaneous gamma/delta T‐cell lymphoma (pcGDTCL) and 132 cases of subcutaneous panniculitis‐like T‐cell lymphoma (SPTCL). Log‐rank P = 0·005. (b) Multivariate survival analysis; survival curves of 37 cases of pcGDTCL and 132 cases of SPTCL using Cox proportional hazards model with diagnosis age, stage, site, sex and race/ethnicity as covariates. The hazard ratio for pcGDTCL was 5·00 (95% confidence interval 1·8–14·3, P = 0·005). Cox proportional hazards modelling revealed that patients with pcGDTCL were at significantly higher risk of death than those with SPTCL (HR 5·00, 95% CI 1·8–14·3, P = 0·005) (Fig 1b). Increasing age (HR 1·31 per 10 years, 95% CI 1·0–1·7, P = 0·04) and stage (HR 1·52, 95% CI 1·1–2·1, P = 0·023) were all significant. Multivariate Cox proportional hazard models for pcGDTCL revealed decreased survival with stage IV disease compared with stages I–III (HR 1·65, 95% CI 1·07–2·54, P = 0·022) after controlling for age at diagnosis, stage, sex and race/ethnicity. Increasing age was associated with increased risk of death in SPTCL (HR 1·36 per 10 years, 95% CI 1·1–1·7, P = 0·005) after controlling for stage, tumour location, sex and race. We found that pcGDTCL and SPTCL demonstrate significant differences in demographics and survival. There were marked differences in age and sex distribution between the two diseases: patients with pcGDTCL were more likely to be older and male than patients with SPTCL. Notably, the 5‐year DSS for pcGDTCL in our study was markedly better than that found by Willemze et al. in 2008.3 This deviation may reflect the small number of cases in our analysis, miscoding of patients or insufficient follow‐up to capture late deaths. Another likely primary complicating factor is accurate diagnosis and classification, which can be extremely challenging in pcGDTCL. Gamma/delta TCR subunits can be expressed on other lymphomas. including mycosis fungoides (MF), although the significance of low levels of gamma/delta TCR expression is controversial.4 5 Overlap of panniculitis‐like lymphomas with lupus panniculitis, MF and peripheral T‐cell lymphoma not otherwise specified may create additional diagnostic ambiguity.6 Differentiating these two diseases is crucial because they require different treatments. Therapies for SPTCL are primarily based on immunosuppression. Treatment for pcGDTCL is almost exclusively based on combination chemotherapy, although allogeneic stem cell transplantation may be considered.7 In conclusion, pcGDTCL and SPTCL are different diseases with distinct epidemiology and clinical courses. While these conclusions are limited by small sample size, they may offer some guidance in constructing a differential diagnosis and providing prognostic guidance. we thank the American Society of Hematology for their support of this project as part of the HONORS research grant, which funded data collection and analysis.