P052 How should we be identifying high-risk cutaneous squamous cell carcinoma in the UK? A dual-centre retrospective cohort study

Emilia Peleva,Ling Li,Sharanpal Jeetle,Eleni Ieremia,Zoe C Venables,Rakesh Anand,Li Ling Choo,Abirami Ganesh Kumar,Rebecca Lumley,Chaerin Im,Anna Nielsen-Scott,Rubeta Matin,Catherine Harwood

DOI: https://doi.org/10.1093/bjd/ljae090.079

IF: 11.113

2024-06-28

British Journal of Dermatology

Abstract:Abstract The incidence of cutaneous squamous cell carcinoma (cSCC) is increasing by 5% per year in the UK. Poor outcomes (local recurrence, nodal metastasis, distant metastasis and disease-specific death) are uncommon but are associated with significant morbidity and mortality. Accurately identifying cSCC at risk of poor outcomes is key to informing clinical decision making. Currently available approaches to cSCC risk stratification include UICC8 staging (Union for International Cancer Control, 8th edition), Brigham and Women’s Hospital staging (BWH), British Association of Dermatologists (BAD) 2020 cSCC guidelines, and a newly published clinicopathological cSCC risk calculator. Which should we use for cSCC risk stratification in a UK population? To address this question, we performed a retrospective cohort study of cSCCs treated over 12 months (2015–2016) in two UK tertiary healthcare centres. Following review of clinical notes, histology was reviewed and patients were recalled where required. A total of 1485 histology-confirmed primary cSCCs from 1272 patients were included, with a median follow-up of 3.8 years (interquartile range 1.4–5.8). The median age at diagnosis was 78 years (interquartile range 71–86), 67% were male, 35% had previous skin cancer and 281 (19%) tumours occurred in immunocompromised patients. Poor outcomes occurred in 4.9% of tumours, including local recurrences in 3.4%, nodal or parotid metastases in 2.4%, distant metastases in 0.5%, and cSCC-specific death in 0.8%. Overall, 95% of poor outcomes occurred within 3 years. For all four risk classifications there was a statistically significant difference (P < 0.05) in frequency of poor outcomes in high- vs. low-risk tumours. However, their performances varied considerably: BAD guidelines (low vs. high or very high risk) showed the highest sensitivity for predicting poor outcomes (86% vs. 11–45% for other classifications) but had low specificity (28% vs. 81–98%). The c-index measures the discriminative ability of risk score systems and was highest for the cSCC risk calculator at 0.69 (95% confidence interval 0.63–0.75), but this was not significantly higher than for the other classifications: 0.62 for UICC8, 0.63 for BWH, and 0.65 for BAD. Only the BAD classification incorporates immune status, and showed the highest sensitivity in immunocompromised patient cohorts. This study provides a detailed evaluation of the performance of four cSCC risk classification systems in a dual-centre UK cohort of almost 1500 primary cSCCs. Furthermore, it includes a high proportion of cSCCs from immunocompromised patients, who are often under-represented in skin cancer research. It highlights that no single risk classification was optimal in terms of sensitivity and specificity in predicting poor outcomes. With the rising incidence of cSCC, further improving approaches to cSCC risk classification is now an urgent priority in order to rationalize decision making, including intensity of surveillance and optimal patient selection for adjuvant and neoadjuvant ­therapies.

dermatology