Current clinical risk stratification systems are limited in their predictive value for metastasis in cutaneous squamous cell carcinoma (CSCC). Progress requires large population-based datasets integrating clinical, pathological, and molecular data and efficient study design, because the volume of patients is large and the event rate is low. This study describes two nationwide cohorts and the collection of CSCC samples with long-term follow-up and sufficient metastatic events to enable prognostic modelling. Linked data of the Netherlands Cancer Registry and the Dutch Nationwide Pathology Databank were used to collect two nationwide CSCC cohorts and perform nested case–control studies. In the discovery cohort (first CSCC diagnosis: 2007–2009), each metastatic case was matched to a non-metastatic control with similar metastatic risk. In the validation cohort (first CSCC diagnosis: 2017–2018), each case was matched to both a random and a risk-matched control. Tissue sections were prepared for hematoxylin and eosin staining, RNA/DNA sequencing, and immunostaining. The discovery cohort included 19,120 CSCC patients with ten years of follow-up and 472 samples (236 case–control sets) with a median time to metastasis of 1.1 (IQR 0.5–2.1) years. The validation cohort included 25,921 CSCC patients with at least five years of follow-up and 555 samples (~ 185 sets with 2 types of controls, with a median time to metastasis of 0.70 (0.3–1.5) years. This design enables the development of absolute risk prognostic models with sufficient number of events. Clinical, histopathological, and molecular data can be combined, paving the way towards more personalized treatments for CSCC patients.
Steijlen et al. (Thu,) studied this question.