Background: Cutaneous T-cell lymphoma (CTCL) is a slowly progressive malignancy for which there is no cure. Therefore, accurate prediction of prognosis is important for the conduct of clinical trials and for counselling of individuals.

Objective: To improve prediction of survival in patients with CTCL.

Methods: Prognostic factors including tumour-node-metastasis (TNM) criteria and the CTCL Severity Index (CTCL-SI) were analysed using a Weibull model for multivariate analysis in a sample of 62 patients with classical CTCL (mycosis fungoides and Sézary syndrome). The Brier score was used to quantify the quality of individual prediction.

Results: Estimated 5-year survival rate (SR5) differed according to TNM stage: stage IA, 100% (95% confidence interval 70-100%); IB-III, 86% (73-100%); IVA, 54% (32-91%); IVB, 0% (0-52%). In a multivariate analysis, two independent prognostic factors were identified: lymph node (P = 0.036) and blood involvement (P = 0.015). A probability of survival model showed correlation of CTCL-SI with survival in patients with CTCL-SI > 20 according to the following formula: SR5 = 124-2 x (CTCL-SI)%. Calibration of SR5 against CTCL-SI-independent CTCL subsets revealed underestimation of Sézary syndrome. When CTCL-SI parameters were adjusted accordingly, the probability of survival model did not change significantly, while SR5 values became adequate. In addition, CTCL-SI was shown to be superior to TNM by 30% regarding individual predictive power.

Conclusions: Probability of survival in CTCL can be accurately predicted by a CTCL-SI-based survival rate formula. Careful monitoring of lymph node and blood compartments and quantification by CTCL-SI are reliable tools for follow-up of patients with CTCL and allow progression-adjusted prediction of prognosis.