Clinical outcome

To relate (changes in) TIL density with clinical outcome, patients were dichotomized based on the development of disease recurrence within the first year after radical cystectomy. The date of radical cystectomy was chosen as t = 0. The time-point of one year was chosen as most clinically relevant interval following cystectomy. If a recurrence occurs after one year, we consider restarting chemotherapy with palliative intent. Additionally, in clinical trials comparing neoadjuvant chemotherapy plus radical cystectomy versus upfront radical cystectomy, disease-free survival curves appear to diverge primarily in the first year after cystectomy [28, 29]. Nevertheless, to exclude bias due to the chosen interval of 1 year, we also performed univariate cox proportional hazards analyses to relate fold changes in TIL density to time to recurrence.

Patients received follow-up according to standard hospital procedures. In the first year, patients were followed up every 3 months with alternately a CT scan of thorax and abdomen or a chest X-ray. In the second year, patients underwent a CT scan every 6 month and, in the third year, a CT scan once a year. All recurrences were confirmed by imaging. Patients who did not have sufficient follow-up data (n = 3) or died within one year without developing a recurrence (n = 3) were excluded from analyses comparing patients with versus without an early recurrence. In Kaplan Meier curves for disease-free survival, these patients were included, but they were censored at the time of last follow-up or non-cancer related death.

We correlated TIL density with disease recurrence rather than overall survival as the number of deaths was relatively low and only part of the patients had access to immunotherapy after the development of metastatic disease.