Predictive Role of Computed Tomography Texture Analysis in Patients with Metastatic Urothelial Cancer Treated with Programmed Death-1 and Programmed Death-ligand 1 Inhibitors

Background: Reliable biomarkers to predict the response of metastatic urothelial cancer (mUC) to programmed death-1 and programmed death-ligand 1 (PD-1/PD-L1) inhibitors are being investigated. Texture analysis represents tumor heterogeneity and may serve as a predictor of response in mUC.

Objective: To assess the predictive ability of computed tomography (CT) texture analysis for progression-free survival (PFS) in patients with mUC treated with PD-1/PD-L1 inhibitors.

Design, setting, and participants: Forty-two postplatinum patients with mUC treated with PD-1/PD-L1 inhibitors from 2013 to 2018, including those with measurable disease per RECIST 1.1 who had contrast-enhanced baseline or first follow-up CT within 3mo after starting treatment, were included. PFS was calculated based on serial follow-up CT scans. Eleven patients with follow-up of <12mo without progression were excluded. Texture features of measurable lesions on baseline and first follow-up CT were extracted using commercially available software (TexRAD; Feedback Plc, Cambridge, UK) using different spatial scaling factors (0, 2-6).

Outcome measurements and statistical analysis: Stepwise logistic regression analysis was conducted to identify patients with PFS <12mo, and performance was assessed using receiver operator characteristic curves.

Results and limitations: Of 31 included patients, 18 had PFS <12mo. Twenty-five baseline CT and 29 first follow-up CT scans met the inclusion criteria. In patients with PFS <12mo, entropy and mean were higher on first follow-up CT (p=0.02 and p=0.005, respectively). A predictive model including mean and entropy on first follow-up CT yielded 95% sensitivity, 80% specificity, 90% positive predictive value, 89% negative predictive value, and 90% accuracy (area under the curve=0.963) to identify patients with PFS <12mo. Limitations include retrospective nature and small sample size.

Conclusions: CT texture analysis can help predict early progression with high accuracy soon after starting PD-1/PD-L1 inhibitors. Studies investigating the correlation of texture analysis with survival endpoints may help validate texture analysis as a biomarker of PD-1/PD-L1 inhibitors' treatment response.

Patient summary: Computed tomography texture analysis can help predict durability of response in patients with metastatic urothelial cancer early during treatment with programmed death-1 and programmed death-ligand 1 (PD-1/PD-L1) inhibitors.