Fig. 3: PD-L1 immunohistochemistry feature derivation and prediction of response.
a, Analysis pipeline to extract image-based IHC texture starting from scanned PD-L1 IHC slides. b, Normalized-value distributions of GLCM and pixel intensity-derived image features stratified by response for the best performing summary statistic in each GLCM class. Features indicated by the red asterisk emerged as salient features in the LR fit with n = 42 responders and n = 63 nonresponders. c, Three representative PD-L1 IHC slides from n = 105 samples corresponding to the maximum (top), median (middle) and minimum (bottom) of the GLCM autocorrelation distribution, with low power, high power, stain intensity and pixel-wise GLCM sample patches. d, Correspondence of the example GLCM features in c between low, medium and high bins of TPS. The interior box-and-whisker bars show the mean as a white dot, the IQR (25–75%) as a black bar and the minimum and maximum as whiskers up to 1.5 × IQR for n = 105 samples. e, Response prediction performance using LR classifiers with PD-L1 features including TPS (LR PD-L1-TPS), pixel and GLCM autocorrelation image features (LR IHC-A), only the complete GLCM features (LR IHC-G) and the result of aggregating patient-level predictions by averaging classifier outcomes from LR IHC-A, IHC-G and LR PD-L1-TPS (LR Path-Average). The bar height and error bar represent the merged AUC and associated 95% CI based on DeLong’s method51 for n = 105 and n = 52 patients in the multimodal and validation cohorts, respectively.
