Composite versus individual biomarkers for predicting clinical benefit to PD-1 blockade in NSCLC 87 3 3Dhistech P1000 scanner. For manual scoring, PD-L1 and CD68 IHC images were uploaded onto Slide Score, a web platform designed for the manual scoring of digital slides using a scoring sheet (www.slidescore.com). CD8, PD-1T, CD3 TILs, CD20+ B cells and TLS were digitally scored as described below. Digital quantification of CD8 and PD-1T TILs Digital image analysis was performed by a trained MD (K.H.) and supervised by an experienced pathologist (K.M.) using the Multiplex IHC v1.2 module from the HALOTM image analysis software, version 2.3.2089.69 (Indica Labs). Researchers were blinded for clinical outcome. For the classification of CD8 lymphocytes in single stains, a computationally derived cut-off of 0.3 optical density (OD) was used, reflecting the intensity of the staining. This cut-off was established by manually optimizing the detection of CD8 positive stained cells in FFPE samples. An image analysis algorithm utilizing a 0.3 OD cut-off was generated for automated analyses of CD8 lymphocytes in subsequent FFPE samples. The quantification of PD-1T TILs followed a previously described methodology25. The frequency of CD8 and PD-1T TILs were determined as the number per mm2 tumor area. Tumor areas were digitally annotated as described previously25. PD-1T TIL data from 94 samples were used from previous work25 (Table S1). For regional analysis of CD8 lymphocytes, classifiers were trained to distinguish stromal and tumoral regions, allowing for the separate quantification of CD8 lymphocytes in these distinct regions. The percentage of CD8 lymphocytes within tumoral regions (i.e. intra-tumoral (IT)) relative to total CD8 TILs was subsequently calculated (Table S1). Scoring of tertiary lymphoid structures The quantification of TLS and the combined number of TLS and lymphoid aggregates (TLS+LA) per mm2 tumor area was performed using the HALOTM image analysis software, version 2.3.2089.69 (Indica Labs). This analysis was conducted on a CD20CD3 double immunostaining, following a previously established methodology25. TLS and TLS+LA data from 91 samples were used from previous work25 (Table S1). CD20 and CD3 quantification by digital image analysis Digital quantification of CD20 and CD3 expression involved the measurement of the total area with CD20 and CD3 expression, relatively. This was performed using a pre‑established image analysis algorithm from the Area Quantification version 1.0 module of HALOTM image analysis software (Indica Labs)25. The resulting CD20positive and CD3-positive areas were normalized per mm2 tumor area. Cell numbers
RkJQdWJsaXNoZXIy MTk4NDMw