Extended Data Fig. 3: Alternative immune deconvolution methods reveal consistent TME findings and T-cell repertoire identifies clonal differences in responders.
a. (left) Concordance between CytoPro and ssGSEA methods for selected signatures related to tertiary lymphoid structures (*** indicates p-value < 0.001) (right) Illustration of relationship between CytoPro signature (Z-normalized) and raw ssGSEA score for Neutrophils (linear regression, shaded region corresponds to 95% CI, N=342). b. Correlation between IHC CD8 staining (positive cells per total area) with ssGSEA CD8 T cell staining as a validation for ssGSEA measurement of CD8 T cells (r=0.61; p < 0.001, Pearson Correlation; shaded region corresponds to 95% CI, N=342). c. Hierarchical clustering of cell-type contributions estimated by ssGSEA. d. Associations between ssGSEA signatures and PFS (Cox proportional hazards model, 95% CI, and log-rank test, n=344 patients). Note significant association with Tfh (T follicular helper) cells and improved outcomes in Avelumab (N=176, p < 0.001; log-rank test) which is not significant in Placebo and has a significant interaction by treatment arm (N=166, p=0.02). In contrast, Neutrophils are associated with worse outcomes for patients receiving avelumab (p = 0.003; log-rank test).
