Fig. 4.

Clinical relevance of B2M aberrations in two independent cohorts. a Analysis workflow for both Van Allen (105 patients pre-anti-CTLA4 treatment) and Hugo (38 patients pre-anti-PD1 treatment) data sets. For both data sets, we analyzed whole exome sequences of paired tumor and normal biopsies using the same pipeline used to analyze our cohort. b Illustration of three non-responders in the Van Allen data set with nucleotide mutations in B2M accompanied by loss of the wild-type allele. Gaps in the top chromosome depict the deleted region in each patient. Exons in B2M are shown as a horizontal blue rectangle, with mutations found in each patient highlighted in red. c Kaplan–Meier survival curves for patients in the Van Allen data set with (red) and without (black) B2M LOH. Log-rank p value is shown (p < 0.01). Inset shows the frequency of patients with B2M LOH in non-responders vs. responders and long-term survivors. One-sided Fisher’s exact p value is shown (p < 0.03). d Identical analysis performed for the Hugo data set as in c