Fig. 3

a ACC mutation frequencies. The heat map lists the frequencies of the 38 most commonly mutated genes across 3 ACC studies: COMETE (n = 45, cutoff = 4) [16], TCGA (n = 91, cutoff = 9) [8], and NIH (n = 43, cutoff = 7 [the current study]), as well as the average percentage of the three studies (ACC column, n = 179), and the cancer average as recorded in the TCGA PanCancerAtlas, which includes 33 cancers and 11.315 cases [https://gdc.cancer.gov/]. The heat map is sorted according to the average percentage of the three studies – the ACC column (i.e. number of patients with the mutation / total number of patients). The four genes shown in italic bold on shaded background (TP53, CTNNB1, NF1 and AFF1) are included in the 576 COSMIC Cancer Gene Census (CGC) Tier 1 genes, while the three genes in bold also on a shaded background (MUC16, MUC4 and PABPC1) are among the 147 CGC Tier 2 genes [REF.: COSMIC, Catalogue of Somatic Mutations in Cancer, https://cancer.sanger.ac.uk/cosmic]. b Correlation between the mutation frequency of the 38 cancer genes in 3A in ACC and in cancer in general both plotted as the average of their occurrence in ACC (Y-axis) and all cancers (X-axis). The diagonal identity line and the dashed linear regression line (R = 0.78, p = 1.6 E-08) are shown. The average mutation occurrence in ACC for 36 of the 38 genes, including TP53, fall below the identity line, indicating the mutation frequency is lower in ACC compared to the average in all cancers. The two exceptions are CTNNB1 and HGC6.3. c Median transcript length: comparing ACC (from 3 datasets) to all cancers in TCGA to all coding genes. Bar graph showing differences in median transcript length; comparing all coding genes from Ensembl Biomart (n = 37,416), the Tier 1 cancer-related genes in the COSMIC database (n = 576), and the most frequently mutated genes in the three ACC datasets (n = 38). The median with interquartile range is shown for each dataset. The transcript median length was statistically different between the 3 datasets: all coding genes vs. COSMIC (p < 0.0001); all coding genes vs. ACC (p < 0.0001); COSMIC vs. ACC (p < 0.01). The non-parametric Kruskal–Wallis statistic was applied for comparison of median values between the three groups