Supplementary Figures 1 - 7 from A Surprising Cross-Species Conservation in the Genomic Landscape of Mouse and Human Oral Cancer Identifies a Transcriptional Signature Predicting Metastatic Disease

Onken, Michael D.; Winkler, Ashley E.; Kanchi, Krishna-Latha; Chalivendra, Varun; Law, Jonathan H.; Rickert, Charles G.; Kallogjeri, Dorina; Judd, Nancy P.; Dunn, Gavin P.; Piccirillo, Jay F.; Lewis, James S.; Mardis, Elaine R.; Uppaluri, Ravindra

doi:10.1158/1078-0432.22454642.v1

- No file added yet -

Supplementary Figures 1 - 7 from A Surprising Cross-Species Conservation in the Genomic Landscape of Mouse and Human Oral Cancer Identifies a Transcriptional Signature Predicting Metastatic Disease

journal contribution

posted on 2023-03-31, 18:06 authored by Michael D. Onken, Ashley E. Winkler, Krishna-Latha Kanchi, Varun Chalivendra, Jonathan H. Law, Charles G. Rickert, Dorina Kallogjeri, Nancy P. Judd, Gavin P. Dunn, Jay F. Piccirillo, James S. Lewis, Elaine R. Mardis, Ravindra Uppaluri

PDF file - 486KB, Figure S1 (A) Average statistics for depth of coverage at 20X, 30X and 40X of all MOC lines. Robust reads were obtained in all samples with a range of 97-98 percent for 20X, 93-95 percent for 30X and 88-92 percent for 40X. (B) Putative SNVs, nsSNVs (Table S1 and S2) and Indels (Table S4) of all MOC lines. Figure S2 Oncoprints from cBio of AKAP9, MED12L, THSD7A, MUC5B, MYH6, LAMA1, LRP2, and 3 RAS (Table S6) genes representing other candidate tumor promoters as compared to the 3 RAS genes. Note that for AKAPs and MED components, we found 9 AKAP family member mutations in 20.4 percent of tumors, with AKAP9 changes in 7 percent (Fig. 1D). Six components of the mediator complex were mutated in 14.7% of cases, with MED12L changes in 5% (Fig. 1E). Figure S3 Number of nsSNV's per node negative (N0) and node positive (N+) tumor in (A) all TCGA OSCC (note that patient TCGA-D6-6516 (N0) with 1463 mutations is not included in this graph) and (B) TCGA OSCC patients who had smoking history reported. There was no significant difference in the average number of mutations between N0 (3272 nsSNVs) and N+ (3097 nsSNVs) patients regardless of smoking status (Tables S8, S9). This analysis showed 17 genes commonly mutated in mouse indolent and human N0 tumors and 55 common genes mutated in mouse aggressive and human N+ tumors (Supplementary Table S10). However, none of these common genes were mutated at high frequency in the human N0 or N+ datasets (Supplementary Table S11 and 12). Finally, comparing N0 and N+ tumors from human TCGA data also showed that specific mutations occur infrequently in both the metastatic and non-metastatic tumors (data not shown). Figure S4 SAM plotsheet of MOC line microarray data with estimated miss rates for delta=4.68 Figure S5 (A) GSEA data of OCAMP-A on UW/FHCRC data with (B) first condensation of enriched genes. (C) GSEA data of OCAMP-A on MDA data with (D) first condensation of enriched genes. (E) GSEA data of OCAMP-A on TCGA data with (F) first condensation of enriched genes. Figure S6 (A) Disease specific survival (DSS) after weighted voting classification of OCAMP-B signature on the MD Anderson dataset shows worse outcome for those with aggressive classification (p=0.028). Figure S7 (A) Disease specific survival (DSS) and (B) overall survival (OS) after weighted voting classification of OCAMP-B signature on the UW/FHCRC dataset shows worse outcome for those with aggressive classification (p<0.01).

History

ARTICLE ABSTRACT

Purpose: Improved understanding of the molecular basis underlying oral squamous cell carcinoma (OSCC) aggressive growth has significant clinical implications. Herein, cross-species genomic comparison of carcinogen-induced murine and human OSCCs with indolent or metastatic growth yielded results with surprising translational relevance.Experimental Design: Murine OSCC cell lines were subjected to next-generation sequencing (NGS) to define their mutational landscape, to define novel candidate cancer genes, and to assess for parallels with known drivers in human OSCC. Expression arrays identified a mouse metastasis signature, and we assessed its representation in four independent human datasets comprising 324 patients using weighted voting and gene set enrichment analysis. Kaplan–Meier analysis and multivariate Cox proportional hazards modeling were used to stratify outcomes. A quantitative real-time PCR assay based on the mouse signature coupled to a machine-learning algorithm was developed and used to stratify an independent set of 31 patients with respect to metastatic lymphadenopathy.Results: NGS revealed conservation of human driver pathway mutations in mouse OSCC, including in Trp53, mitogen-activated protein kinase, phosphoinositide 3-kinase, NOTCH, JAK/STAT, and Fat1-4. Moreover, comparative analysis between The Cancer Genome Atlas and mouse samples defined AKAP9, MED12L, and MYH6 as novel putative cancer genes. Expression analysis identified a transcriptional signature predicting aggressiveness and clinical outcomes, which were validated in four independent human OSCC datasets. Finally, we harnessed the translational potential of this signature by creating a clinically feasible assay that stratified patients with OSCC with a 93.5% accuracy.Conclusions: These data demonstrate surprising cross-species genomic conservation that has translational relevance for human oral squamous cell cancer. Clin Cancer Res; 20(11); 2873–84. ©2014 AACR.