The mutational landscape of the SCAN-B real-world primary breast cancer transcriptome

Abstract

Breast cancer is a disease of genomic alterations, of which the panorama of somatic mutations and how these relate to subtypes and therapy response is incompletely understood. Within SCAN-B (ClinicalTrials.gov: NCT02306096), a prospective study elucidating the transcriptomic profiles for thousands of breast cancers, we developed a RNA-seq pipeline for detection of SNVs/indels and profiled a real-world cohort of 3,217 breast tumors. We describe the mutational landscape of primary breast cancer viewed through the transcriptome of a large population-based cohort and relate it to patient survival. We demonstrate that RNA-seq can be used to call mutations in genes such as PIK3CA, TP53, and ERBB2, as well as the status of molecular pathways and mutational burden, and identify potentially druggable mutations in 86.8% of tumors. To make this rich dataset available for the research community, we developed an open source web application, the SCAN-B MutationExplorer (http://oncogenomics.bmc.lu.se/MutationExplorer). These results add another dimension to the use of RNA-seq as a clinical tool, where both gene expression- and mutation-based biomarkers can be interrogated in real-time within 1 week of tumor sampling.

Keywords: RNA-seq; breast cancer; mutation; survival; transcriptome.

Figure 1. Study design

Study design flow diagram for DNA‐seq‐informed optimization of RNA‐seq variant calling.

Figure EV1. Overview of frequently mutated genes in targeted DNA‐seq and RNA‐seq across 275 ABiM samples

1. A, B

   Waterfall plot of the 20 most mutated genes (rows) across 275 ABiM samples (columns) in (A) targeted DNA‐seq and (B) RNA‐seq. Genes are ranked by variant frequency. Samples are sorted by histological subtype and alteration occurrence. Mutations are colored by predicted functional impact.

Figure 2. Overview of non‐synonymous mutations in terms of base substitution signatures, molecular subtype, and protein impact

1. A

   Contribution of base change types to the overall SNV composition in the ABiM cohort for captured DNA regions and mRNA in the captured DNA regions, as well as SCAN‐B whole mRNA.

2. B

   Number of non‐synonymous mutations per sample. Bars are colored by PAM50 subtypes Luminal A (dark blue), Luminal B (light blue), HER2‐enriched (pink), basal‐like (red), Normal‐like (green) and Unclassified (gray).

3. C–F

   Lollipop plots showing the location, abundance, and impact of SNVs in (C) TP53, (D) PIK3CA, (E) PTEN, and (F) ERBB2 on the respective encoded protein. Protein change labels are shown for the most mutated amino acid positions, with residues ordered left to right by mutation frequency within each label.

Figure 3. Overview of frequently mutated genes across 3,217 SCAN‐B samples

Waterfall plot of the 20 most frequently mutated genes (rows) across 3,217 SCAN‐B samples (columns). Genes are ranked from top to bottom by mutation frequency. Samples are sorted by histological subtype and alteration occurrence. Mutations are colored by predicted functional impact.

Figure 4. Binary heatmap of mutation status of important breast cancer pathways in 3,217 samples

Binary heatmap of mutation status of important BC pathways in 3,217 samples. Samples with wild‐type (wt) pathway status (defined as all member genes being wt) are colored blue, those with mutated pathways (at least one member gene mutated) are colored red. Samples and pathways were clustered using Euclidean distance and Ward linkage. Reactome IDs for the pathways can be found in Table EV4.

Figure 5. Impact of gene mutations on overall survival across treatment groups

1. A–E

   Overall survival (OS) of patients with tumors containing mutations in the genes (A) TP53, (B) PIK3CA, (C) ERBB2, and (D) PTEN. (E) OS by PTEN‐MutExp genotype (“low” defined as PTEN mutation or PTEN expression in the lower quartile across the cohort, “normal” otherwise) stratified by groups receiving no systemic treatment (n = 336), endocrine therapy only (Endo only; n = 1,579), endocrine‐ and chemotherapy (Endo + Chemo ± any; n = 914), as well as HER2 treatment with any other treatment or none (HER2 ± any; n = 348). Specific treatments in these groups are detailed in Table EV5. In each Kaplan–Meier plot, wild‐type (wt) and normal cases are plotted in blue, mutated (mut) and low cases are plotted in red, the log‐rank P value is given, and the hazard ratio (HR) for mutation/low is given with a 95% CI and after univariable and multivariable (MV) Cox regression adjustment. Covariables included in the MV analysis were age at diagnosis, lymph node status, tumor size, and the variables denoted by the following symbols: ¶, ER, PgR, HER2, and NHG; ¤, ER, PgR, and NHG; $, HER2 and NHG. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; NHG, Nottingham histological grade; PgR, progesterone receptor.

Figure EV2. Impact of gene mutations on overall survival across clinical subgroups

1. A–E

   Overall survival (OS) of patients with tumors containing mutations in the genes (A) TP53, (B) PIK3CA, (C) ERBB2, and (D) PTEN. (E) OS by PTEN‐MutExp genotype (“low” defined as PTEN mutation or PTEN expression in the lower quartile across the cohort, “normal” otherwise) stratified by the clinical patient subgroups HoR⁺/HER2⁻ (HoR⁺ when ER⁺ and PgR⁺, HoR⁻ otherwise; n = 2,134), HoR⁺/HER2⁺ (n = 230), HoR⁻/HER2⁺ (n = 104), and TNBC (n = 137). Specific treatments in these groups are detailed in Table EV4. In each Kaplan–Meier plot, wild‐type (wt) and normal cases are plotted in blue, mutated (mut) and low cases are plotted in red, the log‐rank P value is given, and the hazard ratio (HR) for mutation/low is given with a 95% CI and after univariable and multivariable (MV) Cox regression adjustment. Covariables included in the MV analysis were age at diagnosis, lymph node status, tumor size, and the variables denoted by the following symbols: ¤, ER, PgR, and NHG; #, NHG. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HoR, hormone receptor; NHG, Nottingham histological grade; PgR, progesterone receptor; TNBC, triple‐negative breast cancer.

Figure 6. Impact of pathway mutations on overall survival across treatment groups

1. A–E

   Overall survival of patients with tumors containing mutations in pathways (A) WNT signaling, (B) Hedgehog signaling, (C) NOTCH2 signaling, (D) p53 independent DNA damage repair, (E) TGFβ signaling, stratified by groups receiving no systemic treatment (n = 336), endocrine therapy only (Endo only; n = 1,579), endocrine‐ and chemotherapy (Endo + Chemo ± any; n = 914), as well as HER2 treatment with any other treatment or none (HER2 ± any; n = 348). Specific treatments in these groups are detailed in Table EV4. In each Kaplan–Meier plot, wild‐type (wt) cases are plotted in blue, mutated (mut) cases are plotted in red, the log‐rank P value is given, and the hazard ratio (HR) for mutation is given with a 95% CI and after univariable and multivariable (MV) Cox regression adjustment. Covariables included in the MV analysis were age at diagnosis, lymph node status, tumor size, and the variables denoted by the following symbols: ¶, ER, PgR, HER2, and NHG; ¤, ER, PgR, and NHG; $, HER2 and NHG. See Table EV3 for Reactome pathway IDs. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; NHG, Nottingham histological grade; PgR, progesterone receptor.

Figure EV3. Impact of pathway mutations on overall survival across clinical subgroups

1. A–E

   Overall survival of patients with tumors containing mutations in pathways (A) WNT signaling, (B) Hedgehog signaling, (C) cell cycle, (D) p53 independent DNA damage repair, and (E) TGFβ signaling, stratified by the clinical patient subgroups HoR⁺/HER2⁻ (HoR⁺ when ER⁺ and PgR⁺, HoR⁻ otherwise; n = 2,134), HoR⁺/HER2⁺ (n = 230), HoR⁻/HER2⁺ (n = 104), and TNBC (n = 137). Specific treatments in these groups are detailed in Table EV4. In each Kaplan–Meier plot, wild‐type (wt) cases are plotted in blue, mutated (mut) cases are plotted in red, the log‐rank P value is given, and the hazard ratio (HR) for mutation is given with a 95% CI and after univariable and multivariable (MV) Cox regression adjustment. Covariables included in the MV analysis were age at diagnosis, lymph node status, tumor size, and the variables denoted by the following symbols: ¤, ER, PgR, and NHG; #, NHG. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; NHG, Nottingham histological grade; PgR, progesterone receptor.

Figure EV4. Impact of PTEN mutation and expression on overall survival across treatment groups

1. A–C

   Overall survival of patients with tumors containing a (A) PTEN mutation (PTEN‐Mut), (B) a PTEN mutation and/or PTEN expression in the lower cohort‐quartile (PTEN‐MutExp), or (C) PTEN expression in the lower cohort‐quartile (PTEN‐Exp) in the clinical patient subgroups HoR⁺/HER2⁻ (HoR⁺ when ER⁺ and PgR⁺, HoR⁻ otherwise; n = 2,134), HoR⁺/HER2⁺ (n = 230), HoR⁻/HER2⁺ (n = 104), and TNBC (n = 137). Specific treatments in these groups are detailed in Table EV4. In each Kaplan–Meier plot, wild‐type (wt) and normal cases are plotted in blue, mutated (mut) and low cases are plotted in red, the log‐rank P value is given, and the hazard ratio (HR) for mutation/low is given with a 95% CI and after univariable and multivariable (MV) Cox regression adjustment. Covariables included in the MV analysis were age at diagnosis, lymph node status, tumor size, and the variables denoted by the following symbols: ¤, ER, PgR, and NHG; #, NHG. ER, estrogen receptor; HoR, hormone receptor; HER2, human epidermal growth factor receptor 2; NHG, Nottingham histological grade; PgR, progesterone receptor; TNBC, triple‐negative breast cancer.

Figure 7. Impact of tumor mutational burden on overall survival across treatment groups

Overall survival stratified by tumor mutational burden (TMB) across treatment groups in 3,217 patients. Samples were classified as TMB‐high if the amount of non‐synonymous mutations per expressed MB (rnaMB) was ≥ the median number of non‐synonymous mutations per rnaMB across the whole SCAN‐B cohort (0.082 mutations per rnaMB) and TMB‐low otherwise. In each Kaplan–Meier plot, TMB‐low cases are plotted in blue, TMB‐high cases are plotted in red, the log‐rank P value is given, and the hazard ratio (HR) for TMB high is given with a 95% CI and after univariable and multivariable (MV) Cox regression adjustment. Covariables included in the MV analysis were age at diagnosis, lymph node status, tumor size, and the variables denoted by the following symbols: ¶, ER, PgR, HER2, and NHG; ¤, ER, PgR, and NHG; $, HER2 and NHG; #, NHG. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HoR, hormone receptor; NHG, Nottingham histological grade; PgR, progesterone receptor; TMB, tumor mutational burden; TNBC, triple‐negative breast cancer.

Figure EV5. Impact of tumor mutational burden on overall survival across clinical subgroups

Association of tumor mutational burden (TMB) with overall survival in 3,217 patients, and within the biomarker patient subgroups ER⁺, ER⁻, PgR⁺, PgR⁻, HER2 amplified, HER2 normal, Ki67‐high, Ki67‐low, NHG Grade 1–3, HoR⁺/HER2⁻, HoR⁺/HER2⁺, HoR⁻/HER2⁺, TNBC, and molecular subtypes Luminal A and B, HER2‐enriched, and basal‐like according to the PAM50 gene list. In each Kaplan–Meier plot, TMB‐low cases are plotted in blue, TMB‐high cases are plotted in red, the log‐rank P value is given, and the hazard ratio (HR) for TMB‐high is given with a 95% CI and after univariable and multivariable (MV) Cox regression adjustment. Covariables included in the MV analysis were age at diagnosis, lymph node status, tumor size, and the variables denoted by the following symbols: ¶, ER, PgR, HER2, and NHG; ^, ER, PgR, HER2; ¤, ER, PgR, and NHG; $, HER2 and NHG; #, NHG. ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HoR, hormone receptor; NHG, Nottingham histological grade; PgR, progesterone receptor; TMB, tumor mutational burden; TNBC, triple‐negative breast cancer.

Figure 8. The SCAN‐B MutationExplorer

The SCAN‐B MutationExplorer web‐based application for interactive exploration of mutations, and their association with clinicopathological subgroups and overall survival. As an example, generation of the image used in Fig 2 is shown.