A Genomic Analysis of Esophageal Squamous Cell Carcinoma in Eastern Africa

Abstract

Background: Esophageal squamous cell carcinoma (ESCC) comprises 90% of all esophageal cancer cases globally and is the most common histology in low-resource settings. Eastern Africa has a disproportionately high incidence of ESCC.

Methods: We describe the genomic profiles of 61 ESCC cases from Tanzania and compare them to profiles from an existing cohort of ESCC cases from Malawi. We also provide a comparison to ESCC tumors in The Cancer Genome Atlas (TCGA).

Results: We observed substantial transcriptional overlap with other squamous histologies via comparison with TCGA PanCan dataset. DNA analysis revealed known mutational patterns, both genome-wide as well as in genes known to be commonly mutated in ESCC. TP53 mutations were the most common somatic mutation in tumors from both Tanzania and Malawi but were detected at lower frequencies than previously reported in ESCC cases from other settings. In a combined analysis, two unique transcriptional clusters were identified: a proliferative/epithelial cluster and an invasive/migrative/mesenchymal cluster. Mutational signature analysis of the Tanzanian cohort revealed common signatures associated with aging and cytidine deaminase activity (APOBEC) and an absence of signature 29, which was previously reported in the Malawi cohort.

Conclusions: This study defines the molecular characteristics of ESCC in Tanzania, and enriches the Eastern African dataset, with findings of overall similarities but also some heterogeneity across two unique sites.

Impact: Despite a high burden of ESCC in Eastern Africa, investigations into the genomics in this region are nascent. This represents the largest comprehensive genomic analysis ESCC from sub-Saharan Africa to date.

Figure 1.. Workflow for ESCC specimens from Tanzania.

Stepwise progression from recruitment of participants, specimen fixation, shipment, DNA and RNA extraction, and molecular analyses.

Figure 2.. Summary of recurrently mutated genes.

Each column represents one patient sample, and columns are ordered according to gene mutation frequency. The overall frequencies of synonymous and non-synonymous mutations are summarized at the top of the figure. Percentage of the cohort harboring a mutation(s) in a given gene is depicted at the bottom left of each figure. (2A) Somatic point mutations in the Tanzania cohort (n=61). (2B) Somatic point mutations in the Malawi cohort (n=59). (2C) Significantly mutated genes in the Tanzania and Malawi cohorts.

Figure 3.. TP53 mutation rate and rescue in Tanzania and Malawi cases.

For cases from Tanzania, RNA and DNA Integrated Analysis (RADIA) was used to improve somatic mutation detection by analyzing the patient matched normal and tumor DNA along with the tumor RNA-Seq data to identify “RNA Rescue” mutations. The plot along the bottom of the figure shows all positions at which somatic variants in TP53 were detected.

Figure 4.. COSMIC Signatures in the Tanzania cohort.

Mutational signature frequencies in specimens from Tanzania, ordered by frequency. COSMIC mutational signature 5 was the most common, consistent with other reports in squamous cell cancers.

Figure 5.. Transcriptomic analysis of esophageal squamous cell carcinoma from Tanzania and Malawi.

(5A.) t-SNE projection of pan-cancer RNA-Seq data, showing that both the Tanzania and Malawi datasets cluster with other squamous tumors from a combined cohort of clinical and TCGA. (5B.) Heatmap of expression levels of six different TP63 isoforms, three Delta and three TA, in which columns represent individual samples in Tanzania and Malawi datasets and rows represent individual isoforms. (5C.) Final clustering solution (k=2), based on expression of the 3,000 most varying genes across combined Tanzania and Malawi datasets. (5D.) KEGG pathways from Gene Set Enrichment Analysis results for cluster 1 versus cluster 2 contrast, based on whole transcriptome.