Mutations in tumor-adjacent tissues could reveal earliest events leading to malignancy

MD Anderson researchers have identified early genomic events that may lead to malignant transformation. They’ve found that tissue adjacent to tumors that appears to be pathologically normal may harbor cells with mutations that could lead to cancer.

“These findings provide important insights to the earliest changes that lead to malignancy and could help with early detection of cancer or even prevention,” says Paul Scheet, Ph.D., professor and chair of the Department of Epidemiology at The University of Texas MD Anderson Cancer Center

Mosaicism refers to these subpopulations of cells with somatic mutations. One type of mutation is somatic copy number alterations (sCNAs), in which a cell has only one copy of the gene (the maternal or paternal allele was deleted) or three copies of the gene (the maternal or paternal allele was duplicated).

“These large chromosomal alterations, which span many genes, are not compatible with normal development and likely confer a proliferative advantage for these clones, representing the earliest genomic alterations on the path to malignancy,” says Scheet.

Scheet is senior author of an article recently published in Nature Biotechnology that investigates sCNAs in matched samples of tumors and tissue adjacent to tumors from the Cancer Genome Atlas (TCGA), a National Cancer Institute resource that has molecularly characterized over 20,000 samples of 33 different types of cancer. 

Oncogenic patterns of somatic copy number alterations

Because normal tissue has a lower proportion of cells with mutations compared with tumor tissue, the research team used a haplotype-based approach (looking at a group of genes inherited together from a single parent) in the hapLOH algorithm, previously developed by Scheet’s research team, to identify sCNAs present at cell fractions of less than 10%. The hapLOH algorithm identifies statistically significant allelic imbalances (more copies of a chromosome from one parent instead of the expected 1:1 ratio).

The sCNAs were categorized as a gain, a loss, a copy-neutral loss of heterozygosity (i.e., the cell has two copies of the genes from one parent and no copies from the other parent), or undetermined. This allowed the team to study sCNA patterns across different types of cancer and from different samples within an individual patient.

The team found that tumor-adjacent tissue samples had significantly more sCNAs than blood samples. For cancer sites with 10 or more tumor-adjacent tissue samples, tissues adjacent to head and neck squamous cell carcinoma (HNSC; 25%) and bladder urothelial carcinoma (18%) had the highest rates of sCNAs.

The team identified similar oncogenic patterns of chromosomal alterations in tumor and tumor-adjacent tissues. For example, both blood and tumor-adjacent tissue samples had gains in chromosomes 8 and 12, which are common across cancers. Gains of 1q, which have been shown in epithelial tumors, were also seen in tissues adjacent to breast invasive carcinoma and lung adenocarcinoma. Co-occurring gains of 8q, 13q, and chromosome 20 have been reported in esophageal carcinoma, rectum adenocarcinoma, and stomach adenocarcinoma; tissue samples adjacent to these sites had at least one of these gains (and sometimes all three). sCNAs in HNSC-adjacent tissue samples were associated with mutations in established HNSC-promoting genes.

Among the tumor-adjacent tissue samples that had overlapping sCNAs with the tumor tissue, the research team found instances of independent mutation. Of samples with overlapping sCNAs, 21% (19/70) of blood samples and 9% (23/247) of tumor-adjacent tissue samples showed a mirrored allelic imbalance with the matching tumor tissue (for example, one shows a gain in maternal alleles, and the other shows a gain in paternal alleles).

“This finding suggests that these mutations originated independently in the same organ and that these genes are particularly important for the cell population,” said Yasminka Jakubek, Ph.D., instructor in the Department of Epidemiology at MD Anderson and lead author of the article.

Among these tumor-adjacent tissue samples, independent mutations included gains of established oncogenes: H3F3A in lung adenocarcinoma; CARD11, EGFR, and JAK2 in HNSC; and FLT3 in stomach adenocarcinoma. One patient with lung squamous cell carcinoma showed two independent losses of APC, which has been linked to colorectal cancer.

Future cancer genetics research

The findings from this study will affect how other cancer researchers interpret TCGA data, a heavily utilized resource. Most cancer researchers who use TCGA contrast gene expression changes in a patient’s tumor tissue sample with the matching tumor-adjacent tissue sample, which is assumed to be normal.

However, as Scheet points out, “There are degrees of normal. Our study shows that this pathologically ‘normal’ tissue may actually have oncogenic phenotypes that will eventually result in malignant transformation.”

This finding also could affect criteria for sample collection. “To provide tissue samples that are on the more normal side of the continuum, researchers could consider obtaining matched tissue samples that are not adjacent to tumors but are from farther away or in other tissues,” says  Jakubek.

The team’s next steps include identifying other scenarios with tumor-adjacent tissue samples to continue to explore the rates of sCNAs and their prognostic value.