Skip to main content
Premium Trial:

Request an Annual Quote

MSK Study Suggests Germline Tests May Miss Mosaic Cancer Risk Mutations in Some Patients


NEW YORK – A small subset of unselected cancer patients may carry risky germline variants in some, but not all, of their cells — cancer susceptibility mosaicism that may lead to missed pathogenic or likely pathogenic variants on available germline genetic tests, according to research presented Tuesday at the Association for Molecular Pathology virtual annual meeting.

Diana Mandelker, director of Memorial Sloan Kettering Cancer Center's diagnostic molecular genetics lab, shared insights from her team's analysis of 35,311 cancer cases that were molecularly profiled using MSK-IMPACT — a hybrid capture-sequencing and analysis pipeline that interrogates more than 500 cancer-related genes, including 62 genes containing moderate- to high-penetrance cancer susceptibility variants.

Because the pipeline produces insights for patients' tumors as well as their matched normal germline samples, Mandelker noted, it was possible to dig into MSK-IMPACT data to search for mosaicism.

"The only way to really differentiate these mosaic variants is tumor-normal sequencing, which is essential to enhance our sensitivity for detecting these mosaic variants in our cancer patients," she emphasized.

Broadly speaking, mosaic germline mutations can arise when a mutation crops up in a single cell at an early stage of development, she explained, imbuing all of the cells that eventually develop from the altered cell with the same attributes.

"In broad terms, mosaicism is defined as the presence of a variant in a subset of an individual's cells … when a mutation occurs early in development," Mandelker said. "So the mutation was not present in the parents, but occurs during embryogenesis and, as such, the variant is present in a subset of an individual's cell lineages."

She pointed out that past research has uncovered mosaicism in a significant subset of individuals diagnosed with cancer susceptibility syndromes such as retinoblastoma, Li Fraumeni syndrome, neurofibromatosis, or tuberous sclerosis, despite a marked lack of related family histories, for example.

But less is known about the prevalence of mosaic risk variants in unselected cancer patients who do not necessarily meet the criteria for germline testing, Mandelker said, prompting the investigators to take a closer look at variant allelic fractions for dozens of cancer-related genes in in tumor and blood samples profiled with MSK-IMPACT.

Applying bioinformatics to the paired data made it possible to not only distinguish between germline and somatic variants, she explained, but also revealed germline variants that were amplified in the tumor through loss-of-heterozygosity events that wipe out the wild type version of the gene, as well as clonal hematopoiesis events in which somatic variants affect stem cells in the blood lineage.

Mosaic mutations may appear at low frequency in the blood and much higher frequency in the tumor, on the other hand. The team focused in on those alterations by filtering for variants in dozens of cancer susceptibility genes that were enriched in the tumor but were found at relatively low variant allele frequencies — between 1.5 percent and 25 percent — in the blood.

"We had a long list of various quality control metrics to discriminate these candidate mosaic variants from potential sequencing artifacts, circulating tumor cells, et cetera," Mandelker noted.

Using this strategy, the researchers narrowed in on 34 candidate germline pathogenic mosaic variants in cancer susceptibility genes. Nearly half of those appeared in the TP53 gene, though they also highlighted potential mosaicism for variants in RB1, BRCA2, APC, and other genes. And some 85 percent of the potential mosaic variants had undergone loss-of-heterozygosity or a second somatic mutation in the corresponding tumor samples, Mandelker reported.

From there, the team took a closer look at potential mosaic variants in tumor and normal samples from 10 individuals, who had sufficient tissue samples available, using microdissection to tease out normal samples with different histological features.

Those samples were assessed with MSK-IMPACT targeted sequencing, laser capture microdissection, and amplification sequencing to untangle the frequency of suspicious mutations in the blood, which ranged from 1.7 percent to 21 percent. And researchers were able to validate the suspected mosaic mutations in all 10 cases.

In a 57-year-old woman who had been diagnosed with bilateral breast cancer at the age of 37, and later developed a sarcoma that was sequenced on MSK-IMPACT, for example, the researchers identified a pathogenic mutation in MSH6 that turned up with 10 percent frequency in the blood, but 43 percent frequency in her sarcoma tumor sample.

They saw the same MSH6 change in normal lymph node, dermis, epidermis, and skeletal muscle samples from the same individual, at frequencies that varied by tissue type, though those samples did not contain the somatic mutations present in the tumor.

Likewise, in the case of an 84-year-old woman with a microsatellite instability-high ovarian tumor, the team uncovered a risky MSH2 frameshift mutation with 39 percent frequency in the tumor and just 3 percent frequency in the blood. That change, too, occurred as a germline variant in the patient's endometrium, fallopian tube, smooth muscle, and epithelial samples at a range of low frequencies, while a second, distinct MSH2 mutation was found in the tumor alone.

In these and other cases, the mosaic variants identified typically corresponded with the cancer type or types that each individual eventually developed, Mandelker said, although there were hints that some of the tumors may have developed slightly later than they would, on average, in individuals with full-blown cancer syndromes.

The study suggested that such mosaicism may influence the results that individuals receive when they go for cancer susceptibility testing as well.

Mandelker noted that around half of patients with mosaic cancer susceptibility variants met criteria for testing, even though most did not have a family history of cancer. Indeed, 21 of the 34 patients with mosaic risk variants had already gotten genetic testing, but only five individuals had mosaic germline risk variants identified with such testing, all in the retinoblastoma-related RB1 gene. The remaining 16 got negative test results, likely because germline pathogenic variants were found at levels too low to detect in the blood.

"Some seemingly sporadic cancers can be related to mosaic mutations in cancer susceptibility genes," Mandelker concluded. "In our cohort, that was at a frequency of about one in 1,000 cases."