Skip to main content
Premium Trial:

Request an Annual Quote

Study Modeling Clinical WGS for Inherited Cancer Testing Illustrates Benefits, Complications


NEW YORK (GenomeWeb) – Anticipating a future in which whole-genome sequencing becomes a routine part of clinical diagnosis in inherited cancer syndromes, researchers from the University of Texas Southwestern have published a study examining the impact of WGS on the detection of BRCA and other cancer susceptibility mutations.

The results suggest that while WGS clearly adds information above and beyond single-gene or panel tests, the breadth of potential pathogenic variants it uncovers raises questions about how to ensure efficient and accurate interpretation in the context of numerous variants of unknown significance.

Theodora Ross, the study's senior author and director of UT Southwestern's Cancer Genetics Program, told GenomeWeb in an email that the group's intention for the study, called "Project Cancer Family," was to collect empirical evidence of what a hypothetical future cancer genetics clinic would look like "if every patient with a family history of cancer was analyzed with a whole-genome sequence, rather than single-gene or panel testing."

The study, published last week in the journal EBioMedicine, involved several different analyses. Overall, the group examined data from whole genome sequencing, performed by Complete Genomics, for 176 known-BRCA1/2 mutation carriers, and 82 other patients from the UT Southwestern cancer genetics clinic known not to carry a pathogenic BRCA mutation based on testing performed by Myriad Genetics.

As a first evaluation, the researchers looked at sequence data for only the BRCA1/2 mutation carriers and compared the WGS results to determine if they confirmed or conflicted with patients' known, clinically diagnosed mutations.

WGS correctly detected 89.3 percent of BRCA1 mutations and 88.6 of the BRCA2 mutations in this cohort. While WGS also picked up the remaining 11 or 12 percent of mutations, it did not reach a high enough technical quality for clinical reporting. Ross said this was due mainly to the fact that the sequencing performed was research-grade. With more advanced clinical grade sequencing, the concordance would be much higher.

"Based on current [clinical grade] WGS technical quality, there will be little difference in detection of BRCA mutations between WGS and single-gene testing. Since we are not smart enough to know what gene to test for in all of our patients, WGS wins," Ross said in her email.

Indeed, in a second set of evaluations using a subset of the WGS data that covered 163 clinically-relevant genes, Ross and her colleagues were able to detect additional potentially pathogenic gene mutations, both in the non-BRCA1/2 patients and in the BRCA1/2 cohort.

"To our surprise, even in the BRCA patients we found second and third loss-of-function mutations in other known cancer predisposition genes. This supports the idea that we probably don't know enough to justify targeted sequencing [over a broader whole-genome approach,]" Ross wrote.

In BRCA1/2 patients, the researchers identified a total of 1,207 potentially pathogenic variants, or PPVs — defined as non-synonymous variants reaching a low allele frequency threshold  — which averaged out to about seven per patient.

When the team narrowed this field to only loss-of-function variants, the number dropped dramatically, highlighting the fact that the majority of non-synonymous variants identified by WGS in the cancer(?) genetics clinic are likely to be variants of unknown significance.

"I think that this illustrates the challenges that will come with relearning what is pathogenic and what is not," Ross wrote.

For example, Ross and her team found that one identified PPV, present in five different subjects in the BRCA1/2 cohort, was a missense variant in the gene CREBBP, previously reported as pathogenic and diagnostic for a disease called Rubinstein-Taybi Syndrome. In at least one patient with available family history and family member samples, it was clear that neither the patient nor her family members also carrying the variant had any symptoms consistent with the disorder.

Similarly, another missense variant in the gene PRSS1, which is considered pathogenic by clinical labs and is associated with hereditary pancreatic cancer, showed up in nine individuals in the cohort, none of whom reported any pancreatic problems. Further investigation found that one subject's family history, a father who died of pancreatic cancer at age 53, supported the WGS result.

According to the authors, these examples highlight the need for future clinical WGS analyses addressing missense variants to involve detailed medical histories. Moreover, future efforts will also need to consider previously reported disease associations with caution.

"We think we know and then new information comes along and we find out we don't know," Ross said in her email. "The new genetic tools represented by WGS are giving us new information … so it is not the interpretation of WGS per se that is the challenge but the fact that because we can now more fully genotype our patients, some of the old genetic associations from small numbers of families will fall by the wayside. This is going to be a process of unlearning what we learned in the past and hopefully learning from more data from more patients to improve our understanding of genetic predisposition to disease," she said.

In the study, Ross and her team also looked at the same subset of 163 genes in their non-BRCA patient cohort, using the pattern of PPVs in the BRCA1/2 patients as a method for ruling out likely non-pathogenic variants in the other group.

"We thought that since we knew what was pathogenic in the BRCA cohort, we could use that as the control population for what is, and what is not, pathogenic," Ross explained. "If the gene with PPVs was found in the BRCA cohort, then we thought it unlikely to be the cause of cancer … But if the gene with PPVs was only in the non-BRCA cohort, it was more likely to be pathogenic."

Interestingly, the group found that many subjects in the non-BRCA cohort had PPVs in either BRCA1 or BRCA2 — 6 percent and 20 percent, respectively.

None of these were loss-of-function variants, compared to the BRCA1/2 cohort, in which more than 65 percent of BRCA PPVs were LoF.

When the team restricted their WGS analysis to only LoF variants, as a way of simplifying interpretation and increasing the likelihood of variants actually being pathogenic, they found that there were LoF PPVs in 14 genes in 18 of the 82 non-BRCA cases. Importantly, 11 of these variants were in genes implicated specifically in the individual's primary cancer diagnosis. Two also provided a likely genetic diagnosis based on family history.

There were also six individuals in the BRCA1/2 cohort who had LoF PPVs in cancer-associated genes other than BRCA, illustrating the added value of WGS over the methods initially used to determine these patients' clinical BRCA status.

Finally, the researchers expanded their analysis beyond the subset of 163 genes they started with, to all genes annotated in the ClinVar database. According to the authors, in addition to confirming mutations diagnosed with standard clinical tests, variant analysis of the ClinVar gene set identified LoF PPVs in previously unsuspected cancer-associated genes not considered as part of current standard clinical care. Overall, WGS yielded likely genetic cancer risk PPVs in 20 percent of the non-BRCA group, the authors wrote.

Moving forward, Ross said that the team's next goal is to expand this approach, hoping if possible to sequence every patient that comes into the UT Southwestern cancer genetics clinic.

"We see more than 2,000 patients a year … [and] we would love to consent them for a parallel whole-genome sequence together with their clinical tests … We don’t have [funding] for this yet, but there's hope we'll come up with it from somewhere," she wrote.

Ross added that the researchers also plan to put data from the current study into the dbGaP public data base so that other investigators can do their own analyses.

They are also continuing to use WGS and other technologies to pursue a diagnosis for several "mystery families" with cancer susceptibility syndromes that have thus far evaded efforts to determine their cause.