Skip to main content
Premium Trial:

Request an Annual Quote

Rutgers, Foundation Medicine Study Finds Tumor Sequencing Dx May Misidentify Noncancerous Mutations


SAN FRANCISCO (GenomeWeb) – Tumor sequencing tests may pick up not only somatic mutations found in the tumor but also a certain class of non-cancerous mutations that accumulate in hematopoietic stem cells and are associated with age, according to a recent study published by researchers from Rutgers University and Foundation Medicine.

This phenomenon is known as clonal hematopoiesis of indeterminate potential, or CHIP, and is most common in a handful of specific genes. CHIP increases with age and is found in between 5 and 10 percent of the general population over the age of 60. However, according to research published this month in Blood, these mutations can be picked up by tumor sequencing tests and erroneously called as somatic mutations.

"When you find mutations associated with clonal hematopoiesis at low allele frequencies, those have to be tested" to confirm whether they are somatic or not, said Hossein Khiabanian, a senior author of the study and assistant professor of pathology in medical informatics at Rutgers Cancer Institute.

In the study, the team first analyzed 113,079 samples that had been sequenced using Foundation Medicine's FoundationOne assay. The researchers hypothesized that CHIP mutation prevalence would increase with age and be independent of tumor type. They looked across all samples and focused on 257 genes and found four genes with a significantly higher rate of pathogenic mutations in patients over the age of 60 — TET2, DNMT3A, ASXL1, and SF3B1. The four genes were previously known to be associated with CHIP.

In addition, they found that the frequency of mutations in those genes was lower from samples with higher tumor purity, adding evidence that the mutations may originate not in tumor cells but in hematopoietic cells.

To verify this, the researchers focused on the subset of 1,636 samples that came from patients seen at Rutgers Cancer Institute, since the researchers had more clinical data and in some cases matched normal samples to validate whether mutations were somatic or CHIP.

The researchers identified mutations in TET2 and DNMT3A that had been called as somatic mutations in 49 patients. In 36 out of 49 of those cases, the researchers found evidence that the mutations did not originate in the tumor. And, for those 36 patients, the mutations were present at a lower allele frequency than would be expected based on tumor purity. In addition, there was more evidence of tumor-infiltrating lymphocytes in those samples. The researchers also had matched normal samples for 13 patients, and in 10 they detected the same TET2 or DNMT3A mutations in normal blood that had been called in the tumor.

The study illustrates one challenge with tumor-only sequencing, said Michael Berger, associate director of the Center for Molecular Oncology at Memorial Sloan Kettering Cancer Center, who was not affiliated with the study.

Previous research has found that tumor-only sequencing can be problematic in that it does not identify germline cancer predisposition mutations, some of which are clinically actionable.

"It's important to realize that CHIP can be detected in tumor-only sequencing tests and that sometimes a hematological workup might be necessary" to verify whether the mutations are CHIP-associated or truly somatic and present in the tumor, he said.

Berger said in his experience at MSKCC with the MSK-Impact test, which analyzes 468 genes in both tumor and matched normal sample, the team has identified many samples with CHIP that would have been difficult to distinguish had they not also sequenced the matched normal sample.

"By sequencing the tumor and matched, we can identify at the time of sequencing which variants are somatic, which are germline, and which are CHIP-associated," he said.

Berger added that although labs like Foundation Medicine do a good job of distinguishing between somatic and germline mutations using a bioinformatics strategy, "clonal hematopoiesis can be more difficult to catch."

One reason for that, he said, is that the mutations are not in general population databases, since they tend to only occur in older patients. In addition, because the mutations often fall within leukemia-related genes, the mutations that are in databases are found in cancer-related databases like COSMIC, Berger said.

He noted that while there are some known genes that are frequently associated with CHIP, the MSKCC team has also identified CHIP in unexpected genes like KRAS.  For instance, he described one patient with colorectal cancer who was thought to have a KRAS mutation based on tumor-only testing. However, when the patient came to MSKCC, the team discovered that the KRAS mutation was not actually present in the tumor.

In that case, the finding had important clinical implications, since a KRAS mutation would make him ineligible for certain targeted therapies. "It's a difference between being treated with anti-EGFR therapy or not," Berger said.

Jo-Anne Vergilio, senior associate medical director and senior hematopathologist at Foundation Medicine, said that the company is now "working to determine the most helpful way to share [CHIP] findings with treating physicians." She said that the firm does not plan to do matched normal sequencing, but instead is looking into using other parameters, such as "patient age and variant allele frequency of CHIP-associated alterations in comparison to tumor purity," to discriminate between CHIP and somatic mutations.

Vergilio added that recognizing CHIP-associated mutations is important when interpreting tumor sequencing assays to "ensure that therapeutic strategies are directed against the tumor, as opposed to the non-tumor, elements."

The clinical implications of CHIP are mostly not well understood, although there have been some studies associating CHIP with cardiovascular disease. And, several studies have found that cancer patients with CHIP may be more likely to develop therapeutically induced myeloid malignancies, Khiabanian said.

"We don't know why that is, but it's something we're investigating," he said. "How CHIP evolves under therapy and why it's detected in the tumor microenvironment — those are questions we're trying to answer," he added.

Khiabanian said that this study has changed the protocol for evaluating patients at Rutgers. Now, when tumor sequencing results are reviewed by Rutgers' molecular tumor board, if there is evidence suggesting CHIP it recommends sequencing a matched blood sample to confirm.

The MSKCC team is also trying to understand the role of CHIP in cancer, and in January it launched the Precision Interception and Prevention Initiative, which includes a clinic for patients with clonal hematopoiesis. Berger said that patients are referred to the clinic based on a set of criteria that includes not only having a high level of clonal hematopoiesis, but also a particular set of mutations at a particular allele frequency. In addition, he said, the center is especially interested in evaluating cancer patients who would otherwise have a positive prognosis so that they can be more closely monitored for hematological malignancies.

"It's still unclear which subset of people harboring CHIP mutations are at a greatest risk for these diseases," he said.