Skip to main content
Premium Trial:

Request an Annual Quote

Genomic Study Urges Caution in Using Mutational Results from Single Cancer Biopsies to Guide Treatment

Premium

By Julia Karow

Tumors appear to be more heterogeneous than previously thought in terms of the mutations they carry, so relying on a single biopsy to make treatment decisions based on mutational analysis could be problematic, according to a team of researchers led by the Cancer Research UK London Research Institute.

The scientists, who stress that their findings are preliminary because they only studied two patients in detail, analyzed several areas of kidney tumors as well as metastases for mutations and found that about two thirds of the somatic mutations they found were not detectable in every region of the tumor. They published their results this week in the New England Journal of Medicine.

According to Charles Swanton, a professor of cancer medicine at the Cancer Research UK London Research Institute and the senior author of the paper, the study served to answer a fundamental question: How reliable is a single tumor biopsy for portraying the somatic mutational landscape of a tumor?

While it was known that mutations vary across tumor sites, the high percentage of heterogeneous somatic mutations identified in this study came as a surprise, Swanton said, adding that the results call for caution when it comes to using genomic biomarkers to make treatment decisions.

"If you are relying on a single biopsy to stratify a patient for a clinical trial, and two-thirds of the mutations are heterogeneous, you are potentially not going to be treating the whole tumor," he explained, "so that approach, from a clinical perspective, is flawed."

"I think it does raise questions about how we use next-gen sequencing data going forward in a clinical setting."

Marilyn Li, director of the Cancer Genetics Laboratory in the Department of Molecular and Human Genetics at Baylor College of Medicine, agreed.

While the findings of tumor heterogeneity were "not surprising," she noted in e-mail comments that the study "demonstrates that we may need to run targeted sequencing-based cancer testing on tumor biopsies of multiple sites and at multiple stages (such as at diagnosis, during and after treatment, at relapse and metastasis) of the disease."

For their study, the researchers performed exome sequencing on several regions of renal cell carcinoma primary tumors as well as metastatic sites for two patients, using Illumina's Genome Analyzer IIx and HiSeq platforms. In addition, they genotyped the samples using Illumina SNP arrays and profiled their gene expression using Affymetrix arrays. They also performed ploidy and allelic-imbalance profiling of primary tumors from a total of four patients.

In one patient, they performed exome sequencing for two pre-treatment biopsy samples — of the primary tumor and a metastasis — as well as for nine regions of the surgically removed primary tumor post-treatment, and for two additional metastases.

Overall, they found about 130 somatic mutations, both SNPs and indels, of which 40 were ubiquitous, about 60 were shared by some regions, and about 30 were unique to a single region. Among the mutations were a number of genes previously associated with clear-cell carcinoma.

Only a third of the total mutations were present in all the regions of the tumor, suggesting that a single biopsy is not representative of the mutational landscape of the entire tumor.

Based on their results, the researchers constructed a phylogenetic tree of the tumor regions, which showed one branch that eventually evolved into the metastatic sites, and several other branches that remained part of the primary tumor.

They also determined the expression of a 110-gene signature that has been used to classify clear-cell carcinoma into subgroups with good and poor prognosis, and found both subgroups to be present in the same tumor. This suggests that outcomes may not be predicted correctly if they are based on a single tumor biopsy.

To confirm their results, the researchers conducted exome sequencing on a second patient, focusing on nine regions of the primary tumor and one metastasis. They detected about 120 somatic mutations in total, and similar to the first patient, only about a third of those were detected in all regions of the tumor.

The results suggest that clinicians won't be able to rely on a single sample from a tumor, or a single metastasis, to make reliable predictions about the tumor's behavior. "The picture is much more complicated than we thought it would be," Swanton said.

It will now be important, he said, to study what types of mutations occur consistently in all regions of a tumor — so-called truncal mutations — and what role the heterogeneous mutations, or branch mutations, play in terms of fostering metastasis and drug resistance.

To do so, Swanton and his team are now extending their analysis to a larger number of renal cancers, initially up to 10 more.

Although his team has focused on kidney cancer, the results will likely apply to other cancer types as well, he said. Recently, for example, researchers have shown similar heterogeneity for medulloblastoma, glioblastoma, leukemia, and breast cancer.

In the meantime, clinicians need to be careful in relying on genomic biomarkers from single biopsies for treatment decisions. Established biomarkers, such as BRAF, HER2 or EGFR, "clearly function very well," he said, but they are "the low-hanging fruit" — early drivers of the disease biology that are probably present in all tumor sites. "What our work refers to is driver mutations that occur in individual biopsies that may not be such good drug targets," he said.

Basing treatment decisions on mutational results from single biopsies "requires a certain amount of caution at the moment, until we understand more about the evolution and the development and clonal architecture of individual tumors to say categorically that this is the right treatment for this patient," he said.


Have topics you'd like to see covered in Clinical Sequencing News? Contact the editor at jkarow [at] genomeweb [.] com.