NEW YORK (GenomeWeb) — The tumor microenvironment can have a significant effect on genomic analysis of cancerous tissue sample, according to a new study by researchers at the University of California, San Francisco, and Stanford.
Specifically, the purity of a tumor sample, or lack thereof, can skew tests performed by researchers and clinicians, and could confound targeted therapy development, the researchers wrote in a paper published this week in Nature Communications.
Using data from the Cancer Genome Atlas, the team analyzed more than 10,000 tissue samples across 21 cancer types, stratifying the patients by the clinical features of their disease. They obtained gene expression profiles, DNA methylation profiles, and immunohistochemistry (IHC) data, and used four different methods to determine the purity of each sample: ESTIMATE, which uses gene expression profiles of 141 immune genes and 141 stromal genes; ABSOLUTE, which uses somatic copy number data; Leukocytes UnMethylation for Purity (LUMP); and IHC.
The researchers then performed the three most common genomic analyses used in cancer research — correlation, clustering, and differential analysis — on all the samples, to determine whether the variable purity of each sample was having a confounding effect on the results of the analyses.
For example, correlation analysis of bladder carcinoma samples identified high expression levels of the genes JAK3 and CSFR1. But the researchers found that the co-expression of these two genes varied when the purity of the samples differed. When the samples were at their purest, the correlation between the two genes dropped. When the sample included more tissue from the tumor microenvironment, the co-expression rose. Therefore, therapies for bladder carcinoma targeted at the pathway of these two genes may, in fact, be less effective than they would seem.
And in clustering analysis, the team found that variable tumor purity in lung, kidney, and thyroid cancer samples could yield misleading results on the relative expression of CTLA-4 and CD86, which are both targets of immunotherapy.
"Tumor purity is a big problem when you're dealing with fresh tissue from real patients rather than with cell lines, and there has been no systematic analysis of this issue," said UCSF researcher and first author Dvir Aran in a statement. "In the case of immunotherapy, it's an expensive treatment and it can have side effects, so it's important to know which patients are most likely to benefit. If we pay more attention to the immune cells that are actually in tumors we may have more success."
Even the measurement of a sample's mutational burden could vary according to the sample's purity, the team found. Samples with a higher content of normal and immune cells showed a higher mutational burden, whereas samples with higher purity showed a lower overall mutational burden.
"Mutational burden is a useful measure, because it identifies genes and pathways that may lead tumors to respond to conventional targeted drugs," UCSF researcher Marina Sirota said in a statement. "But if it is the greater infiltration of immune cells in a tumor that makes it more sensitive to immunotherapy, we should try to measure that directly as well."
This study shows the importance of understanding a tumor's molecular properties, the team wrote, adding, "If differences between samples from the same cancer type can be attributed to the methods and skills of the sample collector, purity may be a major confounder in all ‘omics’ analyses, and should be accounted for in these studies. Alternatively, if differences are intrinsic to the samples, it is still important to determine which results in molecular analyses are related to differences in overall purity and which are not…. False interpretations in these analyses due to divergent tumor purity levels may have a negative effect on our understanding of cancer biology and on our ability to create treatments."