Despite their utility as diagnostic tools for cancer, microRNA expression profiles can vary depending on the area of a tumor analyzed, raising the possibility that result from a single biopsy could yield misleading results, according to a study published last week in The Journal of Molecular Diagnostics.
“Personalized medicine will be the future of cancer care,” Stefanie Avril, a researcher at the Technical University of Munich and lead author of the study, said in a statement. “However, taking a single tumor biopsy for defining individual treatment is probably oversimplified, and we need to take into account the heterogeneity of tumors.”
According to Avril, the "vast morphological and architectural heterogeneity” of tumors, as well as the heterogeneity in mutational analysis and gene expression, has long been known. Yet when it comes to many of the existing studies examining miRNA expression in cancers, the extent of this variation has not been fully taken into account.
Focusing on breast cancer, Avril said that there is “quite little overlap between different studies looking at microRNA signatures,” noting, for instance, that miRNAs previously associated with hormone receptor status “vary quite a bit between different studies.”
To explore this issue, she and her colleagues prospectively collected 132 tumor samples from 16 patients who had undergone either lumpectomy or mastectomy for large primary invasive breast cancer. Eight to 10 samples were taken from different regions of the tumor, including the central, intermediate, and peripheral zones, she told Gene Silencing News.
The investigators also sampled different tumor-involved lymph nodes if metastases were present.
“Then we looked at four candidate microRNAs that have previously been associated with metastatic potential in breast cancer,” namely miR-31, miR-335, miR-10b, and miR-210, she said.
When the expression levels of the miRNAs were evaluated, the team found “significant intratumoral heterogeneity” within the primary tumors, Avril explained.
Given as the co-efficient of variation, the extent of miRNA heterogeneity in a single patient was about 40 percent within the primary tumor or different lymph node metastases, she said. Variation between patients was about 80 percent for primary tumors and 103 percent for lymph node metastases.
“That already gives a hint that the amount of heterogeneity within one single tumor from one patient is already so high compared to the variation between patients that this might lead to bias if we are comparing samples from different patients or … different tumor subtypes,” Avril said.
She noted that the researchers also found that the level of variation and heterogeneity within a defined tumor zones, as well as between different zones, was the same.
“Intuitively, one might think that if you just look at different samples from the tumor periphery, they might be more like each other than if you compare center and periphery,” samples, Avril said. “But that is not the case, [which] suggests that we cannot avoid sampling bias, for example, by specifically sampling a particular tumor zone.”
In the paper, the team noted several limitations to their work, including the number of cases and individual tissue samples.
“An arbitrary cutoff was set at [greater than] 70 percent tumor cell content to avoid substantial contamination from non-tumor tissue,” they wrote. Additionally, “we analyzed macroscopic heterogeneity, whereas differences on a cellular level were not assessed, as miRNA detection by in situ hybridization poses its own technical challenges and was beyond the scope of this investigation.”
Lastly, the study only assessed tumors 3 centimeters or greater in diameter. “Our data shows no indication that the extent of intratumoral heterogeneity is dependent on the tumor diameter, although it is possible that the extent of intratumoral heterogeneity may be different for smaller tumors,” the team noted.
In the end, the findings suggest that for miRNA profiling studies, “we should take more than just a single sample from each tumor,” she explained. “Ideally, [it should be] two to three samples from the primary tumor, and several tumor-involved lymph nodes if we want to look at metastatic profiles.
Still, the effort benefited from the “systematic and predefined prospective sampling of the tumors in eight to 10 different areas, whereas previous studies assessing intratumoral heterogeneity of biomarkers have commonly only analyzed different areas of one tumor,” the authors wrote.
Overall, researchers “very often only look at single tumor samples or even core needle biopsies to predict response to a certain treatment or find out the best response to a certain treatment for a patient,” Avril said. “We can certainly improve our performance and our predictions by looking at a broader part of the tumor.
“The main point is to spread the sampling as widely as possible."