By Ben Butkus
A group led by clinical researchers from the National Cancer Institute has used a high-resolution DNA melting curve analysis platform from Idaho Technology to detect somatic mutations in multiple cancer types, according to a research paper published this week.
The work demonstrates how inexpensive, moderate-throughput HRM platforms can be used to reliably perform "molecular epidemiology" studies, the goal of which is to determine the frequency of somatic mutations in cancer genes of interest ahead of more detailed and expensive sequencing studies, the study's lead author said this week.
"In the cancer field, there is a whole new world emerging, what we call somatic molecular epidemiology," Stephen Chanock, lead author on the study and an investigator at NCI's Laboratory of Translational Genomics, told PCR Insider.
"In other words, we know that mutations in cancer are very important," Chanock added. "The question is which ones are really important, and are so-called drivers? We have a lot of functional and exciting information on that, but we also need to understand the distribution of these … in populations of tumors."
To that end, Chanock and colleagues in various NCI laboratories turned to HRM curve analysis, a relatively inexpensive technique that has been used successfully for SNP typing and epigenetic studies but, according to Chanock, had never really been studied as a tool for detecting somatic mutations across various cancer cell populations.
"It looked like a suitable technology for sort of a low-throughput [analysis] of … several hundred samples in regions that we are distinctly interested in, that we know have a pretty high prevalence of somatic mutations," Chanock said.
"That's why we wanted to push that to see if we could use it for larger studies," he added. "We have these very large molecular epidemiology studies where we've collected literally hundreds to thousands of [tumor samples], and we wanted to find out whether it works with formalin-fixed material; and with fresh frozen material, which is what you need if you're going to sequence or do some of the high-definition genomics-type studies."
In their study, published online this week in PLoS One, Chanock and colleagues screened 216 fresh frozen pediatric tumor samples and 180 paraffin-embedded tumors from breast, endometrial, and ovarian cancers for somatic sequence alterations in a variety of genes using post-PCR HRM analysis.
The researchers specifically analyzed exons in candidate genes known to harbor established, commonly observed mutations. These genes included PIK3CA, ERBB2, KRAS, TP53, EGFR, BRAF, GATA3, and FGFR3.
They performed pre-HRM PCR on the exons using Taq polymerase, and Idaho Tech's LCGreen PLUS dye on an MJ Research PTC 225 thermal cycler. Next, they performed HRM curve analysis in duplicate on the resulting amplicons in 96-well plates using Idaho Tech's LightScanner instrument, LCGreen Plus+ melting dye, and custom LightScanner software.
Subsequently they performed bi-directional sequence analysis on the samples to confirm the accuracy of the HRM curve analysis.
For the 39 mutations observed in frozen samples, HRM curve analysis yielded sensitivity of 97 percent and specificity of 87 percent. Meantime, for the 67 mutations in paraffin-embedded samples, the technique yielded a sensitivity of 88 percent and specificity of 80 percent.
According to the researchers, their results indicate that HRM analysis with a LightScanner platform "is a promising screening tool for mutation/variant in somatic DNA extracted from either frozen or paraffin-embedded samples;" though they noted that overall accuracy is better in frozen specimens, a result they chalked up to the quality of extracted DNA.
"This is our first foray into whether this is something that has real utility and is worth the investment," Chanock told PCR Insider. "And it is a very interesting tool. It comes with certain caveats. But if we're going to be doing large-scale assessment of what we see in populations of 500 or 1,000 breast tumors … for a few of the places we know we should be looking, this technology is going to help tell us the distribution [of somatic mutations]."
Chanock stressed that the technique "has to be viewed as an intermediary, both in terms of study design and scope. If you just find sequence differences, you still have to go sequence, you still have to go look more closely."
As for the differing results for fresh frozen and paraffin-embedded tissue, Chanock said that it will be the subject of the group's next paper. However, in the current paper, the researchers offered some recommendations for maintaining and even enhancing the screening capabilities of HRM analysis of paraffin-embedded tissue samples.
These recommendations include increasing the amount of total genomic DNA to 30 ng or more, which may increase the HRM analysis success rate up to 96 percent; optimizing pre-HRM PCR primers to reduce GC content; and testing selected amplicons by sequencing a few samples prior to HRM analysis, since more than half of the amplicons that failed sequencing also performed poorly during HRM analysis.
Chanock said that other companies offer HRM analysis platforms, but that the group worked with the Idaho Tech platform because of "the ease and speed with which you can do what I call low-throughput screening."
Have topics you'd like to see covered in PCR Insider? Contact the editor at bbutkus [at] genomeweb [.] com.