A study published this week has shown methylation status in a panel of genes may be useful to non-invasively monitor breast cancer recurrence and treatment response. The two-step multiplex methylation-specific PCR assay, called cMethDNA, detected cell-free cancer DNA in small volumes of sera with a sensitivity and specificity above 90 percent.
In a pilot test, cMethDNA also detected changes in methylation levels indicating recurrence well before traditional imaging methods. The test, which was developed by researchers at the Johns Hopkins Kimmel Cancer Center, was novel in that it employed a unique technique to normalize methylation measurement in cfDNA.
The study was published in Cancer Research. It describes a panel selected from a genome-wide search of genes found to be methylated in serum of breast cancer patients. The panel included three previously described markers — ARHGEF7, TMEFF2, and RASSF1 — as well as seven novel markers — AKR1B1, COL6A2, GPX7, HIST1H3C, HOXB4, RASGRF2, and TM6SF1.
In an interview with PCR Insider, Saraswati Sukumar, lead researcher on the project and co-director of the Breast Cancer Program at the Johns Hopkins Kimmel Cancer Center, was cautiously enthusiastic about the study results. She anticipates that vastly expanding sample sizes may decrease the greater than 90 percent specificity the researchers have observed, but preliminary results suggest no more than a 15 percent decrease.
Initially, Sukumar said her group tried to compare methylated cfDNA to unmethylated levels. However, this is challenging in recurrent metastatic breast cancer, she said. "These are small regions of tumor, they're not large metastatic regions, so we realized that using the unmethylated DNA was becoming impractical in the sense that we wouldn't get reproducible results if the values were very small, because the dynamic range was very low."
Instead, the group tried a different approach. "We decided the best way to do this was to spike the serum right at the beginning; as soon as you take the serum, you spike it with physiological levels of relevant [control] DNA," she said.
The normalization DNA, akin to a spike-in control, consisted of one central portion of phage DNA with 3' and 5' arms corresponding to CpG-free regions from each of the genes in the panel.
Spiking in 50 copies of this construct "tends to give us a fantastic dynamic range, reproducibility, and ability to detect a very low number of copies ... floating in 300 microliters of the serum," Sukumar said. She also added that, as far as she knows, this method is novel to her field of study.
For sample prep, the study compared different commercially available kits to purify cfDNA: Qiagen's QiaAmp MinElute Virus Spin Kit, QiaAmp UltraSens Virus Kit, and Quick-gDNA MiniPrep from Zymo Research. "This was a really unbiased, very in-depth analysis of the three kits," Sukumar said.
The QiaAmp MinElute was found to have the tightest coefficients of variation, and was then used throughout the study. "At the lower concentrations, like 50 copies, we have a slightly higher, maybe 20 percent CV, but as [the] number of copies [increases] it's like 3 percent, which is very impressive," Sukumar said.
For quantification of methylated cfDNA, the first step was a multiplexed PCR for all genes in the panel. The amplification used primers designed for the CpG-free area of the cfDNA fragments.
This was followed by real-time quantitative methylation-specific PCR for each gene and its recombinant control. "By using two different fluorophores, we are able to measure the amplification of the test DNA relative to the spiked DNA," Sukumar said.
The study itself had two parts. The group first tested the panel on a training set of samples from 24 cancer patients and 28 controls. This was followed by a test set of 27 control and 33 recurrent Stage 4 cancers. The panel detected cancer-specific methylated cfDNA with a sensitivity of 91 percent and a specificity of 96 percent. Additionally, in a pilot study of 29 patients, the assay faithfully reflected patient response to chemotherapy.
Sukumar is now comparing the assay in a head-to-head study against CTC, CA15-3, and CEA, three currently available tests for following patients who have metastatic breast cancer, she said. "I feel this will be extremely valuable because it will tell you that in the exact same samples, how do these three different tests behave ... is it possible that there is power in one or two tests put together that would give you the most accurate readings as far as the patient's response to therapy is concerned," Sukumar said. This study is now in the data analysis stages.
In addition, a probing of The Cancer Genome Atlas has indicated the cMethDNA panel could be used for other types of cancers. Sukumar said the panel includes genes shown by TCGA data analysis to be differentially methylated in colorectal and lung cancer, but not in other cancers. "These tumors have very high methylation for the exact same genes we have used for breast [cancer] detection," she said. By TCGA query, patients also appear to show a pattern of low methylation levels in non-cancerous tissues, which can help provide the test with better dynamic range.
Sukumar is currently seeking collaborators who study these types of cancers and can provide serum. "Right now [the TCGA-based correlation] looks terrific, but unless we study the serum ... we can't be sure," she said. "All we are looking for is 300 microliters of serum, which is just hardly a teaspoonful ... this is a very low requirement in order to do the test."
Sukumar's group has a patent on the cMethDNA assay through Johns Hopkins. She hopes to out-license the assay so it can be developed into a breast cancer diagnostic. "We would love to do that, but we haven't had any takers so far. We are talking to a number of people about possible interest, and we've had a number of meetings with different people, but I don't see anything coming to fruition as yet," she said.
She noted that her group has just completed a study of 140 patients who were in a clinical prospective trial. They collected subjects' blood at baseline and multiple intervals during their treatment, "and we know how they responded to the treatment," Sukumar said.
Much of the direct funding to develop cMethDNA came from Avon. "We've been trying this test for 10 years or more, but they have been steady funders, keeping hope in us, keeping faith in us," Sukumar said.
In an email statement to PCR Insider, the foundation said that it was "thrilled that the Johns Hopkins 10-gene test is sensitive and specific in detecting breast cancer recurrence early."
"The Avon Foundation has provided approximately $35 million to 70 projects developing new blood tests to assess breast cancer risk and assess recurrence. Dr. Sukumar and her colleagues' work looks very promising and is one of the most advanced projects in our portfolio of projects in this area," said Marc Hurlbert, executive director of the Avon Breast Cancer Crusade.