A pair of recently published studies demonstrates that a whole-genome amplification-based single-cell sequencing technique known as MALBAC —multiple annealing and looping-based amplification cycles — can be applied in the clinic for prenatal genetic diagnosis and monitoring circulating tumor cells in cancer patients.
The studies utilize MALBAC, a whole-genome amplification method developed by Harvard University's Sunney Xie, and originally published in Science about a year ago. The technique seeks to improve on the amount of genome that can covered in whole genome amplification methods while also reducing amplification bias.
The two studies also build on previous presentations by Xie, who described preliminary results of his work at the Beyond the Genome conference last year in San Francisco.
In one new study published in December in Cell, Xie and colleagues at Peking University's Third Hospital validated the approach on oocytes that were collected from eight healthy volunteers and fertilized via intracytoplasmic sperm injection.
The group tested the method on the first and second polar bodies, which are byproducts of the in vitro fertilization cycle, and on cells from the blastocyst, sequencing a total of 183 cells.
Sequencing the polar bodies eliminates the possibility of damaging a potentially viable embryo, but the genome of the fertilized egg cannot be measured directly and must be deduced. Additionally, sequencing polar bodies will not detect any aberrations contributed by the father.
In the Cell study, the team sequenced 52 female pronucleus cells, 67 PB1 cells, and 64 PB2 cells to 0.7x coverage on the Illumina HiSeq 2000 using MALBAC for whole-genome amplification. On average, genome coverage of each cell was 32 percent.
Next, the group combined sequencing data from cells of the same individual to increase sequencing depth to between 4x and 37x per genome and genome coverage to between 70 percent and 97 percent. Then they used an algorithm similar to population SNP calling to determine heterozygous SNPs from each donor and determine haplotype.
By phasing the SNPs and figuring out the donor's haplotype, the researchers hypothesized that they would be able to deduce the alleles of the female pronucleus. To demonstrate that this could in fact be done, they compared results from sequencing of the two polar bodies with sequencing of the female pronucleus of the same donor to see if they could use sequence information from the polar bodies to deduce the actual sequence information of the female pronucleus. Their predictions were accurate in 40 out of 44 cases, and in the four mismatches three had "severe DNA degradation" and one had an "intrinsic chromosome abnormality."
The technique can also be used to assess aneuploidy as well as look for variants causative of Mendelian disease. The researchers determined that sensitivity and specificity for determining aneuploidy increase for lower resolutions, reaching 97.8 percent sensitivity and 99.5 percent specificity at a 5-megabase resolution. Going down to a 0.5-megabase resolution, sensitivity and specificity are 97.8 percent and 99.5 percent, respectively.
Fuchou Tang, an assistant professor at Peking University's Biodynamic Optical Imaging Center and an author of the Cell study, told CSN that the group is now in the process of conducting a clinical trial of the method. They are recruiting around 30 women who are either carriers of known Mendelian diseases, are more than 30-years old, or have experienced recurrent miscarriages. He anticipates the trial will last around one year.
Additionally, Tang said the researchers are testing a targeted sequencing approach to identify point mutations associated with Mendelian disease.
Moving forward, he expects both single-cell sequencing of polar bodies and blastocyst cells to be useful, depending on the specific situation, since both have advantages and disadvantages.
By selecting a few single cells from the blastocyst for sequencing, "we will detect all of the aneuploidies, no matter if they are from the mother, or from the father, or de novo formed after the fertilization step," he said. "We will also detect all of the gene mutations."
However, "removing a few embryonic cells at the blastocyst stage is more invasive compared to removing the two polar bodies at the fertilized egg stage," he said. Additionally, to detect point mutations, cells from the blastocyst must be sequenced much deeper than the polar bodies, making that method more expensive. "We need to sequence every blastocyst-stage embryo to very deep coverage — at the moment, about 60x for each embryo — to detect the point mutations in the embryo's genome," he said. Assuming that around five blastocysts are acquired, that is approximately 300x sequencing depth for each woman, "which is very expensive at the moment."
But, with polar body sequencing, the "polar bodies from different fertilized eggs [can be sequenced] together to around 30x to 60x for each woman, which is much lower and cheaper than blastocyst stage sequencing," he said.
Circulating tumor cells
In a second study published last month in the Proceedings of the National Academy of Sciences, a different group of researchers, also led by Xie, applied MALBAC and single-cell sequencing to circulating tumor cells from lung cancer patients.
The researchers sequenced the exomes of 24 single CTCs from four patients with adenocarcinoma and compared them to the exomes from the patients' primary and/or metastatic tumors. In one patient, the cancer transitioned from lung adenocarcinoma into small-cell lung cancer metastasis in the liver.
In general, the researchers found that the SNVs and indels in the CTCs were more similar to the metastatic tumor than the primary tumor. For instance, sequencing of patient two's six CTCs identified 106 out of 146 nonsynonymous SNVs/indels in the metastatic tumor and sequencing of patient three's five CTCs identified 145 out of 170 nonsynonymous SNVs/indels in the metastatic tumor.
They also found mutations that could explain clinical progression. For example, in patient one, they identified PIK3CA mutations in seven out of eight of the CTCs, despite being present in very low abundance in the primary tumor. The mutation has been implicated in drug resistance of erlotinib, marketed by Roche as Tarceva. "Consistently, patient one underwent rapid disease progression in the liver metastasis after [one month] of EGFR TKI treatment with erlotinib," the researchers wrote.
To examine the technique's ability to assess copy number variations, the researchers sequenced the whole genome of the eight CTCs from patient one to .1x coverage.
Analyzing copy number variation gave a different picture than analyzing point mutations. "Surprisingly, we found that all CTCs of patient one exhibited reproducible gain and loss CNV patterns (an average of 83 percent of the gain and loss regions was shared between any two CTCs)," the authors wrote.
Additionally, the CNV pattern more closely resembled the metastatic tumor than the primary tumor. To confirm that the cells were indeed circulating tumor cells rather than cells from the metastatic tumor, the researchers analyzed an EGFR mutation present in both the metastatic tumor and the CTCs, observing that while the mutation was homozygous in bulk sequencing of the metastatic tumor, it was 50 percent heterogeneous in the CTCs.
Looking at five other patients, they found that their results were consistent.
"The reproducible CNV patterns that are characteristic of different cancers might allow noninvasive cancer diagnostics and classification through sequencing of CTCs," the authors wrote.
James Hicks, a research professor of cancer genomics at Cold Spring Harbor Laboratory whose team is also working on single-cell sequencing of prostate cancer patients, said that the study adds to the growing body of evidence that single-cell sequencing is a good method for tracing the lineage of cancer.
Single-cell sequencing and "using the copy number profile is a way of accurately following rare cells or rare segments of cells," he told CSN.
Ultimately, he thinks such methods could be used to identify biomarkers and monitor patients. Low-pass sequencing to identify copy number variants will confirm that the CTCs "reflect the lineage of the tumor," he said. Additionally, deeper targeted sequencing can be done on select point mutations to cancer-related genes like TP53, PTEN, or EGFR.
While he said the method itself was good, one surprising result, he said was the fact that the researchers found consistent copy number profiles, which is very different from what his group has found in sequencing CTCs from breast cancer and prostate cancer patients.
Additionally, going forward he said more work will have to be done to understand how sequence data from CTCs can help guide clinical decisions with regard to treatment.
There needs to be "more studies in multiple different cancers to get a picture of what CTCs are going to tell us relative to following disease progression during therapy," he said. "We know that these are cancer-like cells, but [we still don't know] what decisions to make based on their number and genotype."