By Monica Heger
Whole-genome sequencing followed by ultra-deep targeted resequencing of B-cell chronic lymphocytic leukemia tumors throughout the course of treatment shows how patients' mutational profiles change in response to treatment and disease progression, according to an ongoing study.
Anna Schuh, a hematologist at the University of Oxford, reported preliminary results from the study at last week's Biology of Genomes meeting in Cold Spring Harbor, NY.
She and her team sequenced two patients at five different time points during the course of treatment and found four distinct classes of mutations: responsive mutations, which are present at the beginning of the disease, but disappear after treatment; resistant, "mother" mutations, which are present at low levels throughout the entire course of the disease; expanding mutations, which start out at low levels, but expand during remission; and emerging mutations, which are not present at the beginning, but are ultimately responsible for disease aggression.
Schuh and her team sequenced the whole genomes of two patients, both with a poor-prognosis form of the disease. One patient had a more latent form of the disease, surviving for nine years after diagnosis, while the second patient had a more aggressive disease and passed away three years after diagnosis.
The team used Illumina's HiSeq 2000 to sequence to 30-fold coverage each patient's normal genome once and their tumor genomes at five time points throughout treatment. Additionally, they used Illumina's MiSeq to perform ultra-deep targeted sequencing of the mutated genes at each time point and in the normal samples. All sequencing was done at Illumina's UK facility.
The goal was to identify the underlying genetic architecture responsible for the patients' poor prognosis. B-cell CLL is characterized by a poor survival rate and poor response to treatment. One-third of CLL patients have a mutation in the tumor suppressor gene TP53, while another third have an incredibly complex tumor genome. Yet, for the remaining third, there is no clear genomic explanation, said Schuh. The two patients she and her team sequenced were from this third category.
The first patient, who had an indolent form of the disease, was sequenced at five different time points: after treatment with chlorambucil, a first-line anti-cancer drug for CLL; after treatment with the chemotherapy drug FC, comprised of fludarabine and cyclophosphamide; during relapse after not having responded to FC; following treatment with an experimental second-generation monoclonal drug, ofatumubab, under development at GlaxoSmithKline; and during another relapse. Despite having a less-aggressive form of the cancer, this patient never achieved complete remission but did respond well to the second-gen drug, Schuh said.
The second patient was also sequenced at five time points, throughout a similar treatment course. However, this patient did achieve remission after being given the chemotherapy FCR, which is similar to FC, except with the addition of rituximab, and was sequenced during the remission.
Sequencing revealed between 4,000 and 6,000 somatically acquired SNVs and indels per tumor sample, about half of which occurred in genes. Plotting those mutations based on the time point at which they were identified "revealed changes in the mutational profile induced by treatment," Schuh said. "You can see where the patients received medicine and what happened."
For instance, the first patient's tumor genome had a much more stable mutational profile than the second patient's tumor genome, which had dramatic dips and spikes in mutational frequency following treatment and relapse.
After the initial whole-genome sequencing, Schuh and her team sequenced the 40 or so genes in which mutations were found at any given time point. They sequenced all 40 genes in each of the 12 samples (the five treatment time points plus the normal genome) from the two patients on the MiSeq to about 15,000-fold coverage.
The targeted, ultra-deep sequencing confirmed the four classes of mutations identified by the whole-genome analysis. The most surprising aspect of the study was the identification of background "mother mutations," said Schuh. Because those mutations were present at low levels at all time points, they seemed to be mutations that the patient could live with.
These mutations were primarily found in immunoregulatory pathways, so Schuh said they were likely driving or enabling the other more aggressive cancer mutations that appeared in the patient later on.
[ pagebreak ]
The cells with those background mutations could be cancer stem cells, she said. While the results would have to be confirmed in additional patients, they suggest that drug development should focus on these mutations, because by the time the emerging mutations develop, the disease has progressed so far that the tumor is resistant to therapy, said Schuh.
Only one gene that has previously been implicated in CLL was found to be mutated in the two patients, but treatment proved successful at killing cells with those mutations, Schuh said.
The team also found mutations in a gene that has been previously linked to other types of leukemia that was present in low levels at the beginning of cancer, but expanded significantly throughout the course of the disease.
Emerging mutations, which were the mutations that led to the aggressive cancer in the second patient, were found in known cancer pathway genes, such as MEK1. Currently, drugs that would target MEK1 are in development, suggesting a possible treatment avenue for these patients.
However, Schuh said that going forward it would be important to focus drug development not just on mutations to genes like MEK1 that contribute significantly to disease progression, but also on the mutations that develop early — the cancer stem cells.
"The moment you develop MEK1 mutations, that's a bad sign," she said. "But if you can get rid of those mother mutations, you might be able to cure the patient."
Overall, she said the study illustrated that there are "not enough targeted agents," and it confirms that "the longer you wait, the more cancer-related mutations you get."
Elaine Mardis, co-director of the Washington University Genome Institute, said the study was intriguing and that she was surprised by Schuh's results, which differed from those of cancer sequencing studies that she and her team have done.
Schuh's study is similar to the work that the Wash U team has been doing with acute myeloid leukemia samples, but the Wash U group has not observed the background mutations that Schuh's group found. "The findings were completely different," Mardis said. "We can't detect mutations in the remission samples, so that was surprising to me."
The differences could be explained by the fact that the two groups are working with different types of leukemia, she said, but added that it would also be important to verify the results in additional samples because the findings could also be caused by contamination. With leukemias and other liquid tumors, keeping the tumor DNA and normal DNA separate is much more difficult than when working with solid tumors, Mardis said.
Sequencing in Clinical Trials
Schuh said that her group has acquired an additional four patient samples with B-cell CLL and plans to perform whole-genome sequencing and ultra-deep targeted sequencing, using the same method as on the first two patients.
Their next step is to begin sequencing in a clinical trial of GlaxoSmithKline's ofatumubab to determine "patterns of response and resistance to treatment," she said.
Schuh predicted that next-gen sequencing will play an important role in clinical trials in order to stratify patients. By using sequencing to decide the course of treatment in a trial, drug companies will be able to increase the power of their studies at a lower cost, since fewer patients would be needed to achieve statistically significant results.
Eventually, she thinks sequencing will move into the clinic for diagnostic use, and expects that targeted sequencing will be the first application of the technology in that setting. While Schuh is a clinician, as part of her role at the University of Oxford, she is evaluating new technology for use in molecular diagnostics.
The main problems with whole-genome sequencing for diagnostics are the cost, analysis, and turn-around time, she said. Clinicians have to "make a diagnosis quickly. They can't fiddle around with the bioinformatics for six weeks."
Schuh suggested that smaller, lower-throughput sequencers such as the MiSeq would be useful in the clinic because of their focus on targeted deep sequencing and quick turnaround times. Because targeted sequencing could be done on a patient quickly, if the targeted approach came back negative, then it would be appropriate to send a sample to a larger lab for whole-genome sequencing, she said.
Eventually, though, in about five to ten years, "everyone will do whole-genome sequencing," she predicted.
Have topics you'd like to see covered in Clinical Sequencing News? Contact the editor at mheger [at] genomeweb [.] com.