Sequencing the genome of a cancer bladder patient who responded exceptionally well to a cancer drug that failed in most other patients in a clinical trial, researchers have identified mutations that might serve as predictors of sensitivity to the drug.
The approach – sequencing the tumor genomes of "outlier" patients – might prove useful more generally to salvage cancer drugs that only work in specific patient groups, though pinpointing the right mutations might be challenging.
"Although single-patient anecdotes are often dismissed as failing to provide meaningful clinical evidence, this example illustrates the potential for such cases to inform future clinical development of drugs in molecularly defined populations," the authors, who published their study online in Science last month, noted.
"Too often in a clinical trial, we focus on the mean improvement, for the average patient. This paper really shows how one can apply a lot of the latest biological methods to study the patient that does something unexpected," said Atul Butte, an associate professor of Stanford University School of Medicine, who was not involved in the study.
"We have only done this twice so far, and in both cases had surprising success," said Barry Taylor, an assistant professor at the Helen Diller Family Comprehensive Cancer Center at the University of California, San Francisco, one of the senior authors of the study.
He said that the interpretation of tumor genomes is still difficult because of their great complexity. Mutations of interest found in those genomes sometimes mask the key driver lesions that may determine drug response, "so I don't anticipate to have that much success all the time."
Also, it might be that mutations found in an outlier patient are extremely rare and would not be of much use in predicting drug response in other patients. "But they nonetheless may tell us something critical about the biology and the genetic basis of these kinds of responses," he said.
The recently published study is part of a larger project involving the Taylor lab and the group of David Solit at Memorial Sloan-Kettering Cancer Center to study the genomics of several hundred patients with high-grade bladder cancer, with the aim to characterize actionable mutations that could be targeted with specific drugs. "Targeted therapies have not made their way into bladder cancer treatment in any meaningful way, so this is one of our major interests," Taylor said.
The publication focuses on the tumor genome of a single patient with metastatic bladder cancer who was enrolled in a phase II clinical trial of the drug everolimus and had a durable and complete response. The drug is sold by Novartis under the brand names Afinitor, Zortress, and Certican and is currently being used to treat renal cell cancer and a few other tumor types, and also as an immunosuppressant in organ transplants.
Her case was noteworthy because dozens of other bladder cancer patients in the same trial "showed really no objective response by traditional criteria," Taylor said, and the trial failed to achieve its endpoint for progression-free survival.
The results of the genome analysis, he said, "led to the salvage of everolimus, which in the absence of this finding probably would have been abandoned in bladder cancer."
After conventional molecular tests, including a copy number variation assay and some targeted sequencing, failed to explain the patient's profound sensitivity to the drug, the researchers decided to sequence her entire tumor genome as well as a control sample from her blood.
"It was impossible to know in advance which specific genomic lesions this patient harbored that contributed to enhanced everolimus sensitivity, such as a structural rearrangements or other events that that would not have been easily detectable with more targeted approaches," Taylor said, explaining why the team opted for whole-genome sequencing.
Illumina's FastTrack service performed the sequencing on the HiSeq platform, using paired-end 100-base pair reads and sequencing the genome to about 40x coverage. The data were analyzed by both Taylor's and Solit's groups.
The researchers found 140 non-synonymous mutations in the tumor genome, of which two particularly caught their interest: a two-base deletion in the TSC1 gene and a nonsense mutation in the NF2 gene. Mutations in both genes have been associated with dependence on the molecular target of everolimus.
They then sequenced the two genes in a cohort of 96 high-grade bladder cancers and found the TSC1 gene to be mutated in five additional samples, while NF2 was not mutated in any. In a functional assay, NF2 knockdown made cells with a mutation in TSC1 especially sensitive to everolimus.
To explore whether TSC1 mutations alone are linked to everolimus response, the researchers sequenced the gene in 13 additional patients participating in the trial. They found three additional tumors with nonsense mutations in TSC1, of whom two had a minor response to the drug, and one additional tumor with a missense mutation in the gene that also showed a limited response. Eight of the nine patients whose disease progressed had wildtype TSC1.
In general, patients with TSC1 mutations remained on the drug longer and their tumors took about twice as long to recur.
Next, the researchers want to test TSC1 as a biomarker for everolimus response in another clinical trial that will include a selected population of patients that harbor mutations in the gene, as well as maybe TSC2 and NF2, Taylor said. This trial, which is still in the design phase, could involve not only bladder cancer patients but also other cancer types where mutations in those genes are known to occur, he said.
According to Solit, the other senior author of the study, the main limitation for the planned trial is the development of an assay that can detect TSC1 mutations in clinical samples and be performed in a CLIA lab. He said his team hopes to have such an assay in place by the end of the year.
In addition, the teams are pursuing studies to understand the biology of the TSC1 and NF2 mutations further, for example using in vitro and mouse models of bladder cancer.
Taylor said they are also interested in using their approach in other outlier patients from clinical trials of single or combination therapies of targeted drugs. "It's really become a very intriguing approach, and certainly an approach that’s been used by others, even outside of the cancer domain, looking to exploit an extreme phenotype, be it drug response or any observable clinical phenotype," he said.
While their initial focus is on bladder cancer, the researchers are also thinking about applying the approach to other tumor types, as well as to mouse models of human cancer. "Outlier responses in patients are quite rare and often very hard to get your hands on, but preclinical trials of single and combination inhibitors in mouse models of human cancer also produce not only drug resistance but profound drug sensitivity and may be a fertile area for this kind of approach," Taylor said.
The results might not always be straightforward, though. "I think we are going to find a variety of these kinds of findings in other genomes where it's not one gene but the interaction of multiple lesions that do confer these kinds of drug sensitivities," he said.
Also, translating the results into clinical use will be challenging. Studies of BRAF inhibitors in melanoma, for example, have shown that "not all patients, even with the same kinds of mutations, will respond similarly to the drug," he said. "That's the kind of complexity we need to start working our way through."