A little over a year after setting up shop, the Next Generation Diagnostics group at the Novartis Institutes for Biomedical Research in Cambridge, Mass., has begun to support the drug maker's oncology research with a variety of next-generation sequencing-based approaches.
Officially launched in November 2012, one of the group's primary goals is to apply new technologies, in particular next-gen sequencing, to understand the molecular basis of therapeutic response in cancer patients, according to Wendy Winckler, the group's executive director, who presented some of its early work at the Advances in Genome Biology and Technology conference in February.
In particular, the group collaborates with Novartis researchers involved in drug development to generate retrospective genomic profiles of patients from clinical trials in order to find correlations between their genotype and their treatment outcome.
In addition, the group works with Genoptix, Novartis' San Diego-based molecular diagnostics firm, to develop strategies for new diagnostic cancer tests. "Our mission there is to develop innovative laboratory and analytic methods that can help bring all this promise that we in the research space recognize with next-generation sequencing and make that more widely accessible in terms of commercial diagnostics," Winckler told Clinical Sequencing News, though Novartis is not disclosing any specifics of the collaboration at the moment.
Since its inception, the NGD group, which is still recruiting, has grown to almost 40 people organized into three teams: samples, genomics, and bioinformatics. What is fairly unique about this setup is that the teams are all in the same space, working on the same projects. "I think that is a different model than what has been taking place at many other companies, where frequently the sequencing is done in one place, and the analysis might be done somewhere else," Winckler said. "We see that there are a lot of beneficial synergies in having our analysts work hand in hand with our wet lab biologists. It helps us to innovate our methods faster, it helps our analysts to better understand the data they're seeing."
Bioinformatics accounts for about half the entire NGD group and includes software engineers focusing on the LIMS system and the next-gen sequencing pipeline; computational biologists with expertise in cancer genetics; and biostatisticians involved in the interpretation of NGS data from clinical trials. Bioinformatics is still "the rate-limiting step in trying to do this type of analysis, so we wanted to be properly staffed for that," Winckler said. "As sequencing becomes easier, the analysis and interpretation is still a challenge."
The genomics team focuses on a range of techniques, including whole-genome sequencing, targeted sequencing, RNA-seq, and non-sequencing methods such as digital PCR. Much of the group's sequencing is currently done on Illumina platforms, though it has worked with a number of sequencing vendors and is keeping an eye on new developments. "We're constantly monitoring the field and are very open to trying new and exciting technologies as they come out and as they offer us advantages over existing platforms," Winckler said.
The samples team has expertise in biobanking, tissue handling, DNA and RNA extraction, nucleic acid quality control, and assays such as fluorescent in situ hybridization and immunohistochemistry. So far, it has put a lot of effort into methods for optimizing the quantity and quality of nucleic acids extracted from patient samples, while the genomics group has been working on reducing the amount of input material required and developing protocols that tolerate the types of degradation seen in formalin-fixed paraffin-embedded samples. Nearly all of the patient samples the NGD group analyzes are FFPE samples, Winckler said, and the amount is often "quite limited."
The clinical trials projects are just starting to get off the ground because the group, which currently does not operate a CLIA-certified lab, first had to develop a stringent quality management system for processing patient samples, which "took a while to establish," Winckler said.
To characterize the samples, she and her colleagues draw from a portfolio of assays, ranging from unbiased discovery approaches to sensitive mutation-specific assays, "so we can use the right tool, depending on the experimental question posed by any particular trial," she said.
In some studies, for example, they look for drug response signatures in RNA-seq data, while in others they perform exome or targeted sequencing to find events that correlate with drug sensitivity or resistance. In some cases, for example exceptional responders, the NGD group might even employ whole-genome sequencing "to take a deep dive into the tumor genetics," Winckler said.
Many studies involve a 600-gene pan-cancer assay that contains actionable genes and other interpretable loci, sequenced at 300x coverage. But the group has also developed disease-specific assays, for example a "small but comprehensive" non-small cell lung cancer panel that allows them to look for point mutations, copy number changes, and translocations in all currently actionable lung cancer genes.
The advantage of this type of assay is that it only requires small amounts of input DNA. "That makes it enabling for clinical research, where the amount of patient biopsy material is often quite limiting," Winckler said. Up until now, clinical trials teams often had to make choices about what genes to test for because there was not enough material for all of them. "We're hoping that having a comprehensive test like this will allow us to get all of the genetic information that's meaningful and interpretable today from a single assay," Winckler said.
The group is also exploring plasma-based sequencing assays, for example from circulating tumor DNA, to monitor therapy response, though Winckler said the amount of tumor DNA varies a lot between tumor types, tumor stages, and different patients with similar diagnoses. "Given this, there is a ton of research that remains to be done to understand how generalizable that type of approach is, and how it could be used and brought into the clinic," she said.
The choice of assay for clinical trials studies largely depends on the particular trial. For a type of cancer for which not a lot of genetics is known, for example, or to look for a resistance mechanism when there is no good a priori hypothesis, the researchers might opt for a more comprehensive strategy, like exome sequencing. On the other hand, "if we're just trying to profile those things that are targeted by existing drugs, then using a small, targeted approach at deeper coverage and more sensitivity makes sense," she said.
On the data analysis and interpretation side, one challenge has been how to report results back to a wide range of collaborators, an issue the entire field of clinical next-gen sequencing has been grappling with, Winckler said.
Some results are "black-and-white events" that are easy to communicate, for example the presence of an ALK translocation in lung cancer, "but there is also a tremendous number of shades of grey in oncology sequencing reporting right now," for example mutations in actionable genes whose significance is not known yet. "There is going to be a tremendous amount to learn over the next few years as these next-generation sequencing assays become more and more prevalent in oncology, and I'm very interested to see how this continues to evolve in the diagnostic industry," Winckler said.
Her group's collaborators vary in how "genomics-savvy" they are, and how much data they want returned. To accommodate them all, the group is establishing reporting protocols that "can hit these different scenarios in a more standardized way," she said.
On the preclinical side, along with other NGS-groups at Novartis, the NGD group has been involved in characterizing more than 500 patient-derived tumor xenograft, or PTX, mice that Novartis has created, which represent the spectrum of tumor types seen in the US.
Novartis uses the results – from the pan-cancer assay, exome sequencing, and RNA sequencing – to design "mouse clinical trials" that represent the genetic diversity seen in actual patient trials, and to investigate which mouse genotypes are associated with treatment response. "This can help us to generate hypotheses for patient stratification and then, later, human trials," and can provide leads for new drug combinations, Winckler said.