Novartis scientists have generated the PDX Encyclopedia, which contains over 1,000 patient-derived tumor xenograft models spanning a range of common solid cancers. They'll use this collection for in vivo drug screens designed to mimic human clinical trials, which they hope improves candidate therapy profiling.
Seeking to reduce the number of preclinical drugs that fail along the road to regulatory approval, scientists at the Novartis Institutes for Biomedical Research (NIBR), headquartered in Cambridge, MA, have generated an extensive collection of patient-derived tumor xenograft (PDX) models. Called the PDX Encyclopedia (PDXE), it augments NIBR's Cancer Cell Line Encyclopedia, established in 2012 (Nat Med 2015;21:1318–25).
“Our goal is to develop cancer therapeutics with a much higher probability of success in patients,” says William Sellers, MD, vice president and global head of oncology at NIBR, “and we recognized the limitations of doing so with in vitro systems.” The PDXE currently contains over 1,000 models, representing a spectrum of solid cancers, and genomic landscape analyses indicate “close alignment between our models and human data as described by The Cancer Genome Atlas,” Sellers says.
The researchers are harnessing their collection to carry out PDX clinical trials (PCT) that mirror human studies in design: In a given PCT, each mouse receiving the therapy of interest bears a unique tumor xenograft from an individual patient. By treating a group of such mice, the therapy's efficacy against the cancer type in question can be determined, and “we can capture the heterogeneity of responses between patients,” explains Hui Gao, PhD, a senior investigator at NIBR.
So far, PCTs have yielded data “highly consistent with what's seen in humans,” Sellers says. For instance, BRAF-mutant PDXs responded well to BRAF inhibition—and even better with the addition of a MEK inhibitor. Ideally, he adds, PCTs will prove predictive of new therapeutic indications; to that end, “we're using this system to profile all of our clinical candidates and additional compounds.”
Sellers and his team also validated cell line–derived results suggesting that high levels of two proteins, DR5 and caspase-8, predict sensitivity to TAS266, a novel antibody that activates DR5 signaling, thereby triggering apoptosis. In an initial PCT assessing melanoma response to TAS266, just 18% of mice appeared susceptible. A retrospective biomarker analysis then revealed that the response rate to TAS266 was actually 80% in the subset of mice with elevated DR5 and caspase-8.
Importantly, the researchers found that therapeutic activity in vitro wasn't necessarily seen in vivo, and vice versa. “The disconnect was surprising,” Gao says. Novartis's investigational IGF1R inhibitor LFW527 appeared to increase the efficacy of the MEK1/2 inhibitor binimetinib (MEK162; Array BioPharma) in colorectal cancer, non–small cell lung carcinoma, and prostate adenocarcinoma cell lines. When this combination was tested in relevant PCTs, no such synergy was observed—the modest response rate achieved with binimetinib in colorectal cancer “actually worsened” when LFW527 was added, Gao notes.
“It turns out that prior to our analysis, this combination was tried in the clinic, with negative results,” Sellers says. “IGF1R inhibitors have long been touted and always look terrific in vitro, but they have yet to work out in vivo.” On the other hand, a clinical investigation of Novartis's CDK4/6 inhibitor, LEE011, combined with BRAF inhibition is under way, based on encouraging PCT results that weren't seen in cell line studies.
The researchers will continue expanding the PDXE, and hope to eventually add difficult-to-establish models, such as glioblastoma and prostate cancer, to the collection. They're also exploring ways to address the limitations of PDXs, chiefly that the mice, being immunodeficient, can't be used to assess candidate immunotherapies.
“Every model system is imperfect in its own way,” Sellers says. “We'll use the PDXE in ways best suited to its strengths. Take drug combinations, for instance—the number of permutations is well beyond what could be tested in humans. We think our system will prove very useful here; it should also help significantly with biomarker validation.” –Alissa Poh
For more news on cancer research, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/content/early/by/section.