Notwithstanding a sudden setback over its lead cancer drug, Infinity Pharmaceuticals has the tools and imagination to become a success.
By Kevin Davies
July 20, 2009 | The creation of Infinity Pharmaceuticals in 2001 had all the right ingredients: Distinguished scientific founders, including Harvard chemist Stuart Schreiber and Broad Institute director Eric Lander; an experienced management team including ex-Millennium Pharmaceuticals veterans; and a cutting-edge approach to chemical synthesis and drug discovery (see, “Conquering Infinity with Chemical Genetics,” Bio•IT World, Feb 2003). When the company recruited Julian Adams, chief developer of Millennium’s cancer drug Velcade, it seemed like the icing on the cake.
Last April, however, Infinity hit a major bump in the road when a review of the first few dozen patients in its Phase III trial for its lead compound, IPI-504, seemingly on course for approval around 2011, prompted the company to instantly halt the trial.
Just hours earlier, I was listening to Adams and John Keilty, Infinity’s VP IT and Informatics, discuss Infinity’s evolution from a discovery to a clinical company. Both men were relaxed and upbeat. Since Adams’ arrival five years ago, Keilty says Infinity has found its direction. “Everything’s been focused on one purpose: What is the best way to get drugs to patients?” That mission has been accompanied by a wholesale operational rethink, led by informatics/IT, to leverage what was learned in the past.
“Drug discovery doesn’t happen overnight,” says Adams. “It can happen overnight, but it’s the drug development that takes five years.” Infinity continues to have high hopes for IPI-504, which was in an international Phase III trial for GIST (gastrointestinal stromal tumor). The drug targets the chaperone Hsp90, a key molecule in regulating protein folding. Inhibiting the chaperone drives unfolded proteins in the cancer cell to the proteasomes for degradation. At ASCO in June, Infinity released more promising data on IPI-504 for non-small-cell lung cancer and other indications.
The discovery team (some 60 scientists) examines cutting-edge mechanisms for novel cancer targets, looking for genes and pathways that are differentially expressed in cancer cells and druggable. “We take a lot of inspiration from the original concepts from Schreiber,” says Adams, but adds it is absurd trying to build and screen huge chemical libraries.
“I don’t care if you have 5 million compounds, I’m unimpressed!” says Adams. “We have abandoned the synthesis part of it—we just go directly to nature, and have found a number of natural products that represent the starting point for our discovery programs. [When] Nature makes a molecule, we say, ‘God did a nice job—we can do better,’ without any hubris! God didn’t plan for good PK, the FDA, and all that. So we have to add some bells and whistles.”
In Keilty’s view, there are “no tricks” to expediting drug development. “I say this as the IT guy—I think technology gets you [only] so far. What we’ve really focused on is, how can we put tools in the hands of folks to empower them? Technology, in and of itself, doesn’t do it for us.”
Nevertheless, Keilty points to important capabilities in molecular modeling, using third party tools such as MOE and Schrodinger for ligand-receptor docking. His colleague Tom Tibbitts works with discovery and clinical teams, looking at modifications to improve drug-target binding. A lingering issue is the ability to dock small molecules into a protein’s active site, or what Adams calls “the solvation/desolvation problem—and no one has solved that.” It’s not just a matter of high-performance computing but the development of the proper algorithms.
Over the past few years, Keilty’s Informatics team (see, “Informatics Infrastructure”) has grown from supporting a pure discovery platform—collecting terabytes of data from registering and tracking compounds—to building a relational database for clinical data.
Five years ago, before implementing electronic data capture, or EDC, Infinity began filing everything electronically to the FDA—before it was the law—beginning with the IND for IPI-504. “We were the first fully electronic regulatory submission to the oncology division of the FDA,” says Keilty proudly. “It was an unbelievable company-wide effort. Some folks didn’t sleep for a month as we ensured every PDF was perfect. We actually downloaded the software from the FDA…” His voice trails off for effect. After evaluating a lot of alternative platforms, he settled on ISI [Image Solutions]. “Over time, we’ve built other tools to complement their system.”
Infinity is at 150 submissions and counting. Worldwide regulatory filing IT is handled by a younger Infinity employee. “[Our Hsp90 program is] unpartnered,” says Keilty. “We use CROs, particularly for Europe, and for external regulatory submissions. Basically, we pack our own parachute. We do everything from discovery phase to regulatory filings and the intended launch—upon approval.”
Infinity’s first trial was performed on paper. “Mercifully only 18 patients,” recalls Adams, but he quickly figured they needed EDC, which would let them interface with the web-based data collection systems at the clinical sites. “EDC is much simpler than so many of the tools used in discovery,” says Keilty. Medidata edged out offerings from Phase Forward, DataTrack, and others, and Keilty was able to get EDC up and running at the Dana-Farber Cancer Institute a month after signing the contract.
One reason for choosing Medidata was that Keilty had moved from a Java-centric shop to a mostly .NET shop. “We were able to take their system and expand on it,” says Keilty. “Regardless if it comes off the shelf or not, there are a lot of steps to make sure it works exactly as it was set up to do.” Keilty’s team has become adept at knowing what to look for.
Adams and Keilty soon realized that some of their discovery tools, like Spotfire, could be applied to looking at clinical data as well. Each night, the EDC system exports SAS datasets over secure FTP to Infinity. “This is such a rich data source, both for prospective and retrospective analyses… [If] I see an abnormal lab value, it’s nice to look across studies to see if we’ve seen that lab value before.”
Today, Infinity collects hundreds of fields of data, from age and body weight to lab chemistries and imaging results, in a clinical data warehouse designed by Keilty’s Informatics team—the Clinical Data Integration Platform (CDIP). Scientists can query the data in any number of ways, looking for gender differences, lab values, CT scans, tumor shrinkage, EKG monitoring, and so on.
This could expand to genomic differences too. “We haven’t ruled that out,” says Adams. At ASCO, Infinity reported better Phase II results for IPI-504 in patients with non-small cell lung cancer with the wild-type epidermal growth factor receptor than mutated forms. Eventually, Adams anticipates sequencing patients’ entire genomes before validating a multiplexed set of say 5-50 genes that defines a more homogenous sub-population. “If you could correlate that with better responses, we would absolutely go in that direction,” says Adams.
But Keilty cautions that next-gen sequencing is just a commodity. “We really want to be biomarker agnostic. We need something that’s really usable in the clinic.” It may be genome sequencing, it may be circulating tumor cells, or something else. “All we care is that this is a way to separate the patient populations into those that will respond, and those that don’t.”
Another homegrown application—a transactional database called iTRAC—tracks all the processes around a clinical trial. Keilty calls it a “clinical trial management system on steroids.” A big component of the IPI-504 Phase III trial was imaging. Although challenging, all the imaging is analyzed centrally as the data arrive at Perceptive Informatics in Waltham, Mass. Perceptive turned around the results within five days after review by independent radiologists, which became the primary blinded input. Trial managers logged into iTRAC to see if an imaging read was missing and ensure the information was on schedule. Electronic queries were generated automatically, saving Infinity staff auditing hassles. Adams received an email every morning providing a snapshot of the number of patients in the study, the proportion in screening, randomization, and so on.
“The same person doing regulatory submissions is also using iTRAC,” says Keilty. He can call up the CRO and find a missing read if necessary. “We can control the study without really being the one doing a lot of the work.” But normalizing those eclectic sets of data across studies and sites poses a challenge. If certain lab values look strange, the team needs to be able to retrieve similar incidences and “pull those data together instantly. We’ve done a good job of that.” Over time, Keilty says iTRAC will become an important data source. If Infinity wants to expand into a new disease area, then it can be used to formulate a feasibility survey to identify the right people to run the trial.
For all Infinity’s intellectual and informatics capabilities, Adams admits that not all hypotheses bear out. His preliminary conclusion from the IPI-504 setback is that patients were receiving too high a dose of the drug. Until then, Infinity was planning to expand headcount this year by one third to more than 200 staff. “We want to be very pragmatic about this,” says Keilty. “We’re not Novartis. How can you do the most with the people you have?”
Keilty says the company’s three lead programs all stemmed from intellectual insights. Adams, he says, pragmatically applies his deep understanding of biology and chemistry. “He can look at something and say, ‘If we threw this functional group on here and tweaked this a little bit, I bet that works.’ It’s kind of annoying—but mostly he’s right!” Adams admits he doesn’t mind occasionally breaking a Lipinski Rule or two. But he’s not infallible, and there have been times when the pharmaceutical group has “totally made a fool out of me.”
Infinity has “every intention to stay in oncology,” says Adams. The key, he says, lies in the ability to serially sample tumors, which is “the single biggest impediment to understanding how one’s drug is impacting the tumor. The future’s going to be in circulating tumor cells,” says Adams. “I want to have one of those [Mass General CTC Chip] machines in here—I’ll even be a beta tester! Serially sampling the tumor—I want to collect those cells, grow them up, see what genes are being displayed, [learn] what’s the impact of drugs on an ongoing basis. And I want to know when to switch drugs because [the patient has] just developed resistance.”
Infinity has some promising cancer programs entering the clinic, including the Hedgehog pathway inhibitor, IPI-926, in Phase I, which targets an embryonic growth pathway that has been hijacked in adult cancers and could explain why some cancer cells metastasize.
Despite the industry’s travails, Adams sees a silver lining. “It’s raining talent with the consolidation in the industry. There are divestitures of pipelines from failing companies who basically can’t afford to stay alive,” or are simply looking to trim their portfolios. “If we license it, we have to love it as if we discovered it.”
John Keilty, Infinity’s “IT guy,” leads an informatics staff of a dozen spanning everything from software engineering, discovery and development to clinical and business. Flexibility has been vital in assembling the team’s infrastructure, applying staff to different areas. “Do we have a Linux cluster? Do we have modeling software? Are we able to do virtual screening? We have all those capabilities, and if the need arises, we’ll make it happen. But this is a lean and mean operation, and we just try to be very careful where we apply the most expensive resource of all, which is the human resource.”
There are three full-time IT staff handling the network architecture and building expansion. The informatics team is tightly integrated, with core areas including clinical informatics (data management, regulatory submissions); discovery informatics (molecular modeling); cheminformatics; and bioinformatics. There’s also business informatics, currently focused on supporting the finance team but likely to change as the commercial group expands. “We’re Sarbanes Oxley compliant too. It’s a pretty eclectic mix.”
The IT group also handles EDC user management. Says Keilty: “We’ve leveraged a core software engineering team, 1 full-time database, 1 part time, and three engineers. These teams can leverage other core competencies within the team. So far it’s worked pretty well.” K.D.
This article also appeared in the July-August 2009 issue of Bio-IT World Magazine.
Subscriptions are free for qualifying individuals. Apply today.