July | August 2006 | Pharsight, a market leader in clinical trial design and simulation software and services, announced in June a five-year CRADA with FDA’s CDER. The agency will use Pharsight software [primarily Pharsight Knowledgebase Server (PKS)] to build a data warehouse and informatics infrastructure for its nascent disease modeling program. This is likely an important deal for Pharsight and a significant step in the FDA effort to incorporate modeling in drug approval decision-making.
It’s also more demonstrable support — though lacking funding — for the Critical Path Initiative (CPI). The stated purpose of the collaboration is to develop PKS into a repository for the data needed for modeling and simulation, to support Clinical Data Interchange Standards Consortium (CDISC) data formats, and to interact with other FDA databases.
Pharsight will use FDA’s feedback to further develop PKS to meet industry analysis requirements, and further develop DMX (Pharsight’s visualization and collaboration tool) to support Agency-sponsor interactions for the visualization and communication of model-based product profiles.
The CRADA announcement caps a busy period for Pharsight. The company finished its second consecutive profitable year in March, despite a sluggish fourth quarter and flat revenues year-to-year. In mid May, it was awarded a patent (No. 7,043,415) for a graphical method of building PK/PD models. This interface, already deployed in Pharsight’s Trial Simulator, lets users place and connect PK/PD objects that are then automatically converted into a computational model.
Now if only its stock would budge — Pharsight shares are trading around $1.50 at the time of this writing and have been under $2.50 for the past year. Market enthusiasm for life science tool providers has been lukewarm, at best, and even more guarded for software plays.
While Pharsight has enjoyed a leadership position among trial simulation software and services providers, it’s only in the past few years that commercial software has taken hold in the clinical trial. Although PD/PK analysis is required, the use of modeling PD/PK profiles to design trials (dosing, patient segmentation) or do competitive drug comparisons is fairly recent. Now, the outcry over poor pipeline productivity is accelerating the use of simulation tools.
“If you look at Pharma right now, modeling is going on all over the company,” says Dan Weiner, Pharsight SVP, software products. “Not just by the kineticist and clinical pharmacologist. You’ve got formulation developers trying to optimize formulations. You’ve got preclinical scientists doing modeling, trying to understand a drug’s activity and pick lead candidates. All these different scientists are using different tools, the tools don’t communicate. They’re not using a common database, so if one department wants to try to utilize results of another department, they have to reenter data, rebuild models.”
Pharsight is betting its PKS product, central to the CRADA, will knit together disparate modeling efforts and perhaps even expand its market beyond core clinical trial pharmacokineticists.
Modeling and simulation got a public boost after being singled out in the CPI in 2004. Now it’s getting another shot in the arm with FDA’s efforts to actually use the tools. Bob Powell, head of pharmacometrics in FDA’s Office of Clinical Pharmacology, is a driver behind the modeling program and the CRADA. It probably didn’t hurt Pharsight that Powell is a former employee. At one time, he led its consulting business.
With Powell and others’ help, FDA is pushing two related programs. One is a call for so-called end of Phase IIa (EOP2a) meetings, during which sponsors have a nonbinding meeting with FDA and bring their data. FDA will build a model based on sponsor data and relevant historic FDA data from other trials and perform simulations to guide next steps (dosing, trial size, attrition) and even suggest go-no-go decisions.
“We’ve had quite a bit of interest,” says Powell, but only a few recent requests for meetings. (See the figure for the proposed EOP2a process.) He expects interest and the number of EOP2a meeting requests to grow as the program receives more visibility.
Powell is also piloting a program to build a database of models, again based on FDA data, that would eventually be made broadly available to sponsors. Currently, his team is looking at “disease models in Parkinson’s disease, Alzheimer’s disease, non-small-cell lung cancer, diabetes, and osteoarthritis.” Data from FDA and in some cases NIH are being used to build these models.
Much of the effort is around biomarkers. “We have very targeted questions for all these efforts,” says Powell. For example, in the case of lung cancer, FDA is looking at MRI imaging to measure tumor volumes. Other models look at other pathology and physiological parameters. “We need a warehouse designed for this kind data, and drive support for CDISC.”
Powell points to FDA’s recent creation of an Office of Translational Sciences — his group, Clinical Pharmacology, now falls under it — as more evidence of the agency’s commitment to innovative technologies. In any case, FDA’s current modeling and simulation pilot program, the recent CRADA, and growing acceptance of its software inside Big Pharma are positive signs for Pharsight.
The company was incorporated about 10 years ago. “Initially, it was to focus on modeling software and maybe even then as a replacement for the NONMEM program [UCSF-developed population PK/PD program, now licensed by Globomax] that’s been out there,” says Weiner. Early feedback from the pharmaceutical industry prompted the addition of a consulting arm focused on study design. “There was emphasis even then on trying to improve the failure rate in drug studies,” says Weiner.
Even in the early days, Pharsight had a trial simulator program. In 1998, Weiner “had a separate software company SCI — Scientific Consulting Inc.” that owned a strong product, WinNonlin, for PK/PD modeling. Pharsight acquired SCI to bolster its software line. Shortly after, “We acquired MGA — Mitchell Gauthier and Associates. They owned a product called ACSL (advanced computer simulation language) and also had a [toxicology] simulator product,” says Weiner.
What emerged from the amalgam over time is the current product line spanning data management, analysis, and visualization (WinNonlin; Trial Simulator; WinNonMix, a population PK modeling tool; Drug Model Explorer, DMX, a visualization and collaboration tool; most recently Pharsight Knowledgebase Server, PKS).
Mark Hovde, SVP, marketing, brags there are 3,000 WinNonlin users in industry and an equal number in academia. “It’s the dominant tool by far, answering basic questions like drug half life, bioavailability, basic PK curve fitting type analysis,” he says, “and then trial simulator is probably in the low hundreds of users.”
Pharsight now hopes PKS can expand its horizon. It is the company’s first enterprisewide tool — the others are desktop based. Weiner and Hovde hope it will not only capture their hardcore customers in the clinical trial PD/PK analysis and modeling community but also help the company expand into the preclinical and post-marketing space.
“So you give the patient the drug and take his blood at one hour, two hours, three hours, build a nice curve of time versus concentration. PKS is a data repository that is specially equipped for that kind of time-stamped data, and to make it easy for other analysis tools to work on it in a regulatory-compliant way,” gushes Hovde. “It’s the backbone underneath an efficient model-based drug development organization. So you have Oracle Clinical in the clinical space, kind of the big gorilla in the infrastructure; then you’ve got your LIMS systems that are back a step, and in the middle of that is PKS.”
Pharsight has far to go before catching up with Oracle in terms of seats. Weiner estimates there are close to 1,000 PKS seats sold, including inside 16 to 18 big customers. But this early link with the FDA and chance to influence FDA preferences on modeling data formats and approaches can’t hurt.