OR WAIT null SECS
Peek into a future where drugs are approved faster and R&D is less costly thanks to technology.
With the new year upon us, this column would like to to look back at the past one to review some of the important trends. In fact, the past year was a tipping point in the adoption of electronic technology in clinical trials. The industry has been on an accelerated path of adoption of electronic data capture (EDC) for several years, but I believe that we will look back on 2007 as the year in which the most significant changes in attitudes and adoption occurred.
Paul Bleicher MD, PhD, is the founder and chairman of Phase Forward, 880 Winter Street, Waltham, MA 02451, (888) 703-1122, email@example.com, www.phaseforward.com. He is a member of the Applied Clinical Trials Editorial Advisory Board.
But rather than focus on the recent past or near term, I would like to paint a picture of clinical development as it might be at some distant future point, enabled by technology and fashioned by a variety of current and future pressures. I know some of you will tell me that I wrote a column back in October on The Black Swan that discussed the impossibility of prediction. I freely admit that I am not a futurologist and I do not expect this world to ever come about just as I describe it in the following paragraphs. In fact, these ideas are somewhat of a reductio ad absurdum: an extreme extension that would likely be limited by the slow pace in pharma and other factors. The future we all will likely experience will probably be much closer to the present than what I am about to describe.
The process of drug development and drug approval is under pressure from a variety of different and sometimes contradictory forces. The forces of change that are most relevant to this discussion are drug safety, access to new medicines, and the costs of developing new medicines. The one with the most potential to create substantial change in the approval process is drug safety.
We are currently in a cycle where concerns over public safety are at the forefront of public and regulatory discussion. In the United States, several major pharmaceutical companies have been under attack financially for safety issues in widely prescribed medicines. The public is not sufficiently educated in the meaning of risk and benefit in prescription drugs to understand safety issues in perspective and demands safer drugs through lawsuits and legislation.
Unfortunately, it is quite evident that the clinical trial process today is insufficient to identify, characterize, and calculate all of the safety issues in a new medication. The number of patients studied in a typical NDA is too few to identify serious adverse experiences, let alone demonstrate whether they are statistically associated with the drug. In addition, reliance on spontaneous and literature reporting of adverse experiences is flawed and not adequate alone for safety surveillance. The current trend is to examine such safety issues, and even risk/benefit considerations, through careful monitoring of patients postapproval through longitudinal analyses and registries.
On the completely opposite end of the spectrum, the public also consistently demands and hopes for the development of new, life-saving drugs for serious conditions. These demands are intensified as new health threats emerge. Of course, anyone reading this column is well aware of the tremendous costs for identifying and testing a new therapeutic agent. This is compounded by the trend to narrower indications and more "personalized" medicines.
Meanwhile, on the technology front, in parallel with the emergence of EDC, we are witnessing a similar adoption of technologies for the delivery and monitoring of medical care. However, the adoption curve may well be different for such technologies, as the economic pressures and value achieved by these technologies aren't necessarily apparent to those who might pay for them. The technologies that are most relevant to my argument are universal access to the Internet and cell phones, electronic health records (EHR), and personalized home monitoring.
Universal access to the Internet is arriving rapidly in the developed world. While not every family has a computer in their home, most have free access through public libraries, at least in the United States. This access will make many things possible, including the monitoring of individual patient experiences with drug therapies.
The emergence of EHRs is already complete in some countries, where virtually 100% of patient encounters are recorded electronically. However, in the United States, estimates range in the 20% to 25% range and are increasing slowly. The potential for patient encounter data entered into EHR systems is enormous. Such data can be used to look at longitudinal physician-entered data on drug safety and efficacy without requiring specific and separate entry into registry databases. Unfortunately, even established EHR systems often are set up for scanned, faxed or blocks of textual data entry that prevents efficient use of the data for such purposes without substantial intervention. However, changes in health care and a focus on cost effectiveness and outcomes by payors may make it more likely for such systems to be developed and/or installed with the intention to extract and examine longitudinal data.
Finally, the availability of personal home monitoring technologies is increasing. These include patient/provider Web-based communications, ambulatory blood pressure, glucose, peak respiratory flow monitoring, and many other emerging technologies. Along with standardized tests such as home pregnancy and home coagulation tests, there are companies developing a variety of assays for home use, and it is conceivable that even pharmacokinetic sampling could be performed in the future through home devices.
As the airport poster (for Accenture, I believe) says, "This is where it gets interesting." Given a highly creative and rational regulatory environment, one can speculate on a different model for drug approval than exists today. Fanciful as it is, you will see that the model addresses the pressures previously outlined and depends on the technologies mentioned. So, let me describe for you how technology can enable a very different process for drug approvals.
First, there are many existing and developing preclinical, in silico technologies and databases that allow for more effective screening and analysis of drug candidates for known, likely toxicities and safety issues. While these are in their infancy, it is easy to imagine a time when companies will have a substantial amount of information and some confidence about the safety issues to monitor/address in clinical development. Further, Phase 0 clinical trials will also provide information on ADME and tissue distribution that will compliment this information. However, it is difficult to foresee much of a change in the need for Phase I clinical trials.
In this future state, I suggest that drugs might be approved based upon clinical trials that are closer to Phase II in style than Phase III. Initially, this could be used for approval of urgently needed therapies and would only be applicable for some types of drugs. These clinical trials would enroll patients from a more narrowly defined patient population than a typical Phase III, so that variance is kept to a minimum. The endpoints of the clinical trials would be based upon (also dependent on the therapy and the indication) ambulatory endpoints and patient-reported outcomes. For example, rather than measuring peak flow or blood pressure as endpoints at a three month outcome visit, these could be measured frequently or continuously by the patient using home-based devices and communicated via the Internet to a central server. In addition, it might even be possible to correlate this information with drug blood levels, obtained through home-based blood sampling. Certainly, the use of ePRO technologies could link the data, at the moment it was collected, to patient outcome data.
The statistical analysis tools for comparing continuously monitored data exist but might need to be extended, and regulatory reviewers would need to accept such endpoints as a basis for approval. The availability of such data from ambulatory patients, along with adapted statistical techniques, would allow the determination of clinically meaningful, statistically-significant endpoints (for effective drugs, that is) with far fewer patients than would be needed for standard Phase III clinical trials. Potentially, it may be possible to correlate some endpoints in these patients directly with ambulatory pharmacokinetic measurement, to examine pharmacokinetic/pharmacodynamic correlations in ambulatory patients instead of Phase I subjects. In addition, safety parameters could be followed much more closely in these patients, and again possibly correlated with drug levels.
The result of these trials might be a greatly accelerated approval, shaving years off of a clinical development plan of today with far fewer patients tested. This would get much needed drugs into the hands of clinicians and patients earlier, and would likely be able to better pinpoint the subset of patients who benefit from the drug. The earlier and faster approval would reduce costs of development, and mitigate the downside of developing drugs for smaller patient populations.
The obvious response to this vision, from the traditional perspective, is that the drug would not have been tested in broader patient populations and physician settings, and that sufficient data would not have been achieved to ensure safety. Both of these issues can be addressed in the postapproval phase, as discussed in the next section. The safety issue is an interesting one. Certainly, recent history suggests that clinical trials are often insufficient for identifying the safety issues that arise from real-world use after approval. The difference between monitoring 700 and 2000 patients for unusual and rare adverse experiences is certainly real, but both dwarf in comparison to data collection from hundreds of thousands or millions of patients.
Approval on the basis of smaller numbers of patients in Phase II is only acceptable if large numbers of patients can be routinely followed and analyzed in "real time" after approval. Thus, patients using drugs that are approved using this route would be closely followed through registries, EHR data, cell phone, and Web-based instruments for patient-reported outcome data collection and other longitudinal data sources which would be designed to collect and structure data in standardized formats for such analysis. Mechanisms for centralizing the data on regular basis (likely daily or weekly) and a requirement to continuously monitor and report on the data using data mining and data visualization tools would allow for the study of the efficacy, safety, and cost-effectiveness of the drugs in actual use, rather than the artificial setting of a clinical trial.
Few would question that the data so obtained would be more clinically relevant than that obtained from even Phase III clinical trials. Such data could be automatically collected from very large numbers of patients, allowing for identification and study of associations with rare adverse experiences. Although the data would not be as tidy as that obtained in a clinical trial, the quantity of important data would permit proper analysis for conclusions about safety, efficacy, and cost considerations.
The obstacles to changing drug development to allow for this type of accelerated approval/monitoring process are very significant, but the tide seems to be slowly moving in this general direction. In addition, there are other logistical issues around patient privacy that would need to be overcome.
The one gating factor in this is the adoption and use of technologies that would allow immediate access to trial, patient experience, and longitudinal data. As an industry, we have made significant progress toward the first two of these (as mentioned in the opening paragraph), and there are many pilots underway on the latter. Nevertheless, the vision I have put forward would be entirely dependent on extensive and standardized adoption of technologies across the health care community, which most believe is many years away. Until then, it is likely we will continue depend on the tried and true method of drug approval and safety assessment.