COMPUTER-ASSISTED NDAs HAVE "POTENTIAL" TO CUT BACK APPROVAL TIMES, FDA REVIEWERS INDICATE IN SURVEY OF FDA AND INDUSTRY EXPERIENCE WITH CANDAs

FDA reviewers believe that the use of computer-assisted NDAs (CANDAs) may lead to shorter approval times, according to a survey sponsored by the Pharmaceutical Manufacturers Association on the use of CANDAs and conducted by Lexington, Kentucky-based Statistical Consultants, Inc. Presenting the preliminary results of the survey at a joint PMA/FDA meeting on CANDAs held July 16-17, Statistical Consultants President Dennis Haack, PhD, noted that "there's general consensus among the reviewers that we talked with that the potential to reduce the approval time is definitely there by use of CANDAs." Haack cautioned that the survey does not provide conclusive proof that CANDAs reduce approval time. "Our experience is, that as far as the information obtained from this particular survey [goes], the jury is still out," Haack said. "Although, there are indications that the approval time or review time had been reduced, many of the CANDAs that we were talking about...are still within the review process." A future survey would have to be conducted on the CANDAs after they are approved in order to see whether approval times were affected, Haack noted. The survey, entitled "The CANDA Project Evaluation Report," was developed by the PMA's CANDA evaluation task group. Syntex Associate Director-Human Pharmaceutical Regulatory Affairs Carol Grundfest noted that invitations were sent out in March 1989 to 13 companies with experience in a total of 24 CANDA projects to participate in interviews for the survey. Grundfest was formerly with PMA and played a key role in developing the survey. CANDA sponsors were given questionnaires to determine their views on designing and implementing CANDAs. The consultant firm asked FDA reviewers about their individual experiences in using CANDAs during drug reviews. Haack noted that over 75 company representatives and over 40 FDA reviewers were interviewed. The report will be released in September by PMA. A possible reduction in approval time from CANDA use may come "from a number of factors," Haack said. "One of those factors is a reduction in 'dead time.'" He explained that "presumably, if the reviewers have access to data, there will not be a need to go back to the sponsors and ask for re-presentation of the data." Another factor that could shorten approval times, Haack suggested, is the ability of CANDAs to increase a reviewer's comfort level with the product and the data. "A number of reviewers indicated that by having access to the data...the possibility exists for a shorter review time or shorter approval time because of the greater chance, the earlier chance, to have a high comfort level with the product [and] with the submission," Haack said. CANDAs can also help in an FDA reviewer's preparation of the review document for the NDA, Haack reported. He noted that reviewers like to take key information from a CANDA, such as a summary of the pivotal studies, print it out on a wordprocessor, and cut and paste the information into the review document. "The ability to do that reduced the time that it took to prepare the review document, and was a feature, [a] capability that the reviewers were very much interested in and wanted to see," Haack told the PMA/FDA meeting. During a Q&A session, CDER Office of Management Director Robert Bell said he believed "that the computer-assisted NDA will in fact cut down review time." Bell was responding to a question on what incentives sponsors have in investing in computerized NDAs. Bell pointed out that, so far, "a subset" of five CANDAs have been submitted simultaneously with paper versions and subsequently approved. "The overall average [review] time on those five [NDAs] is 17 months, instead of the normal 28 [months]," Bell noted. Statistical Consultants' interviews with FDA reviewers revealed that "nearly all of the reviewers...wanted to see other CANDA submissions," Haack noted. "We thought that was a very important result because it leads to the fact that we think that the computers and computer-assisted submissions are here to stay." They also discovered "the desire to be at the forefront of technology both from the company and the agency standpoint." One benefit to CANDA systems is that the "process of interacting back and forth regarding use of the CANDA created a very good atmosphere of communication between the sponsor and FDA," Haack noted. The survey also found that a CANDA project caused "a remarkable increase in communication within the sponsor company" because of the necessary interaction of different groups within the firm and with FDA, Haack said. However, the survey did not find an increase in communication within FDA as a result of CANDA usage. One way to improve communication within FDA, Haack said, "would be to have a CANDA users group, and there was interest expressed in this within the agency." Such a group, he remarked, might facilitate communication on CANDAs within a division as well as across divisions. Bell expressed his support for the idea of having a CANDA users group. "The idea of communicating between us what we're doing, what we have done, what we see coming, and how we can help each other, I think, is awfully important. I think a users group would be very beneficial." The survey also looked at specific features across CANDA systems and whether they were useful or not useful to FDA and/or industry. Statistical Consultants VP Steve Albrecht noted that awareness and use of CANDA features by a reviewer depended on the reviewer's computer experience. He said that, in general, statisticians have a strong computer background, whereas clinicians have a weaker background. The evaluation found that the systems and users manuals were problems for reviewers. "Manuals in the computer industry are as a whole...not well written," Albrecht noted. The reviewers "tool for help" was calling the companies. "By and large...the FDA was very happy with the responses they got in response to their questions." The survey's findings on use of electronic mail in CANDAs was suprising to the interviewers. "We thought electronic mail was used extensively and used effectively to augment the review process and it was in some cases," Albrecht said. However, in certain circumstances, reviewers used electronic mail and "they found that it took just exactly the same amount of time to get a response back from the company as if they had just sent them the request." He added that "unattended electronic mail translates into down time immediately." A key feature in many of the CANDAs is a full text search capability. "If you want to look at an adverse event across documents, across studies, full text search is the only way to do that quickly, and that's where the functionality of a CANDA really shines." Case report form access was "extensively useful at the FDA and very easy to use," Albrecht noted. The feature "allows you to sort and go through thousands and thousands...of pages automatically." The note storage feature was not used much by reviewers. Statistical Consultants also provided advice to industry on CANDA development. Haack said that there are two extremes that companies should avoid. "The first extreme is that you don't give any consideration whatsoever to an electronic-aided submission. The other extreme is to, independent of interaction with the agency, develop a complete CANDA and submit that CANDA." The consultant firm also recommends that companies use their CANDAs in-house so that clinical personnel can use them during the drug development process. In-house use, Haack pointed out, would allow company personnel to become familiar with the systems, debug them, and possibly help reduce the drug development time. FDA's Center for Biologics Evaluation and Research Associate Director for Research and Regulatory Coordination Elaine Esber, MD, presented a "Points to Consider" document on computer-assisted submissions for license applications. The document provides information on what CBER will require a manufacturer to do before and when filing a computer-assisted product license application or an establishment license application.