COGNEX EXPANDED ACCESS PROPOSED BY FDA AS PART OF "PROGRAM OF FURTHER STUDY"; WARNER-LAMBERT WILL SUPPORT NDA STUDIES WITH OTHER DATA BY APRIL 1

FDA is proposing that additional studies of Warner-Lambert's Alzheimer's drug Cognex (tacrine) be designed to include expanded patient access as well as allow the rapid collection of data. In a March 22 letter to Parke-Davis Pharmaceutical R&D Division President Pedro Cuatracasas, MD, FDA Neuropharmacological Drugs Division Director Paul Leber, MD, said that expanded access is one of "a number of steps we hope Warner-Lambert will take with respect" to the Cognex NDA. The letter from Leber is part of an effort by FDA to clarify the status of Cognex in the aftermath of FDA's Peripheral and Central Nervous System Drugs Advisory Committee meeting March 15 ("The Pink Sheet, March 18, p. 2). The committee was unable to make a recommendation to approve the drug because of unresolved questions about efficacy from two pivotal trials that were designed for expedited review of the drug. The letter to Warner-Lambert is put in the context of "recent discussions" between the company and FDA Office of Drug Evaluation I Director Robert Temple, MD. The letter from Leber suggests that the company consider discussing with the agency a "program of further study that would allow both expanded access and the rapid collection of useful scientific data. A possible design would be a good-sized, relatively low-tech, parallel design study of several doses of tacrine with provision for occasional blood level sampling." The letter notes that Leber and Temple are "prepared to meet with [company representatives] at your earliest convenience to consider further analyses that may shed light on which patients are most likely to respond significantly to tacrine (e.g. examine effects of compliance, body weight, and habitus, renal function, and other demographic features)." Leber added that FDA is "especially interested in the possibility that good response is related to the presence of the 'slow metabolizer' phenotype for debrisoquin hydroxylase." Additionally, the letter expresses interest in a further exploration of "the available evidence" concerning lecithin's role in Alzheimer's "in general" and in the second Cognex pivotal trial (the Levy trial) "in particular." The agency also wants to look at the status of ongoing controlled tacrine studies. Warner-Lambert indicated that it is still not ready to give up on the data submitted to FDA in the Cognex NDA, despite the advisory committee's unwillingness to recommend approval. The company said that it would consider any of FDA's proposals for additional studies as independent from the ongoing review of the Cognex NDA. The company said March 22 that it plans to submit a supplemental data package to the agency by April 1 and that it will meet with FDA officials face-to-face 10 to 14 days after that date. The company stressed that it has had no substantive discussions with FDA to date about expanded access. Warner-Lambert's plans apparently changed during the week after the advisory committee meeting. The company initially reported on March 18 that it was going to supply additional data to the agency "within the week." That data package, according to the company, was to have included blood plasma levels from patients in its second Cognex pivotal trial, a crossover study conducted in the U.K. on 92 patients who received both lecithin and tacrine. The submission was intended to show the "non-role" of lecithin in confounding the efficacy results of Cognex. At the March 15 meeting, the FDA advisory committee had difficulties with both of Warner-Lambert's pivotal efficacy trials. The first pivotal trial (970-1) was an enriched, double- blind, placebo-controlled, parallel group, randomized trial conducted at 16 U.S. centers. The second pivotal trial (970-33), a crossover design, was conducted by Raymond Levy, Maudsley Hospital Institute of Psychiatry, London, under the auspices of the U.K. firm Shire Pharmaceutical. Warner-Lambert acquired the data from Shire in April 1990. Because the committee did not resolve the efficacy questions from the two studies, it did not move on to discuss formally questions of Cognex' safety and marketability. Warner-Lambert has several ongoing studies with Cognex from which the company might be able to collect data that resolve some of the advisory committee questions. The company said that "more than 100" people are continuing drug treatment with Cognex in a "sustained active" phase. Warner-Lambert has ongoing Cognex studies in the U.S., France and Canada. However, none of the studies are of a double-blind, placebo-controlled, parallel group design. At an R&D update last November, the company said it had a U.K. study nearing completion comparing the safety of Cognex b.i.d. and q.i.d., plus two U.S. and French rechallenge studies on patients with elevated liver enzyme levels. The company was also conducting another French study using 100 mg/day. Warner-Lambert says that it has no plans to conduct other higher dose studies at this point. The Shire study included dosing of up to 120 mg/day; the U.S. trial included doses of up to 160 mg/day before it was halted; and the controversial 1986 Summers tacrine trial used levels of up to 200 mg/day. Warner-Lambert began its public relations damage control with a March 18 statement. The company said it "believes that it is in everyone's interests -- patients', caregivers', FDA's and the company's -- to see Cognex approved. And we believe the way to do that is to resolve the confusion" of the advisory committee's outcome by, among other things, providing more data. Warner-Lambert is likely to pursue an aggressive lobbying campaign in Washington. The company waged a very successful legislative lobbying campaign in 1987-88 to win passage of a three-and-one-half-year patent term extension for Lopid (gemfibrozil). Outside forces are also continuing to exert pressure on FDA about Cognex, a fact that led the agency to note the public hue and cry for approval in its critique of the Summers study published in the Jan. 31 New England Journal of Medicine ("The Pink Sheet" Feb. 4, T&G-6). At the public session of the advisory committee meeting, seven speakers asked FDA to approve tacrine. In a March 18 editorial in The Wall Street Journal, entitled "The AIDS Hoax," the newspaper laid the blame for the advisory committee's decision not to recommend approval squarely at the feet of FDA and its drug approval system. In the editorial, the Journal claimed that the defeat of tacrine -- a drug on fast-track approval and designated a "1A" therapy by the agency -- was comparable to FDA's earlier mishandling of AIDS drugs. The editorial concludes: "The lesson here is clearer than ever: It is going to require direct intervention from the White House to make the U.S. drug-approved system serve patients first." The fall-out from Cognex -- public relations, political and otherwise -- followed a marathon 14-hour session at which the Peripheral & CNS Drugs Advisory Committee adjourned at 10:15 p.m. without resolving whether the two pivotal trials established efficacy and were "adequate and well-controlled." The session, at various times confusing and contentious, saw the advisory committee take 10 votes on the efficacy of Cognex (see box, page 8, for encapsulated votes and tally). One of the votes on 970-1 was critical, reversing the committee's position from affirmative to negative on whether the protocol proved a beneficial effect from tacrine in the double-blind, parallel group portion of the trial. Part of the contentiousness arose from presentations by two statisticians, Chuck Davis, PhD, University of Iowa, and FDA's Jonathan Levine. Both statisticians disputed all or parts of Warner-Lambert's statistical analyses. In the end, the committee could decide only that the dose- titration portion of Warner-Lambert's 970-1 protocol showed some statistically significant "small effect" as measured by one standardized cognitive assessment, the ADAS-C. Committee member Sid Gilman, MD, University of Michigan, called it "a weakly beneficial effect" and noted that "a good trial should include more scales [for assessing benefit] than one." Michael Brooke, MD, MacKenzie Center for Health Sciences, University of Alberta, said there were "lots of reasons giving me great doubt" about the effect of tacrine, and noted that any small effect that might be demonstrated he "could not rule out" as coming from "some parallel effect." The committee found that the positive effect found in protocol 970-33 was "confounded" by a number of complicating factors, including the concomitant use of lecithin and the crossover design itself. Committee Chairman Ferris prefaced a discussion of the specifics of the trial's results by noting that his "informal feeling" was that he had "a vague sense that the level of enthusiasm is lower" than for 970-1. He also commented on what he felt was the "post hoc nature to the conclusions" reached by the study. FDA's Leber tried to get the committee to use 970-33 in some form or other to support the first pivotal trial: "How would you argue that this study could be used?" Gilman responded that the Mini-Mental State (MMSE) cognitive assessment was "interesting. Put together with 970-1, I inferred a weak effect." 970-1, a fast-track protocol agreed upon by FDA, Warner- Lambert and the National Institute of Aging in 1987, started with 632 patients. In the six-week double-blind dose-titration portion of the trial, patients were randomized to two week dosing levels with placebo, 40 mg/day or 80 mg/day (placebo-40-80; 40-placebo- 80; and 40-80-placebo). After two weeks on Cognex, patients had to show a four-point improvement in the Alzheimer's Disease Assessment Scale (ADAS-C). Of 231 responders (the enriched population), 90 patients responded best to 40 mg while 141 patients responded best to 80 mg. Patients were then rerandomized to a six week double-blind phase with placebo (112 patients) or best dose (103) and were assessed via ADAS-C and another primary objective scale, the Clinical Global Impression of Change (CGIC), every two weeks. A six-week sustained activity phase followed, with both groups receiving their best dose. Again, 80 mg was the best dose for more patients than 40 mg. The study found that 41% of patients receiving Cognex showed an improvement of four points or more on the 11-point ADAS-C scale versus blinded placebo. Typically, Alzheimer's patients decline four ADAS-C points every six months as their condition deteriorates. However, there was no statistically significant improvement as measured by the CGIC scale, which is administered by physicians. Over a 22-week period, 175 patients showed a one- point gain in a secondary, subjective cognitive assessment. During the first 13-week period of Levy's study (970-33), 94 patients received either Cognex plus lecithin or placebo. After a three-week washout period, 64 patients were then crossed-over to placebo and Cognex plus lecithin. Dosing levels were increased in 20 mg increments from 40 to 120 mg/day. The patients received 10.8 mg/day of lecithin to "ensure adequate plasma levels of choline" in the blood, Parke-Davis CNS Clinical Development Director Stephen Gracon said. The results of 970-33 found that 44% of the 92 patients receiving Cognex improved by three points or more on the 30-point Mini-Mental State Examination assessment scale versus 11% on placebo. Gracon explained that three points is "comparable to one or more years of expected decline." The study employed five assessment scales; three were considered primary. At the end of 30 weeks of study, those on Cognex plus lecithin who crossed-over to placebo were "no less worse than at the start," Gracon said. At the conclusion of the company's data presentation in the early afternoon, Warner-Lambert Pharmaceutical Research Division VP and Chairman Ronald Cresswell, PhD, provided several caveats: "We recognize that tacrine is not a cure for Alzheimer's disease"; "The studies that were presented are not perfect. We recognize that one of the primary outcome measures of 970-1 was not met. We believe that is balanced by the fact that there was a very positive outcome of the ADAS and that there were a lot of positive outcomes on a lot of subjective outcome measures"; "The studies do leave us with a less than perfect understanding of the fine points with respect to benefit and risk. We would like to know more and clearly we will be monitoring this carefully and several studies in this area will be done." Cresswell stated that the company believed that the studies still met "the regulatory requirements for substantial evidence of well-controlled studies." Cognex safety data was discussed by Parke-Davis Senior Director Clinical Development-CNS Jan Wallace, MD. Addressing the longstanding concerns about the incidence of elevated liver enzyme levels in patients receiving Cognex, Wallace said that 48% of patients treated with the drug were within normal ALT (alanine aminotransferase) limits; 24% had "greater than three times the UNL (upper normal limit); and 2% had greater than 20 times UNL." Women had a 50% higher chance of elevated ALT. The average times for onset of elevated ALT was 35 days of treatment, with 95% of patients showing the increases by day 92. The time to peak ALT values was 64 days with a 25-day recovery period from peak. To date, Wallace said, all patients with elevated ALT "have returned to normal." FDA concern over the elevated ALT led to two closed sessions, in 1987 and in July 1989, before the Peripheral and CNS Drugs Advisory Committee and a one-time halt to Cognex clinicals. In October 1987, U.S. trials were stopped after six weeks. The trials were resumed in February 1988 without the 100, 120 and 160 mg/day doses. Warner-Lambert consultant Paul Watkins, MD, University of Michigan, reported that the highest level of elevated ALT was greater than 50 times UNL. Noting that "the sensitivity of individual patients varies greatly," and that it is hard to identify high risk patients at the start of therapy, Watkins suggested that monitoring of blood enzyme levels is "very important." FDA's Leber noted that the dosing levels for Cognex in 970-1 were low compared to the "dramatic results" found in patients in the 1986 study conducted by William Summers, et al., where patients were receiving 200 mg/day. He asked Watkins "would you be frightened" about elevated ALT if the drug is approved at these levels and physicians titrated their patients up to higher doses. Watkins replied that there is "no question of dose relation to the severity of injury; however, toxicity has shown up at both low and high doses. But, we would have to suspect the problem would be worse." FDA's Temple asked Watkins about the incidence of jaundice in Cognex patients. There were two cases confirmed in 1,500 patients. Citing a potential for non-reversible liver damage, Temple said that monitoring is "very important" and suggested that liver biopsies should be used, "especially on patients with low, smoldering" elevated ALT. The committee also considered whether the data was sufficient to determine a persistence of effect, as measured by ADAS-C, for tacrine. Gilman said no, that 970-1 was "too brief," adding there was "not enough data to say clearly [there was] a persistent effect." He added that there was, however, "some suggestion of a lessening of effect over time." FDA's Leber said the agency needed some opinion on the matter of sustainability of Cognex therapy because FDA would have to consider the duration of the drug's use for labeling. Gilman responded: "We have no data." Ferris later asked if blood level data was "informative," to which Gilman replied in the affirmative, saying the blood plasma data suggested an "important dose effect here." Leber noted, however, that during the sustained phase of the trial, Cognex responders "were still losing four-and-one-half ADAS points," suggesting that the improvement from tacrine early in the clinical trial was short-lived. Asked by FDAer Temple to explain their reasons for voting "no" on whether 970-33 was adequate and well-controlled, the committee outlined numerous problems with the study. These included concerns about the crossover design of the trial, its concomitant use of lecithin, the appearance of two different baseline measurements, and the use of multiple objective cognitive assessments. Statistician Paul Meier, PhD, The University of Chicago, a sit-in guest with the committee, also questioned the study's patient randomization methods. He suggested that the different baselines could "mean a loss of impartial randomization." All told, Ferris described the concerns as "a host of reasons why we don't unequivocally accept" 970-33. Warner-Lambert defended the use of lecithin, pointing to a study recently published by the committee's permanent chairman, Leon Thal, MD, Veterans Administration Medical Center, San Diego, that found only one incidence of effect on blood plasma levels from lecithin. Elkan Gamzu, PhD, Cambridge Neuroscience Research VP- development and the principal data presenter for Cognex, said it was "highly improbable that lecithin raised the plasma choline level significantly or even at all." He noted that the lecithin used in 970-33 was "not of very high level purity anyway" to show up in the blood plasma levels. Ferris pointed out that "our opinion" about 970-33 "doesn't hinge on lecithin. I don't think [it's] a pivotal issue," he said, "but it doesn't entirely go away by referencing [the ingredient's] lack of effect in literature." He later went on to say: "I don't know if a single smoking gun exists, but cumulative problems" with 970-33 were responsible for the committee's "no" vote. Calling the 970-1 protocol "a gamble" because of the short-cuts it allowed, Ferris indicated that the committee would have been more persuaded by the data if it had shown "clean, clear-cut" evidence of substantial improvement in the assessments of cognitive function among Alzheimer's patients. In an eleventh hour effort to move the committee forward from efficacy to votes on tacrine's safety and marketability, FDA's Temple noted that "we have in some instances relied on only one adequate and well-controlled study" e.g. with AZT, timolol and the Parkinson's drug Eldepryl. "Is there any basis," he asked, "for us considering one study here." Gilman responded: "I don't see how we could urge you to do that on the basis of what we've heard." Herbert Weingartner, PhD, NIA, concurred as did Antonio Delgado- Escueta, MD, University of California Los Angeles. He told Temple: "No, we need a cleaner design" to be able to accept only one study. FDA's Leber, trying to pin the committee down on any short- term fixes to their unresolved questions, asked "Are there any shorter-term, remedial actions" FDA or Warner-Lambert could take. The consensus was no. Ferris added: "If we dismissed the lecithin problem, it wouldn't be enough to change my opinion." As the meeting wound down and Warner-Lambert's hopes dimmed, Cresswell remarked that the drug is the only treatment that has shown any effect on Alzheimer's and therefore should not be denied to those patients. "In the meantime," while the company provides more data to FDA he said, "if you deny that drug to a large population that has no other treatment, you deny it for at least one-and-one-half to two years." Acting Committee Chairman Steven Ferris, PhD, New York University Medical Center, responded: "I don't think we should be influenced by what I would portray as an emotional side to this story," noting that the committee had to make a "considered judgment" based solely on the available evidence.