Cover Image

PAPERBACK
$34.00



View/Hide Left Panel

14
Review of the FDA Task Force Report "Fialuridine: Hepatic and Pancreatic Toxicity

On November 12, 1993, the Food and Drug Administration (FDA) released a report "Fialuridine: Hepatic and Pancreatic Toxicity" that had been prepared by a special task force. This report was presented together with an oral report by the authors to the Institute of Medicine (IOM) Committee to Review the Fialuridine (FIAU/FIAC) Clinical Trials on July 25th, 1994 (FDA, 1993a).

OBJECTIVE OF FDA REVIEW

The deaths of 5 of 15 participants in the clinical trial H3X-MC-PPPC, evaluating the safety and efficacy of FIAU in patients with hepatitis B virus (HBV) infection, raised the question of whether differences in process could have averted the disaster. A special Task Force of FDA undertook a retrospective assessment of FIAU and FIAC with the stated objective "to determine what data and what analyses of data were available to sponsors and the FDA at the start of the H3X-MC-PPPC study, and whether improvement in the rules governing design, analysis and reporting of data from clinical trials is warranted."

From the outset the FDA Task Force recognized and prefaced their presentation with the recognition that this was a retrospective review initiated in response to the deaths and restricted to the three clinical studies undertaken before the PPPC study. The focus was therefore an evaluation of possible changes in process and design and not a determination of whether or not the outcome of the PPPC trial could have or should have been predicted.

TASK FORCE COMPOSITION

The Task Force was comprised of the following individuals: Ann Witt, Special Assistant to Deputy Commissioner for Operations, Office of Commissioner; Roger Williams, M.D., Associate Director, Science and Medical Affairs, Center for Drug Evaluation and Research; and Leland Pierce, M.D., Clinical Investigations Branch, Office of Compliance, Center for Drug Evaluation and Research. One external advisor, Willis Maddrey, M.D., a hepatologist from the University of Texas Southwestern Medical Center in Dallas, Texas, aided the Task Force.

This was a small internal FDA Task Force that operated independently of the major clinical researchers in HBV disease. Its membership was intentionally restricted to avoid



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials 14 Review of the FDA Task Force Report "Fialuridine: Hepatic and Pancreatic Toxicity On November 12, 1993, the Food and Drug Administration (FDA) released a report "Fialuridine: Hepatic and Pancreatic Toxicity" that had been prepared by a special task force. This report was presented together with an oral report by the authors to the Institute of Medicine (IOM) Committee to Review the Fialuridine (FIAU/FIAC) Clinical Trials on July 25th, 1994 (FDA, 1993a). OBJECTIVE OF FDA REVIEW The deaths of 5 of 15 participants in the clinical trial H3X-MC-PPPC, evaluating the safety and efficacy of FIAU in patients with hepatitis B virus (HBV) infection, raised the question of whether differences in process could have averted the disaster. A special Task Force of FDA undertook a retrospective assessment of FIAU and FIAC with the stated objective "to determine what data and what analyses of data were available to sponsors and the FDA at the start of the H3X-MC-PPPC study, and whether improvement in the rules governing design, analysis and reporting of data from clinical trials is warranted." From the outset the FDA Task Force recognized and prefaced their presentation with the recognition that this was a retrospective review initiated in response to the deaths and restricted to the three clinical studies undertaken before the PPPC study. The focus was therefore an evaluation of possible changes in process and design and not a determination of whether or not the outcome of the PPPC trial could have or should have been predicted. TASK FORCE COMPOSITION The Task Force was comprised of the following individuals: Ann Witt, Special Assistant to Deputy Commissioner for Operations, Office of Commissioner; Roger Williams, M.D., Associate Director, Science and Medical Affairs, Center for Drug Evaluation and Research; and Leland Pierce, M.D., Clinical Investigations Branch, Office of Compliance, Center for Drug Evaluation and Research. One external advisor, Willis Maddrey, M.D., a hepatologist from the University of Texas Southwestern Medical Center in Dallas, Texas, aided the Task Force. This was a small internal FDA Task Force that operated independently of the major clinical researchers in HBV disease. Its membership was intentionally restricted to avoid

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials introducing bias; however, its relative isolation from other components of the medical scientific community meant that only a portion of available expertise and background information was utilized. METHODOLOGY The report evaluated preclinical animal studies submitted as part of the FIAU IND and three clinical trials (R89, R90, and R91) for information suggestive of hepatic or pancreatic toxicity comparable to that observed in the PPPC trial. Animal pharmacology and toxicology assessments were based on existing guidelines for 14-day, 28-day and 6-month human studies in nonexpedited drug development. A focus of the assessment was to look for evidence of hepatic or pancreatic toxicity in animal models. The methods of using pre-existing criteria for animal toxicology and pharmacology data seem appropriate to this IOM committee. The clinical study review was based on "selected laboratory observations" and "clinical events." Selected Laboratory Observations The Task Force defined a selected laboratory observation as any instance of an elevation in serum levels of the hepatic enzymes aspartate aminotransferase (AST) or alanine aminotransferase (ALT), greater than three times the baseline level. Although the Task Force acknowledged that investigators and sponsors often considered such changes indicative of a beneficial drug effect in patients with hepatitis B infection, for the purposes of the Task Force analysis, all observations meeting this definition were instead assumed to be evidence of drug-induced liver inflammation. The definition was constructed following discussions with Willis Maddrey, a hepatologist with expertise in drug-induced hepatotoxicity, with the statement: "Although arbitrary, the ratio appears reasonable based on the natural history of chronic hepatitis associated with HBV infection." Clinical Events These were defined as any events "suggestive or indicative of hepatic or pancreatic damage within 6 months of the final dose of FIAC or FIAU that required inpatient or outpatient evaluation by a health professional." The Task Force clearly states that this definition did not require a causal attribution to the drug and could be related to the underlying disease. This analysis thus involved a worst-case assessment that presupposes that all untoward events are attributable to drug effects. The IOM committee has several major objections to the definition of a rise of AST or ALT to > 3.0 times baseline at any point in a 6-month period as an adverse drug related hepatic event.

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials The Task Force has used the principle of percentage change from baseline as their prime measure and criticized the investigators for looking at and presenting absolute values. This approach is certainly not the recognized, conventional approach. From a conceptual perspective, it suffers from being numerically biased to subjects with low baseline levels. As a generalization any scientific comparison of percentage change from a different baseline is open to criticism and introduces this bias. Increases in ALT or AST were considered in isolation and not in conjunction with DNA markers of HBV disease. The concept of a flare as a marker of therapeutic response was dismissed in this worst-case analysis. Given the complexity of this disease process, no single piece of information should be considered in isolation, and when possible, all alternative explanations, both good and bad, should be considered. It was notable that the Task Force interpretation was not in accordance with inferences drawn not only by investigators and sponsors but also by the Antiviral Drug Division of FDA that was charged with monitoring the FIAU trials. The focus on changes in ALT and AST appears to be inappropriate as a marker of FIAU hepatic toxicity. On examination of the hepatic enzyme data in those subjects who died of this toxicity, a unique feature was that despite the presence of extensive metabolic acidosis, encephalopathy, and coagulopathy, changes in AST and ALT were unremarkable and there was a lack of temporal association between rises in these enzyme markers of hepatic necrosis and other evidence of acutely progressive liver dysfunction. In the opinion of the IOM committee, a rise in ALT or AST did not provide a useful marker of the FIAU hepatotoxicity responsible for the deaths in the PPPC trial. It is important to note that if decisions to alter trial design or stop the PPPC study had been made based on elevated ALT or AST levels in prior studies, they would have been made on an erroneous inference. The committee also finds problematic the approach of using a worst-case interpretation of all clinical events within 6 months of an investigational drug exposure. Application of this approach has major implications for all drug development, particularly those drugs that are being developed for life threatening diseases which have their own natural history of progression. For example, the FDA report advocates considering as drug-induced any negative change in health while taking the drug or during the 6 months after stopping it. This definition should be evaluated from the perspective that most instances of acute hepatocellular damage occur at the time of taking a drug. In this instance, patients with a chronic viral hepatic infection are being evaluated over a prolonged period off this investigational drug and often while receiving other therapy. With a variable disease process, the longer the follow-up, the higher will be the frequency of cases meeting this definition. Most approved medications for any serious disease would fare very badly in this sort of analysis. Before accepting such arbitrary criteria, it is essential to evaluate this methodologic approach in subjects with similar entry criteria but not receiving an investigational new drug. At the present time, there is no methodologic approach that adequately takes into account the inherent variability of the natural history of the disease and allows discrimination of superimposed drug-related events or separate events owing to different drugs. The danger of accepting a worst-case analysis and extending the time of observation to 6 months is that all new drug development for rapidly progressing diseases (e.g., cancer, AIDS, progressive

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials neurological disorders) is likely to be compromised by the inappropriate assignation of disease progression to drug toxicity. Each drug trial must be evaluated in the context of the disease it is meant to treat. In the present case, the IOM committee believes that despite the usual difficulty in ascribing causal relationships to adverse events, the remarkable reproducibility of major FIAU-induced hepatic and pancreatic toxicity in those patients in the PPPC trial, the known ability of ddI to induce pancreatitis, and the natural history of HBV infection progressing to chronic liver failure does permit attribution of causation to the observed toxic events. In the committee's view, attribution of all adverse hepatic and pancreatic clinical events to FIAC or FIAU as proposed by the FDA Task Force would inappropriately ignore potentially important information. RESULTS Animal Pharmacology and Toxicology Studies The FDA Task Force concluded, and the IOM committee agrees, that the animal studies were comprehensive, with the use of appropriate numbers of species, routes of administration, and in certain studies, measures of drug levels in blood. There was no evidence from studies in animals from which hepatic or pancreatic toxicity might have been anticipated. Although we are aware that the FIAU experience is stimulating efforts to model the toxic effects of FIAU on the pancreas and liver in animal models and that new animal model criteria may be developed, it is clear that, the basis of known animal toxicity models available prior to the PPPC trial, there was no way of anticipating the major human toxicity subsequently observed. Overview of Laboratory Events in Studies Prior to H3X-MC-PPPC In the three studies analyzed by the Task Force, R89-001-01 with FIAC, R90-001-01 with FIAU and R91-010-01 with FIAU, 79 patients received one of these investigational drugs. Of those, 24 (29 percent) had an increase in AST or ALT peak-to-baseline ratio of >3.0 at some point during therapy or follow-up. The implication of the report, if not the explicit conclusion, was that if these data had been collected and analyzed in this fashion at the time of the trials they would likely have led FDA, if not the investigators, to view FIAU as a liver toxin. As indicated previously, the IOM committee did not consider this variable a valid measure of drug-induced hepatotoxicity and considers inferences and recommendations based on it to be inappropriate. In this section of the report, the Task Force authors also noted that many of the subjects who developed the unequivocal hepatic and pancreatic adverse effects of FIAU had a history of drug challenge and then rechallenge. The FDA Task Force concluded that "readministration was likely to be associated with severe hepatic and/or pancreatic damage and death." It is unequivocal that most subjects who had the major FIAU toxicity syndrome had a reexposure

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials to the drug and cumulative toxicity might have occurred. However, a minor degree of microvesicular steatosis occurred in at least one subject without reexposure. The numbers of subjects with or without reexposure are too small to discriminate between the alternative hypotheses of a total cumulative dose and first exposure sensitizing a subject to later exposure. In considering the cumulative dose explanation, it is relevant that the investigators considered the implications of extending the duration of drug treatment by reducing the daily drug dosage so that the overall total dosage was not unduly increased in transitioning between one study and the next. Despite these changes in administration, total dosage and duration of dosing are highly correlated, making it difficult to discriminate between them as possible causes of toxicity. There was one example of an individual who erroneously received 3.3 times the normal dose on the first day without immediate ill effects, and still had 13 additional days of protocol treatment. Identification of the mechanism of toxicity and the role of FIAU incorporation into nuclear and mitochondrial DNA may in the future lead to a better understanding of the role of total dose, dose duration, and reexposure to dose to the risk of toxicity. Attribution of Toxicity to Adverse Clinical Events The FDA Task Force observed that none of the 10 adverse clinical events they identified in patients from the trials conducted prior to the PPPC trial were attributed to FIAU toxicity by the investigators or sponsors. The report offers a number of reasons for that: The drug effect might mimic the disease process or the effect of another drug. There was "no prior stated expectation as to the likely incidence of mortality." There was no stated prior expectation of liver enzyme increases in the test population. Because there was a delay in the adverse response, it was reasonable to attribute the adverse response to other, more temporally proximal factors. Changes in liver enzymes levels were ascribed to a "flare." Each adverse effect appeared to have been considered separately. The Task Force noted that "under current rules governing design, analysis and reporting of data from clinical trials, the sponsor is not required to conduct a cumulative analysis." The FDA Task Force goes on to state the need for a worst-case analysis, assuming all adverse effects are drug related. Phase II efficacy and toxicity studies of drugs used to treat complex diseases like AIDS, HBV and cancer are inherently difficult, due to the need to use small numbers of subjects from highly heterogeneous populations. Despite defined entry criteria, it is exceptionally difficult to anticipate the natural history of the disease. If the investigational drug targets the diseased organ, it can be exceedingly difficult to discriminate disease from toxicity of the drug. Any recommendations based on the experience of FIAU must be viewed from the broad perspective of drug discovery. A worst-case analysis as proposed by the Task Force would have the potential to severely compromise future drug development for all diseases with high morbidity and mortality, since the adverse effects of the disease will necessarily be attributed to the drug instead.

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials Investigators, sponsors, and the Antiviral Drug Division of FDA were working with the hypothesis that an increase of AST or ALT level followed by a fall in DNA markers of HBV was associated with therapeutic response. Presumably, a rise in enzyme level without change in HBV DNA would be attributed to adverse hepatic drug effects. However, this combined analysis was not done by the FDA Task Force. In more general terms, the ability of the disease process to influence measures of drug effect creates a confounding variable for all Phase II studies. At the present time, there is no appropriate methodology that can be used to anticipate the extent of variability in disease states seen in studies with small sample sizes. The uniqueness of FIAU toxicity is in providing the first example of a drug that induces progressive multiorgan failure that occurs at long intervals after stopping drug administration. The fact that it has now been identified means that surveillance for delayed organ toxicity must be incorporated into study designs for drug development in the future. It should, however, be appreciated that correct attribution of delayed-onset toxicity is increasingly difficult the longer the follow-up and the sicker the patient group initially studied. In the opinion of the IOM committee, there is a need for cumulative collection of data on adverse events by the sponsor, but that a worst-case analysis is inappropriate. Each case should be interpreted to the full extent of the available information (e.g., pathology reports and autopsy reports), and the cumulative data should be systematically reviewed for unexpected trends. Events reasonably attributed to the underlying disease or other concomitant therapy should not be automatically attributed to the investigational drug. It is recognized that attribution of cause requires judgment. It is the opinion of the IOM committee that the clinicians directly responsible for patient care are those most likely to exercise appropriate judgement. We now turn to another section of the FDA Task Force Report, in which the authors review adverse clinical events associated with the studies immediately preceding the disastrous PPPC trial. In the following sections, we summarize the Task Force review and comment where it differs from our own. Review of Adverse Clinical Events in Trial R89-001-01 This study was sponsored by Oclassen and conducted at the University of California, San Diego (n = 10) and the University of Washington (n = 2) between November 1989 and May 1990. It was a dose ranging Phase I trial of FIAC in patients who were seropositive for both HIV and cytomegalovirus (CMV). The study only used the two lower doses of the four doses originally proposed, and was stopped due to lack of activity against CMV. As noted in Chapter 6, four subjects from this study died in the 6 months following the trial. One developed a severe axonal neuropathy followed by a terminal pneumonia, one died of Kaposi's sarcoma, and the third died of AIDS dementia. These three were not attributed to FIAU toxicity by either the investigators or the Task Force. The fourth case, patient 107, was a 32-year-old drug abuser who was also positive for HBV. He received FIAC at 1 mg/kg/day for 36 days. His serum AST at the start was 43 IU/liter, which rose to 93 after 14 days of therapy. Serum ALT was 44 IU/liter at baseline and rose to 135 after 14 days of therapy. Seven weeks later, when the patient was on zidovudine (AZT), methocarbamol, and

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials amitriptyline, he was admitted with nausea, vomiting, and coagulopathy. The patient deteriorated, with development of jaundice and ascites preceding death. An autopsy showed intrahepatic cholestasis, pancreatitis, and fat necrosis. The study investigators concluded that the patient died of underlying disease or possibly of AZT- or amitriptyline-induced hepatic failure. The FDA Task Force designated this a drug-associated clinical event and considered it relevant because FIAC is predominantly metabolized to FIAU. It is possible that this attribution is correct, but two cardinal features of FIAU toxicity, namely, lactic acidosis and microvesicular steatosis, were not described. The later histological review by Dr. Hyman Zimmermann, who described hepatocellular necrosis with fatty metamorphosis, is described as equivocal. The IOM committee reviewed the available information and designated the probability of causation by FIAC as uncertain. Review of Adverse Clinical Events in Trial R90-001-01 This study was sponsored by Oclassen and was conducted as a Phase I safety study of FIAU at the University of California, San Diego, the University of Washington, and the National Institute of Health (NIH). The study population consisted of patients with serologic evidence of HIV and herpes virus infections; some, but not all, also showed serologic evidence of hepatitis B virus infection. The design was a 14-day, parallel group, dose-ranging study (0.1 to 1.7 mg/kg/day). After the trial had begun approval was obtained for a second course of treatment for those patients whose DNA polymerase markers for HBV fell during the first 14 days of treatment but subsequently recovered. Of the 43 subjects studied, 30 were infected with both HIV and HBV. Thirty-four completed the study and four had repeat therapy. Three of those four who had repeat therapy died within 6 months of their last dose of FIAU, and one was hospitalized with pancreatitis. The study was terminated at the fourth dose level (1.7 mg/kg/day) because of patients' complaints of fatigue. Several cases were reviewed in depth: Subject 401 This 33-year-old patient was a drug abuser who was infected not only with HIV and HBV but also with hepatitis C and D viruses. This patient took FIAU at 1 mg/kg/day for 13 days in April and completed the study in May. He developed progressive liver failure and was readmitted to the hospital 91 days after receiving his last dose of drug with coagulopathy, encephalopathy, and low blood glucose. There was no evidence of lactic acidosis or pancreatitis. Liver histology at autopsy revealed end-stage liver disease with bridging fibrosis. A coincidental autopsy finding was a left renal carcinoma. This patient was identified as a clinical event by the FDA Task Force. In the opinion of the IOM committee, a more reasonable explanation is end-stage liver disease secondary to multiple hepatitis viral infections. There was no evidence of lactic acidosis, pancreatic dysfunction or microvesicular steatosis. Inclusion of this subject among those with drug-related adverse events is unjustified.

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials Subject 406 This 45-year-old patient was HIV and HBV positive and had a prior peripheral neuropathy attributed to treatment of his HIV infection with ddC. He received FIAU at 1.0 mg/kg/day for 14 days in June 1991 and again at 0.5 mg/kg/day for 17 days in June 1992. Before and after taking FIAU, the patient also received AZT, ddI, famotidine, desipramine, and naproxen. In August 1992, he was admitted to his local hospital with abdominal pain and serum amylase of 1,166 and lipase of > 4,000. An autopsy following his death indicated hemorrhagic pancreatitis. Liver histology indicated minimal steatosis with bridging fibrosis. The FDA Task Force describes this as a clinical adverse event potentially related to FIAU therapy. The study investigators attributed the pancreatitis to the patient's current ddI therapy rather than the more distant FIAU trial. In the opinion of the investigators and the IOM committee, it is not possible to confidently assign a single drug a causative role. It is possible that FIAU contributed to the pancreatic toxicity, but the undoubted association of ddI therapy and fulminant pancreatitis, together with the absence of lactic acidosis, microvesicular steatosis, and fulminant liver failure make this a less likely attribution. Subject 408 This patient was a 38-year-old patient with HIV and HBV infections, allergies, eczema and hypertension. A cirrhotic liver was observed when the patient had a splenectomy in 1988. He received FIAU at 1.0 mg/kg/day for 12 days in November 1991 and for 14 days in January 1992. After the FIAU study, he received ddI and fluconazole in addition to AZT, enalapril, and beclomethasone. In April 1992, he developed liver failure, ascites, jaundice, and coagulopathy. He had bleeding esophageal varices and died of liver failure. At autopsy he was found to have advanced micro- and macronodular cirrhosis with active destruction of hepatocytes, fibrosis, cholestasis, and bile duct proliferation. There was no acute pancreatitis or evidence of microvesicular steatosis. The FDA Task Force identified this death as a clinical event potentially related to FIAU. The clinical investigators and the IOM committee consider a more likely explanation to be progressive end-stage liver disease related to longstanding hepatitis B infection. Subject 101 This 31-year-old patient with HIV and HBV infections received FIAU at 1.0 mg/kg/day for 13 days in May 1991. In October 1991, the patient was found to have a peripheral neuropathy requiring narcotic pain control. He subsequently died of Pneumocystis carinii pneumonia in September 1992. This patient was considered not to have a hepatic or pancreatic event by the FDA Task Force. The IOM committee concurs.

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials Subject 409 This 36-year-old HIV- and HBV-positive patient received FIAU at 0.5 mg/kg/day for 14 days in December 1991 and again in April 1992. In August 1992, while the patient was being treated with ddI for his HIV infection, he was admitted to the hospital with right upper abdominal pain and elevated serum amylase levels. The patient improved after ddI treatment was stopped. The FDA Task Force identifies this subject as having a hepatic or pancreatic clinical event that could be related to FIAU. In the opinion of the IOM committee, a more plausible relationship is to ddI, a well-known pancreatic toxicant. It is feasible that FIAU produces pancreatic toxicity, but this patient did not have any other evidence of the FIAU toxicity syndrome. To summarize the adverse events in the R90 trial, none of the 5 patients who had major adverse events had unequivocal evidence of the FIAU toxicity syndrome, even in retrospect. The implication that these were early examples of the toxicity that ultimately killed 5 patients in trial PPPC stretches the facts too far. Review of Adverse Clinical Events in Trial R91-010-01 This study was a dose-ranging study of FIAU in patients infected with HBV free of HIV. The objective was to progress from the previous study by extending the treatment duration from 14 to 28 days and using a smaller daily dose ranging from 0.05 mg/kg/day up to a ceiling of 0.5 mg/kg/day. With the longer treatment, follow-up was extended to 6 months. All 24 subjects enrolled were at NIH and completed drug therapy without incident, but in the following 6 months one patient died and three others had clinical events that were reviewed. Subject 4D This was a 59-year-old patient with chronic HBV infection who received FIAU at 0.5 mg/kg/day for 28 days in August 1992. In October 1992 the patient was admitted to the hospital with abdominal bloating and a peripheral neuropathy. Following right upper quadrant pain, the patient underwent laparoscopic cholecystectomy without finding gallstones. The patient subsequently developed ascites and a persistent severe metabolic acidosis, despite normal hepatic enzyme and bilirubin levels. The patient developed coagulopathy, encephalopathy, and renal failure and died in January 1993. At autopsy there was a nodular architecture, focal bile duct proliferation, lymphocyte infiltration, hepatocyte necrosis, and marked microvesicular steatosis. The pancreas was inflamed. The FDA Task Force identifies this patient as having hepatic and pancreatic toxicity attributable to FIAU. The IOM committee concurs and suggests retrospectively that this is the index patient, i.e. the first clear case of FIAU toxicity as it is now understood. At the time that these events were taking place, the investigators were acutely aware that they were observing

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials a novel and confusing situation. Despite discussions with other senior staff at NIH and with the IRB of record, causation remained uncertain and the lack of a close temporal relationship to FIAU therapy put it low on their list of potential causative mechanisms. Compelling evidence to suggest a late hepatotoxin-induced syndrome unrelated to FIAU was provided by comparison of liver biopsies obtained at the time of the patient's laparoscopic surgery with prior biopsies and with that obtained at autopsy. The biopsy specimen obtained laparoscopically two months after completing FIAU therapy was reported in the investigators' detailed report to the IOM committee to show ''chronic active hepatitis with marked bridging hepatic fibrosis. There were small amounts of fat similar in extent to those in biopsies obtained before therapy'' (Hoofnagle, 1994b). In contrast, at the autopsy at NIH the findings were described as "micronodular cirrhosis with mild inflammatory activity and with moderate to marked microvesicular fat." This represented a marked change in hepatic morphology occurring long after exposure to FIAU but almost immediately after exposure to anesthesia and numerous other therapeutic agents over the intervening time. It did, however, increase their awareness of this unusual syndrome, so that when it was next seen, it was immediately recognized and a correct association was identified. Subject 6B This 59-year-old patient with HBV infection received FIAU at 0.25 mg/kg/day in May 1992. Four months later, the patient had an episode of right upper abdominal pain. No cause was found and the patient recovered. The FDA Task Force identifies this subject as having a clinical event which should be attributed to FIAU in a worst case analysis. In the opinion of the IOM committee, there is insufficient evidence to causally ascribe these symptoms. Subject 1C This 56-year-old patient with HBV infection received FIAU at 0.05 mg/kg/day for 28 days in June 1992. In October 1992 the patient developed right upper quadrant pain and had a biliary stone removed from the cystic duct. A liver biopsy showed a moderately active micronodular cirrhosis. The Task Force identified this subject as having a significant clinical event related to FIAU therapy, but acknowledged that interpretation was difficult because of the cholelithiasis. The IOM committee considered the gallstone the more plausible explanation for the abdominal pain, since other features of FIAU hepatic and pancreatic toxicity were not present. Subject 6A This 27-year-old male with HBV infection received FIAU at 0.1 mg/kg/day for 28 days in April 1992. By January 1993, he had an elevation in his hepatic enzymes (94 IU/liter to 235

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials for AST, 244 to 730 for ALT), and a liver biopsy in February 1993 showed moderate active hepatitis with lobular inflammation and bridging fibrosis but no evidence of microvesicular steatosis, which was confirmed in an independent histology review by Zimmermann subsequent to trial PPPC. The FDA Task Force identifies this subject as having probable evidence of an adverse clinical event related to FIAU therapy. In the opinion of the IOM committee and the investigator, the more likely explanation is that this patient's underlying hepatitis had become more active. The relationship of this change in activity to FIAU therapy is uncertain. Subject 3B This 23-year-old woman with HBV infection had a liver biopsy in April 1992 which demonstrated chronic persistent hepatitis without fibrosis. In May 1992 she received 28 days of FIAU at 0.25 mg/kg/day. Following this, her serum liver enzymes rose from 48 IU/liter to 156 for AST and 69 to 191 for ALT. Subsequently, in November 1992, she had peak levels of 1326 and 934 respectively and her serum bilirubin level peaked at 3.6 mg/dl. A liver biopsy at that time showed severe chronic active hepatitis with bridging fibrosis and lobular inflammation. By July 1993, her liver enzymes levels had returned to normal. The FDA Task Force identifies this patient as one having an adverse clinical event occurring within 6 months of taking FIAU. In the opinion of the IOM committee, the more likely explanation is that this patient had a flareup of her hepatitis or a flare reaction indicative of viral clearing. There was no evidence of lactic acidosis or pancreatic involvement to indicate a full FIAU toxicity syndrome. Identification of FIAU in the liver might provide further information. In summary, in trial R91, the index case of FIAU toxicity was first retrospectively identified. All other clinical events identified by the Task Force have more plausible explanations than FIAU toxicity. RECOMMENDATIONS The FDA Task Force has made a series of recommendations on (1) clinical design and execution, (2) reporting requirements, (3) FDA record-keeping, and (4) further review of FIAU toxicity. These recommendations have since become the basis for a number of proposed changes to the Code of Federal Regulations (CFR) governing the IND process. These deserve review and discussion before being implemented as a unilateral decision. Clinical Trial Design and Execution The FDA Task Force recommends greater attention in the design of clinical trials to the possibility that drug toxicity could mimic the underlying disease process or effect of other drugs. Specifically, they recommend

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials Strong consideration of control groups. Prospective evaluation of endpoints. Assessment of the expected incidence of death or serious events in the study population in the absence of the investigational drug. Inclusion of observed effects on identified end points in the investigator brochure with presumption of drug-related toxicity. Consideration of Control Groups The IOM committee agrees that there is a major need to develop methodology for creating appropriate control groups for Phase I and Phase II clinical trials. The combination of the variable natural history of diseases and small sample sizes make the design of control groups for evaluation of drug response extremely difficult in the early stages of drug development. Essentially two alternatives are used, each with major limitations. One option is to randomize subjects to either placebo or standard therapy or to the investigational drug. In an ideal world this remains the "gold standard." However, the major objection to this approach is that the natural history of the disease is too variable to permit valid inferences to be drawn from studies with small samples, so that the number of subjects required for this approach to be valid is usually large. This approach is optimal for larger Phase III comparative trials, but it is generally not considered practical for Phase I and Phase II trials. The alternative approach is to use historical controls from previous studies, groups of untreated patients, or patients given conventional therapy. In the majority of instances, however, the size of the control group is small, the selection criteria are subtly different, the location and timing of the study are different, and the effect of standard drug therapy is hard to dissociate from the underlying disease process. The IOM committee recommends funding research to explore new methodologic approaches to the design of appropriate comparative patient groups with which to compare new information from Phase I and Phase II trials. A major focus could be to develop techniques enabling the use of historical data from previous trials to match patients and events with study participants, thus allowing for more accurate and objective attribution of adverse events to an investigational drug or other causes. Matching of subjects should not be confined to consideration of entry criteria, but should also include personal demographics, disease extent, disease complications, concomitant therapy, or any other relevant dimension. There is now a substantial body of data that in essence is a national repository and resource. This information is provided in Phase I, II, and III drug trials with stringent quality control to the FDA through the IND process. It has the potential to be integrated to provide large series of subjects that could be matched not only for entry criteria but also for disease extent and severity assessment and other potential confounding variables. In recent years, substantial advances have been made in multivariate analysis and in determining the probabilities of associations within complex data sets. It is therefore conceivable that a more sophisticated individualized basis of selecting comparative groups from an expanded, documented prior series of patients could be developed to incorporate not only entry criteria but estimates of disease pathology at the time of the study. It is contemplated that recent advances in multivariate methods for the detection of associations

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials between variables could be coupled with Bayesian approaches to provide a more systematic, dynamic, objective, and accurate attributions of events than those provided by current approaches. Prospective Evaluation of End Points The IOM committee agrees that protocol design should define prospective end point measures in study design and should also define the rules for stopping a study. These should be individualized to the study context and not be rigidly defined to embrace all study designs. Assess Expected Incidence of Death or Serious Events in The Study Population in The Absence of The Investigational Drug The IOM committee agrees with the principle that observed events should be compared with the expected events in the study population, but we wish to emphasize our prior comments that selection of comparison groups is difficult. A subsequent section of this report demonstrates how little such an assessment would have added to the conduct of the FIAU trials. This topic is one that requires the development and validation of new methodologic approaches. Observed Effects on Identified End Points Should Be Included in The Investigator Brochure and Presumed to Be Drug-Related Toxicity. The IOM committee does not agree that all observed events in a clinical trial should be included in the investigator brochure and presumed to be related to the investigational drug. This worst-case analysis has the potential to be counterproductive to drug discovery by trivializing event analysis and constitutes an overreaction to the FIAU tragedy. The current approach of informed interaction and discussion between the sponsor and FDA on a case by case basis is considered to be an effective and appropriate mechanism. In the PPPC trial, the investigator did inform all study participants of the death of the previous trial's patient 4D, and stopped the study within 12 hours of their recognizing a similar pattern of severe hepatotoxicity in a second patient. This was an appropriate and rapid response. The fact that other patients had already been exposed to the drug and subsequently developed complications, even though the drug was discontinued, was surprising and unfortunate, but not a failure of the decisionmaking process. Extent of Follow-Up The FDA Task Force recommends that greater attention be given to establishing adequate follow-up periods for subjects in clinical trials after the conclusion of the study.

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials Specifically, the protocol submitted to FDA should include the length and type of follow-up. The IOM committee agrees that the experience of delayed severe hepatic and pancreatic toxicity caused by FIAU indicates the need for greater consideration of extended follow-up. However, it must be realized that the longer the follow-up the greater the chance that non-drug-related events are confounding variables in drawing causal inferences. The IOM committee recommends flexibility in study design in terms of requirements of end point measures, the use of independent review committees, and the development of new methodologic approaches as described above to integrate and anticipate adverse events to be expected in the absence of the investigational drug. Given the current inability to predict outcome, the statement of the Task Force that "Sponsors should, therefore, be responsible for developing and evaluating an adequate data base for any investigational drug to help assess whether adverse events are likely to occur at protracted intervals after cessation of drug administration" is outside the current state of knowledge and the current ability of sponsors. In the absence of new approaches, it is an unrealistic expectation. Adequate Safety Monitoring The FDA Task Force recommends that "sponsors describe their safety monitoring and evaluation program and if inadequate, require them to develop extramural resources for this purpose." The IOM committee wishes to comment that an irony of the FIAU experience is that the safety monitoring in place in all studies reviewed was of exceptionally high quality, and lack of it was not the cause for the disaster. The IOM committee endorses the general concept of routine evaluation of events occurring to trial subjects by independent safety monitoring committees. This has been proven effective in the management of multicenter studies and provides a mechanism for reducing bias from investigator or regulatory authority. A broader application of this mechanism is strongly advocated; however, its presence in the FIAU studies seems unlikely to have altered their outcomes. Reporting Requirements The FDA Task Force recommends that there should be "full, complete and timely reports of all deaths and serious adverse experiences and safety discontinuations during the IND/NDA process". Specifically they wish to establish the following requirements: To receive information, on a semiannual basis, regardless of attribution of cause, of all serious adverse events that occur within 6 months or the prescribed follow-up period (whichever is longer), including both U.S. and non-U.S. experiences, along with a reasonable analysis by the sponsor. The FDA Task Force recommends that a worst-case scenario be assumed by the sponsor, whose analysis should then focus on refuting this assumption of drug toxicity. They acknowledge that reports of data from double-blind studies should not be unblinded for this reporting purpose.

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials Autopsy reports should be provided when available. Any toxicity detected or suggested should be acknowledged in each new study protocol and in the updated investigators' brochure. The IOM committee strongly endorses the effective reporting of information during new drug discovery to FDA. The committee wishes to point out that implementation of the above recommendations will not only double the FDA workload (i.e., change from annual to semiannual reports) but will exponentially increase the amount of information presented by requiring the cumulative, worldwide experience to be reported. This has the potential to create an administrative nightmare without guaranteeing enhanced patient safety. The IOM committee also agrees that all serious events observed should be reported and that the sponsor and investigators should provide a reasonable analysis of each, using all available information. The current requirements for 3- and 10-day safety reports when an event is ascribed to a drug-related event are considered appropriate (even though the time frames are short) and events considered not related to the study drug should be included in the periodic IND reports. The IOM committee also agrees that the sponsor should provide a cumulative report of all known adverse events to date at the time of each annual report to facilitate critical analysis. The IOM committee does not agree that a worst-case analysis should be performed for all routine reports or that all adverse events should automatically be incorporated into new protocol description and drug information brochures. This process would essentially be unduly restrictive to the search for new drugs for the treatment of life-threatening diseases. It is recognized that the current working model in which an investigator working with a sponsor presents information for FDA review creates a situation in which two groups address the same information with opposite biases. As pointed out in the FDA Task Force report, an "unavoidable feature of clinical trials is that investigators and sponsors are optimistic about the safety and effectiveness of the drug they are testing." It is equally valid that FDA review provides a pessimistic counterbalance. In the opinion of the IOM committee, the adoption of a worst-case analysis as a routine requirement would tip the balance too far in a pessimistic direction. In our view, all available information should be routinely screened by an independent data monitoring system developed by the sponsor and investigator in agreement with FDA. Such a monitoring group could also take advantage of new input if new methods are developed for collating prior experience in Phase I and Phase II studies and anticipating patient outcome in the absence of an investigational drug. A major concern identified by the IOM committee is that adoption of reporting requirements would decrease the signal-to-noise ratio in interpretation of clinical studies. A further consideration that is relevant not only to what data should be included in the investigators' brochure but to what information should be made available to institutional review boards and prospective patients via consent form is that relevant adverse event information will be at risk of being trivialized if judgment is not exercised in data evaluation.

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials Summarization of All Existing Safety Data "The FDA Task Force recommends that sponsors be required to prepare and submit summaries of all safety data generated during the IND process." They explain that the timing of such overviews should be negotiated with FDA, that such overviews should extend to related classes of drugs and be linked to semiannual reports and that sponsors be required to submit a final report of an IND within 90 days if it is requested by the FDA. As previously indicated, the IOM committee endorses full cumulative data reporting to FDA on all events by the sponsor. It agrees that FDA should have the option of requesting an interim report if new information becomes available within 90 days of a request. However, the nature of information being provided to FDA reviewers by the sponsor (e.g., the definition of end point measures) needs to be agreed upon before the study begins and not on a post hoc basis. Furthermore, because an increasing number of events are going to be considered, the division of FDA working concurrently with the sponsor should also be responsive to causal attribution by independent safety monitoring boards and should not rely solely on a worst-case scenario analysis. FDA Record Keeping The FDA Task Force noted that its current systems for collecting and retrieving data in an IND are largely manual. Because the review of safety data is a critical element in the evaluation of a new drug, they recommend that FDA "assess the feasibility of and resource requirements for updating its tools and systems." The IOM committee acknowledges that implementation of this recommendation will require significant infusion of new resources. It strongly endorses this investment. One possible avenue that should be explored is a joint funding venture with other organizations, both public and private, to explore new approaches to patient safety management in Phase I and Phase II drug trials. The organization of and access to available information from previous studies could be a major resource for such an approach, but the only way to have this information readily available would be through a computerized relational data base. Further Review of Fiau Toxicity The FDA Task Force recommends that "the Public Health Service appoint a panel of qualified experts in relevant fields to evaluate the pathophysiology of FIAU toxicity." The IOM committee strongly endorses this opinion and recommends that research resources be identified and targeted at complementary studies to current research in this area. It is clear that a delayed drug-specific toxicity first identified in humans was the primary cause of this disaster. Other drugs may induce similar adverse effects. The role of a newly identified mechanism causing cumulative changes in mitochondrial or nuclear DNA in disease pathophysiology is still uncertain. A clearer understanding of the mechanism of this process is a pre-requisite to preventing similar tragedies in the future. Many current antiviral drugs and

OCR for page 104
Review of the Fialuridine (FIAU) Clinical Trials cancer chemotherapeutic strategies, as well as the introduction of gene therapy, have the potential to induce changes in nuclear and nonnuclear DNA. There is therefore an urgent need to develop model systems that predict clinical outcomes if we are to see development of drugs free of this toxicity in the future.