Review of the Fialuridine (FIAU) Clinical Trials (1995)

Viruses, subcellular organisms that cause both acute self-limited disease in humans and long-term chronic disease, have become the focus of intense research in the last two decades. Composed of DNA or RNA inside a protein shell, these pathogens include the influenza virus that causes the common "flu", and the human immunodeficiency virus (HIV) that is responsible for the many manifestations of AIDS. Some vital infections are eliminated or controlled by the immune system, though often not until the victim suffers through an initial bout of illness. Others are not so easily managed: these viruses often mutate at a high rate in response to the defense mechanisms of the host immune system and cause persistent or chronic disease; viruses can hide in cells in the body or in parts of cells that may not be readily accessible to drugs aimed at killing these organisms; and viruses develop resistance to drugs by changing or mutating with time. Intense research in the last decade has yielded encouraging results in the war against these viruses, but although some persistent vital infections such as HIV and hepatitis C virus (HCV) can be contained with treatment, one can rarely eliminate the virus or completely rid patients of these infectious agents.

Chronic vital diseases such as HCV and hepatitis B virus (HBV) infection are important causes of morbidity and mortality in immunocompromised patients (cancer patients and HIV-positive patients such as those enrolled in the early fiacitabine [FIAC] and fialuridine [FIAU] trials at the National Institutes of Health [NIH) as well as immunocompetent patients (such as those enrolled in the final FIAC trial at NIH). Cytomegalovirus (CMV) is a pathogen that causes chronic disease in immunocompromised but not immunocompetent patients. Infection with HIV is an example of the most lethal of the common vital diseases in human. HIV infection is of particular importance, since the rate of chronicity after acute infection is almost 100 percent and an effective vaccine to prevent infection is lacking. These problems apply equally to HCV, which affects 1-2 percent of the U.S. population. Fortunately, the natural history of chronic HCV infection is much more benign than that of HIV or HBV infection (see below). Nevertheless, HCV disease can result in hepatic failure, hepatocellular carcinoma, and even death, although the length of time to the development of these complications is prolonged (the mean times from transfusion, the presumed time of HCV acquisition, to the development of cirrhosis and hepatocellular carcinoma are 25 and 30 years, respectively).

Infection with HBV differs from that of these other pathogens in that an effective vaccine has been available for more than a decade and the rate of chronicity after acute infection is low. Nevertheless, HBV remains an important global health problem, with a world-wide prevalence of 5 percent, and a U.S. prevalence of 1 percent. Vaccination programs

in the United States targeted at those individuals at greatest risk of infection (health care workers, intravenous drug users, and infants of infected mothers) have failed for two fundamental reasons: difficulty identifying and accessing patients at risk and non-compliance with vaccination regimens once they are instituted. Even in educated population at high risk of infection (e.g., health care workers), compliance with vaccination programs is only 50 percent. Compliance in other risk groups is much lower. Because of these limitations, universal vaccination of all newborns and prepubescent teenagers has been advocated and has begun to be instituted. Proof of the efficacies of these programs is awaited.

FIAC and subsequently FIAU were drugs first tested against CMV and varicella-zoster virus (VZV), which are common pathogens in HIV-positive patients. Little efficacy against these viruses was seen, but marked suppression of HBV infection was observed in patients infected with both HIV and HBV. This led to considerable interest in examining the effects of these drugs in the treatment of chronic HBV infection. To assess the importance of treating patients with chronic HBV infection, an understanding of the prognosis for untreated patients with this disease is necessary.

The risk of chronicity after acute infection is largely determined by the age at which infection is acquired. With perinatal acquisition, chronic infection is almost universal (Stevens, Beasley, Tsui and Lee, 1975). With infection in healthy adults, the rate of chronicity is extremely low (less than 5 percent), but these carriers remain at risk of developing hepatocellular carcinoma (HCC) (Seeff, Beebe, Hoofnagle, Norman, Buskell-Bales, Waggoner, Kaplowitz, Koff, Petrini, Schiff, et al, 1987; Norman, Beebe, Hoofnagle and Seefe, 1993). Since most HBV infections world wide are acquired in the early neonatal period, when the risk of chronicity is high and effective vaccine programs are not yet in place, there remains a large population in whom chronic HBV infection is endemic.

Symptoms of HBV infection are variable. With acute HBV infection, patients are frequently fatigued and jaundiced (i.e., their skin turns yellow) and they complain of abdominal pain. Acute infection in adults rarely results in life-threatening complications; death from acute liver failure occurs in less than 1 percent of adult patients. The vast majority of patients recover completely and develop protective immunity (i.e., they cannot be reinfected with this virus if they are subsequently exposed). Fewer than 5 percent of adults will develop chronic liver disease, often with only minor complaints of fatigue, joint pains, poor appetite and abdominal pain. Children who acquire HBV infection from their mothers are often asymptomatic and develop problems only after prolonged infection. Chronic infection, however, results in scarring of the liver (with consequent impairment of the flow of blood through the liver, back-pressure into the veins of the intestine, and an increased propensity for life-threatening bleeds). Other complications include retention of fluid in the abdomen (ascites), increased susceptibility to life-threatening bacterial infections, and mental confusion (encephalopathy). Once these complications occur, the life expectancy of the patient is very limited and liver transplantation is the only therapeutic option.

HBV infection has been linked epidemiologically to the development of HCC. In males with perinatally acquired infection, the life-long risk of development of HCC is 40 to 60 percent (Beasley, Hwang, Lin and Chien, 1981; Wright, Venock and Millward-Sadler, 1992). Screening programs may detect malignant lesions at a stage when they are still respectable, but they have failed to demonstrate a positive survival benefit. Once patients have symptoms related to liver cancer (abdominal pain or weight loss), the benefit of surgical resection or cancer chemotherapy is limited.

Knowledge of the natural history of disease without therapy is essential when making decisions regarding the need for therapeutic intervention. Many studies have evaluated the long-term consequences of infection. Perhaps that most applicable to the patients included in the H3X-MC-PPPC trial is a study of patients evaluated for antiviral therapy at Stanford University from 1974 to 1981. Three hundred seventy-nine patients were considered for treatment with either interferon or adenine arabinoside (Ara-A; like FIAC and FIAU, Ara-A is a nucleoside analog) (Weissberg, Andres, Smith, Weick, Nichols, Garcia and Robinson 1984). Five-year survival was determined by telephone contact and cause of death was determined by autopsy whenever available. The majority of patients were men (as in the PPPC trial) and histology at the time of entry into the study varied (approximately one-third had chronic persistent hepatitis, one-third had chronic active hepatitis, and one-third had cirrhosis). Five-year survival was 97, 86, and 55 percent in these three histologic groups respectively. Most patients died of liver failure, and only four (8 percent) died of HCC. Since an autopsy was rarely performed, the number of deaths from HCC may have been underestimated. Patients with hepatic decompensation (i.e., patients who have already developed complications because of their infection) were included in this natural history study, though patients with disease this advanced would not have been eligible for the PPPC trial.

The most important conclusion from this and other natural history studies of HBV infection is that when infection becomes chronic, significant liver damage can occur and the risk of progressive liver failure is greater in those with an advanced histologic stage of disease. The morbidity related to chronic infection is hard to determine since subjective end points (fatigue, weight-loss, abdominal pain etc.) are more difficult to assess than objective end-points related to mortality (i.e. rates and causes of deaths). Regardless of the range of morbidity and mortality assessed in different studies, it is clear that liver disease associated with HBV infection may progress to liver failure, the development of HCC, and death.

An important feature noted in studies of the natural history of chronic HBV infection is that of the spontaneous loss of viral replication (indicated by the disappearance from the blood of HBV envelope antigen, a viral protein highly correlated with active infection, and the loss of HBV DNA in serum). This occurs in 2.5 to 25 percent of patients (Hoofnagle, Dusheiko, Seefe, Jones, Waggoner and Bales, 1981; Perrillo, Campbell, Sanders, Regenstein and Bodicky, 1984; Perrillo, 1992). Viral clearance is hypothesized to be mediated by the immune system and is frequently associated with a transient rise in the levels of aminotransferases (aspartate aminotransferas [AST] and alanine aminotransferase [ALT], enzymes that are present in liver cells and are markers of liver injury when they present in quantity in blood). This so-called flare phenomenon is also seen in patients with a positive response to interferon therapy (see below).

The first responsibility of an investigator and drug sponsor (usually the pharmaceutical company) assessing the utility of a new drug is to determine whether it is safe. The second duty is to determine whether the drug is efficacious. Finally, the practical realities of everyday patient care come to the fore. Most available antiviral agents have had to be given by injection, either intravenously or subcutaneously, because they are poorly absorbed from the intestine. Approaches to treatment of viral diseases have generally been either to stimulate the natural host defenses against infection (by drugs such as interferon) or to inhibit directly the ways that viruses divide and multiply (by nucleoside analogs, molecules which are similar to the building blocks of DNA and that become incorporated into the viral DNA as the virus divides, inhibiting and interrupting this replicative machinery).

Important advances have been made in the therapy of some viral diseases, notably those belonging to the herpes virus family. Herpes simplex virus is controlled effectively with the oral nucleoside analog acyclovir, and CMV infection is treated effectively with the intravenously administered nucleoside analogs ganciclovir and phosphonoformate (foscarnet). An effective oral agent against CMV disease is still being sought. Long-term administration of nucleoside analogs such as ganciclovir (used against CMV) requires the placement of permanent venous access lines, often utilizing the operating room, and twice-daily home therapy with visiting nurses. These IV lines also carry with them potential complications of systemic and life-threatening bacterial infections. The costs of such parenterally administered antiviral agents have not been examined closely, but it is readily apparent that an orally active agent would have a positive impact on most of the problems outlined above. This brings us to the question of what treatments for chronic HBV were available when the FIAC/FIAU trials were begun.

Interferon, a naturally occurring protein produced by lymphocytes and fibroblasts in response to chronic viral infection, has been tested in the treatment of chronic HBV infection since the mid-1970s (Greenberg, Pollard, Lutwick, Gregory, Robinson and Merigan, 1976). Initially serum-derived and subsequently synthesized interferons have been shown to effectively inhibit viral replication in approximately one-third of treated patients (Greenberg et al., 1976; Perrillo, Schiff, Davis, Bodenheimer, Lindsay, Payne, Dienstag, O'Brien, Tamburro, Jacobsen, et al., 1990). Recombinant interferon alfa-2b is the only treatment for chronic HBV infection in the United States approved by the Food and Drug Administration (FDA), but treatment results in a response rate of only 37 percent (Perrillo et al., 1990). Selection of patients who have HBV DNA levels of less than 100 pg/ml may yield response rates of 50 percent. Nevertheless, a large number of patients either fail to meet the criteria for treatment or do not respond to therapy. Moreover, treatment is expensive (approximately $5,000 for a 16-week course of 5 million units per day), must be given by the parenteral route, and is associated with significant, even if largely reversible, side effects including flu-like symptoms, fatigue, muscle aches, low-grade fevers, and depression. Many of these side effects overlap with the complaints of patients with chronic viral infections and, although not life-threatening, may lead to considerable morbidity and the cessation of therapy. Hence, there is a need for an effective orally available therapy.

Several potential alternatives to interferon in the treatment of chronic HBV infection have been evaluated during the last decade. Ara-A, and its soluble monophosphate, Ara-AMP, as well as nucleoside analogs effective in the treatment of HIV infection (dideoxycytidine [ddC] and dideoxyinosine [ddI]) are agents that have been studied carefully as therapy for chronic HBV infection. All have been found to be only partially effective, must be given intravenously or intramuscularly, and are associated with significant adverse effects, most notably reversible peripheral neuropathy (Ara-A) and pancreatitis (ddI) (Weller, Bassendine, Craxi, Fowler, Monjardino, Thomas and Sherlock, 1982; Bassendine, Chadwick, Salmeron, Shipton, Thomas and Sherlock, 1981; Garcia, Smith, Weissberg, Eisenberg, Bissett, Nair, Mastre, Rosno, Roskamp, Waterman, et al., 1987; Scullard, Pollard, Smith, Sacks, Gregory, Robinson and Merigan, 1981; Kassianides, Hoofnagle, Miller, Doo, Ford, Broder and Mitsuya, 1989). The other drugs examined have been largely ineffective. Ribavirin is an orally available nucleoside analog used in the treatment of respiratory syncytial virus infection. In vivo activity against HBV (Tong, Huang, Lefkowitz, Lee, Co and Lo, in press) is limited and its toxicity is significant (hemolytic anemia, gastrointestinal side effects). Thymosin, an immune modulator, must be administered parenterally, and appears to be safe, but data from a recent randomized controlled clinical trial are still being analyzed for proof of efficacy against HBV. Other drugs such as 3'-thiacytidine are currently under investigation in the treatment of chronic HBV infection. Experience in more than 2,000 patients treated with this nucleoside analog suggests that this drug is safe and that it has at least transient efficacy in inhibiting HBV replication (Dienstag, Perillo, Schiff, Bartholomew, Vicary and Rubin, 1994).

With this background, it is easy to understand the interest generated by FIAU when this drug was shown to inhibit HBV replication and, in early trials, appeared to have minimal side effects.

Immunological activation with T-cell destruction of infected hepatocytes has been the postulated explanation for the rise in ALT and/or AST levels; the so-called flare phenomenon, seen in association with spontaneous and therapeutically induced clearance of virus. In either circumstance, this rise in enzyme levels is frequently accompanied by conversion from a replicative state (HBV DNA positive) to a nonreplicative state (HBV DNA negative). A rise in liver enzymes levels may be seen either preceding or concurrent with the inhibition of viral replication (Hoofnagle et al., 1981; Perrillo et al., 1984). In one report, 6 of 23 untreated males experienced this rise (elevations at least twice the average of two previous determinations; mean increase, 4.9-fold) during the spontaneous clearance of HBV (Perrillo et al., 1984). Interferon is believed to act both by immune stimulation and by a direct antiviral mechanism. In patients on interferon therapy, the flare typically occurs between the second and third month of treatment and is associated with a fall in HBV DNA levels. Degree of rise in ALT levels that determines a flare differs in various studies. Retrospective analysis of the U.S. multicenter interferon trial identified a two fold or greater rise in ALT levels in 60 percent of those who responded; this phenomenon was seen in 37 percent of patients overall (Perrillo et al., 1990). In one study from Europe (Alexander, Brahm, Fagan, Smith, Daniels, Eddelston, Williams,

1987), AST levels rose to 10 to 20 times the upper limit of normal accompanied by a ''hepatitis-like'' illness before the complete normalization of liver enzyme levels. This syndrome was rare in patients who failed to respond to therapy. In a trial at the National Institutes of Health (NIH), a rise in aminotransferase levels (to values at least two fold the pretreatment levels) was noted in the majority of responders to interferon (Hoofnagle, Peters, Mullen, Jones, Rustgi, Di Bisceglie and Hallahan, 1988). Thus, this phenomenon is well-recognized as a positive event either in the course of the natural history of disease or in the course of interferon therapy. The transient rise is typically associated with a fall in HBV DNA levels and heralds a permanent clearance of virus. Most studies define the flare as at least a two fold rise in ALT above baseline values, although the degree of observed elevation may be much greater (Alexander et al., 1987). It is thus difficult to distinguish, on the basis of an elevation in ALT levels alone, drug-induced hepatotoxicity from a positive response to antiviral therapy. Although an FDA task force retrospectively proposed that ALT or AST elevations greater than three times the baseline should be more suggestive of drug-induced hepatic injury than of a flare signaling a positive response to treatment, the task force did acknowledge the difficulty in clearly distinguishing between these two events. It should be emphasized that although the flare is believed to be due to immune-mediated lysis of infected hepatocytes expressing viral antigens on the surface membrane, without confirmation in an animal model of hepatitis B, this remains only a hypothesis.

The pathophysiology of the flare associated with nucleoside analogs such as Ara-A is even less well understood. There appears to be a qualitative difference between the phenomenon occurring with Ara-A and that occurring with interferon, in that the rise in aminotransferase levels in the former case occurs after the drug has been stopped (Weller et al., 1982; Bassendine et al., 1981). Moreover, the information regarding the flare associated with nucleoside analogs is derived from very few cases. Since these drugs have no known immunomodulatory properties, the explanation for the phenomenon occurring in these circumstances is that the drug lowers HBV DNA levels to a point where host immune mechanisms can effectively clear the remaining virus, with associated necrosis of infected hepatocytes. There are no experimental data to support this hypothesis. Regardless, based on knowledge at the time of the PPPC trial, it is understandable that the investigators would interpret the rise in transaminase levels seen in some patients as a positive sign indicative of viral clearance.

A recent trial of another nucleoside analog, 3'-thiacytidine (lamivudine), in the treatment of chronic HBV infection demonstrated a rise in liver enzyme levels (AST and/or ALT) during therapy (greater than two times the baseline level) in 50, 36, and 27 percent of patients treated with 25, 100, and 300 mg per day, respectively (Dienstag, Perillo, Schiff, Bartholomew, Vicary and Rubin, 1994). Elevations in liver enzyme levels three times the baseline level were less frequent (less than 20 percent) (Jules Dienstag, personal communication, November 1994). This rise in liver enzyme levels was associated with a fall in HBV DNA in 8 of the 11 patients. The investigators concluded, that even in the absence of known immunomodulatory effects of this drug, treatment related ALT rises were common (Dienstag et al., 1994). Whether this rise occurs as a consequence of therapeutically induced viral clearance or as a consequence of spontaneous fluctuations during the course of disease is difficult to determine. Regardless of the mechanism, information available subsequent to the PPPC trial suggests that ALT in ALT

levels elevations three times the baseline level may be seen in HBV-infected patients treated with nucleoside analogs other than FIAU.

Nucleoside agents were initially developed in the 1950s and 1960s as potential anticancer agents. In the late 1960s, the potential use of nucleoside analogs for the treatment of infectious diseases became apparent, with flucytosine, an antifungal agent, being the most prominent example. Other agents which followed, such as adenosine arabinoside, cytosine arabinoside, and iododeoxyuridine, demonstrated antiviral activity against herpes-related viruses, thereby creating an opportunity to treat viral diseases. Since most of these agents were initially developed as anticancer drugs, their major side effect profiles are similar to antineoplastic agents; specifically, they target rapidly dividing cells, such as those in the skin, gut, and bone marrow. Therefore, the most common adverse events noted with antiviral nucleoside agents are nausea, vomiting, diarrhea, cytopenias (low blood counts), and rash.

All of the agents approved by the FDA for use against the human immunodeficiency virus are nucleoside analogs, most of which were originally developed as drugs to potentially treat cancer. Zidovudine, which was developed in 1964, is most commonly associated with nausea, loss of appetite, low blood counts, and muscle aches. Elevation in liver enzymes is seen in a small proportion of patients. Among the other approved antiretroviral agents available for use, didanosine(ddI) is most commonly associated with gastrointestinal disturbances, most notoriously pancreatitis, which may be fatal. The incidence of pancreatitis is dose-related, with a rate of 5 to 9 percent occurring at currently established doses but up to a 30 percent rate at very high doses. Peripheral neuropathy is observed on occasion but hepatotoxicity is uncommon. On the other hand, zalcitabine (ddC) and stavudine (D4T) are most often associated with peripheral neuropathy, in a dose-limiting fashion, and only rarely associated with pancreatitis. Hematologic and hepatic abnormalities are less common in ddC and D4TR recipients than those taking zidovudine.

Microvesicular steatosis, multiple small fat droplets within liver cells (hepatocytes), has also been seen in association with nucleoside analogs used in the treatment of AIDS (ddI and azidothymidine [AZT]) (Lai, Gang, Zawacki and Cooley, 1991). Nonspecific fatty changes in the liver (hepatic steatosis) are seen in about 50 percent of all liver biopsy specimens from patients infected with HIV and are common in biopsy specimen from patients with chronic viral hepatitis as well. An observation of microvesicular steatosis, however, is suggestive, although not conclusive proof, of toxic liver injury. Microvesicular steatosis of the liver has been described in association with alcohol, drugs (tetracycline, sodium valproate), Reye's syndrome (a disease of children linked epidemiologically to viral disease in association with salicylate therapy), acute fatty liver of pregnancy, and metabolic disorders of fatty acid metabolism (3-hydroxyacyl-coenzyme A dehydrogenase deficiency) (Treem, Rinaldo, Hale, Stanley, Millington, Hyams, Jackson and Turnbull, 1994). The association with nucleoside analogs was made before the FIAU trials at NIH, and indeed complicates the interpretation of possible FIAU hepatoxicity since many patients in the R89 and R90 trials were on ddI and/or AZT.

Scattered case reports of steatosis and lactic acidosis together have been associated with HIV infection. By January, 1993, 15 cases had been reported in the literature, 3 cases reported as abstracts, and 2 additional cases reported to the FDA. Fourteen (70 percent) of the cases were fatal. Seventeen of the 20 patients had been on zidovudine at the time the acidosis began; the other three had received zidovudine in the past but had not taken the drug for over four weeks prior to the onset of acidosis. The original reports ascribed the syndrome to HIV disease itself, but by 1993, investigators were beginning to ask whether antiretroviral therapy might be the cause of this syndrome, a hypothesis unfortunately not confirmed until the cases of FIAU deaths were reported.

Viruses, subcellular pathogens composed of DNA or RNA inside a protein shell or "envelope," are the cause of many acute and chronic diseases. Chronic infection by hepatitis-B virus remains an important global health problem despite the availability of an effective vaccine. World-wide prevalence is estimated at 5 percent, including about one million Americans. Acute infection rarely results in life-threatening complications, and most healthy adults resolve acute infection. A small proportion of adults (but a large proportion of neonates) fail to clear the virus and develop chronic liver disease. Symptoms are still relatively mild (fatigue, poor appetite, abdominal pain), but chronic infection gradually impairs liver function and increases the risk of cirrhosis and liver cancer.

Treatment of viral diseases involves either stimulating natural host defenses with substances like interferon, or directly interfering with viral replication with nucleoside analogs such as acyclovir or AZT. Interferon was approved by FDA in 1992 for use against HBV, but it is effective in only about a third of patients, must be given parenterally, and produces significant adverse side effects. Nucleoside analogs tried prior to FIAC and FIAU, all of which required IV administration, were either ineffective or produced unacceptable toxicity. The prospect of an effective oral antiviral drug thus made FIAC and FIAU attractive candidates for development.

Also introduced in this chapter, because it is the center of much post-tragedy discussion about whether liver failure in the PPPC victims should have been anticipated, was the "flare" phenomenon often associated with natural or therapy-induced clearance of HBV. Viral clearance is hypothesized to be immune-mediated and is frequently associated with a transient rise in serum aminotransferases (AST, aspartate aminotransferase, and ALT, alanine aminotransferase). These enzymes are present in liver cells and are generally thought to be indicators of liver injury when present in quantity in blood. Elevations in ALT or AST in patients being treated for HBV infection are thus open to conflicting interpretations - indicating either successful therapy or a toxic effect of the liver.

Finally, we have briefly reviewed what was known about toxic effects produced by nucleoside analogs other than FIAC and FIAU, in part because some have suggested these effects should have provided ample warning that FIAU would produce dangerous effects and in part because many of the subjects in the trials we review here took these drugs before, during or after their FIAU treatment, confounding interpretation of adverse events.