Cardiac dysfunction associated with a nucleotide polymerase inhibitor for treatment of hepatitis C

Chronic hepatitis C virus (HCV) infection affects more than 3% (170 million) of the world's population and is a major cause of liver cirrhosis and hepatocellular carcinoma.1 Historically, HCV treatments have included pegylated (Peg) interferon (IFN)-α and ribavirin (RBV), but these agents are limited by restricted efficacy and frequent side effects.2 New regimens comprised of direct-acting antivirals (DAAs) that target different steps in the HCV life cycle are in development and some have received breakthrough therapy status by the U.S. Food and Drug Administration (FDA).3-5

One DAA in development was BMS-986094 (formerly INX-08189), a nucleotide analog HCV nonstructural 5B polymerase inhibitor.6 BMS-986094, either alone or combined with RBV, was well tolerated and exhibited dose-dependent antiviral activity in treatment-naïve HCV genotype 1–infected patients over 7 days of treatment.7 In part A of a phase II study, the combination of BMS-986094 with Peg-IFN-α and RBV appeared to be well tolerated in patients with HCV genotype 2 or 3 infection.8 After review of safety data, BMS-986094 was evaluated in IFN-free combinations with daclatasvir (DCV) and RBV in part B of this study. Subsequently, dosing was terminated after 34 patients had been dosed in part B, immediately after a patient experienced rapidly progressive heart failure and expired. Other cases of cardiotoxicity were later identified.

This report is a retrospective review of adverse cardiac events and additional clinical findings noted in the index case and in patients enrolled in part B of the BMS-986094 study. We also comment on an evaluation from nonclinical studies of potential mechanisms underlying the observed cardiotoxicity and on potential implications of ongoing development of DAAs for the treatment of HCV.

Patients and Methods

Results

At the time of dosing termination on August 1, 2012, 34 patients had received therapy in part B for approximately 1-6 weeks (Table 1). Twenty patients received 200 mg of BMS-986094 with either DCV (n = 14) or RBV (n = 6), 9 received 100 mg of BMS-986094 with either DCV (n = 4) or RBV (n = 5), and 5 received 50 mg of BMS-986094 with DCV and RBV (Fig. 1). Fourteen (39%) treated patients were noted to have evidence of cardiac dysfunction, as defined by left ventricular (LV) ejection fraction (LVEF) <50%: 6 had severe LV dysfunction (LVEF <30%), and 8 had moderate LV dysfunction (LVEF 30%-50%; Fig. 1) at one or more evaluations in the 6-month period following drug discontinuation.

Index Case

The case that led to study drug discontinuation for all phase II subjects was a 25-year-old white male with a history of chronic HCV infection, opioid dependence, and mild depression, who received 200 mg of BMS-986094 and DCV for 40 days in part B of the study. No significant clinical events were reported during the initial three on-treatment visits (Fig. 2). This patient's week 2 ECG demonstrated a newly acquired, nonspecific ST abnormality that was considered not clinically significant at the time; creatinine and troponin levels remained within normal limits. Creatine kinase was in the normal reference range (30-294 U/L) at baseline and week 2 (141 and 227 U/L, respectively) and just above the normal reference range on weeks 1 and 4 (300 and 312 U/L, respectively). On day 39, the patient reported nausea and vomiting and was prescribed prochlorperazine. The following day, he was hospitalized for shortness of breath and was noted to have pulmonary edema, acute renal failure, and shock liver. A TTE demonstrated an LVEF <10%. Laboratory data on admission included: creatine kinase, 397 U/L (normal range: 30-200 U/L); troponin, 0.02 (normal range: <0.04); creatinine, 2.3 mg/mL (normal range: 0.7-1.3 mg/dL); alanine transaminase, 3,861 IU/L (normal range: 13-69 IU/L); and aspartate aminotransferase, 4,909 IU/L (normal range: 15-46 IU/L). Over the next 2 weeks, escalation in therapy included an intra-aortic balloon pump, extracorporal membrane oxygenation, intravenous dopamine, and continuous renal replacement therapy. Despite these measures, cardiac function did not improve and life support was withdrawn by the family approximately 2 weeks after hospitalization.

At autopsy, pathological evaluation of the explanted heart demonstrated diffuse elongation and thinning of ventricular cells associated with fine interstitial fibrosis, very limited areas of necrosis, and limited small areas of mononuclear inflammation (Fig. 2). The degree of myocarditis appeared insufficient to account for severe biventricular dysfunction.

Cardiovascular Evaluation of Patients After Treatment Discontinuation

After dosing termination, all patients were required to have serial TTE evaluations performed locally; 80% of patients had five or more TTEs entered into the clinical database through 6 months after dosing was terminated. Of the 20 patients (including the index case) treated with 200 mg of BMS-986094, 4 had severe systolic dysfunction (defined as nadir LVEF <30%) and 7 had moderate systolic dysfunction (nadir LVEF ≥30%-<50%). Of the 9 patients dosed with 100 mg of BMS-986094, 2 had severe systolic dysfunction and 1 had moderate systolic dysfunction. None of the patients dosed with 50 mg of BMS-986094 had cardiac dysfunction. LV systolic dysfunction was observed regardless of concomitant HCV therapy (DCV or RBV). Duration of treatment on study was longer for patients with moderate or severe LV systolic dysfunction (n = 14) than for those with normal LV function (defined as LVEF >50%; n = 20): 86% of the former group of patients received ≥20 days of treatment, compared to 35% of the latter.

Figure 3 illustrates the LVEF over time for patients with abnormal TTE. The nadir LVEF occurred during the first several weeks after treatment discontinuation in most patients and, in some cases, was preceded by a normal LVEF—in 2 patients, the nadir was reached approximately 10 weeks after treatment discontinuation. Improvement of LVEF was documented in 4 of 6 patients with severe LV systolic dysfunction and in 6 of 8 patients with moderate LV systolic dysfunction. The last available LVEF was >50% in 7 of these patients. In addition to the index case, only 1 patient, treated with 200 mg of BMS-986094 and DCV, did not improve based on last data available. In this patient, cardiac evaluation revealed significant coronary artery disease requiring coronary artery bypass surgery.

ST-segment abnormalities were frequently noted in ECGs of patients with TTE evidence of LV dysfunction. A retrospective evaluation identified three main abnormalities: ST depressions, T-wave inversions, or loss of T-wave amplitude. ST depression appeared to be the most specific and least sensitive finding: It was observed only in 2 patients with severe cardiomyopathy. T-wave inversions appeared to be most sensitive and were observed in 86% of cases and 20% of patients with normal LVEF. Loss of T-wave amplitude was observed in 79% of cases and in 60% of patients with normal LVEF.

Increases in BNP were observed in patients with severe and moderate systolic cardiomyopathy and in patients with normal LVEF (Fig. 3). Based on available retrospective and prospective data, the following patients had BNP >100 pg/mL: 5 of 6 with LVEF <30%, 4 of 8 patients with LVEF 30%-50%, and 2 of 20 with LVEF >50%. The index case did not have BNP increase >100 pg/mL during therapy, and no BNP data were collected during his hospitalization.

The index case and other patients experienced renal function changes based on serum creatinine levels (Fig. 3). Most patients, regardless of severity of systolic dysfunction, experienced increased serum creatinine, which improved over time after treatment discontinuation. The highest recorded serum creatinine value remained within the normal reference range in the majority of patients. In addition to the index case, who required renal replacement therapy, 1 patient required temporary hemodialysis.

Discussion

This is the first clinical report of cardiotoxic changes associated with a novel nucleotide analog polymerase inhibitor developed for the treatment of chronic HCV infection. Although most participants did not have symptomatic heart failure, the frequency of cardiac dysfunction after exposure suggests possible drug-related cardiotoxicity. The most common findings were ST-segment and T-wave abnormalities on the surface ECG. Plasma BNP levels were raised over baseline, but, in most cases, were not significantly elevated above normal reference ranges. Most cases exhibited recovery of cardiac function after ending BMS-986094 treatment. Pathological analysis of the myocardium from the index case showed profound sublethal myocyte injury with very limited areas of necrosis and no appreciable infiltration of polymorphonuclear cells, consistent with, but not diagnostic of, a toxic cardiomyopathy. In the earlier phase of this study (part A; BMS-986094 combined with Peg-IFN and RBV), there was no clear cardiac signal. This may have been resulted, in part, from the lower doses utilized in this earlier phase. In addition, subjects had been off BMS-986094 therapy for 3-6 months at the time the index cases had been reported and the initial echocardiograms were performed to assess cardiac function. Although standard ECG monitoring performed in part A did not identify cardiac signals at the time the study was ongoing, retrospective review of these ECGs focusing on nonspecific ST/T-segment abnormalities described in this article suggested a possible dose relationship, although the data were limited to draw any conclusions.

As recently reported, a series of nonclinical in vitro and in vivo investigative studies were conducted for further characterization of potential mechanisms associated with BMS-986094 cardiotoxicity.10-12 In monkeys, the investigators described partial-to-complete reversibility of dose- and time-related ventricular dilatation, decreased ventricular function, and ST-segment changes.12 These nonhuman primate observations mirrored some of the clinical findings associated with BMS-986094 described in this article. The reversible nature of the events suggests transient, rather than permanent, injury to the cardiac myocytes. The actual mechanism remains unknown, but these studies indicate that BMS−unknown_hyphen;986094 did not appear to be a direct mitochondrial toxicant; however, some potential changes in cardiac energy utilization were described.11 In vitro, BMS-986094 was associated with concentration- and time-related cytotoxicity in human cardiomyocytes.13

Further evaluation of the potential mechanisms of cardiac dysfunction and its relevance to other DAAs requires further study. Of note, the cardiotoxicity of BILN 2061 (Boehringer Ingelheim), a reversible inhibitor of HCV NS3/NS4A serine protease, halted its clinical development. Initial pathological examination suggested mitochondrial dysfunction with minimal myonecrosis, although metabolomic assessments were not performed.14 Consistent with our findings for BMS-986094, biomarkers of cardiac dysfunction (i.e., cardiac troponins and BNP) were insensitive measures of BILN 2061-associated myocardial injury, and for both agents, cardiac function normalized over time. Currently, it remains unclear whether cardiotoxicity is a general effect of DAAs. To date, we are unaware of unexpected cardiac events in patients treated with other investigational therapies with a similar chemical or molecular structure to BMS-986094. Nevertheless, ongoing studies are limited in size, and, without the appropriate monitoring for cardiac safety signals, an agent presumed to have less cardiotoxicity could potentially yield increased cardiac dysfunction when used in a larger population of patients—particularly in those with underlying heart disease or who are older and at risk for cardiovascular disease (CVD). Notably, 75% of U.S. patients with HCV were born between 1945 and 1965, which is the demographic with the highest prevalence of CVD.15 For HCV DAAs currently on the market since 2011, a review of the FDA Adverse Event Reporting System quarterly reports at the time of this publication did not reveal any new reports of serious cardiac signals for these novel therapies.16

Appropriate screening and monitoring methods to identify cardiac dysfunction remain unclear. In this study, several traditional cardiac screening tests were performed and only a small fraction of patients with evidence of cardiac dysfunction experienced symptoms of heart failure. Plasma levels of BNP, a sensitive marker of myocardial stress, were modestly elevated in the majority of patients with cardiac dysfunction, and no patients had elevated troponin levels. Therefore, these biomarkers likely have little utility as screening methods. Nonspecific ECG abnormalities were noted in cases of cardiac dysfunction, but were only apparent in hindsight, and it remains unknown whether they have the associated sensitivity or specificity needed for screening. Therefore, screening methods need to be developed for use in patients treated with agents that may increase cardiotoxicity.

Moving forward, postmarketing surveillance for novel DAAs may be needed for cardiovascular safety given that the population is shifting to an older demographic increasingly at risk for CVD.17 Although BMS-986094 development was discontinued based on the seriousness of the clinical events reported in the phase II study, it could be postulated that safety issues from a less-toxic medication might only become apparent after the drug is marketed and used to treat larger patient populations. With the introduction of the Critical Path Initiative,18 the FDA highlighted the need to establish more-definitive systems to detect and respond to drug toxicity involving patients, regulatory agencies, and industry. Postmarketing approaches to identifying potential toxicities include use of registries, electronic safety monitoring systems, and patient-reported outcomes.

In the case of BMS-986094, patients exposed to BMS-986094 in phase I and II trials have been offered the opportunity to be enrolled into a registry (Longitudinal Assessment of Cardiovascular and Renal Health in Patients with Hepatitis-C [CARE-Hep C]; NCT01846494) created to understand the incidence of cardiac and renal dysfunction related to the study medication and to arrange for appropriate referral in cases of cardiac dysfunction. This registry may improve our understanding of the BMS-986094 safety profile and may also help determine the incidence of HCV-associated cardiac dysfunction. Furthermore, CARE-Hep C could serve as a template for the observation of patients undergoing treatment with novel DAAs.

In conclusion, we report that a novel polymerase inhibitor developed for the treatment of chronic HCV infection may cause a toxic cardiomyopathy. The majority of patients with echocardiographic evidence of cardiac dysfunction did not have heart failure symptoms and none had significant abnormalities in cardiac troponin or BNP levels. Several patients developed concomitant renal failure and some required hemodialysis. Results of prospective nonclinical mechanistic studies suggest that BMS-986094 is not a direct mitochondrial toxicant, but that alteration of cardiac energy generation or utilization may be involved. Ongoing and future clinical trials of DAAs may benefit from careful surveillance for cardiac and renal abnormalities.