Reata’s failed BEACON trial for bardoxolone methyl published in NEJM – excess hospitalizations and deaths due to heart failure – November 14, 2013

We remember October 18, 2012 – the day the BEACON trial for Reata’s bardoxolone methyl was terminated – as one of the saddest days in diabetes drug development history. As a reminder, Reata and Abbott had been developing the drug as the first potential disease-modifying treatment for chronic kidney disease, one of the most devastating diabetes complications. BEACON was terminated due to an excess of deaths and serious adverse events, and we have been ardently awaiting the detailed data.

The results of the trial have been published in NEJM, and, the data appear fairly clear-cut. After a median drug exposure of only seven months in the bardoxolone arm and eight months in the placebo arm, there was no benefit of bardoxolone on the primary composite outcome of end-stage renal disease (ESRD) or cardiovascular death (HR=0.98, 95% CI=0.70-1.37; p=0.92), and there was a very unambiguous excess of hospitalization for heart failure or death due to heart failure events (HR=1.83; 95% CI=1.32-2.55; p<0.001). There was also a significant increase in the secondary composite outcome of nonfatal MI, nonfatal stroke, hospitalization for heart failure, or CV death (HR=1.71; 95% CI=1.31-2.24; p<0.001). The excess of heart failure and secondary composite endpoint events all occurred very early on – mostly within the first four weeks. After that, the Kaplan Meier curves are nearly parallel – this led us to wonder whether the excess of events were completely attributable to the short drug exposure or whether pre-existing conditions played a role. Nonetheless, the randomization was robust, the signal very clear, and a plausible speculation on mechanism (discussed below), which makes the result very hard to ignore.

The authors state that the mechanism for bardoxolone methyl increasing heart failure is unclear – they did not perform ECGs prior to randomization since the heart failure result was unexpected (which means it is difficult to accurately ascertain whether the hospitalizations for heart failure were actually due to heart failure). Weight declined by a substantial ~5 kg (11 lbs; ~5%) in the bardoxolone arm compared to placebo, although the authors could not tell whether it was loss of fat, intracellular water, or interstitial water. Serum albumin and hemoglobin levels fell (it is not clear how much), which may suggest that patients experienced fluid retention leading to hemodilution. Consistent with this hypothesis, bardoxolone also increased blood pressure (between group differences of 1.5/2.1 mmHg). In combination with an increase in heart rate (by 3.8 bpm compared to placebo), the authors state that this constellation of factors were likely to produce heart failure in patients at risk. The trial’s DSMB, after recognizing the early increase in heart failure events, attempted to identify patient characteristics associated with elevated heart failure risk to potentially modify eligibility requirements, but was not able to do so.

Drs. Jonathan Himmelfarb (University of Washington, Seattle, WA) and Katherine Tuttle (University of Washington, Seattle, WA) wrote an accompanying editorial outlining lessons learned from BEACON (and also speculate on a potential mechanism for the results): they note that 1) more extensive preclinical research (or better attention paid to preclinical research) could have alerted researchers to the agents’ safety risk prior to clinical trials; 2) bardoxolone was such a potent transcription factor that it is not surprising that it had off-target effects; and 3) the increases in albumin:creatinine observed in phase 2 should have alerted investigators to potential safety issues. More details on these and other considerations below.

• In Drs. Jonathan Himmelfarb and Katherine Tuttles’ accompanying editorial outlined lessons learned from BEACON:
  
  • The first lesson Drs. Himmelfarb and Tuttle drew from the bardoxolone story was that more extensive preclinical research could have alerted researchers to the agents’ safety risk before clinical trials were conducted. A study in diabetic rats found that bardoxolone analogues were associated with increased occurrences of kidney injury, hypertension, proteinuria, and weight loss – similar to those seen in some clinical trials. Unfortunately, this preclinical study was published just a few weeks after Reata and Abbott terminated the BEACON study on October 18, 2012.  At the 13^th Annual Rachmiel Levine Diabetes and Obesity Symposium, Dr. Mark Cooper (Baker IDI Heart and Diabetes Institute, Melbourne, Australia) argued that Reata “jumped into” the phase 3 BEACON study without having done the necessary basic science research, citing the same rodent study as Drs. Himmelfarb and Tuttle (Zoja et al., Am J Physiol Renal Physiol 2012).
  • Drs. Himmelfarb and Tuttle also noted that it is unsurprising bardoxolone has off-target effects, since it is a potent activator of a transcription factor. Indeed, in addition to activating Nrf2, bardoxolone stimulates peroxisome proliferator-activated receptor γ (PPAR-γ). Drs. Himmelfarb and Tuttle explain that peroxisome proliferator-activated receptor (PPAR) γ activity might contribute to fluid retention and heart failure, especially in people with advanced diabetic kidney disease.
  • Drs. Himmelfarb and Tuttle also called for researchers to be careful when developing a drug for diabetic kidney disease that increases (instead of decreases) people’s albuminuria.    
• The BEACON authors explain that it is not entirely surprising that the safety signals did not emerge in BEAM given that the patient-months of drug exposure was ~10-times that in BEAM and that the study population was much sicker. They also acknowledge that the drug formulation changed between studies, but did not comment on whether this was a relevant factor. Consistent with phase 2 BEAM results, bardoxolone patients did see a significant mean increase in eGFR from baseline (5.5 ml/min/1.73 m² – a bit less than the 8.2-11.4 ml/min/1.73 m² increase observed in BEAM, which enrolled patients with higher baseline eGFR) compared to a decline in placebo of 0.9 ml/min/1.73 m², with the 6.4 ml/min/1.73 m² difference being statistically significant (p<0.001), but this change apparently was not enough to delay progression to ESRD. This was somewhat surprising, though the authors do note that since only 46% of planned events had accrued by the time the trial stopped, this analysis was very underpowered (not that it would have mattered in the end, given the safety data).
• As a reminder, there had been tremendous enthusiasm for bardoxolone methyl as a disease-modifying treatment for chronic kidney disease (CKD) because of the current lack of effective treatment options for CKD, so the news of its termination came as a major setback for the diabetes arena, and especially for CKD patients as well as payers hoping that the compound would bend the cost curve. It had been widely viewed that bardoxolone could hit blockbuster status and Abbott had partnered with the company last year, investing $450 million for rights in most international markets, followed by a later investment of $400 million for a follow-up portfolio. Results from the phase 2 BEAM trial seemed especially promising because gains in eGFR for the mid- and high-level doses tested persisted for the entire 52 weeks of treatment and even after treatment discontinuation, suggesting that the drug was disease-modifying and could reverse CKD.
• As context for how grave a condition stage 4 CKD is, a 2004 longitudinal study on CKD outcomes in the Kaiser Permanente network found that the five-year mortality rate of patients with stage 4 CKD was 46% – worse than many cancers. Furthermore, patients with stage 4 CKD were twice as likely to die during stage 4 as they were to progress to ESRD and dialysis (Keith et al., Arch Intern Med 2004). Needless to say, the costs of ESRD are very high, in addition to patient suffering that goes alongside it.
• We are hopeful that some further development in this sphere will emerge from various companies pursuing compounds including AbbVie and Lilly in particular. AbbVie has a endothelin-receptor antagonist (atrasentan) that initiated its phase 3 SONAR trial in May 2013. Lilly has one undisclosed phase 1 small molecule candidate, and three phase 2 candidates: LY2623091, a mineralocorticoid receptor antagonist; LY2382770, a TGF-beta mAb; and LY3016859, a TGF-alpha/epiregulin mAb (an inhibitor of two epidermal growth factor receptor ligands). Concert Pharmaceuticals has a candidate (CTP-499) in phase 2. Pfizer has a phosphodiesterase inhibitor in phase 2 (PF00489791), and Vascular Pharma plans to advance its candidate VPI-2690 into phase 2 and to file an IND in the US in 2H13.

-- by Jessica Dong, Hannah Deming, and Kelly Close