Drug Baron

Bardoxolone and BG-12: A Tale of Two CITIEs (Covalently-Interacting Therapeutic IntErventions!)

| 1 Comment

Until 18th October, bardoxolone methyl (Abbott/Reata) and BG-12 (Biogen) had two things in common: the nrf2 transcription factor complex as their molecular target; and their candidacy for blockbuster status.

Then Reata halted clinical development of bardoxolone, following an excess of unspecified serious adverse events and mortality among the treated patients in the Phase 3 BEACON trial in chronic kidney disease.

Almost simultaneously, the FDA paused for breath in their assessment of BG-12, delaying a decision on approval – although Biogen argue that the delay is unrelated to the sudden collapse of the bardoxolone programme.

Should Biogen shareholders be concerned?  Is this the end of the road for nrf2 activators?

Ironically, the presumed similarity of mechanism is also less certain than has often been assumed.  Both compounds can react covalent with nrf2 and modulate its activity, but the extent to which nrf2 is central to the beneficial effects or the adverse events seen with bardoxolone is an open question.

If the data with BG-12 supported approval before the Reata announcement then nothing should change.  And nrf2 remains an attractive target for non-covalent modulators – both bardoxolone and BG-12 suggest, but don’t yet prove, that this pathway is an important one in a range of disease processes.

 At the start of this month, a poll of the best candidates in clinical development to be approved, and then to go on and establish themselves in the marketplace as blockbusters, would likely have included bardoxolone methyl and BG-12 (the dimethyl ester of fumaric acid).

Both drug candidates have interesting histories.

Bardoxolone methyl was initially entered into clinical development by Reata Pharmaceuticals as an anti-cancer agent.  But during Phase I, the effect on serum creatinine levels suggested it had a much more impressive effect on kidney function than on cancer, and the Reata team cleverly switched horses entering Phase 2 in chronic kidney disease.

Impressive effects on glomerular filtration rate led to what was ultimately one of the most valuable Phase 2 deals ever struck, as Abbott paid $450m in cash for ex-US rights to the drug in September 2010, leaving Reata with the valuable US rights and control of the development programme.  As more and more positive data came in, Abbott added a further $400m in late 2011 to license related preclinical molecules.  Although it is never a procession to approval, the stars nevertheless seemed favorably aligned.

Bardoxolone shares more than a passing resemblance to Super Glue

Likewise, dimethyl fumarate did not start life as a treatment for multiple sclerosis (MS).  Instead, the first human exposure to this very simple molecule came from its widespread use as a fumigant and anti-fungal treatment particularly for leather.

While undoubtedly effective, it hit the headlines in 2007 when leather furniture manufactured in China were imported into the EU, where dimethyl fumerate had been banned since the mid-1990s.  Sachets of the compound sewn into the chairs led to at least sixty cases of severe allergic dermatitis in Finland, and more than a hundred cases in other EU countries including the UK.

Interestingly, while concentrations in air as low as 1ppm can lead to extensive, pronounced eczema that is difficult to treat, the same compound administered orally leads to anti-inflammatory effects in psoriasis.  Like bardoxolone, BG-12 started in one indication and switched when MS patients with psoriasis reported dramatic improvements their MS symptoms.  By April 2011, Biogen was reporting impressive top-line results from the DEFINE and CONFIRM Phase 3 trials in more than 2,500 patients with relapsing-remitting multiple sclerosis

Not only did oral dimethyl fumerate dramatically reduce relapse rates (the primary end-point) it also reduced lesions detected by MRI as well as a range of measures of disability progression.  Importantly, these benefits were seen without any indication of allergic sensitization – or, in contrast to the outcome from the Phase 3 studies with bardoxolone methyl, any increase in serious adverse events.

Two drugs, sharing a molecular target, both with anti-inflammatory effects reached Phase 3.  One, apparently, has a clean safety profile, but the other is associated with a significant increase in severe adverse events and even mortality.

There are exceptions, but as a general rule compounds that react to form covalent bonds with their target(s) are very high risk to develop

The simplest explanation for the difference attributes the adverse events with bardoxolone methyl to idiosyncratic toxicity (that is, molecule-specific effects).  After all, BG-12 and bardoxolone share little structural similarity even though they both modulate nrf2 activity.

But its not so easy to dismiss the bardoxolone data without understanding the nature of the adverse events that have been seen.  Abbott and Reata have been coy about disclosing the nature of the adverse events, but the focus has become the risk of malignancy.   Nrf2 has a complex role in oncogenesis, with components acting as tumour suppressor genes that are frequently mutated in human cancers, while nrf2 itself controls expression of myriad cytoprotective pathways.  On the other hand, interactions between nrf2 components and bcl2, as well as stimulation of autophagy, suggest that nrf2 activators may lead to immortalization and hence promote transformation.

While these pathways are well studied in vitro, it is difficult to know what the balance between pro- and anti-cancer effects of the nrf2 pathway will be in vivo.  More importantly, the multi-component nature of the nrf2 complex means that different activators could plausibly affect the pro- and anti-cancer pathways differently.

Until the full details of the adverse events in the trials with bardoxolone are made public, there will always be a suspicion that the nrf2 activation turned out to exacerbate oncogenesis.

Does that mean, then, that BG-12 is also unsafe?  After all, it too is supposed to activate nrf2.

No – for at least three reasons.  Firstly, more patients with MS have been exposed to BG-12 in the DEFINE and CONFIRM trials alone than had received bardoxolone in all the clinical trials performed to date.  And the data, so far, suggests there is no excess of malignancy or any other serious adverse event.  As DrugBaron noted previously, the data should trump the hypothesis.

Secondly, studies of the gene expression patterns activated by nrf2 differ depending on the activator that is used.  In addition to bardoxolone and BG-12, there are numerous other putative activators of nrf2, such as sulforaphane – an abundant antioxidant in broccoli and related vegetables.  Each of these compounds induce a different subset of cytoprotective genes suggesting that there is more than one mode of molecular activation of the nrf2 complex.  It is perfectly possible therefore that both bardoxolone and BG-12 could be nrf2 activators but yield different cellular responses and hence different safety profiles.

The third reason, though, and perhaps most compelling, is the difference in reactivity of the two agents.  Both bardoxolone and BG-12 activate nrf2 by covalently reacting with the cys-151 residue in KEAP-1 – an amino acid residue that nature has carefully evolved to be exquisitely sensitive to the prevailing redox environment.   But, whereas bardoxolone seems to have a degree of specificity, BG-12 will react with the first activated thiol it meets.

For once in drug development, better specificity may not lead to better safety.

After oral administration, BG-12 reacts rapidly with glutathione, an abundant thiol-containing metabolite, such that free levels of the active species (monomethyl fumarate) are undetectable in blood.  By contrast, bardoxolone survives to activate nrf2 in multiple tissues because it is not inactivated by circulating glutathione.

Fortuitously, then, BG-12 may be short acting and even localized to leukocytes in the blood, while bardoxolone may yield a global and long-lasting upregulation of nrf2.

It is not yet proven – and given the difficulty of such studies, it may never be proven beyond doubt.  But as long as the data continues to support the efficacy and safety of BG-12 there is no need at all to cast a shadow over its future on the basis of findings with bardoxolone.

Indeed, both compounds will react with sites other than components of the nrf2 complex.  Bardoxolone shares more than a passing resemblance to Super Glue – the reactive site on the molecule is a cyanoacrylate and, like BG-12, will undergo Michael addition reactions very readily with a range of nucleophiles.  The extent to which these unknown (though not unexpected) additional targets contribute to either the efficacy seen with both these molecules, or indeed the adverse events associated with bardoxolone therapy is entirely unknown and difficult to predict.

On a more positive note, surely these clinical programmes have at least validated nrf2 as a target for new anti-inflammatory drugs?

After all, the impact of BG-12 on relapse rates in MS, not to mention progression of disability, is impressive and bardoxolone dramatically reduced serum creatinine levels (interpreted as an increase in kidney function) in patients with chronic kidney disease.  The statistics in each case leave almost no room for doubt.

The statistics may be rock solid, but the biological interpretation of the Phase 2 data from the bardoxolone programme is on considerably shakier ground.  Critics have noted that patients treated with bardoxolone lost so much lean body mass that the reduction in creatinine levels could plausibly reflect lower production rather than increased clearance.  Serum creatinine may be a useful surrogate for glomerular filtration rate, but only if other parameters are unaffected.  The halted BEACON trial would have settled this question, providing evidence of clinically-relevant impact as well reduced serum creatinine levels, but it was by no means a certainty that robust clinical benefit would have been delivered.  And with the trial halted, the efficacy data to date cannot be taken as validation of nrf2 as a target in diabetic nephropathy.

For BG-12, the Phase 3 data is already in and demonstrates clinical benefit.  But here the question is the lack of specificity.  BG-12 is beneficial in MS, but how much of that effect is mediated through nrf2? Most likely some of it – maybe even all of it – but with such an important question unanswered it is not yet time to consider the nrf2 pathway clinically validated in any disease.

As usual, the data and not the theory has cast the dice: BG-12 remains on track for blockbuster status, while bardoxolone is added to the grim roll call of expensive Phase 3 failures for 2012.

But there are some lessons to be learned.

There may be exceptions, but drugs that react to form covalent bonds with their target(s) are very high risk.  Even if they work safely, as with clopidogrel, a non-reactive analogue (such as Brilinta™ ticagrelor) usually does better.  Covalent reactions put the time-scale of drug action at the mercy of the half-life of the target, and any off-target effects are usually more pronounced.  Even BG-12 may eventually be trumped by a non-covalent nrf2 activator (if indeed nrf2 activation is important for the safety and efficacy profile of BG-12),

While nrf2 may not be clinically validated, these two drug programmes have further stimulated a huge research effort, and the theory (if not yet the data) continues to highlight this pathway as an exciting future source of therapies for a range of diseases.  A better understanding of the different transcriptional outputs that can be achieved using different classes of activators will go a long way towards determining the fate of second-generation non-covalent nrf2 activators.

But perhaps the most important lesson is that even when drugs share a common molecular target the predictions that can be made for one drug from data generated using a structurally distinct agent are very limited indeed.  Biology is complex, and two very different ligands interacting with the same protein can yield almost completely different outcomes (so-called “functional selectivity” – the phenomenon that underpinned FXT’s non-classical somatostatin receptor agonists).

DrugBaron’s opinion, then, is that the findings with bardoxolone should not alter the view of BG-12 – at least until the data challenges this theoretical independence.  Multiple sclerosis patients – and Biogen shareholders – are no doubt hoping that the FDA will reach the same conclusion.

 

 

  • networkpharmacology

    There would appear to be a couple of potential explanations for the above.

    1. The central mode of action of nrf2 is also responsible for the adverse effect.

    2. The off-target binding activity of bardoxolone is responsible for the adverse effect.

    Sadly insufficient information is issued in the public domain to ascertain which is likely (recent post on what these means for biotech investors) although as you point out the results of BG-12 point toward the latter explanation.

    The issue in both cases is often the scrutiny is based on incorrect assumptions. For instance binding affinity below IC50 thresholds on a spectral spread of proteins that are important at the systemic level may be responsible for efficacy in the first place (rather than purely the headline target of nrf2) and may be responsible for the bardoxolone adverse effects in relation to other systems.

    Unless you lower the threshold of binding affinity and gather interaction data across a wider spread of proteins for each compound you would never know this. Also unless you can compute the high importance proteins at the systemic level then it is a bit of a needle in a haystack when seeking which proteins to gather this binding data on.

    The above is now possible but reliant on data and would seem a logical route to establishing the reality and perhaps paving the way for next generation drugs with greater efficacy and less risk of adverse effect.