Last week the United States Food and Drug Administration approved the drug pretomanid for use in combination with bedaquiline and linezolid “for treating a limited and specific population of adult patients with extensively drug resistant, treatment-intolerant or nonresponsive multidrug resistant pulmonary tuberculosis (TB)”.
The phrase “in combination with” is critical since the evidence on which the approval is based is mostly from a trial in which the three drugs were used in combination.
This highlights one of the key challenges in TB research today – disentangling the impact of single drugs from the impact of combinations of multiple drugs.
In this particular case, there is no way to know whether the combination of bedaquiline, linezolid and pretomanid worked because all three drugs made critical contributions, or whether it worked because two of the drugs worked very well and the other just moderately, or not at all.
It is not implausible that the success of this three-drug combination is mainly due to bedaquiline and linezolid. In fact, much improved XDR-TB survival rates have been seen in South Africa in recent years as these two drugs were more widely provided and both are now included in the World Health Organisation’s list of preferred drugs for the treatment of MDR-TB. Critically, these improved outcomes were seen without pretomanid.
Though the quality of the scientific evidence still isn’t what it should be, few would disagree today that bedaquiline and linezolid are highly effective drugs that should be included in any drug combination to treat MDR or XDR-TB, providing, of course, the patient is not resistant to either.
By contrast, we really do not know much at all about pretomanid. The key trial on which its approval was based, called Nix-TB, was a single-arm trial of 109 people – in other words, there was no blinding and no comparison with a control regimen. While single-arm trials are generally a bad idea – because it provides much less reliable evidence – it was an appropriate design in this case. That is because when the trial started there was no real standard of care for the treatment of XDR-TB – with five-year survival in South Africa being below 30%. Randomising patients to such odds would have been unethical.
The Nix-TB trial was rightly hailed as a breakthrough when its first results were reported in October 2016.The fact that most trial participants sick with XDR-TB were cured – and cured in six months without hearing loss (common with some older TB treatments) – was a major leap forward. Dr Francesca Conradie and others involved with the NixTB trial conducted here in South Africa were part of something historic and extraordinary.
But, nevertheless, it can both be true that the Nix-TB trial was a major breakthrough and that one of the drugs in the NixTB trial regimen might not work very well or at all.
We have hardly any published evidence from randomised controlled trials (RCTs) in which pretomanid has been compared to other drugs. For example, we do not have a trial in which XDR-TB patients were randomised to receive bedaquiline, linezolid and either pretomanid or delamanid (delamanid has a similar action to pretomanid but has been used much more widely). A head-to-head comparison of delamanid and pretomanid is obviously needed, yet somehow such a trial has not yet been prioritised. We also do not have pretomanid trials similar to the phase III RCTs conducted for delamanid and bedaquiline (the latter is still under way).
To the contrary, some of the more recent developments of pretomanid have raised red flags. Both in the STAND and now the SimpliciTB trials, the MDR-TB section of the trials are essentially single-arm trials. That is to say all MDR-TB patients in these trials receive regimens containing pretomanid – meaning there is no group of MDR-TB patients in the trial with which to make comparisons. Such a single-arm design was appropriate in NixTB, but is shocking to see in an MDR-TB trial where good control regimens exist. That pretomanid’s developer, the non-profit TB Alliance, has chosen this path is disappointing.
In addition, as part of the FDA process some rather disappointing results emerged from the STAND trial – indicating that the pretomanid-containing regimen in that trial could not be shown to be non-inferior to the standard treatment for drug sensitive TB (DS-TB) and that the pretomanid-containing regimen was associated with substantially more deaths and more severe adverse events than standard DS-TB treatment (see page 115 in this dossier). The FDA would, of course, have taken into account that the risk benefit calculations look quite different for XDR-TB than in DS-TB, which is probably why bad news like this was not a deal-breaker for an XDR-TB approval. As an aside, it is unacceptable that this data has only come to light through the FDA process and that it has not yet been published in a medical journal.
On a more positive note, Doctors Without Borders (MSF) has stepped into the evidence vacuum and is conducting an RCT called TB-PRACTECAL in which regimens containing pretomanid are being compared to a WHO-recommendations-compliant local standard of care which does not contain pretomanid. This trial, expected to report findings in 2021, is probably our best bet for knowing whether pretomanid can be used to treat MDR-TB. For the time being, using pretomanid to treat MDR-TB, even in a clinical access programme as some have suggested, seems to be putting patients at unnecessary risk given that we have other good drugs available about which we know a lot more. An XDR-TB and hard-to-treat MDR-TB clinical access programme on the other hand would be a good thing, given that this group of patients really don’t have many other viable options.
Hopefully pretomanid turns out to be a safe and extremely effective drug. We desperately need it to be.
However, until we have more evidence, we should keep our minds open to the possibility that pretomanid may not be effective or may only be moderately effective – and hope no major safety issues emerge as more people are exposed to the drug.