I listened with interest to the 19 July 2017 FDA Workshop entitled “Development of New Tuberculosis Treatment Regimens-- Scientific and Clinical Trial Design Considerations.” Most of the materials from the meeting are available online (see link below my signature).
I was interested in this workshop both because I want to see new TB drugs progress but also because I wanted to listen for insights from TB-related strategies that would help with the problems we face in evaluating some antibacterial antibiotic combinations.
My more detailed notes are found below my signature but let me summarize here my two main insights.
Science of TB. Here I was delighted to learn that assays for a cell wall component (lipoarabinomannan, LAM) hold real promise as a way to get real-time insight into mycobacterial load and response to therapy. Very cool!
Science of Combinations. Sadly, I did not learn any clever new tricks for evaluating combinations in an efficient fashion. Outside of beta-lactam + beta-lactamase (BL-BLI) combinations, the core problem of showing the contribution of each component remains difficult. A certain amount of this can be addressed with preclinical and Phase 1 data, but a simple path to clear Phase 3 data remains elusive. This is an ongoing problem for certain types of add-on therapies (e.g., immune modifiers) and is not really a regulatory issue — there is simply no reason to use an add-on if you can’t show the benefit. This is going to require further contemplation!
In any event, many thanks as well to our colleagues at FDA for both holding this meeting and for making it possible to listen from a distance – the quality of the streaming was excellent, including the opportunity to participate in a fire alarm!
Best wishes, --jr
John H. Rex, MD | Chief Medical Officer, F2G Ltd. | Chief Strategy Officer, CARB-X | Expert-in-Residence, Wellcome Trust | Follow me on Twitter: @JohnRex_NewAbx
Upcoming events of note:
More detailed notes
Materials online: https://www.fda.gov/Drugs/NewsEvents/ucm548365.htm
The morning talks covered overall need, the current pipeline, preclinical tools, PK-PD, diagnostics, and data standardization. These decks speak well without commentary and I will offer only two further points, both from Debra Hanna’s talk. First, she mentioned that CPTR have submitted data to FDA showing excellent inter-laboratory reproducibility of hollow-fiber systems. Second, she described used of assays for lipoarabinomannan (LAM), a cell wall component of TB. As with assays such as galactomannan for aspergillosis, it appears that this marker offers both diagnostic potential as well as a real-time way to assess the impact of therapy on bioburden.
The afternoon’s integrated discussions of the problem of evaluating the component elements of a combination regimen were my greater interest.As you know, combination therapy is fundamental to TB treatment and I hoped for discussions at the workshop that might generate ideas for development of antibacterial combinations. For avoidance of doubt, I am not talking about combinations such as beta-lactam + beta-lactamase inhibitors (BL-BLI) in which the BLI drops the MIC from (say) > 128 mg/L to 0.25 mg/L.
Rather, I am interested in the setting in which the new agent has an independent activity but can’t be tested on its own as monotherapy. This may follow for several reasons and brief monotherapy might be possible, but the overall effect is that the pivotal tests of the new agent must be done in combination with other drugs.
This, in turn, leads to the combination rule which says that you need to show how each individual component contributes to the combination. FDA highlighted this with their list of questions to the participants where the #1 questions for the morning and afternoon, respectively, were:
Although it was the talk from FDA that summarized the rules around combinations, you should not view this as a regulatory problem. It really is not — the regulatory statement of the need to show the contribution of each component is no more than common sense. If you have a standard of care therapy (SOC) that works reasonably well, then what you need is a demonstration that SOC + New is better than SOC alone. And, the definition of better needs to be clear and compelling!
Combined with the ideas from the following talks (Spigelman, Wells, Vernon, Starke), the following points stood out for me. I’ll foreshadow by saying that (frustratingly!) none of these ideas seemed to offer a powerful new insight for antibacterial development.
First and most fundamental, antibacterial development benefits greatly from the fact that definitive monotherapy is often possible where this simply is not a path for a new TB drug. Brief monotherapy for TB is at most possible in Phase 1 — after that a combination is required. Hence, compromises that might be accepted for a new TB drug will not be accepted for a new antibacterial.
Second, preclinical PK-PD predictions must be confirmed in man. The TB community has made great strides, especially with hollow-fiber systems and these data can contribute to the demonstration that each component plays a role, but you still need a clear .
Third, the fundamental design of the TB community’s Unified Development Pathway is at heart the same as we generally use for new antibacterial agents and reduces to a randomized non-inferiority study vs. a standard regimen in the UDR (usual drug resistance setting) plus a (randomized if possible) trial vs. best therapy in the MDR/XDR setting. This should sound very familiar!
Fourth, superiority-based add-on strategies can be pursued only so long as there are no good alternatives. But, once you have active choices it becomes harder and harder to rely on this approach – and recent experience shows that we’ve probably already reached a point with antibacterial agents where such trials cannot reliably be run. This is great for patients but hard for developers!
Finally, all of this leaves me still frustrated in my hunt for reliable ways to evaluate antibiotic alternatives that are not appropriate as monotherapy.We can borrow bits from here and there, but core problem remains hard. The prototypical example here for me would be a virulence inhibitor – such an agent does not have an MIC, is less (or not) amenable to PK-PD-based dose selection, and like new TB drugs fails the test of “Would you use this agent by itself?” In that setting, the workshop did not suggest a path other than showing some form of superiority when the agent is added to a properly dosed and otherwise active regimen. To my eye, this creates a very high-risk situation for developing such agents.