Monday, May 15, 2017
Last week, a remarkable paper was published in Nature (subscription required – sorry). The paper is remarkable because it takes a simple, practical and elegant approach to understanding how drugs penetrate Gram-negative bacteria. This problem has confounded antibiotic discovery for decades. The penetration problem has recently been highlighted by the Pew Charitable Trust Roadmap for antibiotic discovery and by the proposed OMEGA project. The problem has been so challenging that many scientists, including yours truly, did not believe that there were universal rules that one could follow to achieve penetration of antibiotics into Gram-negatives. While we may not be 100% wrong, it is now clear that there are a few important such rules for E. coli, an important Gram negative pathogen.
Richter et. al. from the Department of Chemistry and the Institute for Genomic Biology at the University of Illinois used a very sensitive method (tandem mass spectrometry) to actually measure the penetration of various molecules into E. coli. They then studied a collection of 100 molecules including antibiotics with known activity against E. coli and those that were active against Gram positive bacteria but not against Gram negatives. Based on these data, the authors were able to show a strong correlation with the presence of a positive charge at neutral pH or primary amine. They also realized that even though a positive charge might be required, it was not sufficient since not all such molecules were able to penetrate E. coli.
To further explore the requirements for penetration, the authors carried out a search of compounds with primary amines for 297 different molecular characteristics. Those characteristics that were associated with penetration could be identified. The authors then went further and measured the penetration of compounds where specific characteristics differed – one at a time. In this way, they showed that rigid compounds penetrated better than flexible ones and that flat compounds were better than globular compounds. Using this information, they were then able to take a natural product that was active against Gram positive bacteria, and by altering it chemically along the lines of their predictive model, they were able to create a new compound with activity against both Gram positive and Gram negative bacteria.
What is surprising to many of us is that the data provided by the authors differs considerably from the conclusions we had surmised based entirely on retrospective analyses of marketed antibiotics. The great thing about science is that progress means change and with progress more questions always arise. Its perpetual.
So, Hermione, there are at least some rules after all. While the paper by Richter and coworkers is a breakthrough – no question about that – it is not going to be the whole story. And the authors clearly recognize this. How about Klebsiella? It is closely related to E. coli. What about E. coli that are resistant already either via increasing their ability to efflux compounds out of the cell or by limiting the ingress of compounds? Then there is the problem of Pseudomonas – a pathogen with not only the ability to radically limit ingress of molecules compared to E. coli, but also one that more readily pumps them back out again. What about Acinetobacter? – a complete unknown here. Will these rules apply to all Gram negatives or are they specific to E. coli?
Richter and coworkers have shown us a way forward. We now have to fill in the rest of the blanks.
Monday, May 8, 2017
I take pen (computer keyboard) in hand today along with John Rex who has agreed to be my co-author of this blog. Because John is currently very busy, the retired guy (me) gets the last word.
Last week, the Infectious Diseases Society of America (Boucher et al) published a white paper on the development of antibiotics for resistant pathogens, narrow spectrum indications and unmet needs. An editorial by George Drusano accompanies the paper. These are important discussions that are based on the myriad of workshops and advisory boards held by the FDA over the last several years. The white paper represents clearly the views of the IDSA. We are grateful that the IDSA has taken this step (full disclosure – JR is a co-author).
A summary (credit to JR here) of core concepts of the paper might be –
1. AMR continues to advance and narrow-spectrum drugs could be valuable tools to address this problem.
2. But, developing narrow-spectrum drugs is surprisingly difficult when you want to do so for less common pathogens or forms of resistance – and these are precisely the settings where we most need new drugs.
a. There are a few (but only a few) settings where narrow-spectrum is / would be pretty easy: S. aureus (skin) and GC (STDs).
b. Everything else is / will be hard.
3. ALL approaches to the general problem of narrow-spectrum drugs have significant flaws.
a. There is no simple approach: 100+ people attacked this for a day during July 2016 and failed to generate any strong alternatives.
4. We thus must find ways to use a collection of relatively flawed tools as the basis of registration.
5. The alternative action of not finding this pathway is also a form of action and would be unacceptable.
6. In this/the/an alternative plan, comparative clinical data will be needed on each new agent but generating these comparative data is (and should be!) very hard.
a. The needed patients are uncommon … and we want this to remain true, as if they are common then we’ve failed with infection prevention.
b. Neither non-inferiority-based nor superiority-based approaches are routinely tractable.
c. Diagnostics will not solve the problem in that they don’t create patients with the target pathogen – they only help spot them.
d. That said, screening for such rare patients would be facilitated by layering study of such drugs on top of a clinical trial network that is actively running UDR-focused studies.
7. Excellent preclinical PK-PD programs are now possible and need to be better leveraged.
a. Multiple isolates and models.
b. PK in patients with the target syndrome.
c. Is there a way to use these as an informative Bayesian prior? Not clear, but perhaps worth further study.
8. To resolve these issue, and despite their flaws, we will need to use elements from these two categories of clinical research tools:
a. The concept of validated Animal Models (approaching the ideas of the Animal Rule to the extent possible).
b. Validated external controls (e.g., reasonably contemporaneous controls who could actually have been enrolled).
9. At a practical level, we need to validate ertapenem for HABP-VABP
a. This appears to be the best way to provide a companion for a narrow-spectrum anti-pseudomonal.
b. There is a paper coming out on this from the Ambrose group.
10. And to succeed in a sustained fashion, we must make pull incentives work.
One quibble – (see this prior blog) the paper uses the term “superiority studies” in a very specific way. They mean prospective randomized controlled trials designed to determine if one treatment is superior to another. The IDSA argues that such trials are not feasible outside a large outbreak of infection caused by virtually pan-resistant bacteria – something we work hard to avoid. They also point out the difficulty of randomizing patients to treatments that, in that situation, are likely to be inferior. We agree. But there are other ways of running superiority trials. As the paper also points out, the use of validated external controls (something I have been pushing for the last two years) can help here. Whether or not such a trial will be able to enroll sufficient patients to achieve statistical superiority or not is not yet clear – but the external controls will clearly increase the power of such a study. (On a personal note, I am very gratified that the idea of validated external controls seems to have taken hold among my colleagues).
The other aspect of the paper we would like to emphasize involves the use of animal models. We agree with the FDA and the IDSA that exploring models that might more clearly reflect the human situation as compared to our current mouse models is a good idea. But to achieve the validation of any new model, as clearly stated by both the IDSA and Drusano, will take years of effort and is not a guaranteed result. All agree that the existing models are well validated and are likely to provide reliable dosage predictions when combined with appropriate human pharmacokinetic studies.
Wednesday, April 26, 2017
Recently, Rep. Jan Schakowsky (D-IL) and other democrats introduced legislation aimed at controlling drug prices, and, for us, providing for a prize for the introduction of a new antibiotic to market that provides activity against resistant infections or other key lifesaving characteristics. It is great that someone in congress wants to address the market failure for antibiotics with a government funded pull incentive. We’ve all been waiting for this news for a long time. But – it looks like we’ll have to keep on waiting. This bill is so full of poison pills that my only hope is that it never passes as is.
In Title III of the bill, the prize is defined as $2B to come from the 2018 budget and to be available for 10 years. That fund is to provide for up to three prizes. Really? It would be enough for one prize, but not three. Therefore, the appropriation would have to be renewed after every award. Right. Of course, the goals are a little vague. They include reduction of readmission, an improved diagnostic or better programs in antibiotic stewardship. Where was the goal for a new antibiotic again?
The prize is to be doled out by the Director of NIH. Although the NIH has a small number of antibiotic experts in its ranks, it has been pretty much the opposite of an antibiotic R&D powerhouse for decades. Why should the NIH control the award? Why not BARDA that has great experience in this area?
Then, the company has to set a “reasonable price” for the product (whatever that means) and must at the same time give up exclusivity. How does that work? Obviously, if you no longer control the patent on the product and you are immediately beset by generics, there is no motivation to spend money on marketing or on anything else. At that point, you’re in commodity mode where your greatest expense is manufacturing and distribution. In addition, any marketing materials (why do you need those if you’ve given up exclusivity?) must be submitted to NIH, CDC and FDA before release. That guarantees a delay of 10 years.
Then there is the usual request for study by the National Academies (again) on the utility of push and pull incentives to stimulate research and development and on de-linking costs from pricing. There is no doubt that push and pull incentives could have an effect on pricing as has been clearly demonstrated in the detailed analysis in the O’Neill reports from the UK. I still question whether these mechanisms “should” have too great an effect on pricing since high prices help to control use and overuse is not a good thing.
I want to spend some time, again, on this idea of de-linking. I think I may be alone among my colleagues in wanting to remove this jargon from our vocabulary. First, you don’t really remove cost considerations from pricing since you are paying for some of the research in your push incentives – so one way or another, we pay. Second, the term fails to take into account the important aspect of physician education that must occur with the introduction of any new antibiotic to the market. Physicians have to understand the advantages and disadvantages of any such product, its appropriate use, when it should not be used, and how to make these critical decisions. So far, this has been driven substantially by the company marketing the product, and involved collaboration with physician thought leaders, pharmacists and others in the health care system. If the company did not fund these activities, who would? Has anyone thought about where this activity would come from?
It is very disturbing, frustrating and demoralizing (but not surprising) to see such an important idea, pull incentives to stimulate antibiotic research and development, receive such a bludgeoning at the hands of the political party that should know better.
Thursday, April 13, 2017
Today the FDA held an important Advisory Committee meeting to discuss approaches to the clinical development and approval of antibiotics designed for specific pathogens like Pseudomonas aeruginosa and Acientobacter baumanii. The meeting materials can be found here.
(OK. I gave up just after 3:30 pm – so I may have missed something really important).
Once again, the make-up of the committee left much to be desired. There was no PK/PD person with expertise in antibacterial drugs on the panel. It became obvious during the discussion that this was a serious omission. The safety expert came from a parasitology background and was obvious unfamiliar with antibacterial development. Lynn Marks, as the industry representative, was, mostly, quiet (I’m not being derogative here). One pleasant surprise was the consumer representative who seemed knowledgeable and asked relevant questions.
The meeting began with Ed Cox and Sumathi Nambiar reviewing prior recent FDA meetings examining this question. Both emphasized how difficult it has been to identify a means to develop these agents in the content of all our prior experience with antibacterial drugs. In particular, Sumathi noted the disadvantages of both the superiority approach and the NI approach to these agents. One issue that folks have with the superiority approach is that it is time limited. That is, after the first one or two drugs, such trials will no longer be possible.
Robin Isaacs later presented the Entasis plan for a non-inferiority trial of their new B-lactamase inhibitor-sulbactam combination for the treatment of Acinetobacter infections. He also showed that about 60% of Acinetobacter globally are multiply antibiotic-resistant. Robin noted that the mortality of serious Acinetobacter infections like HABP/VABP and bacteremia approach 40% when treated with standard of care and 80% if untreated. Given the 40% number, he justifies an NI margin of 20% in an NI trial with 28 day all cause mortality as the endpoint. All the speakers noted that a rapid diagnostic only makes enrollment slightly easier and may limit the use of prior antibiotics. Robin said that a rapid diagnostic was not necessary especially if up to 48 hours of prior therapy could be allowed. Entasis believes that such a trial, that would target geographic areas where Acinetobacter has a high incidence like Taiwan and Thailand.
The FDA presented information that had previously been shown by Polyphor regarding POL7080, an antibiotic targeting Pseudomonas aeruginosa specifically. Polyphor declined to present themselves (?). Their plan for a NI trial with a 10% NI margin suggested that they would have to enroll over 3000 patients to complete the trial.
For me, though, the highlight of the day was the Infectious Diseases Society presentation by Trish Perl. She suggested using validated external controls to bolster trial data. Although Trish was not specific, this seems most likely to occur in the context of a superiority trial. I have been saying this for years now. By carrying out such a trial with a small number of patients where there is a 4:1 randomization to best available therapy, we can use the active controls to validate the external controls we have used. The FDA seems to be uncomfortable with external controls, but might be more comfortable with a validation plan as suggested by Trish and IDSA.
Another highlight was John Rex’s public comments. I paraphrase here. He suggests we will need four lines of evidence. Animal models. PK in man. Safety profile. Some clinical data consistent with the above. Stewardship guarantees lack of overuse. If we have no path – we’ll lurch from crisis to crisis. We live this now. We lack drugs for some bugs. If the drug is available we can cautiously improve our understanding of the agent (given the limited data used for approval).
Mike Green asked about “provisional” approvals or conditional approvals – something that, without a surrogate endpoint, the FDA is unable to do under current law. This is a topic we discussed at the pathogen-specific workshop last year. There we suggested that one could return to a product post-market if data warranted it. But we should be clear that the drug would be approved – not conditionally approved. If at post-market follow-up a significant problem was identified, the drug could then be reviewed at another advisory committee and might then required label changes or might even have to be withdrawn. This is obviously a cumbersome way to do this. A change in regulations here would be welcome – but would apparently require legislation – an area where we are currently paralyzed.
Another issue that arose was the one of diagnostic accuracy – especially in patients with HABP/VABP. This is not like patients with cancer where we have a tissue diagnosis and we can even carry out receptor typing prior to initiating therapy. The problem is that if we enroll patients where our diagnostic accuracy is poor, the “noise” engenders a tendency for all outcomes to be the same. For me, this also raises the problem of mortality as the endpoint for this indication. We believe that 50% of this mortality is unrelated to the infection being studied. This also will drive any trial towards the null. For an NI trial this is good for the sponsor but not for the patient and physician. For a superiority trial, this is not good for the drug and may result in non-approval of an efficacious agent. Jack Bennet emphasized this dilemma for us.
In answer to the questions posed by the FDA, a number of panel members supported the idea of using a clinical trials network. There was general support for non-inferiority trials on the panel in contrast to many of the speakers in the public comment section. Many panel members wanted post-market evaluations to be an important component of the approval of pathogen-specific agents.
In conclusion, the consensus seems to be that we must go here. There was also agreement that this is a tough problem and that the best we can achieve is some kind of “happy medium.” But - “You won’t get a recipe from us.”