Thursday, December 6, 2012

Writing Grants - Antimicrobial Screening - 3



Guest Blogger Lynn Silver of LL Silver Consulting, LLC 

The three main screening paradigms are 1) in vitro screens for enzyme inhibition, ligand binding, interruption of protein-protein interactions, etc. 2) phenotypic whole cell screens that select for inhibitors or effectors of specific cellular targets and 3) empirical, whole cell kill-the-bug-screens.  

To start with the last - for most grants I’ve seen, empirical screening seems mostly applied to natural products – but this is very inefficient if not coupled with a rapid dereplication system that can identify knowns, since a very high percentage of Actinomycete and fungal fermentation broths contain antibacterial agents and a minuscule number of those will be novel.   Note that most Pharma chemical libraries, and, likely, most other publicly available libraries, have been screened empirically already.  Empirical screens could still be useful if libraries were previously unscreened, or, for natural products, if the producing organisms are themselves novel.  But any empirical screening will require extensive follow-up to establish mechanism and safety.
In vitro assays for inhibition of enzyme activity (or other readout) are prevalent in grant applications.  Standard rules for HTS obtain (in establishing z-factors, etc.), but also important is setting up assays to reflect as well as possible intracellular conditions. For example, if ATP is present in the cell at 2 mM and you use 100 ?M in the assay – you may find ATP-competitive agents that will not show activity intracellularly.

With many screens there will be no true positive control – but that shouldn’t stop you.  It is very important, however, to run many negative controls,  antibacterials, possible interfering compounds – as false positives and high hit rates will complicate screening and require a good deal of counterscreening to eliminate.   In writing the grant, discuss planned controls, possible pitfalls and remedies.  As Dave noted in his initial write-up, many in vitro screening hits will not have the physicochemical properties necessary to enter the cell and will lack antibacterial activity.  This is especially true for Gram-negatives which have a tougher permeability barrier and more contribution from efflux pumps than do Gram-positives.  What will you do about it?  How do you plan to optimize in vitro hits to gain antibacterial activity – and, importantly, show that any antibacterial activity you do develop is related to enzyme inhibition?  In the same vein – you may find that your hit has antibacterial activity – but you must, very early on, establish that the antibacterial activity is indeed due to inhibiting your selected target – because this is very rarely true with screening of synthetic libraries.  Some pointers on connecting in vitro to whole cell activity may be found in O’Neill and Chopra (2004. Exp. Opin. Invest. Drugs 13:1045-1063) and Silver (2011. Clin. Microbiol. Rev. 74:71-109). 

One way of making the connection is to have a phenotypic assay which will give a whole cell readout that indicates a specific kind of effect on cellular physiology.  For example, inhibitors of cell wall synthesis should cause Gram-negative bacteria to form spheroplasts in osmotically stabilized medium.  And extraneous toxic properties of the hit should prevent spheroplast formation – so spheroplast formation is a good indication that a step (or steps) in cell wall synthesis has been inhibited.  In fact, the spheroplasting assay was used at Merck and at other pharmaceutical companies from the early 1960’s as a primary screen for cell wall inhibitors.   Thus phenotypic assays may be used as secondary to primary in vitro screens but also as another primary screening paradigm.  And primary screening with a whole cell phenotypic assay has the great benefit of selecting for compounds with activity against the whole cell.

Other phenotypic screens may use engineered bacterial strains to turn on a reporter gene if a specific target or pathway is inhibited – for example by taking advantage of stress regulons (see Fischer et al. 2004. Genome Res. 14:90-98), by specifically downregulating a desired target (for example by turning on antisense; see Singh et al. Curr. Opin. Drug Disc. Devel.  10:160-166) in order to sensitize to inhibitors or using a pair of strains engineered so that one should be sensitive to an inhibitor while the other is resistant (see Testa. 2003. Curr. Pharm. Biotechnol.4:248-259 for a screen for inhibitors of the non-mevalonate isoprenoid synthetic pathway).  Concepts of cell based screening are discussed in Mills and Dougherty (2012. pp. 901-929. In Antibiotic Discovery and Development.  Eds. Dougherty and Pucci. Springer.)  As with any antibacterial screening methodology, hits must followed up with tests of specificity (hitting the desired target[s] only), selectivity (hitting bacteria and not their hosts) and testing and optimization for the many layers of necessary pharmacological requirements for development (which presumably Dave or another guest blogger will discuss.

In writing screening grants, first read as much of the literature on screening as you can and recognize that there has been much of it done over the past 50 or more years.  It raises the reviewers’ hackles to read a dismissive introduction that says antibacterial screening has always been done empirically and now you will do it rationally.  It has been done rationally – BUT it must be noted that almost all the classes of antibacterial drugs in clinical use were FOUND by empirical and not rational methods.  Thus it would behoove us all to consider why such approaches haven’t worked so well (see Silver. 2012.  Clin. Microbiol. Rev. 74:71-109 and Gwynn et al., 2011. Ann. N. Y. Acad. Sci. 1213: 5-19) and approach new projects and grant-writing with an aim of improving on the past.

Friday, November 30, 2012

Suspending Disbelief - The FDA Reboot and Nosocomial Pneumonia


Yesterday I spent the day in front of my computer screen watching theatre of the macabre – the FDA Anti-Infectives Drug Advisory Committee’s latest review of Theravance’s telavancin for nosocomial pneumonia.  I watched this in light of all the activity going on at FDA to make their trial designs more feasible and to draw companies back into antibiotic R&D (more on this below).  Yet 95% of what was said at the meeting focused on the completely infeasible design of an all cause 28 day mortality endpoint with a 10% non-inferiority margin (see my previous blog on this). I noticed that Tom Fleming wasn’t there (I hope that’s a permanent change).  But that didn’t stop one of the statistician committee members from justifying a no vote by saying that there was no clear evidence that the comparator drug, vancomycin, worked better than placebo.  Of course he’s right – but there is no clear evidence that parachutes work better than placebo either! 

NO ONE WILL EVER RUN TRIALS IN NOSOCOMIAL PNEUMONIA UNDER THE FDA’S CURRENT GUIDANCE.  NO ONE.

Those of you who have been following this blog (and some who haven’t) are familiar with the now-famous FDA reboot announced by Janet Woodcock of the FDA in May of this year.  Changes in FDA requirements for trial designs focusing on feasibility of non-inferiority designs as well as approaches to the rapid development of antibiotics active against multiply-resistant pathogens have been discussed now at two Brookings meetings and a separate FDA task force meeting.  I attended both Brookings meetings.  In fact, the change in FDA attitude seemed to be foreshadowed by Ed Cox’s presentation at the Gordon Research Conference on New Antibacterial Discovery in April of this year.  Since then, I have been on a high drinking in all of the optimistic and forward-looking pronouncements from FDA top management. 

In this context, I was astounded, flabbergasted, disappointed, distressed and angry watching the AIDAC meeting yesterday. I had hoped that FDA might even hint that they were going to drop this completely unrealistic trial design requirement for nosocomial pneumonia.  But no. In light of yesterday’s meeting, if this is the FDA reboot for antibiotic development, what we need is the nuclear option.

 Of course, this is not the first time the FDA has pulled the rug out from under us and it won’t be the last. But what this means is that I cannot pass the red face test recommending that clients even approach FDA to discuss trials in pneumonia – at least not before the clients have discussed designs with European regulators. It also means that until the FDA comes out with some concrete proposals, (which may have to wait another year or two for actual guidance documents), the reboot, as far as I’m concerned, is on hold.

And I don’t understand the FDA’s dilemma here.  We don’t have to wait for guidance. In fact, I think hasty development of guidance is a bad idea. A quick way forward would be a workshop or series of workshops or even an AIDAC meeting where they present potential designs and get comments.  These meeting presentations and summaries frequently serve as “FDA current thinking” for sponsors to plan trials and provide a starting point for discussion with FDA.  Of course, the FDA may not be ready for such public meetings as yet. But right now – we have NOTHING other than optimistic discussions that have now been undermined by FDA action at their own advisory committee meeting. 

Wednesday, November 28, 2012

Writing Grants - Antimicrobial Screening - 2

 Guest Blogger Lynn Silver of LL Silver Consulting, LLC 


Thanks to Dave Shlaes for the opportunity to contribute to his blog with an entry on grant-writing-and-screening-paradigms.  I have some familiarity with the subject as I worked in antibacterial discovery for 21 years at Merck, including the development and deployment of numerous screens, and I have since been consulting in the pre-clinical discovery sphere.  But, more to the point, I, too, have been doing antibacterial drug discovery peer review for NIH and the EU as well as reviewing for journals. And, in my humble opinion, there is a need for a remedial course in antibacterial screening (and drug discovery in general). Let me emphasize that this is indeed all my opinion – you and others may disagree.

My remit here is to discuss screening paradigms.  I’m not sure if that includes target choice – but I’ll start with a few words on that particular tough nut.  Numerous reviews have covered the various points important for selecting and qualifying targets.  But each author has his or her own POV and will emphasize personal favorites. 

My set: 
1) The target should be essential to the growth, and better, to the survival, of the pathogen [preferably under conditions that can be measured in vitro but that also approximate in vivo conditions].  Genetic proof, preferably with conditional mutants or constructs, is optimum – but demonstration of inviability of knock-outs can suffice.  Don’t pick a target whose product can be supplied by nutrients in the host. I.e. avoid auxotrophic targets in most cases [but there are exceptions].   Anti-pathogenesis targets are a different area; they are interesting – but so far not much has come of them.  Reviewers will be skeptical of them – so if this is your bailiwick, prepare to explain how such an inhibitor would be developed [as a therapeutic? A prophylactic? In combination with a standard antibacterial?]

2) The target should be present in a useful spectrum of pathogens with homologs being structurally and mechanistically closely related.

3) The target should have no human homolog – or there should be a good reason to believe that selectivity [kill pathogen not host] can be achieved.

4) Inhibition of the target should not lead to rapid selection of resistance.  Anl hypothesis to consider: inhibitors of a single-gene encoded target will select for pre-existing resistant mutants in the infecting population.  If the frequency is high enough and the mutants fit enough, then such resistant mutants may compromise therapy.  I have been a proponent of the idea that good monotherapeutic antibacterials target the products of multiple genes or structures produced by multiple gene products (Silver, LL. 2007. Nat.Rev. Drug Disc. 6:41-55) and, as such, select much less frequently for single step, high level target-based resistance than do single-targeted agents.  If you choose a single-gene encoded target, be prepared to defend the choice. 
 
5) Having ways of measuring the activity of the target or function in vitro as well as the phenotype of inhibiting the target in the intact cell are important for screening and/or follow-up. 

6) Ability to produce the target in quantity for structural studies and eventual iterative optimization is a plus - but not a big enough plus to substitute for meeting the other necessary criteria.

7) Druggability.  Does the target have a deep enough, well-defined enough binding site for selective inhibitory ligands?  Does your chemical library contain likely ligands?

8) Read Payne. et al. (2007. Nat. Rev. Drug Disc. 6:29-40) and note that they ran 70 HTS antibacterial campaigns with very little to show for it.  Why will you succeed when GSK failed?

OK – you’ve got a target.  Consider the chemicals to be screened.  Be warned, chemical libraries are often filled with aggregators, detergents, alkylating agents – compounds that will interfere with many types of screens.  Read Babaoglu et al. (2008, J. Med. Chem.51:2502-2511).   And, similarly, the detergents and alkylators and many other types of toxic compounds will have uninteresting, non-specific antibacterial activity.  It is easy to kill bacteria, even multi-drug resistant bacteria, with chemicals.  It is hard to find novel antibacterials that are highly selective for pathogen over host.

Stay Tuned for Part 3 in this series and - coming very soon - FDA and Nosocomial Pneumonia. 

Friday, November 23, 2012

The Pharmacometric Solution: We Told You So!




Paul G. Ambrose, Pharm.D, FIDSA, Sujata M. Bhavnani, Pharm.D, M.S. and George L. Drusano, M.D., FIDSA


It was 11 October 2007 at a US Food and Drug Administration (US FDA) and Infectious Disease Society of America (IDSA) co-sponsored workshop where we first proposed a pharmacometric solution to some of the most vexing problems in antimicrobial drug development [CID 2008;47:S225-31].  First, we described how pharmacometrics could be used to determine the magnitude of treatment effect, which allows for the estimation of non-inferiority margins using contemporary clinical data [AAC 2012;56:1466-70].  This is important, as the alternative is to use historical data.   In our view, the use of historical data is assuredly unwise because of the disparity between modern medical care and that available at the dawn of antimicrobial chemotherapy. Second we described how continuous, numeric endpoints  holds the promise for the evaluation for evaluating the impact of drug exposure on time to event, thereby reducing the number of patients required to detect between-regimen differences, and defining the optimal length of therapy.

In December of last year, the European Medicines Agency (EMA) released a guideline for the development of antimicrobial drugs, which questioned the usefulness of historical data [EMA guideline 2011] and, importantly, explicitly mentioned the pharmacometric solution.  The EMA guideline said “Historical data are often used to estimate the no-treatment effect but the relevance of these data to current medical practice may be questionable.  Sponsors are encouraged to explore alternative and emerging methods for estimating the no-treatment effect (e.g., using pharmacometric-based approaches)."  We applauded the EMA for taking an important step towards ensuring the development of new antimicrobial agents and urged others to support the pharmacometric-based approach [Lancet ID 2012;12:265-6].  In October 2012, IDSA published recommendations on the conduct of superiority and organism-specific clinical trials of antibacterial agents for the treatment of infections associated with drug-resistant bacteria [CID 2012;55:1031-46].   Therein, IDSA suggested a pharmacometric-based approach to defining the no-treatment effect using contemporary data.

The 22nd October EMA workshop further highlighted the rising importance of pharmacometric approaches in the development of antimicrobial regimens for the treatment of drug-resistant bacteria.  There we made two presentations.  The first described how the use of in vitro and animal models as well as mathematical models can be used to identify treatment regimens that hold the promise to slow or prevent the amplification of drug-resistant pathogens, while the second described a paradigm where high-quality pharmacometric information obtained from pre-clinical and clinical data sources supported approval and balanced unmet medical need and data quantity.

The in vitro models provide insight into the relationship between drug concentration and cell kill as well as resistant subpopulation response to drug pressure. The animal models allow the impact of drug penetration to the infection space to be elucidated and also allow some insight into the impact of the immune system, which is completely missing from the in vitro system. The mathematical modeling ties this information together to allow delineation of exposure profiles that are linked to specific endpoints (2 Log kill, resistance suppression, etc). Human pharmacokinetic information, including penetration into specialized spaces such as ELF allow the final bridging between pre-clinical data and the clinic. This also allows the generation of optimal approaches to study patients in the clinic so that exposure-response relationships may be developed.

Given the rising tide of resistance to our existing antimicrobial armamentarium, a risk-benefit approach for the assessment of the effectiveness of is appropriate.  Simply put, it is unacceptable to have an absence of active antimicrobial agents for the management of infections associated with drug-resistant pathogens.  The EMA is listening and acting.  The FDA is clearly listening.