March 2019: Antimicrobial agents

[Seinfeld Voice] What’s the deal with AmpC beta-lactamases?  This month’s CID has a nice review of a topic surrounded by much lore and rather less clinical data: the AmpC beta-lactamase.  Let’s go over the highlights:

First, AmpC beta-lactamase resistance comes in two types.  There is non-inducible resistance, which occurs in bacteria with either chromosomal AmpC genes carrying promoter mutations (seen in E.coli, Shigella and A.baumannii) or AmpC genes carried on plasmids (seen in K.pnuemo, E.coli, Salmonella, and others).  Then there’s the more clinically vexing inducible resistance, seen in E.cloacae, E.aerogenes, S.marcescense, C.freundii, M.morganni P.stuartii, and P.aeruginosa.  At its simplest, the inducible resistance system works like this: a regulatory protein, AmpR, is continually expressed and blocks expression of the AmpC beta-lactamase.  However, when beta-lactams are present and block cell wall assembly, cell wall degradation products accumulate and bind to AmpR, disabling it and allowing production of the beta-lactamase.  Because different beta-lactams produce these degradation products at different rates (presumably because they have affinities for specific PBPs? I’m a bit unclear on this part), some beta-lactams are better “inducers” than others, and in the presence of those inducers the bacteria with an inducible AmpC will shift from a beta-lactam susceptible to resistant phenotype.

So, which beta-lactams are the potent AmpC inducers?  The aminopenicillins, the first-generation cephalosporins, and the cephamycins (e.g. cefoxitin, isolated resistance to which is the traditional marker of an inducible AmpC phenotype).  On the other hand, weak AmpC inducers include pip-tazo, aztreonam, and the third-generation cephalosporins –among which cefepime is particularly attractive, as it not only doesn’t induce AmpC much but is also stable to hydrolysis by AmpC anyway.  This has at least in vitro consequences – resistance to cefepime emerges in E. cloacae much more slowly than resistance to ceftriaxone or ceftazidime.  As for the carbapenems, imipenem potently induces AmpC, but that doesn’t matter much as the carbapenems are all stable to AmpC.

Not all species with inducible AmpC genes are equally likely to demonstrate that inducibility in clinical practice.  So, for which organisms do we need to worry about the AmpC phenotype? (practically speaking: when should we assume resistance to third-generation cephalosporins despite the initial susceptibility testing returning sensitive)?  This is where the data is limited.  The evidence is clearest for Enterobacter, where ceftriaxone resistance has been shown to emerge on treatment in 8-19% of cases; the authors advise avoiding ceftriaxone for Enterobacter except for uncomplicated urinary tract infections.  Though this review points to only two clinical studies of emergent 3rd-gen cephalosporin resistance with inducible AmpC organisms other than Enterobacter, the rates of such resistance were 0%-7% in both cases.  So for the most part, when we talk about clinically relevant inducible AmpC infections, we’re talking about Enterobacter.

How do we detect AmpC beta-lactamases?  There are no CLSI-endorsed critera, but both cloxacillin or boronic acid inhibit AmpC activity and so can be added next to a cephamycin or 3rd-gen cephalosporin disk on a susceptibility plate to perform a double-disk synergy test.  Given that clinically-relevant inducible AmpC infections are limited to just a few bacterial species, though, I think actually ordering that would be pretty boronic.  Instead, it might be wise to just ask the micro lab not to report cephalosporin susceptibilities for Enterobacter.

How should organisms expressing AmpC be treated?  An international survey of ID specialists suggest 58% prefer carbapenems, with most of the rest preferring cefepime and a minority (8%) pip-tazo.  The data comparing outcomes with various regimens are all retrospective and most have small sample sizes.  A pilot randomized controlled trial, MERINO II, is currently underway to compare pip-tazo to meropenem for inducible AmpC organisms, which should shed some light on the issue and provide the basis for a second, properly powered trial.  Cefepime arguably has a better track record than pip-tazo, with few failures reported and those mostly occurring in high-inoculum infections treated with twice daily dosing.  My opinion is that high dose cefepime ought to be adequate for most cases of infection with inducible AmpC organisms and that carbapenems ought to be reserved for severe infections or highly immunocompromised hosts; your mileage may vary. 30838380

Ceftriaxone at home is as good as IV flucloxacillin in the hospital for children with cellulitis.  The Cellulitis at Home or Inpatient in Children from the ED (CHOICE) trial randomized kids aged 6 months to 18 years who had a moderate to severe cellulitis to receive either OPAT with ceftriaxone or inpatient hospitalization with IV flucloxacillin. The primary outcome assessed was treatment failure, defined as change in the empiric antibiotics within 48 hours due to either lack of clinical improvement or an adverse event.  The study had a non-inferiority design with a margin of 15% and took place at a single Australian hospital over two years.

A total 190 patients were randomized, 95 to each arm; two children were excluded after randomization due to being found ineligible for the study, resulting in an intention-to-treat analysis of 188 children.  Treatment failure occurred in 2% of the children who received ceftriaxone OPAT versus 7% of those who received flucloxacillin in the hospital (risk difference -5%; 95% CI -11% to 1%, p=0.09).  This difference was driven primarily by a lower rate of adverse events in the children receiving ceftriaxone (2% vs 11%; p=0.05), which squares nicely withe ton of cephalosporin vs antistaphylococcal penicillin for MSSA papers all showing better tolerability of the cephalosporins.  No differences in acquisition of nosocomial pathogens (MRSA, ESBL-producing gram-negatives, or C. difficile) were observed between the two groups.

So, ceftriaxone administered via OPAT was non-inferior (and very nearly shown to be superior) to inpatient IV flucloxacillin for moderate to severe cellulitis in kids.  But this begs the question – given that PICCs are expensive and produce adverse events of their own, why not send the kids home with a bottle of cephalexin?  We have studies showing that osteomyelitis and endocarditis are readily treated with oral antibiotics, so what data shows that cellulitis - even severe cellulitis – benefits from IV versus oral antibiotic treatment? 30853250

Neither peak nor trough ganciclovir levels correlate with either clinical efficacy in CMV viremia or the incidence of adverse events. The authors retrospectively studied 82 patients who received ganciclovir with therapeutic drug monitoring for CMV infections over a 10-year period. The patients had a median age of 55, half were men, 86% were white, and most had undergone either stem cell (45%) or solid organ (36%) transplantation. The patients received a median dose of 450mg ganciclovir daily for a median 19 days, with therapeutic monitoring and dose adjustment after a median 7 days. Monitoring included a measured peak ganciclovir concentration in 79% of patients, of which 83% were at goal; 71% of patients had a trough concentration, of which 83% were at goal. The authors found no associations between either the peak or trough ganciclovir levels and either clinical efficacy endpoints (e.g. reduction or clearance of viremia) or patient safety outcomes (e.g. cytopenias, nephrotoxicity, or neurotoxicity).

This is a great example of how a “negative” study can still be important and directly impact clinical care. In this case, we learned that therapeutic drug monitoring for ganciclovir is unlikely to be clinically useful, not least because peak and trough are usually within target ranges even without monitoring. Moreover, these data suggest that a patient with an infection not responding to ganciclovir at standard doses may not benefit from pushing the dose higher, and similarly that someone who develops an adverse event on ganciclovir might not benefit from a dose reduction. 30602515

Ribaxamas is a new poorly bioavailable oral beta-lactamase for the prevention of C. difficile infection (CDI) in patients treated with beta-lactams.  Here’s a clever idea. Intravenous beta-lactams (particularly the 3rd-gen cephalosporins) cause CDI by diffusing into the gastrointestinal tract to disrupt the GI microbiome.  So what if we could give patients a beta-lactamase that stayed confined to the GI tract, such that IV beta-lactams could maintain therapeutic levels in serum, but would be destroyed as soon as they leaked into the gut?

The authors (who work for the drug manufacturer) performed a multicenter double-blind RTC randomizing patients expected to receive at least 5 days of ceftriaxone for lower respiratory tract infection to receive either ribaxamase (150mg by mouth four times daily) or placebo for the duration of ceftriaxone treatment plus 72 hours after finishing therapy.  The primary outcome of interest was the incidence of CDI within the following 4 weeks after treatment.  Patients, investigators, study staff, and sponsoring personnel were all blinded to the treatment assignment during the study.

Over a year of recruitment, the researchers randomized 413 patients (ribaxamase, n=207; placebo, n=206).   A total two patients in the ribaxamase arm and seven patients in the placebo arm developed CDI (1% vs 3.4%; difference 2.4% with 95% CI -0.6 to 5.9%).  Interestingly, while adverse events were similar between groups, deaths were numerically higher in the ribaxamase arm (11 deaths versus 5 deaths in the placebo arm), which was driven by more cardiac deaths in the ribaxamase group.

The authors conclude that ribaxamase reduced the rate of CDI in patients receiving ceftriaxone for lower respiratory tract infection.  I think we should be skeptical.  They make this claim on the grounds of a one-sided p value of 0.045 generated from a five-patient difference in the outcome of a total sample of 413 (giving a NNT of about 40, by the way).  Well, that’s actually less of a difference than was observed in the relative mortality rates between the two arms (11 vs 5, or a six-patient difference).  Yet the authors don’t conclude that ribaxamase increases mortality in patients given ceftriaxone for lower respiratory tract infection.  It seems to me that we ought to have concluded from this study that ribaxamase both prevents CDI and kills people, or does neither of those things.  Go figure. 30885591