Antimicrobial agents: July 2018

A dosing calculator accurately provides optimal drug dosing for antipseudomonal beta-lactams in the ICU. Last month I mentioned that optimal dosing of antibiotics in ICU patients is tricky on account of multiple factors; honestly, if you’re prescribing antibiotics in the ICU and aren’t a PharmD, you ought to consult with one. This group performed a prospective study validating an antibiotic dosing software tool (ID-ODS™, developed by Optimum Dosing Strategies, a company with which one co-author is affiliated) for cefepime, piperacillin-tazobactam, and meropenem in 49 critically ill patients. The calculated doses achieved drug level target attainment in patients 98% of the time, and recommended doses differed from those in the packages insert 22% of the time (30% for cefepime, 16% for pip-taz, and 25% for meropenem).

The authors note that, had they applied a more aggressive goal of achieving drug serum levels >4-fold above MIC for 100% of the treatment duration (versus the standard goal of 50% time above MIC), target attainment would have dropped to 65% of patients, largely driven by failures to reach target attainment in the pip-taz group. I wonder if greater PK/PD variability might be another contributing factor to the oft-observed inferiority of pip-tazo versus carbapenems for serious infections due to ESBL-producing organisms. 29967022

While we’re on the subject, another paper from this month’s AAC looked at antipseudomonal beta-lactam levels in septic patients with increased renal clearance and also found more variability with pip-taz. Quoting that paper: “Measured creatinine clearance adequately explained changes in drug concentrations in population pharmacokinetic models for cefepime, ceftazidime, and meropenem, but not for piperacillin.” 29987138 

Do we need to routinely measure therapeutic levels of isavuconazole? Probably not. Each of the first three mold-active azoles (itraconazole, voriconazole, and posaconazole) were found to need therapeutic drug level monitoring for optimal efficacy after their approval – which you’d expect, PK/PD variation being more significant in real-world use than in the healthy volunteers and defined patient populations that make up clinical trials. But what about isavuconazole? This research group (which included three co-authors affiliated with Astellas, isavuconazole’s producer), examined drug levels in plasma samples from 283 patients treated with isavuconazole in the course of routine care. The authors compared the variation in serum levels among these samples with the variations observed in the phase 3 trials that resulted in isavuconazole’s approval.

Mean serum levels in the real-use cases were modestly lower than those from the trials (2.98 vs 3.30 ug/mL; p=0.014). More reassuringly, only 10% of patients from either group had isavuconazole concentrations below ~1ug/ml (the threshold of efficacy for the other mold-active azoles), which is better than any other azole formulation. And in fact, no level of isavuconazole has been shown to clearly correlate with treatment efficacy anyway. The authors conclude that isavuconazole level monitoring may be reasonable to consider in patients with unexplained hepatotoxicity or treatment failure, or who have a reason to have low drugs levels (e.g. obesity, medication noncompliance), but probably does not need to be done routinely. 29735569

Drug level monitoring of Polymixin B, on the other hand, might be valuable. The authors of this study used a previously established PK/PD model for polymyxin B and performed Monte Carlo simulations to determine what proportion of patients with standard dosing would end up with drug exposures (ie 24hr AUCs) that fell into a therapeutic range. They defined the upper bound of their therapeutic range as a drug exposure threshold likely to produce nephrotoxicity (selected via meta-analysis of prior data) and defined the lower bound as drug exposure required for efficacy based on infection studies in mice. The authors’ simulation predicted that standard dosing would result in only 71% of cases falling into this therapeutic window, but that this proportion could be improved to >95% if the dosing was adjusted based on a single 24hr drug level. Of the drug exposures falling outside the therapeutic window, about two-thirds of cases (21% of the whole set) were too high and the one third too low.

So, Polymyxin B drug level monitoring and dose adjustment would primarily reduce the nephrotoxicity associated with that drug, though it might also improve efficacy. Of course, this conclusion comes with clear caveats (i.e. these are the conclusions of a modeling study based on meta-analysis and mouse data). 29760144 

Nelfinavir – an old and long-discarded antiretroviral – might have a second life as a drug for Leishmaniasis. I’m amazed that anyone thought to try treating Leishmania with an ancient protease inhibitor, but someone did, and lo and behold it has potent activity against promastigotes and amastigotes in vitro. The authors demonstrated that nelfinavir was synergistic with both amphotericin B and miltefosine for multiple strains of L. infantum. When given in a mouse model of visceral leishmaniasis, nelfinavir significantly reduced the parasite burdens in liver and spleen and was synergistic with other agents. One thing I found odd is that the authors conclude that nelfinavir might be helpful “as a treatment option in the care of patients with visceral leishmaniasis also infected with HIV.” Why not study it for leishmaniasis treatment generally? I’ve never used nelfinavir, so I don’t have an appreciation for its toxicities, but I can’t imagine they’re much worse than those of the other drugs we use for leishmaniasis. 29969693

Is it time to declare ampicillin plus ceftriaxone the preferred regimen for Enterococcus faecalis endocarditis? This retrospective cohort study compared adults with E. faecalis endocarditis treated at two Mayo clinic campuses who received ampicillin plus either gentamicin (AG) or ceftriaxone (AC). Out of 85 patients (67 in the AG group and 18 in the AC group), 1-year mortality rates were identical despite the patients in the AC arm having more baseline renal impairment and higher Charlson Comorbidity Indices. As one might expect, patients in the AG arm experienced more nephrotoxicity (mean change in creatinine at end of therapy +0.4 versus -0.2 in the AG vs AC arms).

This is not the first study to show equivalent outcomes and fewer significant adverse events with ampicillin plus ceftriaxone rather than ampicillin plus an aminoglycoside (see also the original 2013 Spanish cohort study on this subject: 23392394). Sure, there’s no randomized controlled trial data showing that AC is noninferior to AG, but that’s true the other way around, too. If the best available data says outcomes are equivalent and ampicillin-ceftriaxone is safer, shouldn’t the burden of proof be on the folks who want to keep using aminoglycosides? What do you think? 29969596