Antimicrobial Agents: September 2018

The MERINO trial provides the best evidence yet that carbapenems are preferable to piperacillin-tazobactam for the treatment of serious infections due to ESBL-producing organisms. This has been an ongoing point of debate in the world of gram-negative bacilli. Many variants of the extended-spectrum beta-lactamase (ESBL) appear sensitive to piperacillin-tazobactam on routine susceptibility testing despite exhibiting an inoculum effect (i.e. decreased susceptibility when the ‘bug to drug’ ratio is unfavorable) in vitro. Retrospective studies have reached different conclusions about whether carbapenems provide better clinical outcomes than pip-tazo for ESBL infections, and a randomized controlled trial showed no difference in outcomes between pip-tazo and ertapenem for the treatment of ESBL UTI (see 28592240). On the other hand, beta-lactams tend to concentrate in urine, so should we extrapolate UTI outcomes to the treatment of more serious infections with ESBL-producing organisms, like bacteremia? Maybe not.

This multicenter, international trial randomized 379 adults with bacteremia due to E. coli or K. pneumoniae resistant to ceftriaxone but susceptible to piperacillin-tazobactam to receive either pip-tazo (4.5g every six hours) or meropenem (1g every eight hours). Patients with polymicrobial bacteremia and those being treated without expectation of cure or short-term survival were excluded. Randomization occurred within 72 hours of blood cultures each drug was given for 4 to 14 days, determined by the treating physician. The primary outcome of interest was mortality at 30 days. Patients in the meropenem arm had higher APACHE II scores (21 vs 18), more diabetes (41% vs 31%) and urinary sources of bacteremia (67% vs 55%), and were less likely to have immune compromise (21% vs 27%); otherwise, the groups were similar.

Mortality at 30 days was higher in piperacillin-tazobactam arm than the meropenem arm (12% vs 4%), a finding that persisted in multivariate analysis (aOR 3.41). Interestingly, all of the deaths were deemed related to the underlying comorbidity (e.g. metastatic cancer) rather than the infection, and most occurred after 14 days (see supplementary materials). The MERINO trial authors actually got on twitter to talk about this, arguing that sepsis has long-lasting negative effects on homeostasis that can tip someone who is already really sick from their underlying disease over the edge. I thought this was plausible; your mileage may vary. None of the secondary outcomes (clinical and microbiologic resolution of infection by day 4, rates of relapsed infection, rate of C. difficile colitis, and rates of subsequent CRE colonization or infection) reached statistical significance, although the results trended in favor of meropenem in every case. The rate of adverse events was similar between groups.

Despite its limitations, the MERINO trial is the highest quality evidence available addressing whether piperacillin-tazobactam is an acceptable alternative to carbapenems for bacteremic ESBL infections. Please note that the study used the maximum dosing of pip-tazo, so the many US hospitals using more conservative dosing (e.g. 3.375g every six hours or 4.5g every eight hours), would expect an even greater disparity in outcomes. 30208454

How safe and effective is outpatient parenteral antimicrobial therapy (OPAT) for injection drug users? Throughout residency and fellowship, several attendings told me that OPAT is not appropriate for injection drug users, with the reasoning being that these patients would access their PICCs to shoot up their drugs and in doing so cause PICC-related complications (e.g. septic thrombophlebitis, endocarditis, line occlusion). However, I could never find much data showing that treatment outcomes in injection drug users were any worse than in nonusers, or any better with inpatient parenteral therapy than with OPAT.

This month’s OFID has a nice review of this controversial topic. The authors identified ten studies, including two prospective observational studies without control groups and a total 800 patients managed with OPAT. The studies included patients who were either actively using injection drugs or had done so in the past 1-12 months; the majority of patients were being treated for osteomyelitis, endocarditis, or skin and soft tissue infections. Most patients were discharged home and received antibiotics either at an infusion center or from a nurse making house calls.

OPAT completion rates ranged from 72% to 100%; durations of therapy ranged 18 to 42 days. The mortality rate was 0% in seven studies and 2%, 5%, and 10% in the remaining three; of note, in the study reporting 10% mortality, the mortality rate was no better in patients not using injection drugs. The studies that included a comparison group noted injection drug users were more likely to have an episode of nonadherence (6% vs 1%) but had no differences in rates of treatment failure, relapsed infection, hospital readmission, or mortality. PICC-related adverse events in drug users occurred in 3-9% of episodes, or at a rate of 1-4 complications per 1000 catheter days; of the three studies reporting rates of PICC misuse, one reported an incidence of 2% and the other two reported an incidence of 0%. Two studies reported cost analyses and OPAT for injection drug users was cost-saving in both cases, to the tune of $12,000-$25,000 per episode.

Bottom line? While OPAT in injection drug users has been poorly studied, the available evidence suggests that treatment outcomes and adverse event rates are comparable to those in nonusers, that PICC misuse is uncommon, and that the practice is likely cost-saving. 30211247

Speaking of OPAT, this month’s CID has a study comparing outcomes of OPAT with and without involvement of an ID specialist. The outcomes of interest were rates of readmission to the ED or hospital and total healthcare costs associated with OPAT. Using a national insurance claims database of adults treated with OPAT over a one-year period, the authors created a propensity score-matched cohort (n=8,200). Relative to patients who received OPAT without ID consultation, patients who received OPAT under the supervision of an ID specialist were less likely to end up in the ED (OR 0.45) or hospital (OR 0.66) in the month after their initial discharge, and their mean cost of care was $1,488 lower. It’s nice to have data proving the value ID specialists bring to our patient’s care. 30247512

Baloxavir marboxil, a new antiviral for influenza, is as clinically effective as osteltamivir (and superior with regard to a few endpoints you don’t care about). Baloxavir is a new class of influenza antiviral targeting the viral polymerase. This study reports the results of a phase 3 trial, CAPSTONE-1, which randomized US and Japanese children and adults with uncomplicated influenza to receive baloxavir, osteltamivir, or placebo.

Adults received one dose of baloxavir with 5 days of placebo, 5 days of oseltamivir with one dose of placebo, or a placebo substitute for both drugs in a 2:2:1 fashion; children were randomized 1:1 to the baloxavir and osteltamivir arms only. Participants had less than 48 hours of fever plus at least one respiratory symptom; pregnant and hospitalized patients were excluded. No symptomatic therapy other than acetaminophen and no other antimicrobials (except antibiotics for secondary bacterial pneumonia) were allowed. The primary outcome was reduction in patient-reported symptom severity scores for seven symptoms; other endpoints included changes in viral RNA titers and duration of viral RNA detection in nasopharygeal swabs.

1064 patients were included in the intention-to-treat analysis. Baseline characteristics were similar between groups. The median time to alleviation of clinical symptoms was shorter in patients who received baloxavir versus placebo (median difference 23 hours, p<0.001), and this was more pronounced in the patients starting baloxavir within 24 hours of symptom onset than in those starting later (mean reduction in symptom duration of 33 versus 13 hours). Median time to defervescence and median time to patient-defined “return to usual health” were also shorter with baloxavir versus placebo.

While the results section studiously avoids any mention of the relative clinical efficacy of baloxavir verus oseltamivir, a Kaplan-Meier in the supplementary materials (Figure S4) shows that outcomes were all but identical. What does get mentioned in the results section are the virologic endpoints, because that’s where baloxavir outperformed oseltamivir, with larger drops in viral RNA titer 24 hours into therapy and shorter median durations of viral detection (24 versus 72 hours, p<0.001). Baloxavir resistance mutations were detected in 9% of patents given the drug and were associated with having infectious virus on day 5 of therapy. Drug-related adverse events were more common with oseltamivir (8%) than baloxavir or placebo (4% each).

In summary, baloxavir is the first in a novel class of influenza antivirals, alleviates symptoms about as well as oseltamivir, and is better tolerated and superior with regards to certain virologic endpoints. Based on the virologic data, it’s plausible that baloxavir could render patients noninfectious sooner than oseltamivir, but that needs a clinical study to determine. Baloxavir was granted Priority Review status by the FDA in June, and an approval decision is expected before the end of the year. 30184455

Those aminoglycoside-impregnated surgical beads might actually be worsening infection outcomes – at least for early prosthetic joint infections. Some surgical subspecialists just love stuffing patients full of aminoglycoside-impregnated beads. In theory this makes sense for a deep surgical site infection – these drugs have greater antibacterial effect at higher concentrations, and the beads ought to let us achieve much higher drug levels locally than a patient could tolerate systemically. And yet, this practice doesn’t have much in the literature (and, to my knowledge, no RTC data) to support it; it is, for the most part, “just a thing we do.”

This month’s JAC has a retrospective cohort study investigating the use of gentamicin-impregnated beads and sponges in the treatment of early prosthetic joint infection (PJI) managed with debridement and implant retention (DAIR). The outcome of interest was early treatment failure, defined as PJI-related death, need for implant removal or repeat DAIR, or need for suppressive antibiotic therapy due to persistent signs of infection, all within 60 days of the initial surgery. Patients treated at three Dutch hospitals from 2006-2016 were included (n=386), of whom 76% received local gentamicin in the form of beads or a sponge. Because of relevant differences between the local gentamicin and control groups (e.g. greater rates of prostheses placed for fracture, longer duration of symptoms, greater incidence of pus, more S. aureus and polymicrobial infections, etc), the authors performed 1:1 propensity score matching to come up with local gentamicin and control groups that were similar (n=77 for each).

Before propensity matching, use of local gentamicin was clearly associated with early failure (43% vs 24%, p = 0.001); after matching, the association was no longer statistically significant in univariate analysis but still trended toward more early failure with local gentamicin use (40% vs 26%, p = 0.06), and in multivariate analysis use of local gentamicin again associated with early failure (OR 1.97; 95% CI 1.12-3.48). In a subgroup analysis examining the technique of local gentamicin administration, only local gentamicin given by antibiotics beads later removed by lavage (i.e. in a planned second surgery performed in the absence of evidence of persistent infection) were not associated with greater rates of failure (9.7% vs 10.5%, p = 0.9).

This is a retrospective study, and propensity score matching is by no means a guaranteed fix for selection bias. However, the conclusion is clinically meaningful: a doubled risk of treatment failure for PJI is nothing to sneeze at. Hopefully this paper sets the stage for the randomized clinical trial I suspect will be needed to actually change surgical practice. 30189006