https://orcid.org/0000-0002-1659-4902
Abstract
Recommended antimicrobial treatment durations for ventilator-associated pneumonia (VAP) caused by Pseudomonas aeruginosa have evolved over the past few decades. In this Viewpoint, we provide a narrative review of landmark trials investigating antimicrobial treatment durations for VAP caused by P. aeruginosa, and appraise iterations of expert consensus guidelines based on these data. We highlight strengths and weaknesses of existing data on this topic and provide recommendations for future avenues of study.
(See the Editorial Commentary by Metersky et al on pages 750–2.)
In 1995, a consensus statement from the American Thoracic Society regarding duration of therapy for hospital acquired pneumonia stated “microorganisms such as Pseudomonas aeruginosa or Acinetobacter spp. have been associated with higher rates of treatment failure, relapse and death” and “while carefully controlled studies documenting duration of therapy have not been reported, in these settings antibiotics should be continued for a minimum of 14–21 days to reduce the chance of relapse” [1]. Nearly 30 years later, although consensus guideline recommendations on the optimal duration of therapy for P. aeruginosa ventilator associated pneumonia (PsA-VAP) have changed multiple times, data to inform practice remain challenging to translate to bedside care. In this viewpoint we review the evolving recommendations for P. aeruginosa VAP, critically appraise the evidence supporting these recommendations, and offer suggestions for future research avenues to address this clinical question.
THE PNEUMA TRIAL AND THE 2004 IDSA/ATS HAP/VAP/HCAP GUIDELINES
The first randomized, controlled trial evaluating the optimal duration of therapy for VAP—the PneumA trial—occurred in 51 French intensive care units from 1999 to 2002 [2]. Patients ≥18 years of age were randomized on day 3 (once it was confirmed the empiric antibiotic regimen displayed in vitro activity against the causative pathogen) to receive 8 or 15 days of antibiotic therapy for VAP. Although investigators in the trial were blinded until day 8, those randomized to the longer duration received an additional week of treatment in an open-label fashion. The primary outcome was 28-day mortality and secondary outcomes, also assessed through 28 days, included pulmonary “recurrence” and antibiotic free days. In this trial, “recurrence” was defined as either relapsing pneumonia with the original index pathogen or pulmonary superinfection with a pathogen not present in respiratory cultures at the time of trial enrollment.
Mortality within 28 days occurred in 37/197 (18.8%) of patients receiving 8 days of therapy compared to 35/204 (17.2%) receiving 15 days (difference 1.6%; 90% confidence interval [CI]: −3.7 to 6.9). Rates of pulmonary recurrence by day 28 were also similar in the 2 arms (28.9% vs 26.0%, respectively; difference 2.9% [90% CI: −3.2 to 9.1]). The mean number of antibiotic free days over the 28-day study period was 4.4 days higher in the shorter duration arm. Among patients who developed pulmonary recurrence, multidrug resistant pathogens were encountered less frequently in those treated with shorter course therapy (42.1 vs 62.0% of pulmonary recurrences, P = .04).
Although these results supported the use of shorter course therapy in the overall study cohort, the findings were murkier for the subgroup of patients infected with non-lactose fermenting gram-negative bacilli (127 patients in the trial, with 136 total non-lactose fermenting organisms isolated—120 [88%] of which were P. aeruginosa). Although there was no significant difference in mortality in this subgroup, patients receiving 8 days of therapy had a higher rate of pulmonary recurrence (40.6% vs 25.4%; difference 15.2% [90% CI: 3.9 to 26.6]). This difference was primarily driven by relapse due to the original pathogen (32.8% vs 19.0%; difference 13.8% [90% CI: 7.8–19.7]). The findings from this study informed updated HAP/VAP treatment guidelines published in 2005, which recommended that “efforts should be made to shorten the duration of therapy from the traditional 14 to 21 days to periods as short as 7 days, provided that the etiologic pathogen is not [Pseudomonas] aeruginosa” [3]. Based on this trial and the subsequent guideline recommendation, standard practice for VAP duration of therapy over the next decade was to treat for 7–8 days unless the offending pathogen was P. aeruginosa, where 14 days of therapy was often given.
2016: A NEW META-ANALYSIS AND UPDATED GUIDELINE
In 2016, a new guideline panel from the American Thoracic Society and the Infectious Diseases Society of America provided updated recommendations for VAP duration of therapy [4]. These guidelines recommended a 7-day course ;of VAP therapy, regardless of causative pathogen. To support their recommendation, the expert consensus panel performed a meta-analysis of studies comparing short course (7–8 days) versus long course (10–15 days) therapy and found no differences in mortality or VAP recurrence. Furthermore, they performed a meta-analysis on the subpopulations of these trials with non-lactose fermenting gram-negative bacilli (NF-GNB) as causative pathogens and observed no differences for either pneumonia recurrence (odds ratio [OR], 1.42; 95% CI: .66–3.04) or mortality (OR, 0.94; 95% CI: .56–1.59) with shorter course therapy.
Based on these findings, the panel stated that for VAP caused by non-fermenting pathogens, “a different recommendation was not indicated because, even if there is a small increased recurrence rate, mortality and clinical cure do not appear to be affected” [4]. They further highlighted the following limitations of the existing trial data suggesting increased recurrence with shorter courses of therapy in PsA-VAP: (1) that data were aggregated exclusively from post hoc subgroup analyses; (2) that there was a potential bias towards long-course therapy given that recurrence was followed until 28 days post enrollment leading to a shorter time course for post-therapy recurrences to occur in patients receiving longer courses of antibiotics; (3) that studies used broad definitions for recurrence including superinfections, both pulmonary and extrapulmonary, that are not necessarily relevant to the question at hand; and (4) that trial recurrences were often defined microbiologically, potentially equivocating residual airway colonization with recurrent infection (though, notably, follow-up respiratory cultures in the PneumA study were collected bronchoscopically, which is more markedly more specific for VAP than cultures collected via endotracheal aspiration) [5, 6].
Although these recommendations and the supporting rationale are reasonable, there are important inaccuracies and limitations to the analyses performed and the data underpinning the recommendations that merit attention. Specifically, the panel's meta-analysis of recurrence in NF-GNB has been widely cited as evidence that recurrent pneumonia attributable to shortened courses of antibiotic therapy is uncommon or artifactual. Yet a closer investigation of this meta-analysis reveals significant concerns with this conclusion.
The meta-analysis includes data from 4 trials with a total of 265 patients (147 short course therapy, 118 long course) with VAP due to NF-GNB. In total, 127 (48%) of the patients came from the previously discussed PneumA trial, in which receipt of short course therapy was associated with an odds ratio of 2.01 for recurrent pneumonia. The second trial by Medina and colleagues (49 patients, 18%) is based on a 2007 abstract presented at ICAAC that remains unpublished 15 years later [7]. Although we cannot comment on the data due to their unavailability and question the inclusion of unpublished works in a meta-analysis, the results were directionally consistent with those of the PneumA trial (short course was associated with a 2.72-fold increased risk for recurrence). The third trial is a small pilot study by Fekih Hassen et al (30 patients, 11% of meta-analysis), which minimally impacts the analysis given the sample size [8]. In reviewing the data, the denominator listed in the meta-analysis for this trial (30 patients) is the total number of patients enrolled in the trial. However, this trial was not exclusively in patients infected with NF-GNB, and recurrence rates are not reported as a function of index pathogen. Thus, these data are of uncertain relevance to the research question at hand.
The fourth and final study included by the panel in the meta-analysis was a prospective, open-label trial by Capellier et al of 225 patients with early-onset VAP randomized to 8 versus 15 days of antibiotic treatment [9]. This study contributed significantly to the panel's meta-analysis (59 patients, 22%); however, no patients in this trial had VAP due to non-fermenting gram-negative bacilli at the time of enrollment. Of the 143 baseline gram-negative pathogens identified, 30% were Enterobacterales and 66% were Haemophilus spp. For this reason, the findings of this study are immaterial to an analysis of NF-GNB VAP treatment durations, and they should not have been included in the panel's meta-analysis.
Unfortunately, the misapplication of data from this trial goes beyond its inclusion in the meta-analysis. In this trial, there were 30 episodes of secondary pneumonia among the 116 patients randomized to 8 days of antibiotic therapy versus 15 episodes of secondary pneumonia among the 109 patients randomized to 15 days of antibiotic therapy (25.8% vs 13.8%, P = .03). If the panel chose to classify these secondary pulmonary infections as recurrences, the methodologically consistent approach would be to evaluate this outcome in the subgroup of patients with NF-GNB isolated from respiratory cultures at the time of randomization. As no such patients existed in this trial, the panel instead evaluated recurrence rates by calculating the number of patients in each treatment arm who developed a secondary pneumonia caused by NF-GNB divided by the total number of pathogens isolated from all secondary pulmonary infections in each treatment arm. In this trial, among those who received 15 days of antibiotics, there were 15 secondary pulmonary infections with 17 total pathogens, 8 of which were NF-GNB, and among those who received 8 days of antibiotics, there were 30 secondary infections with a total of 42 pathogens, 13 of which were NF-GNB. These numbers (8/17 [47%] vs 13/42 [25%]) align precisely with the numbers presented in the panel's meta-analysis (Figure 1). Therefore, even though treatment with short course therapy was associated with twice as many secondary pulmonary infections relative to longer courses of therapy, the methodologically flawed manner in which data from this trial were extracted and presented makes it appear that short course antibiotic therapy was associated with less recurrent pneumonia.
Forest plot from 2016 IDSA/ATS HAP/VAP guidelines summarizing findings of meta-analysis of NF-GNB VAP and pneumonia recurrence as a function of short versus prolonged course therapy. Abbreviations: ATS, American Thoracic Society; CI, confidence interval; HAP, hospital acquired pneumonia; IDSA, Infectious Diseases Society of America; NF-GNB, non-lactose fermenting gram-negative bacilli; VAP, ventilator-associated pneumonia.
Both the inappropriate inclusion of this study (because no patients in the study had VAP caused by NF-GNB at the time of treatment randomization) and the misapplication of its data (by using an incorrect definition and data source for recurrent pneumonia) in the panel's meta-analysis lead to the erroneous conclusion that there were no differences in rates of pneumonia recurrence as a function of antimicrobial treatment duration in NF-GNB VAP. When this study is excluded from the meta-analysis, short course therapy is associated with significantly increased rates of pneumonia recurrence, with an odds ratio of 2.07 (95% CI: 1.11–3.83) and a Cochrane's Q test showing no heterogeneity (P = .79), consistent with findings from the 2015 Cochrane meta-analysis (Figure 2). Therefore, although the panel's rationale for exercising skepticism on the clinical importance of pneumonia recurrence observed in trial data remains valid, trial data available at the time did suggest an imbalance in recurrence rates between treatment durations, and further evidence was clearly needed to inform practice.
Revised forest plot, using identical data from the 2016 IDSA/ATS HAP/VAP meta-analysis, with Capellier 2012 study removed. Abbreviations: ATS, American Thoracic Society; IDSA, Infectious Diseases Society of America; HAP, hospital acquired pneumonia; VAP, ventilator-associated pneumonia.
THE IMPACT OF THE DURATION OF ANTIBIOTICS ON CLINICAL EVENTS IN PATIENTS WITH PSEUDOMONAS AERUGINOSA VENTILATOR-ASSOCIATED PNEUMONIA (IDIAPASON) TRIAL
The question of PsA-VAP treatment durations took on renewed vigor earlier this year with the publication of the iDIAPASON trial [10]. Given the uncertainties surrounding the subgroup analysis from the PneumA trial, this study was designed to provide a definitive answer to the question of optimal treatment duration for P. aeruginosa VAP. The trial was a randomized controlled trial—with methodology similar to the PneumA trial from decades earlier—comparing 8- and 15-day treatment durations with enrollment limited to patients with P. aeruginosa VAP. This study planned to enroll 600 patients to be able to demonstrate non-inferiority of short course therapy. Due to slow enrollment, the trial was unfortunately terminated early after enrolling only 186 patients in the intent to treat analysis.
The primary outcome in iDIAPASON was a composite of mortality and VAP recurrence occurring during the ICU stay until day 90. Recurrence was defined as clinical suspicion of VAP based on clinical, laboratory, and radiographic criteria with a positive quantitative culture for P. aeruginosa occurring at least 48 hours after stopping trial treatment. In the ITT population, the primary outcome occurred in 35.2% of patients receiving 8 days and 25.5% receiving 15 days of therapy. Non-inferiority was not demonstrated as the upper bound of the 90% CI was greater than the predefined criteria of 10% (difference 9.7%, 90% CI: −1.9 to 21.2%). In this trial, 32/186 (17%) of the patients were inappropriately included, primarily because they did not meet study diagnostic criteria for VAP or because they did not have P. aeruginosa isolated from respiratory cultures. When those subjects were removed for per-protocol analysis, results were similar (difference in the composite of pneumonia mortality and recurrence was 12.8%, 90% CI: .4–25.6%).
Despite randomization, there were notable differences in some clinically relevant markers of illness at baseline between the two arms. Therefore, post hoc adjusted analyses were performed on these 2 populations and the differences between arms became slightly more pronounced (ITT 12.5%, 90% CI: 1.3–23.6; PP 16.3% 90% CI: 3.9–28.8%). When assessed individually 90-day survival was 81.4% in the 15-day group and 75.6% in the 8-day group (HR = 1.37, 90% CI: 0.81–2.33) and recurrence rates were 9.2% and 17% in the two arms (difference 7.9%, 90% CI: −0.5 to 16.8%).
Although iDIAPASON was limited by its sample size and open-label design, these data are the most informative to date on the question of optimal treatment duration of P. aeruginosa VAP. In limiting the trial to P. aeruginosa, the investigators overcame the subgroup limitation of previous trials. By assessing the outcome out to 90 days, the investigators minimized the impact of differential time of follow-up due to one arm receiving a longer duration of therapy. Furthermore, the investigators limited their recurrence definition to relapsing infection due to the study pathogen occurring ≥48 hours after completing the initial course of therapy and adjudicated all potential cases of recurrent pneumonia in a blinded fashion, making this outcome a more clinically relevant and applicable finding. Despite its limitations, this study represents a significant advancement over previous literature on the topic and demonstrates concerning signals regarding the association between short course therapy and adverse outcomes in PsA-VAP.
Although both the PneumA and iDIAPASON trials have evaluated antibiotic treatment durations in PsA-VAP, it is also important for provides to recognize that both trials have limited generalizability to patient care. For example, the PneumA trial excluded patients with a SAPS II score >65, immunocompromised patients, patients without full code status, and those who received empiric antibiotics that were not active against the responsible VAP pathogen. Only one third of patients in the PneumA trial had circulatory shock at the time of enrollment, and fewer than a quarter had acute respiratory distress syndrome, the only two validated predictors for VAP recurrence in systematic reviews to date [11]. The iDIAPASON study had similar exclusion criteria and further excluded patients with prior P. aeruginosa isolated from respiratory cultures, a patient population that is highly enriched in multi-drug resistant isolates, which may necessitate a different treatment approach.
THE PATH FORWARD
As this viewpoint highlights, despite important work conducted on PsA-VAP to date, considerable ambiguity remains regarding the optimal duration of therapy. Available evidence suffers from methodologic limitations and limited generalizability. Limitations notwithstanding, present evidence suggests an increased risk of recurrence with short course therapy, although the clinical importance of this finding remains uncertain. Whether or not the current state of evidence should warrant a blanket recommendation of an extra week of therapy for P. aeruginosa VAP is debatable, but appreciation for the increased risk of recurrence with short course needs to be acknowledged and considered in clinical decision making.
Ultimately, more research is needed, and we would argue that a clearer clinical research question would be better suited to do this. Guidelines already recommend individualized therapy, including prolongation of treatment durations, for patients with a delayed or unfavorable clinical response to treatment [4]. While further study of these patient phenotypes is of interest, an unresolved question is whether a second week of therapy has any benefit in PsA-VAP for patients with a favorable initial clinical response to treatment. The current literature base does not address this as randomization in trials occurs at day 3, and treatment courses continue/stop at day 8 regardless of patient status. Therefore, we propose that future duration of therapy studies should enroll patients at day 7 of treatment (rather than at the outset of infection) and stratify enrollment by clinical response to treatment (eg, randomize patients with a favorable 7-day response to stopping therapy versus receiving an additional 7 days of treatment). As there is a concern for increased recurrence with only 1 week of therapy, double blinding should be considered a critical component of future studies to minimize provider biases that could influence both recurrent pneumonia diagnosis and treatment in the setting of clinical uncertainty.
As in iDIAPASON, the definition of recurrence should be limited to a relapsed infection, caused by the initial VAP pathogen, after cessation of study drug for a period of time. Lastly, future trials of PsA-VAP should reflect the complexity and breadth of modern clinical practice, preferentially enrolling patients at risk for relapse—those with severe critical illness, immunocompromising conditions, and drug-resistant pathogens. These steps would provide important answers to what remains an unsettled research question.
Notes
Financial support. This work received no funding from any agency in the public, commercial, or non-for-profit sectors.
References
Author notes
Potential conflicts of interest. O. R. A. reports grants or contracts from Charles River Laboratories; consulting fees from Shionogi, Charles River Laboratories. K. S. K. reports consulting fees from Merck, Qpex, Allecra, Utility Pharmaceuticals, Melinta, GSK, CARB-X, ARLG; stock or stock options from Merck. E. K. M. reports consulting fees for Advisory Boards from AbbVie, Merck, Shionogi, Cidara, Ferring, Summit, La Jolla, Venatorx, Entasis; participation on DMBD for A Phase I, Randomized, Double-blind, Placebo-controlled, Dose Escalation Study to Evaluate the Safety, Tolerability, and Pharmacokinetics of Intravenous Ertapenem in Combination with Zidebactam (WCK 6777) in Healthy Adult Subjects; role as Treasurer of Society of Infectious Diseases Pharmacists. J. M. P. reports grants or contracts from Merck; consulting fees from Merck, Shionogi, GlaxoSmithKline, Qpex, Abbive, Melinta.
All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
