High-Impact PCCM article summaries: Gram-stain Guided Abx in VAP

Yoshimura et al. Effect of Gram-stain Guided Initial Antibiotic Therapy on Clinical Response in Patients with Ventilator-associated Pneumonia: The GRACE-VAP Randomized Clinical Trial. JAMA Network Open 2022

 

Question: In patients with ventilator-associated pneumonia (VAP), is an empiric antibiotic strategy guided by Gram stain of an endotracheal aspirate non-inferior to antibiotics informed by 2016 IDSA guidelines?

Why ask it: VAP is a common complication of critical illness and associated with significant morbidity and mortality. Antibiotic selection is challenging as both under- and over-treatment of VAP may cause harm. There is conflicting data on whether Gram stain of a respiratory sample can safely inform antibiotic selection in patients with VAP.

Intervention: 206 patients diagnosed with VAP based on a modified Clinical Pulmonary Score ³ 5 at 12 centers in Japan randomized to Gram stain-guided antibiotics (see Comments for details) or guideline-based treatment.

 

Results (all written as Gram stain-guided group vs guideline-based group)

  • Clinical response rate (primary outcome, see Comments for details)
    • 77% vs 72%, p < 0.001
  • 28-day mortality
    • 14% vs 18%, p=0.44
  • Median 28-day ventilator-free days
    • 22 vs 22, p=0.21
  • Median 28-day ICU-free days
    • 19 vs 20, p=0.42
  • Administration of anti-pseudomonal antibiotics
    • 70% vs 100%, p<0.001
  • Anti-MRSA antibiotics
    • 61% vs 100%, p<0.001
  • Coverage rate of initial antibiotic therapy
    • 86% vs 92%, p=0.18
  • Antibiotic escalation
    • 7% vs 1%, p=0.03

 

Conclusion: In patients with VAP, Gram-stain-guided empiric antibiotics was non-inferior to guideline-based antibiotics with respect to clinical response assessed 7 days after the end of therapy

 

Comment:

  • This was a multicenter open-label noninferiority trial with blinded endpoint assessment. Unblinding of the treating team was necessary due to the nature of the intervention
  • Notable exclusions
    • Heart failure or atelectasis diagnosis
    • Receipt of antibiotics for > 24 hrs after meeting inclusion criteria
    • Infection with COVID-19
  • Additional study procedure details (picture also pasted below)
    • Endotracheal aspirate (not a true alveolar sample) used for Gram stain and culture
    • Antibiotic selection in gram stain group
      • GPCs in chains and/or gram-positive bacilli: non-pseudomonal b-lactam
      • GPCs in clusters without GNRs: Anti-MRSA agent
      • GNRs without GPCs in clusters: Anti-pseudomonal agent
      • GPCs in clusters and GNRs: Anti-MRSA agent + anti-pseudomonal agent
      • Negative Gram stain: Anti-MRSA agent + anti-pseudomonal agent
    • Antibiotic selection in guideline-based group
      • Anti-MRSA + single anti-pseudomonal agent (authors note 80-90% of their P. aeruginosa isolates are susceptible to an anti-pseudomonal b-lactam so they did recommend empiric double coverage)
    • Antibiotics were administered for at least 7 days
  • Primary outcome = “clinical response rate” assessed 7 days after end of therapy. Clinical response defined as all of the following
    • Completion of antibiotics within 14 days
    • Improvement or lack of progression of baseline radiographic findings at the end of therapy
    • Resolution of signs and symptoms of pneumonia
    • Lack of antibiotic re-administration owing to pneumonia by the follow-up or test of cure visit
  • The study enrolled 206 out of 208 patients assessed (!!!)
  • The study met their proposed sample size of 100 patients per group (informed by an assumed clinical response rate of 68% and a 20% noninferiority margin)
  • Notable patient characteristics
    • Men: 69%
    • Trauma: 27%
    • Post-arrest: 24%
    • Stroke: 10%
    • Immunocompromised: 3%
    • Previous antibiotics: 28%
    • Sepsis: 34%
    • Median number of days from ICU admission to randomization: 4
  • Most common isolated pathogens:
    • Staphylococcus aureus: 50% (21% of which were MRSA)
    • Klebsiella spp: 17%
    • H influenza: 8%
    • Other notables
      • P. aeruginosa: 5%
      • ESBL-producing Enterobacteriaceae: 3%
      • MDR Gram-negatives: 0
    • It’s not clear to me from data tables how many samples were culture-negative or how many had polymicrobial growth. 14/206 had a negative Gram stain
  • Test characteristics of Gram stain
    • S. aureus
      • Sensitivity 84% and specificity 76%
    • GNRs
      • Sensitivity 83% and specificity 61%
  • In subgroup analysis, there was significant heterogeneity in the effect of Gram stain-guided antibiotics in patients with TBI with better clinical response rates seen in Gram stain-guided antibiotics
  • Numerically more diarrhea in guideline-based group (27 vs 38)
  • Only 4 cases of C.diff (1 in Gram stain-guided group)
  • My take
    • The main issue is generalizability to our practice
      • This study included a lot of patients with trauma and stroke, relied on endotracheal aspirates, had low rates of infection with P. aeruginosa and resistant GNRs, and did not incorporate PCR-based diagnostics.
      • We care for many more immunocompromised patients, perform early alveolar sampling in most patients with suspected VAP, and have access to multiplex PCR-based diagnostics. In our alveolar samples, a high % are culture-negative and P. aeruginosa is the most common bacterial pathogen identified (thanks to Cathy Gao and Marjorie Kang for this data)
      • In this context, I would be hesitant to discontinue anti-MRSA or anti-pseudomonal antibiotics based on Gram stain alone (I would instead wait for the PCR). The results of a Gram stain may add some secondary support to a de-escalation decision informed primarily by PCR testing.
      • Use of Gram stain to guide antibiotics may be of use to clinicians in settings without access to PCR-based testing depending on local microbiology and resistance patterns
      • How to use PCR-based diagnostics to inform antibiotic selection in severe pneumonia is a major area of research (including at NM)
    • Our group has looked at Gram stain to rule-out culture-positive HAP/VAP and found that the combination of a BAL neutrophil threshold of < 50% + a negative Gram stain had a negative predictive value of nearly 100% (Walter, AJRCCM 2018)