Predicting ventilator-associated lower respiratory tract infection outcomes using sequencing-based early microbiological response: a proof-of-concept prospective study

TL;DR

This study shows that early microbiological response measured by sequencing can predict outcomes in patients with ventilator-associated lung infections caused by Acinetobacter baumannii.

Contribution

A proof-of-concept study demonstrating that relative quantification ratios from sequencing can predict clinical outcomes in VA-LRTI patients.

Findings

An RQR ≥1.41 predicted poor outcomes with high accuracy (AUC 0.9471).

Patients with RQR <1.41 had a 100% survival rate at 28 days.

RQR correlated positively with ICU stay duration in survivors.

Abstract

Ventilator-associated lower respiratory tract infections (VA-LRTIs) cause significant mortality. This study was assigned to explore the association between early microbiological responses defined by quantitative targeted amplicon-based next-generation sequencing (QtNGS) and clinical outcomes in patients with Acinetobacter baumannii-dominant VA-LRTI.

A prospective observational study including 34 participants was conducted to assess the probability of predicting clinical outcomes using sequencing-based early microbiological response. Bronchoalveolar lavage fluids (BALFs) were collected at admission and 3 days post-treatment from these patients for QtNGS to determine the relative quantification ratio (RQR) of A. baumannii. Patients were categorized into survival (n=26) and non-survival (n=8) groups. The RQR was calculated as the quantification of A. baumannii determined by QtNGS after treatment to pretreatment. RQR significantly increased on day 4 in the non-survival group (median 5.285), and decreased in the survival group (median 0.1864). Receiver’s operation characteristic curves revealed that an RQR ≥1.41 was predictive of poor outcomes, with an area under the curve of 0.9471 (0.8759-1). The accuracy of the RQR determined by QtNGS was further evaluated by retesting the same specimen using digital droplet PCR, and the linear correlation was confirmed in the RQR calculated by two methods. The 23 patients with RQR<1.41 all survived for 28 days, whereas the survival rate for the 11 patients with RQR ≥1.41 was 27.27%. RQR was significantly and positively correlated with the length of ICU stay in survivors.

The RQR of A. baumannii detected by QtNGS correlates with the prognosis of VA-LRTI patients.

Figures