mLife (Sep 2022)

Rapid, automated, and reliable antimicrobial susceptibility test from positive blood culture by CAST‐R

  • Pengfei Zhu,
  • Lihui Ren,
  • Ying Zhu,
  • Jing Dai,
  • Huijie Liu,
  • Yuli Mao,
  • Yuandong Li,
  • Yuehui He,
  • Xiaoshan Zheng,
  • Rongze Chen,
  • Xiaoting Fu,
  • Lili Zhang,
  • Lijun Sun,
  • Yuanqi Zhu,
  • Yuetong Ji,
  • Bo Ma,
  • Yingchun Xu,
  • Jian Xu,
  • Qiwen Yang

DOI
https://doi.org/10.1002/mlf2.12019
Journal volume & issue
Vol. 1, no. 3
pp. 329 – 340

Abstract

Read online

Abstract Antimicrobial susceptibility tests (ASTs) are pivotal in combating multidrug resistant pathogens, yet they can be time‐consuming, labor‐intensive, and unstable. Using the AST of tigecycline for sepsis as the main model, here we establish an automated system of Clinical Antimicrobials Susceptibility Test Ramanometry (CAST‐R), based on D2O‐probed Raman microspectroscopy. Featuring a liquid robot for sample pretreatment and a machine learning‐based control scheme for data acquisition and quality control, the 3‐h, automated CAST‐R process accelerates AST by >10‐fold, processes 96 paralleled antibiotic‐exposure reactions, and produces high‐quality Raman spectra. The Expedited Minimal Inhibitory Concentration via Metabolic Activity is proposed as a quantitative and broadly applicable parameter for metabolism‐based AST, which shows 99% essential agreement and 93% categorical agreement with the broth microdilution method (BMD) when tested on 100 Acinetobacter baumannii isolates. Further tests on 26 clinically positive blood samples for eight antimicrobials, including tigecycline, meropenem, ceftazidime, ampicillin/sulbactam, oxacillin, clindamycin, vancomycin, and levofloxacin reveal 93% categorical agreement with BMD‐based results. The automation, speed, reliability, and general applicability of CAST‐R suggest its potential utility for guiding the clinical administration of antimicrobials.

Keywords