MSACL 2023 Abstract

Introduction
Today, approximately 1.3 million people die each year worldwide from bacterial antimicrobial resistance (AMR) [1]. AMR is of great concern in the context of bloodstream infections (BSIs) when it evolves into sepsis [2]. If MALDI-TOF has markedly shortened the delay of the pathogen identification step, AMR evaluation is still based on growth tests in the presence of antibiotics. The latter, negatively impacts the diagnosis turnaround time, thus the administration to the patient of active antimicrobial agents. With the aim of streamlining this turnaround time to less than 90min from a positive blood culture, the LC/Zeno SWATH DIA sensitivity and specificity performances were evaluated to concomitantly provide the identity and antibiotic resistance profile of etiological agents related to BSIs based on peptide surrogates.

Methods
We analyzed two collections of clinical isolates composed of 124 Staphylococcus aureus (training set) and 300 Enterobacterales (validation set) strains which were selected for the diversity of both their phenotypic and genetic resistance profiles.
Our methodology consists in a 10-min sample preparation procedure for pelleting the micro-organisms from a positive blood culture aliquot followed by a 15-min concomitant ultrasonic bacterial lysis and tryptic digestion step.
Firstly, the training set of Zeno SWATH DIA runs were conducted on a Waters M-Class UPLC system. The peptides were separated on a C18 micro-LC column (300 µm x 150 mm) at a flow rate of 6 µL/min; 11.5min gradient - 25min total chromatographic time). Secondly, the validation set runs were conducted on the Evosep One System using the manufacturer’s separation method of 11.5min (100 samples/day) and the corresponding column (150 µm x 80mm; EV1109) and flow rate (1.5µL/min). Both LC systems were connected to a 7600 ZenoTOF mass spectrometer operated in Zeno SWATH DIA mode using the OptiFlow Turbo V ion source.
All LC/Zeno SWATH DIA data were processed with DIA-NN (1.8) software in library-free mode using a FASTA file containing the genus and species-specific ribosomal protein sequences as well as resistance mechanism-related protein sequences. The generated library was then used to analyze the raw data.

Results
A comprehensive quality by design approach has been selected to assess the key LC-SWATH DIA parameters having the most influence on the overall performance detection of peptide surrogates used for the identification of the pathogen and the detection of antibiotic resistance-related proteins.
Initially, a screening design allowed us to determine which of the six parameters evaluated (gradient length, TOF accumulation time, number of SWATH windows, fixed/variable windows, MS m/z range, and MS/MS m/z range) had the highest impact on the number of proteotypic peptides detected.
Subsequently, the two most critical parameters (number of SWATH windows and MS m/z range) have been thoroughly optimized using a response surface methodology. The other non-critical parameters were set at the level that produced the best response.
Regarding the results, 100% of the positive blood cultures were correctly identified in both the training set (124/124) and the validation set (300/300).
This study allowed the detection of some of the most common protein effectors involved in bacterial resistance mechanisms such as methicillin-resistance inducing PBP2a and PBP2c proteins in Staphylococcus aureus as well as different types of β-lactamases in Enterobacterales (OXY; OXA-1; OXA-9; CMY; DHA; ESC; ACC; PER; VEB; CTX-M; TEM; SHV; KPC; GES; OXA-23; OXA-48-like; VIM; NDM to mention a few).
Taking a step further, the sensitivity and specificity performances of the LC/Zeno SWATH DIA method have been evaluated. Firstly, for the training set, this workflow enabled us to reach 100% (5/5) specificity (SP) and 100% (5/5) sensitivity (SE) for PBP2c and 99% (98/99) SE and 100% (99/99) SP for PBP2a when comparing the proteomics results to the expected genomic data. Secondly, for the validation set of Enterobacterales, a widely diverse panel of the 26 most prevalent β-lactamases were detected with 98% (297/300) SE and 100% (300/300) SP when compared to both genomic and phenotypic resistance profiles for each strain.
Moreover, time is truly of the essence when it comes to diagnosis and so is the actual throughput in a clinical setting. In this context, we decided to also assess the performances of Zeno SWATH DIA in combination with very short liquid chromatography gradients, namely the Evosep 3.2min gradient (300 samples/day method) on 15 clinical strains out of the validation set.
Successful identification of the positive blood cultures was achieved in 100% of the cases (15/15). We were able to reach an overall SP of 84% and an overall SE of 100% in contrast to the expected genomic data. When comparing these results to those obtained with the 100 samples/day method for the same samples, the latter performed better with an overall SE of 95% and an overall SP of 100%, highlighting the limits of very fast gradients for this application.

Conclusion
We have demonstrated through this study that LC/Zeno SWATH DIA might be an interesting alternative to current standard of care methods in a clinical setting, providing in vitro diagnostic (IVD) classification, as it allows concomitant identification and AMR profiling of pathogens directly from positive blood cultures with a short turnaround time of 90min.

References
[1] Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis. Lancet 2022; 399: 629-55
[2] Kumar, A. et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Crit. Care Med. 34, 1589–1596 (2006)