Identification of Mycobacterium abscessus Subspecies by MALDI-TOF Mass Spectrometry and Machine Learning.

Mycobacterium abscessus is one of the most common and pathogenic nontuberculous mycobacteria (NTM) isolated in clinical laboratories. It consists of three subspecies: M. abscessus subsp. abscessus, M. abscessus subsp. bolletii, and M. abscessus subsp. massiliense. Due to their different antibiotic susceptibility pattern, a rapid and accurate identification method is necessary for their differentiation. Although matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) has proven useful for NTM identification, the differentiation of M. abscessus subspecies is challenging. In this study, a collection of 325 clinical isolates of M. abscessus was used for MALDI-TOF MS analysis and for the development of machine learning predictive models based on MALDI-TOF MS protein spectra. Overall, using a random forest model with several confidence criteria (samples by triplicate and similarity values >60%), a total of 96.5% of isolates were correctly identified at the subspecies level. Moreover, an improved model with Spanish isolates was able to identify 88.9% of strains collected in other countries. In addition, differences in culture media, colony morphology, and geographic origin of the strains were evaluated, showing that the latter had an impact on the protein spectra. Finally, after studying all protein peaks previously reported for this species, two novel peaks with potential for subspecies differentiation were found. Therefore, machine learning methodology has proven to be a promising approach for rapid and accurate identification of subspecies of M. abscessus using MALDI-TOF MS.

Fourier transform infrared spectroscopy; can it be used as a rapid typing method of Neisseria gonorrhoeae?

BACKGROUND: Typing of Neisseria gonorrhoeae is necessary for epidemiologic surveillance, while time consuming and resource intensive. Fourier transform infrared (FTIR) spectroscopy has shown promising results when typing several bacterial species. This study investigates whether FTIR spectroscopy can be used as a rapid method for typing clinical N. gonorrhoeae isolates, comparing FTIR spectroscopy to multi locus sequence typing (MLST), N. gonorrhoeae multi antigen sequence typing (NG-MAST) and whole genome sequencing (WGS). METHODS: Sixty consecutive isolates from a venereology clinic and three isolates from an outbreak were included. Isolates were analysed with FTIR spectroscopy on the IR Biotyper system (Bruker Daltonik) with the IR Biotyper software (version 2.1) with default analysis settings (spectral range 1300-800 cm-1). Four technical replicates of each isolate were analysed in three different runs. The output was a hierarchical cluster analysis (HCA) presented as a dendrogram; a tree-like overview of how closely different isolates are related. FTIR spectroscopy was compared to MLST, NG-MAST and WGS to see if the FTIR spectroscopy-dendrogram grouped the isolates in the same clusters. RESULTS: Fifty-one out of 60 isolates, and the three outbreak isolates, produced at least one spectrum in each run and were included. No agreement between FTIR spectroscopy and MLST or NG-MAST or WGS was shown. The FTIR spectroscopy-dendrogram failed to cluster the outbreak isolates. CONCLUSION: FTIR spectroscopy (spectral range 1300-800 cm-1) is not yet suitable for epidemiologic typing of N. gonorrhoeae. Absence of a capsule as well as phase- and antigenic variation of carbohydrate surface structures of the gonococcal cell wall may contribute to our findings. Future studies should include analysis of a wider range of the spectrum recorded (4000-500 cm-1), and should also explore further mathematical analytic approaches of the similarity between spectra.

Multicentre study on the reproducibility of MALDI-TOF MS for nontuberculous mycobacteria identification.

The ability of MALDI-TOF for the identification of nontuberculous mycobacteria (NTM) has improved recently thanks to updated databases and optimized protein extraction procedures. Few multicentre studies on the reproducibility of MALDI-TOF have been performed so far, none on mycobacteria. The aim of this study was to evaluate the reproducibility of MALDI-TOF for the identification of NTM in 15 laboratories in 9 European countries. A total of 98 NTM clinical isolates were grown on Löwenstein-Jensen. Biomass was collected in tubes with water and ethanol, anonymized and sent out to the 15 participating laboratories. Isolates were identified using MALDI Biotyper (Bruker Daltonics). Up to 1330 MALDI-TOF identifications were collected in the study. A score ≥ 1.6 was obtained for 100% of isolates in 5 laboratories (68.2-98.6% in the other). Species-level identification provided by MALDI-TOF was 100% correct in 8 centres and 100% correct to complex-level in 12 laboratories. In most cases, the misidentifications obtained were associated with closely related species. The variability observed for a few isolates could be due to variations in the protein extraction procedure or to MALDI-TOF system status in each centre. In conclusion, MALDI-TOF showed to be a highly reproducible method and suitable for its implementation for NTM identification.

Multicenter Evaluation of Rapid BACpro® II for the Accurate Identification of Microorganisms Directly from Blood Cultures Using MALDI-TOF MS.

The identification of microorganisms directly from blood cultures using MALDI-TOF MS has been shown to be the most impacting application of this methodology. In this study, a novel commercial method was evaluated in four clinical microbiology laboratories. Positive blood culture samples (n = 801) were processed using a rapid BACpro® II kit and then compared with the routine gold standard. A subset of monomicrobial BCs (n = 560) were analyzed in parallel with a Sepsityper® Kit (Bruker Daltonics, Bremen, Germany) and compared with the rapid BACpro® II kit. In addition, this kit was also compared with two different in-house methods. Overall, 80.0% of the monomicrobial isolates (609/761; 95% CI 71.5-88.5) were correctly identified by the rapid BACpro® II kit at the species level (92.3% of the Gram negative and 72.4% of the Gram positive bacteria). The comparison with the Sepsityper® Kit showed that the rapid BACpro® II kit generated higher rates of correct species-level identification for all categories (p > 0.0001), except for yeasts identified with score values > 1.7. It also proved superior to the ammonium chloride method (p > 0.0001), but the differential centrifugation method allowed for higher rates of correct identification for Gram negative bacteria (p > 0.1). The percentage of accurate species-level identification of Gram positive bacteria was particularly noteworthy in comparison with other commercial and in-house methods.

Standardised surveillance of Clostridium difficile infection in European acute care hospitals: a pilot study, 2013.

Clostridium difficile infection (CDI) remains poorly controlled in many European countries, of which several have not yet implemented national CDI surveillance. In 2013, experts from the European CDI Surveillance Network project and from the European Centre for Disease Prevention and Control developed a protocol with three options of CDI surveillance for acute care hospitals: a 'minimal' option (aggregated hospital data), a 'light' option (including patient data for CDI cases) and an 'enhanced' option (including microbiological data on the first 10 CDI episodes per hospital). A total of 37 hospitals in 14 European countries tested these options for a three-month period (between 13 May and 1 November 2013). All 37 hospitals successfully completed the minimal surveillance option (for 1,152 patients). Clinical data were submitted for 94% (1,078/1,152) of the patients in the light option; information on CDI origin and outcome was complete for 94% (1,016/1,078) and 98% (294/300) of the patients in the light and enhanced options, respectively. The workload of the options was 1.1, 2.0 and 3.0 person-days per 10,000 hospital discharges, respectively. Enhanced surveillance was tested and was successful in 32 of the hospitals, showing that C. difficile PCR ribotype 027 was predominant (30% (79/267)). This study showed that standardised multicountry surveillance, with the option of integrating clinical and molecular data, is a feasible strategy for monitoring CDI in Europe.

Diversity and antifungal susceptibility of Norwegian Candida glabrata clinical isolates.

BACKGROUND: Increasing numbers of immunocompromised patients have resulted in greater incidence of invasive fungal infections with high mortality. Candida albicans infections dominate, but during the last decade, Candida glabrata has become the second highest cause of candidemia in the United States and Northern Europe. Reliable and early diagnosis, together with appropriate choice of antifungal treatment, is needed to combat these challenging infections. OBJECTIVES: To confirm the identity of 183 Candida glabrata isolates from different human body sites using matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) and VITEK(®)2, and to analyze isolate protein profiles and antifungal susceptibility. The minimum inhibitory concentration (MIC) of seven antifungal drugs was determined for the isolates to elucidate susceptibility. DESIGN: A total of 183 C. glabrata isolates obtained between 2002 and 2012 from Norwegian health-care units were analyzed. For species verification and differentiation, biochemical characterization (VITEK(®)2) and mass spectrometry (MALDI-TOF) were used. MIC determination for seven antifungal drugs was undertaken using E-tests(®). RESULTS: Using VITEK(®)2, 92.9% of isolates were identified as C. glabrata, while all isolates (100%) were identified as C. glabrata using MALDI-TOF. Variation in protein spectra occurred for all identified C. glabrata isolates. The majority of isolates had low MICs to amphotericin B (≤1 mg/L for 99.5%) and anidulafungin (≤0.06 mg/L for 98.9%). For fluconazole, 18% of isolates had MICs >32 mg/L and 82% had MICs in the range ≥0.016 mg/L to ≤32 mg/L. CONCLUSIONS: Protein profiles and antifungal susceptibility characteristics of the C. glabrata isolates were diverse. Clustering of protein profiles indicated that many azole resistant isolates were closely related. In most cases, isolates had highest susceptibility to amphotericin B and anidulafungin. The results confirmed previous observations of high MICs to fluconazole and flucytosine. MALDI-TOF was more definitive than VITEK(®)2 for C. glabrata identification.

Identification of airborne bacteria by 16S rDNA sequencing, MALDI-TOF MS and the MIDI microbial identification system.

The aim of this study was to collect and identify airborne bacteria in Norway, Sweden and Finland and to compare three different technologies for identifying collected airborne bacterial isolates: the "gold standard" method 16S rDNA sequencing, MALDI-TOF MS using the MALDI Biotyper 2.0 and the MIDI Sherlock® Microbial Identification System (MIDI MIS system). Airborne bacteria were collected during three different periods from May to October 2009 using air sampling directly on agar plates. A total of 140 isolates were collected during three sampling campaigns, and 74 % (103) of these isolates were analyzed by all three methods. The dominant genera in Norway and Finland were the gram-positive bacteria Bacillus and Staphylococcus, whereas the gram-negative bacterium Acinetobacter was the dominant genus in Sweden. Using 16S rDNA sequencing, MALDI-TOF MS and MIDI MIS analysis, 83, 79 and 75 %, respectively, of the isolates were identified and assigned to order or higher taxonomic levels. In this study, the MALDI-TOF MS combining with the MALDI Biotyper 2.0 classification tool was demonstrated to be a fast and reliable alternative for identifying the airborne bacterial isolates. These studies have increased knowledge about the airborne bacterial background in outdoor air, which can be useful for evaluating and improving the operational performance of biological detectors in various environments. To our knowledge, this is the first time that 16S rDNA sequencing, MALDI-TOF MS and MIDI MIS analysis technologies have been compared for their efficiency in identifying airborne bacteria.

Multilocus sequence typing of serial Candida albicans isolates from children with cancer, children with cystic fibrosis and healthy controls.

Candida albicans is a fungal pathogen, but also a commensal in many individuals. Since detailed molecular studies of children carrying C. albicans are lacking, we longitudinally investigated fecal and tonsillopharyngeal samples from 10 children undergoing treatment for cancer, six children treated for cystic fibrosis (CF), and seven healthy children during the time period of 1999-2008. Multilocus sequence typing (MLST) was performed on 62 C. albicans isolates. Only three of the 23 children (13%) were colonized with genetically unrelated strains in the longitudinal follow-up. We identified 32 different diploid sequence types (DSTs), but only one (409) was shared by two siblings. Most often, the fecal strain types were identical or closely related to the tonsillopharyngeal reservoirs. We found no closely related strain types in children who were hospitalized in the same ward or in children attending the same day care center. There was no sign of resistance to fluconazole, caspofungin, amphotericin B or flucytosine over time. This study shows that both children with cancer or CF, and healthy children usually harbor the same C. albicans strain over time. We did not find indications of clonal spread between children in the same environments, except in a pair of siblings.