The Use of Machine Learning for Image Analysis Artificial Intelligence in Clinical Microbiology.

The growing transition to digital microbiology in clinical laboratories creates the opportunity to interpret images using software. Software analysis tools can be designed to use human-curated knowledge and expert rules, but more novel artificial intelligence (AI) approaches such as machine learning (ML) are being integrated into clinical microbiology practice. These image analysis AI (IAAI) tools are beginning to penetrate routine clinical microbiology practice, and their scope and impact on routine clinical microbiology practice will continue to grow. This review separates the IAAI applications into 2 broad classification categories: (i) rare event detection/classification or (ii) score-based/categorical classification. Rare event detection can be used for screening purposes or for final identification of a microbe including microscopic detection of mycobacteria in a primary specimen, detection of bacterial colonies growing on nutrient agar, or detection of parasites in a stool preparation or blood smear. Score-based image analysis can be applied to a scoring system that classifies images in toto as its output interpretation and examples include application of the Nugent score for diagnosing bacterial vaginosis and interpretation of urine cultures. The benefits, challenges, development, and implementation strategies of IAAI tools are explored. In conclusion, IAAI is beginning to impact the routine practice of clinical microbiology, and its use can enhance the efficiency and quality of clinical microbiology practice. Although the future of IAAI is promising, currently IAAI only augments human effort and is not a replacement for human expertise.

False positive congenital cytomegalovirus saliva screening results with an FDA-approved assay at two institutions.

BACKGROUND: Congenital cytomegalovirus (CMV) infection is a significant cause of childhood hearing loss and developmental delay. Congenital CMV screening was implemented at two large hospital-affiliated laboratories using the FDA-approved Alethia CMV Assay Test System. In July 2022, an increase in suspected false-positive results was noted, leading to implementation of prospective quality management strategies. METHODS: The Alethia assay was performed per manufacturer-provided instructions on saliva swab specimens. After discovery of possible elevated false-positive rates, all positive results were confirmed by repeat Alethia testing on the same specimen, orthogonal polymerase chain reaction (PCR) on the same specimen, and/or clinical adjudication. Additionally, root cause analyses were conducted to pinpoint the source of false-positive results. RESULTS: At Cleveland Clinic (CCF), 696 saliva specimens were tested after initiation of the prospective quality management strategy, of which 36 (5.2%) were positive for CMV. Five of 36 (13.9%) were confirmed CMV positive by repeat Alethia testing and orthogonal PCR. Vanderbilt Medical Center (VUMC) tested 145 specimens, of which 11 (7.6%) were positive. Two of 11 (18.2%) confirmed as positive by orthogonal PCR or clinical adjudication. The remaining specimens (31 from CCF and 9 from VUMC) were negative for CMV by repeat Alethia and/or orthogonal PCR testing. DISCUSSION: These findings suggest a false positive rate of 4.5-6.2%, higher than the 0.2% reported for this assay in FDA claims. Laboratories using Alethia CMV may consider prospective quality management to evaluate all positive results. False-positive results can lead to unnecessary follow-up care and testing, and decreased confidence in laboratory testing.

Multicenter Evaluation of the BIOFIRE Blood Culture Identification 2 Panel for Detection of Bacteria, Yeasts, and Antimicrobial Resistance Genes in Positive Blood Culture Samples.

Diagnostic tools that can rapidly identify and characterize microbes growing in blood cultures are important components of clinical microbiology practice because they help to provide timely information that can be used to optimize patient management. This publication describes the bioMérieux BIOFIRE Blood Culture Identification 2 (BCID2) Panel clinical study that was submitted to the U.S. Food & Drug Administration. Results obtained with the BIOFIRE BCID2 Panel were compared to standard-of-care (SoC) results, sequencing results, PCR results, and reference laboratory antimicrobial susceptibility testing results to evaluate the accuracy of its performance. Results for 1,093 retrospectively and prospectively collected positive blood culture samples were initially enrolled, and 1,074 samples met the study criteria and were included in the final analyses. The BIOFIRE BCID2 Panel demonstrated an overall sensitivity of 98.9% (1,712/1,731) and an overall specificity of 99.6% (33,592/33,711) for Gram-positive bacteria, Gram-negative bacteria and yeast targets which the panel is designed to detect. One hundred eighteen off-panel organisms, which the BIOFIRE BCID2 Panel is not designed to detect, were identified by SoC in 10.6% (114/1,074) of samples. The BIOFIRE BCID2 Panel also demonstrated an overall positive percent agreement (PPA) of 97.9% (325/332) and an overall negative percent agreement (NPA) of 99.9% (2,465/2,767) for antimicrobial resistance determinants which the panel is designed to detect. The presence or absence of resistance markers in Enterobacterales correlated closely with phenotypic susceptibility and resistance. We conclude that the BIOFIRE BCID2 Panel produced accurate results in this clinical trial.

Urinalysis Exhibits Excellent Predictive Capacity for the Absence of Urinary Tract Infection.

OBJECTIVE: To assess predictive value of urinalysis for negative urine culture and absence of urinary tract infection, re-evaluate the microbial growth threshold for positive urine culture result, and describe antimicrobial resistance features. Urine culture is associated with 27% of U.S. hospitalizations, and unnecessary antibiotic prescription is a main antibiotic resistance contributor. METHODS: Urinalyses with urine culture from women ages 18-49 from 2013 to 2020 were studied. Clinically diagnosed urinary tract infection (CUTI) was defined as (1) uropathogen growth, (2) documented diagnosis of urinary tract infection, and (3) antibiotic prescription. Sensitivity, specificity, and diagnostic predictive values were used to assess urinalysis performance in predicting isolation of a uropathogen by culture and in detection of CUTI. RESULTS: Total 12,252 urinalyses were included. Forty-one percent of urinalyses were associated with positive urine culture and 1287 (10.5%) with CUTI. Negative urinalysis exhibited high predictive accuracy for negative urine culture (specificity 90.3%, PPV 87.3%) and absence of CUTI (specificity 92.2%, PPV 97.4%). Twenty-four percent of patients not meeting the CUTI definition were still prescribed antibiotics. Twenty-two percent of cultures associated with CUTI exhibited growth less than 100,000 CFU/mL. Escherichia coli was implemented as causing 70% of CUTIs, and 4.2% of these produced an extended spectrum beta-lactamase. CONCLUSION: Negative urinalysis exhibits high predictive accuracy for absence of CUTI. A reporting threshold of 10,000 CFU/mL is more clinically appropriate than a 100,000 CFU/mL cutpoint. Reflex culture based on urinalysis results could complement clinical judgement and improve laboratory and antibiotic stewardship in premenopausal women.

Tailored Treatment Based on Helicobacter pylori Genetic Markers of Resistance Is Associated With Higher Eradication Success.

INTRODUCTION: Increasing antimicrobial resistance with Helicobacter pylori infection has focused efforts to tailor eradication therapy based on identifying genetic markers of resistance to predict antimicrobial susceptibility. METHODS: In this retrospective study, we report the effect of routine inclusion of antimicrobial susceptibility testing and recommendations for eradication therapy with gastric specimens with H. pylori . RESULTS: The use of a recommended treatment regimen based on genetic markers of resistance was associated with an 84% rate of eradication success and 4.4 greater odds of eradication relative to unrecommended treatment. DISCUSSION: This is the first study describing the use of H. pylori genetic resistance testing as standard of care.

Building the Model.

CONTEXT.­: Machine learning (ML) allows for the analysis of massive quantities of high-dimensional clinical laboratory data, thereby revealing complex patterns and trends. Thus, ML can potentially improve the efficiency of clinical data interpretation and the practice of laboratory medicine. However, the risks of generating biased or unrepresentative models, which can lead to misleading clinical conclusions or overestimation of the model performance, should be recognized. OBJECTIVES.­: To discuss the major components for creating ML models, including data collection, data preprocessing, model development, and model evaluation. We also highlight many of the challenges and pitfalls in developing ML models, which could result in misleading clinical impressions or inaccurate model performance, and provide suggestions and guidance on how to circumvent these challenges. DATA SOURCES.­: The references for this review were identified through searches of the PubMed database, US Food and Drug Administration white papers and guidelines, conference abstracts, and online preprints. CONCLUSIONS.­: With the growing interest in developing and implementing ML models in clinical practice, laboratorians and clinicians need to be educated in order to collect sufficiently large and high-quality data, properly report the data set characteristics, and combine data from multiple institutions with proper normalization. They will also need to assess the reasons for missing values, determine the inclusion or exclusion of outliers, and evaluate the completeness of a data set. In addition, they require the necessary knowledge to select a suitable ML model for a specific clinical question and accurately evaluate the performance of the ML model, based on objective criteria. Domain-specific knowledge is critical in the entire workflow of developing ML models.

Diagnostic stewardship for urinary tract infection: A snapshot of the expert guidance.

The urine culture, the cornerstone for laboratory diagnosis of urinary tract infection (UTI), is associated with a high frequency of false-positive and false-negative results, and its diagnostic threshold is debated. Urine culture takes days to result, and antibiotics are often initiated while awaiting final culture readings. Further, asymptomatic bacteriuria-the presence of bacteria in urine in the absence of UTI symptoms-generally does not warrant treatment. The authors review current expert guidance on the use of urine culture, including approaches to ordering, processing, and reporting of urine cultures, with the goal of reducing unnecessary antibiotic use and misdiagnosis of UTI.

Design and Implementation of Improved SARS-CoV-2 Diagnostic Assays To Mitigate the Impact of Genomic Mutations on Target Failure: the Xpert Xpress SARS-CoV-2 Experience.

In 2020, the U.S. Food and Drug Administration (FDA) enabled manufacturers to request emergency use authorization (EUA) to facilitate the rapid authorization of in vitro diagnostic (IVD) platforms for the detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Uncommon SARS-CoV-2 point mutations could cause nucleocapsid (N) gene target failure (NGTF) when using first-generation Xpert Xpress assays, so improvements were designed and implemented. In response to NGTF reports and with consideration of viral genomic information in public databases, the Xpress assays were redesigned to mitigate the impact of SARS-CoV-2 mutations on qualitative assay performance. The second-generation assays include a third gene target (RNA-dependent RNA polymerase [RdRp]) and redundant oligonucleotide probes for the N2 target. First- and second-generation assay performances were evaluated using a challenge set of samples. A second-generation assay with updated oligonucleotide chemistry received FDA EUA in September 2021. A prototype assay with oligonucleotide chemistry similar to that of the second-generation assay with FDA EUA successfully detected all three gene targets (N2, envelope [E], and RdRp) in all challenge samples (100%; 50/50), including variants with N gene mutations (g.29197C>T or g.29200C>T), which caused NGTF in the first-generation assays. Investigation and reporting of IVD target failures, public sharing of viral genomic sequence data, and the FDA EUA pathway were essential components in facilitating a short cycle time from the identification of mutations that impact the performance of an IVD assay to the design and implementation of an improved IVD assay. IMPORTANCE The SARS-CoV-2 genome has mutated during the coronavirus disease 2019 (COVID-19) pandemic. Some of these mutations have impacted the performance of nucleic acid amplification tests like PCR, which are commonly used as diagnostic tools to detect an infection. The U.S. Food and Drug Administration (FDA) emergency use authorization (EUA) process enables the rapid reformulation and regulatory authorization of improved PCRs. In our experience, the identification of SARS-CoV-2 mutations that impact PCR performance, the subsequent development of improved PCR chemistry, and the use of the FDA EUA regulatory pathway led to improved diagnostic performance during the SARS-CoV-2 pandemic that is able to keep pace with the rapidly evolving genome of SARS-CoV-2.

Meningococcal Urethritis: Old and New.

Neisseria meningitidis is a common commensal bacterium found in the respiratory tract, but it can also cause severe, invasive disease. Vaccines have been employed which have been successful in helping to prevent invasive disease caused by encapsulated N. meningitidis from the A, C, W, Y, and B serogroups. Currently, nonencapsulated N. meningitidis groups are more common commensals in the population than in the prevaccine era. One emerging nonencapsulated group of bacteria is the U.S. N. meningitidis urethritis clade (US_NmUC), which can cause meningococcal urethritis in men. US_NmUC has unique genotypic and phenotypic features that may increase its fitness in the male urethra. It is diagnostically challenging to identify and distinguish meningococcal urethritis from Neisseria gonorrhoeae, as the clinical presentation and microbiological findings are overlapping. In this review, the history of meningococcal urethritis, emergence of US_NmUC, laboratory diagnosis, and clinical treatment are all explored.

As Omicron Takes Hold and Other New Variants Arise, COVID-19 Testing Remains the Universally Agreed Tool to Effect Transition From Pandemic to Endemic State.

The COVID-19 pandemic has caused more than 448 million cases and 6 million deaths worldwide to date. Omicron is now the dominant SARS-CoV-2 variant, making up more than 90% of cases in countries reporting sequencing data. As the pandemic continues into its third year, continued testing is a strategic and necessary tool for transitioning to an endemic state of COVID-19. Here, we address three critical topics pertaining to the transition from pandemic to endemic: defining the endemic state for COVID-19, highlighting the role of SARS-CoV-2 testing as endemicity is approached, and recommending parameters for SARS-CoV-2 testing once endemicity is reached. We argue for an approach that capitalizes on the current public health momentum to increase capacity for PCR-based testing and whole genome sequencing to monitor emerging infectious diseases. Strategic development and utilization of testing, including viral panels in addition to vaccination, can keep SARS-CoV-2 in a manageable endemic state and build a framework of preparedness for the next pandemic.

Partial ORF1ab Gene Target Failure with Omicron BA.2.12.1.

Mutations in the viral genome of SARS-CoV-2 can impact the performance of molecular diagnostic assays. In some cases, such as S gene target failure, the impact can serve as a unique indicator of a particular SARS-CoV-2 variant and provide a method for rapid detection. Here we describe partial ORF1ab gene target failure (pOGTF) on the cobas ® SARS-CoV-2 assays, defined by a ≥2 thermocycles delay in detection of the ORF1ab gene compared to the E gene. We demonstrate that pOGTF is 97% sensitive and 99% specific for SARS-CoV-2 lineage BA.2.12.1, an emerging variant in the United States with spike L452Q and S704L mutations that may impact transmission, infectivity, and/or immune evasion. Increasing rates of pOGTF closely mirrored rates of BA.2.12.1 sequences uploaded to public databases, and, importantly increasing local rates of pOGTF also mirrored increasing overall test positivity. Use of pOGTF as a proxy for BA.2.12.1 provides faster tracking of the variant than whole-genome sequencing and can benefit laboratories without sequencing capabilities.

Partial ORF1ab Gene Target Failure with Omicron BA.2.12.1.

Mutations in the genome of SARS-CoV-2 can affect the performance of molecular diagnostic assays. In some cases, such as S-gene target failure, the impact can serve as a unique indicator of a particular SARS-CoV-2 variant and provide a method for rapid detection. Here, we describe partial ORF1ab gene target failure (pOGTF) on the cobas SARS-CoV-2 assays, defined by a ≥2-thermocycle delay in detection of the ORF1ab gene compared to that of the E-gene. We demonstrate that pOGTF is 98.6% sensitive and 99.9% specific for SARS-CoV-2 lineage BA.2.12.1, an emerging variant in the United States with spike L452Q and S704L mutations that may affect transmission, infectivity, and/or immune evasion. Increasing rates of pOGTF closely mirrored rates of BA.2.12.1 sequences uploaded to public databases, and, importantly, increasing local rates of pOGTF also mirrored increasing overall test positivity. Use of pOGTF as a proxy for BA.2.12.1 provides faster tracking of the variant than whole-genome sequencing and can benefit laboratories without sequencing capabilities.

Genomic heterogeneity underlies multidrug resistance in Pseudomonas aeruginosa: A population-level analysis beyond susceptibility testing.

BACKGROUND: Pseudomonas aeruginosa is a persistent and difficult-to-treat pathogen in many patients, especially those with Cystic Fibrosis (CF). Herein, we describe a longitudinal analysis of a series of multidrug resistant (MDR) P. aeruginosa isolates recovered in a 17-month period, from a young female CF patient who underwent double lung transplantation. Our goal was to understand the genetic basis of the observed resistance phenotypes, establish the genomic population diversity, and define the nature of sequence evolution over time. METHODS: Twenty-two sequential P. aeruginosa isolates were obtained within a 17-month period, before and after a double-lung transplant. At the end of the study period, antimicrobial susceptibility testing, whole genome sequencing (WGS), phylogenetic analyses and RNAseq were performed in order to understand the genetic basis of the observed resistance phenotypes, establish the genomic population diversity, and define the nature of sequence changes over time. RESULTS: The majority of isolates were resistant to almost all tested antibiotics. A phylogenetic reconstruction revealed 3 major clades representing a genotypically and phenotypically heterogeneous population. The pattern of mutation accumulation and variation of gene expression suggested that a group of closely related strains was present in the patient prior to transplantation and continued to change throughout the course of treatment. A trend toward accumulation of mutations over time was observed. Different mutations in the DNA mismatch repair gene mutL consistent with a hypermutator phenotype were observed in two clades. RNAseq performed on 12 representative isolates revealed substantial differences in the expression of genes associated with antibiotic resistance and virulence traits. CONCLUSIONS: The overwhelming current practice in the clinical laboratories setting relies on obtaining a pure culture and reporting the antibiogram from a few isolated colonies to inform therapy decisions. Our analyses revealed significant underlying genomic heterogeneity and unpredictable evolutionary patterns that were independent of prior antibiotic treatment, highlighting the need for comprehensive sampling and population-level analysis when gathering microbiological data in the context of CF P. aeruginosa chronic infection. Our findings challenge the applicability of antimicrobial stewardship programs based on single-isolate resistance profiles for the selection of antibiotic regimens in chronic infections such as CF.

Multicenter Evaluation of the Acuitas AMR Gene Panel for Detection of an Extended Panel of Antimicrobial Resistance Genes among Bacterial Isolates.

The Acuitas antimicrobial resistance (AMR) gene panel is a qualitative, multiplex, nucleic acid-based in vitro diagnostic test for the detection and differentiation of 28 antimicrobial resistance markers associated with not susceptible results (NS; i.e., intermediate or resistant) to one or more antimicrobial agents among cultured isolates of select Enterobacterales, Pseudomonas aeruginosa, and Enterococcus faecalis. This study was conducted at four sites and included testing of 1,224 deidentified stocks created from 584 retrospectively collected isolates and 83 prospectively collected clinical isolates. The Acuitas results were compared with a combined reference standard including whole-genome sequencing, organism identification, and phenotypic antimicrobial susceptibility testing. The positive percent agreement (PPA) for FDA-cleared AMR targets ranged from 94.4% for MCR-1 to 100% for armA, CTX-M-2, DHA, IMP, OXA-9, SHV, vanA, and VEB. The negative percent agreement (NPA) for the majority of targets was ≥99%, except for AAC, AAD, CMY-41, P. aeruginosa gyrA mutant, Sul1, Sul2, and TEM targets (range, 96.5% to 98.5%). Three AMR markers did not meet FDA inclusion criteria (GES, SPM, and MCR-2). For each organism, 1 to 22 AMR targets met the minimum reportable PPA/NPA and correlated with ≥80% positive predictive value with associated NS results for at least one agent (i.e., the probability of an organism carrying an AMR marker testing NS to the associated agent). We demonstrate that the Acuitas AMR gene panel is an accurate method to detect a broad array of AMR markers among cultured isolates. The AMR markers were further associated with expected NS results for specific agent-organism combinations.

Deep Convolutional Neural Networks Implementation for the Analysis of Urine Culture.

BACKGROUND: Urine culture images collected using bacteriology automation are currently interpreted by technologists during routine standard-of-care workflows. Machine learning may be able to improve the harmonization of and assist with these interpretations. METHODS: A deep learning model, BacterioSight, was developed, trained, and tested on standard BD-Kiestra images of routine blood agar urine cultures from 2 different medical centers. RESULTS: BacterioSight displayed performance on par with standard-of-care-trained technologist interpretations. BacterioSight accuracy ranged from 97% when compared to standard-of-care (single technologist) and reached 100% when compared to a consensus reached by a group of technologists (gold standard in this study). Variability in image interpretation by trained technologists was identified and annotation "fuzziness" was quantified and found to correlate with reduced confidence in BacterioSight interpretation. Intra-testing (training and testing performed within the same institution) performed well giving Area Under the Curve (AUC) ≥0.98 for negative and positive plates, whereas, cross-testing on images (trained on one institution's images and tested on images from another institution) showed decreased performance with AUC ≥0.90 for negative and positive plates. CONCLUSIONS: Our study provides a roadmap on how BacterioSight or similar deep learning prototypes may be implemented to screen for microbial growth, flag difficult cases for multi-personnel review, or auto-verify a subset of cultures with high confidence. In addition, our results highlight image interpretation variability by trained technologist within an institution and globally across institutions. We propose a model in which deep learning can enhance patient care by identifying inherent sample annotation variability and improving personnel training.

Clinical and Infection Prevention Applications of Severe Acute Respiratory Syndrome Coronavirus 2 Genotyping: an Infectious Diseases Society of America/American Society for Microbiology Consensus Review Document.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged into a world of maturing pathogen genomics, with more than 2 million genomes sequenced at the time of writing. The rise of more transmissible variants of concern that impact vaccine and therapeutic effectiveness has led to widespread interest in SARS-CoV-2 evolution. Clinicians are also eager to take advantage of the information provided by SARS-CoV-2 genotyping beyond surveillance purposes. Here, we review the potential role of SARS-CoV-2 genotyping in clinical care. The review covers clinical use cases for SARS-CoV-2 genotyping, methods of SARS-CoV-2 genotyping, assay validation and regulatory requirements, and clinical reporting for laboratories, as well as emerging issues in clinical SARS-CoV-2 sequencing. While clinical uses of SARS-CoV-2 genotyping are currently limited, rapid technological change along with a growing ability to interpret variants in real time foretells a growing role for SARS-CoV-2 genotyping in clinical care as continuing data emerge on vaccine and therapeutic efficacy.

Clinical and Infection Prevention Applications of Severe Acute Respiratory Syndrome Coronavirus 2 Genotyping: An Infectious Diseases Society of America/American Society for Microbiology Consensus Review Document.

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged into a world of maturing pathogen genomics, with >2 million genomes sequenced at this writing. The rise of more transmissible variants of concern that affect vaccine and therapeutic effectiveness has led to widespread interest in SARS-CoV-2 evolution. Clinicians are also eager to take advantage of the information provided by SARS-CoV-2 genotyping beyond surveillance purposes. Here, we review the potential role of SARS-CoV-2 genotyping in clinical care. The review covers clinical use cases for SARS-CoV-2 genotyping, methods of SARS-CoV-2 genotyping, assay validation and regulatory requirements, clinical reporting for laboratories, and emerging issues in clinical SARS-CoV-2 sequencing. While clinical uses of SARS-CoV-2 genotyping are currently limited, rapid technological change along with a growing ability to interpret variants in real time foretell a growing role for SARS-CoV-2 genotyping in clinical care as continuing data emerge on vaccine and therapeutic efficacy.

Teleclinical Microbiology: An Innovative Approach to Providing Web-Enabled Diagnostic Laboratory Services in Syria.

OBJECTIVES: Telemedicine can compensate for the lack of health care specialists in response to protracted humanitarian crises. We sought to assess the usability of a teleclinical microbiology (TCM) program to provide diagnostic services in a hard-to-reach region of Syria. METHODS: A semimobile station was equipped with conventional micrograph and macrograph digital imaging systems. An electronic platform (Telemicrobiology in Humanitarian Crises, TmHC) was created to facilitate sharing, interpreting, and storing the results. A pilot study was conducted to identify the bacterial species and antimicrobial susceptibility pattern of 74 urinary clinical isolates. An experience survey was conducted to capture the feedback of 8 participants in the program. RESULTS: The TmHC platform (https://sdh.ngo/tmhc/) enabled systematic transmission of the laboratory records and co-interpretation of the results. The isolates were identified as Escherichia coli (n = 61), Klebsiella pneumoniae (n = 12), and Proteus mirabilis(n = 1). All the isolates were multidrug resistant. The performance of our TCM module was rated 4 (satisfying) and 5 (very satisfying) by 6 and 2 users, respectively. Data security of and cost-effectiveness were the main perceived concerns. CONCLUSIONS: Although we encountered several context-related obstacles, our TCM program managed to reach a highly vulnerable population of 4 million people confined in the northwest region of Syria.