Microscopic interpretation of stained smears is one of the most operator-dependent and time-intensive activities in the clinical microbiology laboratory. Here, we investigated application of an automated image acquisition and convolutional neural network (CNN)-based approach for automated Gram stain classification. Using an automated microscopy platform, uncoverslipped slides were scanned with a 40× dry objective, generating images of sufficient resolution for interpretation. We collected 25,488 images from positive blood culture Gram stains prepared during routine clinical workup. These images were used to generate 100,213 crops containing Gram-positive cocci in clusters, Gram-positive cocci in chains/pairs, Gram-negative rods, or background (no cells). These categories were targeted for proof-of-concept development as they are associated with the majority of bloodstream infections. Our CNN model achieved a classification accuracy of 94.9% on a test set of image crops. Receiver operating characteristic (ROC) curve analysis indicated a robust ability to differentiate between categories with an area under the curve of >0.98 for each. After training and validation, we applied the classification algorithm to new images collected from 189 whole slides without human intervention. Sensitivity and specificity were 98.4% and 75.0% for Gram-positive cocci in chains and pairs, 93.2% and 97.2% for Gram-positive cocci in clusters, and 96.3% and 98.1% for Gram-negative rods. Taken together, our data support a proof of concept for a fully automated classification methodology for blood-culture Gram stains. Importantly, the algorithm was highly adept at identifying image crops with organisms and could be used to present prescreened, classified crops to technologists to accelerate smear review. This concept could potentially be extended to all Gram stain interpretive activities in the clinical laboratory.

Resolving the coronavirus disease 2019 (COVID-19) pandemic requires diagnostic testing to determine which individuals are infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The current gold standard is to perform reverse-transcription polymerase chain reaction (PCR) on nasopharyngeal samples. Best-in-class assays demonstrate a limit of detection (LoD) of approximately 100 copies of viral RNA per milliliter of transport media. However, LoDs of currently approved assays vary over 10,000-fold. Assays with higher LoDs will miss infected patients. However, the relative clinical sensitivity of these assays remains unknown. Here we model the clinical sensitivities of assays based on their LoD. Cycle threshold (Ct) values were obtained from 4700 first-time positive patients using the Abbott RealTime SARS-CoV-2 Emergency Use Authorization test. We derived viral loads from Ct based on PCR principles and empiric analysis. A sliding scale relationship for predicting clinical sensitivity was developed from analysis of viral load distribution relative to assay LoD. Ct values were reliably repeatable over short time testing windows, providing support for use as a tool to estimate viral load. Viral load was found to be relatively evenly distributed across log₁₀ bins of incremental viral load. Based on these data, each 10-fold increase in LoD is expected to lower assay sensitivity by approximately 13%. The assay LoD meaningfully impacts clinical performance of SARS-CoV-2 tests. The highest LoDs on the market will miss a majority of infected patients. Assays should therefore be benchmarked against a universal standard to allow cross-comparison of SARS-CoV-2 detection methods.

Because of its ease of dispersal and high lethality, Bacillus anthracis is one of the most feared biowarfare agents. A better understanding of anthrax pathogenesis is urgently needed to develop new therapies for systemic disease that is relatively unresponsive to antibiotics. Although experimental evidence has implicated a role for macrophages in anthrax pathogenesis, clinical and pathological observations suggest that a direct insult to the host vasculature may also be important. Two bacterial toxins, lethal toxin and edema toxin, are believed to mediate the clinical sequelae of anthrax. Here, I examined whether these toxins are directly toxic to endothelial cells, the cell type that lines the interior of blood vessels. I show for the first time that lethal toxin but not edema toxin reduces the viability of cultured human endothelial cells and induces caspase-dependent endothelial apoptosis. In addition, this toxicity affects both microvascular and large vessel endothelial cells as well as endothelial cells that have differentiated into tubules within a type I collagen extracellular matrix. Finally, lethal toxin induces cleavage of mitogen-activated protein kinase kinases in endothelial cells and inhibits phosphorylation of ERK, p38, and JNK p46. Based on the contributions of these pathways to endothelial survival, I propose that lethal toxin-mediated cytotoxicity/apoptosis results primarily through inhibition of the ERK pathway. I also hypothesize that the observed endothelial toxicity contributes to vascular pathology and hemorrhage during systemic anthrax.

The observed MIC may depend on the number of bacteria initially inoculated into the assay. This phenomenon is termed the inoculum effect (IE) and is often most pronounced for β-lactams in strains expressing β-lactamase enzymes. The Clinical and Laboratory Standards Institute (CLSI)-recommended inoculum is 5 × 10 ⁵ CFU ml ⁻¹ with an acceptable range of 2 × 10 ⁵ to 8 × 10 ⁵ CFU ml ⁻¹ . IE testing is typically performed using an inoculum 100-fold greater than the CLSI-recommended inoculum. Therefore, it remains unknown whether the IE influences MICs during testing performed according to CLSI guidelines. Here, we utilized inkjet printing technology to test the IE on cefepime, meropenem, and ceftazidime-avibactam. First, we determined that the inkjet dispense volume correlated well with the number of bacteria delivered to microwells in 2-fold ( R ² = 0.99) or 1.1-fold ( R ² = 0.98) serial dilutions. We then quantified the IE by dispensing orthogonal titrations of bacterial cells and antibiotics. For cefepime-resistant and susceptible dose-dependent strains, a 2-fold increase in inoculum resulted in a 1.6 log ₂ -fold increase in MIC. For carbapenemase-producing strains, each 2-fold reduction in inoculum resulted in a 1.26 log ₂ -fold reduction in meropenem MIC. At the lower end of the CLSI-allowable inoculum range, minor error rates of 34.8% were observed for meropenem when testing a resistant-strain set. Ceftazidime-avibactam was not subject to an appreciable IE. Our results suggest that IE is sufficiently pronounced for meropenem and cefepime in multidrug-resistant Gram-negative pathogens to affect categorical interpretations during standard laboratory testing.

Resolving the COVID-19 pandemic requires diagnostic testing to determine which individuals are infected and which are not. The current gold standard is to perform RT-PCR on nasopharyngeal samples. Best-in-class assays demonstrate a limit of detection (LoD) of ~100 copies of viral RNA per milliliter of transport media. However, LoDs of currently approved assays vary over 10,000-fold. Assays with higher LoDs will miss more infected patients, resulting in more false negatives. However, the false-negative rate for a given LoD remains unknown. Here we address this question using over 27,500 test results for patients from across our healthcare network tested using the Abbott RealTime SARS-CoV-2 EUA. These results suggest that each 10-fold increase in LoD is expected to increase the false negative rate by 13%, missing an additional one in eight infected patients. The highest LoDs on the market will miss a majority of infected patients, with false negative rates as high as 70%. These results suggest that choice of assay has meaningful clinical and epidemiological consequences. The limit of detection matters.