Rapidly adaptable automated interpretation of point-of-care COVID-19 diagnostics.

BACKGROUND: Point-of-care diagnostic devices, such as lateral-flow assays, are becoming widely used by the public. However, efforts to ensure correct assay operation and result interpretation rely on hardware that cannot be easily scaled or image processing approaches requiring large training datasets, necessitating large numbers of tests and expert labeling with validated specimens for every new test kit format. METHODS: We developed a software architecture called AutoAdapt POC that integrates automated membrane extraction, self-supervised learning, and few-shot learning to automate the interpretation of POC diagnostic tests using smartphone cameras in a scalable manner. A base model pre-trained on a single LFA kit is adapted to five different COVID-19 tests (three antigen, two antibody) using just 20 labeled images. RESULTS: Here we show AutoAdapt POC to yield 99% to 100% accuracy over 726 tests (350 positive, 376 negative). In a COVID-19 drive-through study with 74 untrained users self-testing, 98% found image collection easy, and the rapidly adapted models achieved classification accuracies of 100% on both COVID-19 antigen and antibody test kits. Compared with traditional visual interpretation on 105 test kit results, the algorithm correctly identified 100% of images; without a false negative as interpreted by experts. Finally, compared to a traditional convolutional neural network trained on an HIV test kit, the algorithm showed high accuracy while requiring only 1/50th of the training images. CONCLUSIONS: The study demonstrates how rapid domain adaptation in machine learning can provide quality assurance, linkage to care, and public health tracking for untrained users across diverse POC diagnostic tests.

It can be difficult to correctly interpret the results of rapid diagnostic tests that give a visual readout, such as COVID rapid tests. We developed a computational algorithm to interpret rapid test results using an image taken by a smartphone camera. This algorithm can easily be adapted for use on results from different test kits. The algorithm was accurate at interpreting results obtained by members of the public using various COVID rapid tests and diagnostic tests with similar outputs used for other infections. The use of this algorithm should enable accurate interpretation of rapid diagnostic tests by members of the public and hence enable improved medical care.

Development of a rapid point-of-care test that measures neutralizing antibodies to SARS-CoV-2.

BACKGROUND: After receiving a COVID-19 vaccine, most recipients want to know if they are protected from infection and for how long. Since neutralizing antibodies are a correlate of protection, we developed a lateral flow assay (LFA) that measures levels of neutralizing antibodies from a drop of blood. The LFA is based on the principle that neutralizing antibodies block binding of the receptor-binding domain (RBD) to angiotensin-converting enzyme 2 (ACE2). METHODS: The ability of the LFA was assessed to correctly measure neutralization of sera, plasma or whole blood from patients with COVID-19 using SARS-CoV-2 microneutralization assays. We also determined if the LFA distinguished patients with seasonal respiratory viruses from patients with COVID-19. To demonstrate the usefulness of the LFA, we tested previously infected and non-infected COVID-19 vaccine recipients at baseline and after first and second vaccine doses. RESULTS: The LFA compared favorably with SARS-CoV-2 microneutralization assays with an area under the ROC curve of 98%. Sera obtained from patients with seasonal coronaviruses did not show neutralizing activity in the LFA. After a single mRNA vaccine dose, 87% of previously infected individuals demonstrated high levels of neutralizing antibodies. However, if individuals were not previously infected, only 24% demonstrated high levels of neutralizing antibodies after one vaccine dose. A second dose boosted neutralizing antibody levels just 8% higher in previously infected individuals, but over 63% higher in non-infected individuals. CONCLUSIONS: A rapid, semi-quantitative, highly portable and inexpensive neutralizing antibody test might be useful for monitoring rise and fall in vaccine-induced neutralizing antibodies to COVID-19.

Mass Spectrometric Analysis of Urine from COVID-19 Patients for Detection of SARS-CoV-2 Viral Antigen and to Study Host Response.

SARS-CoV-2 infection has become a major public health burden and affects many organs including lungs, kidneys, the liver, and the brain. Although the virus is readily detected and diagnosed using nasopharyngeal swabs by reverse transcriptase polymerase chain reaction (RT-PCR), detection of its presence in body fluids is fraught with difficulties. A number of published studies have failed to detect viral RNA by RT-PCR methods in urine. Although microbial identification in clinical microbiology using mass spectrometry is undertaken after culture, here we undertook a mass spectrometry-based approach that employed an enrichment step to capture and detect SARS-CoV-2 nucleocapsid protein directly from urine of COVID-19 patients without any culture. We detected SARS-CoV-2 nucleocapsid protein-derived peptides from 13 out of 39 urine samples. Further, a subset of COVID-19 positive and COVID-19 negative urine samples validated by mass spectrometry were used for the quantitative proteomics analysis. Proteins with increased abundance in urine of SARS-CoV-2 positive individuals were enriched in the acute phase response, regulation of complement system, and immune response. Notably, a number of renal proteins such as podocin (NPHS2), an amino acid transporter (SLC36A2), and sodium/glucose cotransporter 5 (SLC5A10), which are intimately involved in normal kidney function, were decreased in the urine of COVID-19 patients. Overall, the detection of viral antigens in urine using mass spectrometry and alterations of the urinary proteome could provide insights into understanding the pathogenesis of COVID-19.

Multidrug-resistant Pseudomonas aeruginosa in healthcare facilities in Port-au-Prince, Haiti.

OBJECTIVES: Pseudomonas aeruginosa is a leading cause of opportunistic infections worldwide, particularly in healthcare settings, and frequently demonstrates resistance to commonly prescribed antimicrobials. Carbapenem resistance is prevalent worldwide, however there are currently limited data available from Haiti. The aim of this study was to characterise and document this phenotype in Port-au-Prince, Haiti, to further inform the need for appropriate infection control, empirical treatment guidelines and laboratory screening measures, both in Haiti and globally. METHODS: A total of 50 P. aeruginosa isolates were characterised by multilocus sequence typing (MLST) and antimicrobial susceptibility testing, of which 8 isolates were also subjected to whole-genome sequencing (WGS) to identify potential genetic correlations of phenotypic resistance. RESULTS: By MLST, 23 sequence types (STs) were identified, including 13 new STs. Nineteen isolates belonged to a single, previously characterised ST (ST654), all of which demonstrated a multidrug-resistant phenotype, including resistance to meropenem, imipenem and ceftazidime; two isolates were also resistant to colistin. WGS revealed the presence of genes encoding several previously characterised resistance determinants in ST654; notably ACC(6')-Ib3-cr and GES-7. Metallo-ß-lactamase genes (blaVIM-5) were also detected in three isolates. CONCLUSION: These findings confirm that drug-resistant clones of P. aeruginosa are present in Haiti, supporting the need for appropriate screening and control measures and confirming that drug-resistant micro-organisms pose a global threat. Further investigations are required to guide appropriate antimicrobial prescribing in this region.

Retrospective clinical evaluation of 4 lateral flow assays for the detection of SARS-CoV-2 IgG.

In a Clinical Laboratory Improvement Amendments laboratory setting, we evaluated severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) IgG detection with 4 lateral flow immunoassays [LFIAs; 2 iterations from BTNX Inc. (n = 457) and 1 each from ACON Laboratories (n = 200) and SD BIOSENSOR (n = 155)]. In a cohort of primarily hospitalized, reverse-transcription polymerase chain reaction-confirmed coronavirus disease 2019 cases, sensitivity at ≥14 days from symptom onset was: BTNX kit 1, 95%; BTNX kit 2, 91%; ACON, 95%; and SD, 92%. All assays showed good concordance with the Abbott SARS-CoV-2 IgG assay at ≥14 days from symptom onset: BTNX kit 1, 99%; BTNX kit 2, 94%; ACON, 99%; and SD, 100%. Specificity, measured using specimens collected prior to SARS-CoV-2 circulation in the United States and "cross-reactivity challenge" specimens, was 98% for BTNX kit 1 and ACON and 100% for BTNX kit 2 and SD. These results suggest that LFIAs may provide adequate results for rapid detection of SARS-CoV-2.

Urinary biomarkers of mycobacterial load and treatment response in pulmonary tuberculosis.

BACKGROUNDControl of the tuberculosis (TB) pandemic remains hindered in part by a lack of simple and accurate measures of treatment efficacy, as current gold standard markers rely on sputum-based assays that are slow and challenging to implement. However, previous work identified urinary N1, N12-diacetylspermine (DiAcSpm), neopterin, hydroxykynurenine, N-acetylhexosamine, ureidopropionic acid, sialic acid, and mass-to-charge ratio (m/z) 241.0903 as potential biomarkers of active pulmonary TB (ATB). Here, we evaluated their ability to serve as biomarkers of TB treatment response and mycobacterial load.METHODSWe analyzed urine samples prospectively collected from 2 cohorts with ATB. A total of 34 study participants from African countries treated with first-line TB therapy rifampin, isoniazid, pyrazinamide, and ethambutol (HRZE) were followed for 1 year, and 35 participants from Haiti treated with either HRZE or an experimental drug were followed for 14 days. Blinded samples were analyzed by untargeted HPLC-coupled high-resolution TOF-mass spectrometry.RESULTSUrinary levels of all 7 molecules significantly decreased by week 26 of successful treatment (P = 0.01 to P < 0.0001) and positively correlated with sputum mycobacterial load (P < 0.0001). Urinary DiAcSpm levels decreased significantly in participants treated with HRZE as early as 14 days (P < 0.0001) but remained unchanged in cases of ineffective therapy (P = 0.14).CONCLUSIONUrinary DiAcSpm, neopterin, hydroxykynurenine, N-acetylhexosamine, ureidopropionic acid, sialic acid, and m/z 241.0903 reductions correlated with successful anti-TB treatment and sputum mycobacterial load. Urinary DiAcSpm levels exhibited reductions capable of differentiating treatment success from failure as early as 2 weeks after the initiation of chemotherapy, advocating its further development as a potentially simple, noninvasive biomarker for assessing treatment response and bacterial load.FUNDINGThis work was supported by the Clinical and Translational Science Center at Weill Cornell College of Medicine (NIH/NCATS 1 UL1 TR002384-02 and KL2TR000458), the Department of Defense (PR170782), the National Institute of Allergy and Infectious Disease grants (NIAID T32AI007613-16, K24 AI098627, and K23 AI131913), the NIH Fogarty International Center grants (R24 TW007988 and TW010062), NIH grant (R01 GM135926), the Abby and Howard P. Milstein Program in Chemical Biology and Translational Medicine, and the Tuberculosis Research Units Networks (TBRU-N, AI111143).

Dual-Pharmacophore Pyrithione-Containing Cephalosporins Kill Both Replicating and Nonreplicating Mycobacterium tuberculosis.

The historical view of ß-lactams as ineffective antimycobacterials has given way to growing interest in the activity of this class against Mycobacterium tuberculosis (Mtb) in the presence of a ß-lactamase inhibitor. However, most antimycobacterial ß-lactams kill Mtb only or best when the bacilli are replicating. Here, a screen of 1904 ß-lactams led to the identification of cephalosporins substituted with a pyrithione moiety at C3' that are active against Mtb under both replicating and nonreplicating conditions, neither activity requiring a ß-lactamase inhibitor. Studies showed that activity against nonreplicating Mtb required the in situ release of the pyrithione, independent of the known class A ß-lactamase, BlaC. In contrast, replicating Mtb could be killed both by released pyrithione and by the parent ß-lactam. Thus, the antimycobacterial activity of pyrithione-containing cephalosporins arises from two mechanisms that kill mycobacteria in different metabolic states.

Opposing reactions in coenzyme A metabolism sensitize Mycobacterium tuberculosis to enzyme inhibition.

Mycobacterium tuberculosis (Mtb) is the leading infectious cause of death in humans. Synthesis of lipids critical for Mtb's cell wall and virulence depends on phosphopantetheinyl transferase (PptT), an enzyme that transfers 4'-phosphopantetheine (Ppt) from coenzyme A (CoA) to diverse acyl carrier proteins. We identified a compound that kills Mtb by binding and partially inhibiting PptT. Killing of Mtb by the compound is potentiated by another enzyme encoded in the same operon, Ppt hydrolase (PptH), that undoes the PptT reaction. Thus, loss-of-function mutants of PptH displayed antimicrobial resistance. Our PptT-inhibitor cocrystal structure may aid further development of antimycobacterial agents against this long-sought target. The opposing reactions of PptT and PptH uncover a regulatory pathway in CoA physiology.

Differentially Detectable Mycobacterium tuberculosis Cells in Sputum from Treatment-Naive Subjects in Haiti and Their Proportionate Increase after Initiation of Treatment.

Recent reports indicate that the sputum of 80% or more of treatment-naive subjects with tuberculosis recruited in England or South Africa contained more viable Mycobacterium tuberculosis cells detected by limiting dilution (LD) in liquid culture than detected as CFU. Efforts to generate such differentially detectable (DD) M. tuberculosis populations in vitro have been difficult to reproduce, and the LD assay is prone to artifact. Here, we applied a stringent version of the LD assay to sputum from 33 treatment-naive, HIV-negative Haitian subjects with drug-sensitive tuberculosis (TB) and to a second sputum sample after two weeks of standard treatment with isoniazid, rifampin, pyrazinamide, and ethambutol (HRZE) for 13 of these subjects. Twenty-one percent had statistically defined levels of DD M. tuberculosis in their pretreatment sputum at an average proportional excess over CFU of 3-fold. Sixty-nine percent of those who received HRZE had statistically defined levels of DD M. tuberculosis in their sputum, and of these, the mean proportionate excess over CFU was 7.9-fold. Thus, DD M. tuberculosis is detectable in pretreatment sputum from a significant proportion of subjects in the Western Hemisphere, and certain drugs or drug regimens, while reducing CFU, may at the same time increase the proportional representation of DD M. tuberculosis among the surviving bacilli. Monitoring DD M. tuberculosis may improve our ability to predict the efficacy of efforts to shorten treatment.IMPORTANCE Measurement of the reduction in CFU in sputum of patients with TB up to 2 weeks after the initiation of treatment is the gateway test for a new TB treatment. Reports have suggested that CFU assays fail to detect the majority of viable M. tuberculosis cells in sputum samples from the majority of patients when the number of M. tuberculosis is estimated by limiting dilution (LD). In an effort to avoid potential methodologic confounders, we applied a modified version of the LD assay in a study of a geographically distinct population. We confirmed that differentially detectable (DD) M. tuberculosis is often found before treatment, albeit at lower proportionate levels than in earlier reports. Strikingly, the prevalence and proportionate representation of DD M. tuberculosis increased during standard treatment. Sublethal exposure to certain antibiotics may help generate DD M. tuberculosis cells or enrich their representation among the surviving bacteria, and this may contribute to the need for prolonged treatment with those agents in order to achieve durable cures.