Relationship between conventional culture and flow cytometry for the diagnosis of urinary tract infection

A fully interpretable machine learning model for increasing the effectiveness of urine screening

OBJECTIVES

This article addresses the need for effective screening methods to identify negative urine samples before urine culture, reducing the workload, cost, and release time of results in the microbiology laboratory. We try to overcome the limitations of current solutions, which are either too simple, limiting effectiveness (1 or 2 parameters), or too complex, limiting interpretation, trust, and real-world implementation ("black box" machine learning models).

METHODS

The study analyzed 15,312 samples from 10,534 patients with clinical features and the Sysmex Uf-1000i automated analyzer data. Decision tree (DT) models with or without lookahead strategy were used, as they offer a transparent set of logical rules that can be easily understood by medical professionals and implemented into automated analyzers.

RESULTS

The best model achieved a sensitivity of 94.5% and classified negative samples based on age, bacteria, mucus, and 2 scattering parameters. The model reduced the workload by an additional 16% compared to the current procedure in the laboratory, with an estimated financial impact of €40,000/y considering 15,000 samples/y. Identified logical rules have a scientific rationale matched to existing knowledge in the literature.

CONCLUSIONS

Overall, this study provides an effective and interpretable screening method for urine culture in microbiology laboratories, using data from the Sysmex UF-1000i automated analyzer. Unlike other machine learning models, our model is interpretable, generating trust and enabling real-world implementation.

Prediction of the bacterial shape in urinary tract infections with the Sysmex UF-1000i analyser: technical note

Background

The aim of our study was to explore the utility of the Sysmex UF-1000i analyzer as a rapid screening tool for urinary tract infections (UTI) and its ability to predict bacterial shape in order to help physicians choose the appropriate empiric treatment.

Materials and methods

This is a retrospective study, including 1023 urine cytobacteriological examinations. Urines were processed according to the recommendations of the medical microbiology reference system (REMIC). Using the Sysmex Uf-1000i analyzer, the authors evaluated bacteria forward scatter (B_FSC) and fluorescent light scatter (B_FLH) in a preliminary discrimination step for UTI caused by bacilli or cocci bacteria.

Results

The authors got 1023 positive samples. Comparing baccili and cocci bacteria, the authors observed a statistically significant difference for B_FSC but not for B_FLH. The values of B_FLH are very close for the four categories of microorganisms compared (bacilli, cocci, bacilli-cocci association, and yeasts). For these same categories, tests show different values for the B_FSC. A separate analysis of the B_FSC values for bacilli shows that their distribution is relatively homogeneous and exhibits a peak between 20 and 30 ch.

Conclusion

Dimensional parameters of bacteria generated by UF-1000i could be a rapid and useful tool for predicting the bacterial shape causing UTI.

A novel approach to screening and managing the urinary tract infections suspected sample in the general human population

Background

Automated urine technology providing standard urinalysis data can be used to support clinicians in screening and managing a UTI-suspected sample. Fully automated urinalysis systems have expanded in laboratory practice. Commonly used were devices based on digital imaging with automatic particle recognition, which expresses urinary sediment results on an ordinal scale. There were introduced fluorescent flow cytometry analyzers reporting all parameters quantitatively. There is a need to harmonize the result and support comparing bacteria and WBC qualitative versus semiquantitative results.

Methods

A total of 1,131 urine samples were analyzed on both automated urinalysis systems. The chemical components of urinalysis (leukocyte esterase and nitrate reductase) and the sediment results (leukocytes and bacteria) were investigated as potential UTI indicators. Additionally, 106 specimens were analyzed on UF-5000 and compared with culture plating to establish cut-offs that can be suitable for standard urinalysis requirements and help to guide on how to interpret urinalysis results in the context of cultivation reflex.

Results

The medians of bacteria counts varies from 16.2 (absence), 43.0 (trace), 443.5 (few), 5,389.2 (moderate), 19,356.6 (many) to 32,545.2 (massive) for particular digital microscopic bacteriuria thresholds. For pyuria thresholds, the medians of WBC counts varies from 0.8 (absence), 2.0 (0-1), 7.7 (2-3), 21.3 (4-6), 38.9 (7-10), 61.3 (11-15) to 242.2 (>30). Comparing the culture and FFC data (bacterial and/or WBC counts) was performed. Satisfactory sensitivity (100%), specificity (83.7%), negative predictive value (100%), and positive predictive value (75%) were obtained using indicators with the following cut-off values: leukocytes ≥40/µl or bacteria ≥300/µl.

Conclusions

Accurate urinalysis gives information about the count of bacteria and leukocytes as useful indicators in UTIs, in general practice it can be a future tool to cross-link clinical and microbiology laboratories. However, the cut-off adjustments require individual optimization.

Rapid Bacterial Detection in Urine Using Laser Scattering and Deep Learning Analysis

Images of laser scattering patterns generated by bacteria in urine are promising resources for deep learning. However, floating bacteria in urine produce dynamic scattering patterns and require deep learning of spatial and temporal features. We hypothesized that bacteria with variable bacterial densities and different Gram staining reactions would generate different speckle images. After deep learning of speckle patterns generated by various densities of bacteria in artificial urine, we validated the model in an independent set of clinical urine samples in a tertiary hospital. Even at a low bacterial density cutoff (1,000 CFU/mL), the model achieved a predictive accuracy of 90.9% for positive urine culture. At a cutoff of 50,000 CFU/mL, it showed a better accuracy of 98.5%. The model achieved satisfactory accuracy at both cutoff levels for predicting the Gram staining reaction. Considering only 30 min of analysis, our method appears as a new screening tool for predicting the presence of bacteria before urine culture. IMPORTANCE This study performed deep learning of multiple laser scattering patterns by the bacteria in urine to predict positive urine culture. Conventional urine analyzers have limited performance in identifying bacteria in urine. This novel method showed a satisfactory accuracy taking only 30 min of analysis without conventional urine culture. It was also developed to predict the Gram staining reaction of the bacteria. It can be used as a standalone screening tool for urinary tract infection.

FlowUTI: An interactive web-application for optimizing the use of flow cytometry as a screening tool in urinary tract infections

Due to the high prevalence of patients attending with urinary tract infection (UTI) symptoms, the use of flow-cytometry as a rapid screening tool to avoid unnecessary cultures is becoming a widely used system in clinical practice. However, the recommended cut-points applied in flow-cytometry systems differ substantially among authors, making it difficult to obtain reliable conclusions. Here, we present FlowUTI, a shiny web-application created to establish optimal cut-off values in flow-cytometry for different UTI markers, such as bacterial or leukocyte counts, in urine from patients with UTI symptoms. This application provides a user-friendly graphical interface to perform robust statistical analysis without a specific training. Two datasets are analyzed in this manuscript: one composed of 204 urine samples from neonates and infants (≤3 months old) attended in the emergency department with suspected UTI; and the second dataset including 1174 urines samples from an elderly population attended at the primary care level. The source code is available on GitHub (https://github.com/GuillermoMG-HUVR/Microbiology-applications/tree/FlowUTI/FlowUTI). The web application can be executed locally from the R console. Alternatively, it can be freely accessed at https://covidiario.shinyapps.io/flowuti/. FlowUTI provides an easy-to-use environment for evaluating the efficiency of the urinary screening process with flow-cytometry, reducing the computational burden associated with this kind of analysis.

UF-5000 flow cytometer: A new technology to support microbiologists' interpretation of suspected urinary tract infections

This case study aims to describe the adoption of an innovative flow cytometer (i.e., UF-5000), which can support the microbiologists' process of diagnosing suspected urinary tract infections (UTIs). The new clinical information provided can be used to improve the identification of both contamination and colonization, thus reducing inappropriate antibiotic prescriptions. In July and August 2017, the Microbiology Laboratory of Alessandria (Italy) conducted a retrospective monocentric study analyzing data about 1,295 urine specimens from inpatients and outpatients with symptoms of UTIs. The results of this study show that the innovative technology can successfully support the diagnostic process in microbiology laboratories and, consequently, the supply of sustainable treatments by hospitals.

Cellular immune responses in amniotic fluid of women with preterm labor and intra-amniotic infection or intra-amniotic inflammation

PROBLEM

Preterm birth is commonly preceded by preterm labor, a syndrome that is causally linked to both intra-amniotic infection and intra-amniotic inflammation. However, the stereotypical cellular immune responses in these two clinical conditions are poorly understood.

METHOD OF STUDY

Amniotic fluid samples (n = 26) were collected from women diagnosed with preterm labor and intra-amniotic infection (amniotic fluid IL-6 concentrations ≥2.6 ng/mL and culturable microorganisms, n = 10) or intra-amniotic inflammation (amniotic fluid IL-6 concentrations ≥2.6 ng/mL without culturable microorganisms, n = 16). Flow cytometry was performed to evaluate the phenotype and number of amniotic fluid leukocytes. Amniotic fluid concentrations of classical pro-inflammatory cytokines, type 1 and type 2 cytokines, and T-cell chemokines were determined using immunoassays.

RESULTS

Women with spontaneous preterm labor and intra-amniotic infection had (a) a greater number of total leukocytes, including neutrophils and monocytes/macrophages, in amniotic fluid; (b) a higher number of total T cells and CD4⁺ T cells, but not CD8⁺ T cells or B cells, in amniotic fluid; and (c) increased amniotic fluid concentrations of IL-6, IL-1β, and IL-10, compared to those with intra-amniotic inflammation. However, no differences in amniotic fluid concentrations of T-cell cytokines and chemokines were observed between these two clinical conditions.

CONCLUSION

The cellular immune responses observed in women with preterm labor and intra-amniotic infection are more severe than in those with intra-amniotic inflammation, and neutrophils, monocytes/macrophages, and CD4⁺ T cells are the main immune cells responding to microorganisms that invade the amniotic cavity. These findings provide insights into the intra-amniotic immune mechanisms underlying the human syndrome of preterm labor.

Progress in Automated Urinalysis

New technological advances have paved the way for significant progress in automated urinalysis. Quantitative reading of urinary test strips using reflectometry has become possible, while complementary metal oxide semiconductor (CMOS) technology has enhanced analytical sensitivity and shown promise in microalbuminuria testing. Microscopy-based urine particle analysis has greatly progressed over the past decades, enabling high throughput in clinical laboratories. Urinary flow cytometry is an alternative for automated microscopy, and more thorough analysis of flow cytometric data has enabled rapid differentiation of urinary microorganisms. Integration of dilution parameters (e.g., creatinine, specific gravity, and conductivity) in urine test strip readers and urine particle flow cytometers enables correction for urinary dilution, which improves result interpretation. Automated urinalysis can be used for urinary tract screening and for diagnosing and monitoring a broad variety of nephrological and urological conditions; newer applications show promising results for early detection of urothelial cancer. Concomitantly, the introduction of matrix-assisted laser desorption ionization-time-of-flight mass spectrometry (MALDI-TOF MS) has enabled fast identification of urinary pathogens. Automation and workflow simplification have led to mechanical integration of test strip readers and particle analysis in urinalysis. As the information obtained by urinalysis is complex, the introduction of expert systems may further reduce analytical errors and improve the quality of sediment and test strip analysis. With the introduction of laboratory-on-a-chip approaches and the use of microfluidics, new affordable applications for quantitative urinalysis and readout on cell phones may become available. In this review, we present the main recent developments in automated urinalysis and future perspectives.

Automated Flow Cytometry: An Alternative to Urine Culture in a Routine Clinical Microbiology Laboratory?

The urine culture is the "gold standard" for the diagnosis of urinary tract infections (UTI) but constitutes a significant workload in the routine clinical laboratory. Due to the high percentage of negative results, there is a need for an efficient screening method, with a high negative predictive value (NPV) that could reduce the number of unnecessary culture tests. With the purpose of improving the efficiency of laboratory work, several methods for screening out the culture-negative samples have been developed, but none of them has shown adequate sensitivity (SE) and high NPV. Many authors show data about the efficacy of flow cytometry in the routine clinical laboratory. The aim of this article is to review and discuss the current literature on the feasibility of urine flow cytometry (UFC) and its utility as an alternative analytical technique in urinalysis.