Automatic patient-level recognition of four Plasmodium species on thin blood smear by a real-time detection transformer (RT-DETR) object detection algorithm: a proof-of-concept and evaluation.

Malaria remains a global health problem, with 247 million cases and 619,000 deaths in 2021. Diagnosis of Plasmodium species is important for administering the appropriate treatment. The gold-standard diagnosis for accurate species identification remains the thin blood smear. Nevertheless, this method is time-consuming and requires highly skilled and trained microscopists. To overcome these issues, new diagnostic tools based on deep learning are emerging. This study aimed to evaluate the performances of a real-time detection transformer (RT-DETR) object detection algorithm to discriminate Plasmodium species on thin blood smear images. The algorithm was trained and validated on a data set consisting in 24,720 images from 475 thin blood smears corresponding to 2,002,597 labels. Performances were calculated with a test data set of 4,508 images from 170 smears corresponding to 358,825 labels coming from six French university hospitals. At the patient level, the RT-DETR algorithm exhibited an overall accuracy of 79.4% (135/170) with a recall of 74% (40/54) and 81.9% (95/116) for negative and positive smears, respectively. Among Plasmodium-positive smears, the global accuracy was 82.7% (91/110) with a recall of 90% (38/42), 81.8% (18/22), and 76.1% (35/46) for P. falciparum, P. malariae, and P. ovale/vivax, respectively. The RT-DETR model achieved a World Health Organization (WHO) competence level 2 for species identification. Besides, the RT-DETR algorithm may be run in real-time on low-cost devices such as a smartphone and could be suitable for deployment in low-resource setting areas lacking microscopy experts.IMPORTANCEMalaria remains a global health problem, with 247 million cases and 619,000 deaths in 2021. Diagnosis of Plasmodium species is important for administering the appropriate treatment. The gold-standard diagnosis for accurate species identification remains the thin blood smear. Nevertheless, this method is time-consuming and requires highly skilled and trained microscopists. To overcome these issues, new diagnostic tools based on deep learning are emerging. This study aimed to evaluate the performances of a real-time detection transformer (RT-DETR) object detection algorithm to discriminate Plasmodium species on thin blood smear images. Performances were calculated with a test data set of 4,508 images from 170 smears coming from six French university hospitals. The RT-DETR model achieved a World Health Organization (WHO) competence level 2 for species identification. Besides, the RT-DETR algorithm may be run in real-time on low-cost devices and could be suitable for deployment in low-resource setting areas.

Which sub-compartments of fat mass and fat-free mass are related to blood viscosity factors?

The size of body compartments is a determinant of several factors of blood viscosity. Red cell aggregation is proportional to fat mass while hematocrit is proportional to both fat-free mass and abdominal adiposity, but which parts of these body components are involved in this relationship is not known. Segmental bioelectrical impedance analysis (sBIA) provides a possibility to delineate the relationships more precisely between various subdivisions of the body and blood viscosity factors, going farther than preceding studies using non segmental BIA. In this study we investigated in 38 subjects undergoing a standardized breakfast test with mathematical modelling of glucose homeostasis and a segmental bioelectrical impedance analysis (sBIA) the relationships between the various compartments of the body and viscosity factors. Blood and plasma viscosity were measured with the Anton Paar rheometer and analyzed with Quemada's model. The parameters better correlated to hematocrit are fat free mass (r = 0.562) and its two components muscle mass (r = 0.516) and non-muscular fat-free mass (r = 0.452), and also trunk fat mass (r = 0.383) and waist-to hip ratio (r = 0.394). Red cell aggregation measurements were correlated with both truncal and appendicular fat mass (r ranging between 0.603 and 0.728). Weaker correlations of M and M1 are found with waist circumference and hip circumference. This study shows that the correlation between lean mass and hematocrit involves both muscle and non-muscle moieties of lean mass, and that both central and appendicular fat are determinants of red cell aggregation.

Liquid chromatography-tandem mass spectrometric assay for the quantification of CDK4/6 inhibitors in human plasma in a clinical context of drug-drug interaction.

The CDK4/6 inhibitors palbociclib and ribociclib are kinase inhibitors used in association with hormonal therapy for the management of patients with metastatic breast cancer. Like most kinase inhibitors, therapeutic drug monitoring may be used for personalize their dosage. To this aim, we developed and validated a sensitive and specific HPLC-MS/MS method for palbociclib and ribociclib quantification in blood samples. We then quantified exposure to palbociclib (plasma trough concentration; Ctrough) in a real-life cohort of patients with locally invasive or metastatic breast cancer (n = 18) at day 15 of the first cycle of palbociclib treatment to characterize palbociclib concentration at steady state (Clinicaltrials.gov identifier NCT04025541, IdRCB n° 2018-A00064-51, 03/07/2018). The geometric mean (± standard deviation [min-max]) of palbociclib plasma Ctrough was 88.58 ng/mL (± 26.4 [46.5 ng/mL - 133 ng/mL]) at day 15. Some covariates, such as drug-drug interactions, could explain the concentration variations observed in our Caucasian cohort. These first results in real-life settings obtained with our HPLC-MS/MS method give important information on palbociclib monitoring and pharmacokinetic variability.

Shear-dependency of the predicted ideal hematocrit.

 The ideal hematocrit is the hematocrit (Hct) value resulting in the highest value of Hct/viscosity (h/η) ratio and can thus be predicted from viscometric measurements with the use of equations such as Quemada's one which yield the determination of the bell-shaped curve of h/η as a function of Hct. In a series of recent papers we applied this approach to various populations, using viscometry at high shear rate (1000 s-1). However the shape of this curve has been reported to be dependent on the shear rate, resulting in a right-shift in this top value when Hct increase. We present here in 11 young recreative athletes the evolution of the predicted top of the h/η curve and optimal theoretical Hct and the discrepancy between theoretical and optimal values over the range of shear rates 1 to 6000 s-1. Results show that the predicted optimal value of both h/η and Hct increases when shear rate increases and that the discrepancy between predicted laquooptimalraquo and actual values decreases and becomes almost asymptotic at very high shear (500 s-1). It is minimal at 2720 s-1. The correlation between predicted laquooptimalraquo and actual values of both parameters describes the same evolution. Therefore, it is better for assessing h/η and its agreement with theoretical values, and for determining the theoretical ideal hematocrit, to measure blood viscosity at shear rates equal or superior to 500 s-1.

Fetal growth retardation and hemorheological predictors of oxygen delivery in hypertensive vs normotensive pregnant women.

Physiological modifications of blood rheology during pregnancy and their alterations in pregnant hypertensive women have been extensively studied in the 1980's. Since vascular resistance is higher in hypertensive pregnant women whose newborns are small-for gestational-age (SGA), we investigated in a personal database if growth retardation of newborns is related to the oxygen delivery index (ratio hematocrit/blood viscosity) and to the difference between hematocrit (Hct) and the prediction of its optimal valued based on Quemada's equation. A sample of 38 hypertensive pregnant women (age 29 yr±1) was compared with 64 controls matched for age and gestational age, studied at 35±1 weeks gestation, extracted from a larger series of 162 pregnant women. On the whole the hypertensive group gave birth to smaller children (p = 0.014). Plasma viscosity correlated with blood pressure (BP) only in hypertensive women (r = 0.403 p < 0.05). The bell-shaped curve of predicted optimal Hct of non hypertensive pregnant women was similar to that of non-pregnant women, but in hypertensive women it was shifted toward higher values (p = 0.07), and the predicted optimal Hct (but not the actual one) was correlated with systolic blood pressure (SBP) (r = 0.349 p < 0.001) and diastolic blood pressure (DBP) (r = 0.218 p < 0.05). The predicted optimal Hct/viscosity (h/η) ratio was higher in hypertensive women whose newborns exhibited a low birth weight (p = 0.03), resulting in a higher discrepancy between actual and model-predicted «ideal¼ values of h/η ratio (p = 0.03) and Hct (p = 0.02) compared with the subgroup with no growth retardation. Therefore, in hypertensive women whose newborns exhibited a low birth weight, hemorheological parameters predicting oxygen supply are shifted to lower values than predicted by the model.

Leg electrical resistance predicts venous blood viscosity and hematocrit.

 We previously reported that whole body bioelectrical impedance analysis (BIA) measurements are correlated to some hemorheologic factors, suggesting a relationship between viscosity factors and electric properties of flowing blood not only in vitro but also in vivo. Recently we reported that with segmental BIA (analyzing the body considered as composed of 5 cylinders) predictive equations for various determinants of blood viscosity were closer than for the wole body. Another widely used BIA technique uses leg-to-leg impedance measurements so that two cylinders (the two legs) are analyzed. We investigated whether impedance measured with this technique (Tanita TBF-300) is also a predictor of blood viscosity factors. From viscometric measurements performed on venous blood drawn in recreative athletes over the range of shear rates 1 to 6000 s-1 (RHEOMETRE Anton Paar CP 50-1), we found a correlation between leg-leg resistance at 50 kHz (Rx[50 kHz]) and blood viscosity at 1000 s-1 (η1000= 0.0051 Rx[50 kHz] + 1.3265; r = 0.521 p = 0.028 yielding a prediction of η1000 (Bland Altman plot: bias 0.05 [RANGE - 0.24; 0.34]. Neither plasma viscosity nor the red cell rheology index «k¼ of Quemada's model are correlated with Rx[50 kHz], but hematocrit (Hct) does (Hct (%) = 0.0217 Rx[50 kHz] + 33.783; r = 0.480 p = 0.044) yielding a prediction of Hct (Bland Altman plot: bias - 0.11, [range - 1.67; 1.45]. The discrepancy between actual and predicted Hct is also correlated with resistance at 50 kHz (r = 0.575 p = 0.031) as does the discrepancy between actual and predicted Hct/viscosity ratio (r = -0.651 p = 0.006). Therefore, as other previously studied methods, leg to leg BIA predicts viscosity, suggesting that blood rheology may influence the passage of an electric current in the legs.

Liquid chromatography-electrospray ionization-tandem mass spectrometry method for quantitative estimation of new imiqualine leads with potent anticancer activities in rat and mouse plasma. Application to a pharmacokinetic study in mice.

Imidazoquinoxaline derivatives (imiqualines) are a new series of anticancer compounds. Two lead compounds (EAPB0203 and EAPB0503) with remarkable in vitro and in vivo activity on melanoma and T-cell lymphomas have been previously identified. The modulation of the chemical structure of the most active compound, EAPB0503, has led to the synthesis of two compounds, EAPB02302 and EAPB02303, 7 and 40 times more active than EAPB0503 against A375 human melanoma cancer cell line, respectively. The aim of this study was to develop and validate a sensitive and accurate liquid chromatography-electrospray ionization-tandem mass spectrometry method to simultaneously quantify EAPB02303 and its potential active metabolite, EAPB02302, in rat and mouse plasma. Analytes were detected in multiple reaction monitoring acquisition mode using an electrospray ionization detector in positive ion mode. Following a liquid-liquid extraction with ethyl acetate, analytes and internal standard were separated by HPLC reversed-phase on a C18 RP18 Nucleoshell column (2.7µm, 4.6×100mm). The method was validated according to FDA and EMA Bioanalytical Method Validation guidelines. The robustness of the method was assessed by introducing small variations in nine nominal analytical parameters. Statistical interpretation was performed by mean of the Student's t-test. Standard curves were generated via unweighted quadratic regression of calibrators (EAPB02303: 1.95-1000ng/mL, EAPB02302: 7.81-1000ng/mL in rat plasma; EAPB02303: 0.98-1000ng/mL, EAPB02302: 1.95-1000ng/mL in mouse plasma). From the analysis of QC samples, intra- and inter-assay precision and accuracy studies demonstrated %R.S.Ds. <12.5% and percent deviation from nominal concentration <7%. Matrix effects (mean matrix factors from 91.8-108.5% in rat plasma; and from 90.4-102.4% in mouse plasma) and stability assays (recoveries >87%) were acceptable and in accordance with the guidelines. No quantifiable carryover effect was observed. The LLOQs were 1.95ng/mL for EAPB02303 and 7.81ng/mL for EAPB02302 in rat plasma, and 0.98ng/mL and 1.95ng/mL for the two compounds in mouse plasma, respectively. This method was successfully implemented to support a mouse pharmacokinetic study following a single intraperitoneal administration of EAPB02303 in male C57Bl/6 mice. The obtained pharmacokinetic parameters of EAPB02303 would be useful to optimize the dosing and the rhythm of administration for subsequent preclinical in vivo activity studies.

Development and validation of a liquid chromatography-electrospray ionization-tandem mass spectrometry method for the determination of urolithin C in rat plasma and its application to a pharmacokinetic study.

Urolithins are microflora human metabolites of dietary ellagic acid derivatives. There is now a growing interest in the biological activities of these compounds. Several studies suggest that urolithins have potential antioxidant, anti-inflammatory, anticancer and anti-glycative activities. Recently, our group investigated the role of urolithins as potential anti-diabetic treatments; among the four urolithins, urolithin C was the most promising compound. The purpose of this paper was to develop a rapid, sensitive and specific liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS/MS) method for the determination of urolithin C in rat plasma. To date, no method is reported for the quantification of urolithin C in any of the matrices. Plasma samples were extracted with ethyl acetate. Urolithin D was selected as the internal standard. The separation was carried out on a C18 Kinetex EVO column (2.1mm×150mm, 2.6µm) using a mobile phase of acetonitrile-1% aqueous formic acid solution (30:70, v/v). A triple quadrupole mass spectrometer in the negative ion mode was used for the determination of the target analyte. The monitored ion transitions were m/z 243→187 for urolithin C and m/z 259→213 for the internal standard. The calibration curve range was 4.95-1085µg/L (r2>0.994). The intra- and inter-day precisions were less than 10%; accuracies ranged from 96.6 to 109%. The mean extraction recovery of urolithins C and D was greater than 91%. No significant matrix effects and no carryover effects were observed. Small changes in LC-ESI-MS/MS conditions did not have significant effect on the determination of urolithin C. Stability tests under various conditions were also investigated. This highly specific and sensitive method was used to analyze samples collected during preclinical pharmacokinetic studies in rats. Glucuronyl and sulfate conjugates of urolithin C were the main metabolites detected in plasma.

Cytotoxicity micropollutant removal in a crossflow membrane bioreactor.

The application of membrane bioreactor (MBR) technology was investigated with the aim of evaluating its potential for cytostatic drug and cytotoxicity bioremoval. The toxicity removal was assessed from biomarker test. CP removal of up to 80% was achieved under the operating conditions studied (HRT of 48 h and a SRT of 50 days). The increase of TMP was associated with an increase of supernatant toxicity as if fouling led to retention of the toxicity. Peaks of supernatant cytotoxicity were correlated with peaks in supernatant humic acid contents. It may suggest that molecules with a toxic effect may be adsorbed or entrapped in humic acids substances. Our study then points out that advances in wastewater treatment using an MBR can provide a suitable process for lowering CP concentrations before discharge into the aqueous environment. However, a tertiary treatment is necessary if complete elimination of toxicity is targeted.

Detection of hemoglobin-based oxygen carriers in human serum for doping analysis: confirmation by size-exclusion HPLC.

BACKGROUND: Hemoglobin-based oxygen carriers (HBOCs) are being developed as potential substitutes for the oxygen-carrying functions of erythrocytes, but athletes may obtain and experiment with HBOCs as an illicit means of enhancing oxygen transport. An electrophoretic technique has been developed to screen for the presence of HBOCs in blood samples (Lasne et al. Clin Chem 2004;50:410-5). Interest has focused on complementary methods that can provide legally defensible scientific evidence for the presence of HBOCs in blood samples collected for doping control. METHODS: The aim of this research was to develop a size-exclusion SEC-HPLC technique to identify in plasma or serum samples the presence of HBOCs that are currently under development. This method was also used to detect a polymerized bovine hemoglobin (Hemopure) after infusion in 12 healthy males. RESULTS: The chromatograms of all HBOCs tested were clearly separated from the 54-min peak associated with human hemoglobin dimers. It was possible to differentiate between the different HBOC products based solely on their chromatographic profiles, provided they were at high concentrations. Differences were discernible not only based on the presence (or absence) of peaks, but also the separation between respective peaks. The profiles for serum samples collected from the men immediately after infusion of Hemopure showed a distinctive profile. The shape of the chromatographic profile remained consistent for at least 48 h. CONCLUSIONS: Under the analytical conditions reported here, SEC-HPLC was able to separate native hemoglobin from the modified hemoglobin molecules present in each of the HBOC products studied. In tandem with electrophoretic screening, SEC-HPLC provides evidence of the presence of HBOCs and can therefore be regarded as a method that satisfies the criteria for use in an antidoping control setting.