Inter-assay diagnostic accuracy of cerebrospinal fluid kappa free light chains for the diagnosis of multiple sclerosis.

Background: Cerebrospinal fluid (CSF) kappa free light chain (κFLC) measures gained increasing interest as diagnostic markers in multiple sclerosis (MS). However, the lack of studies comparing assay-dependent diagnostic cutoff values hinders their use in clinical practice. Additionally, the optimal κFLC parameter for identifying MS remains a subject of ongoing debate. Objectives: The aim of this study was to compare same-sample diagnostic accuracies of the κFLC index, κIgG index, CSF κFLC/IgG ratio, and isolated CSF κFLC (iCSF-κFLC) between two reference centers using different methods. Methods: Paired serum and CSF samples were analyzed for κFLC and albumin concentrations by Freelite®-Optilite (Sint-Jan Bruges hospital) and N Latex®-BNII (Ghent University hospital). Diagnostic performance to differentiate MS from controls was assessed using ROC curve analysis. Results: A total of 263 participants were included (MS, n = 80). Optimal diagnostic cutoff values for the κFLC index (Freelite®-Optilite: 7.7; N Latex®-BNII: 4.71), κIgG index (Freelite®-Optilite: 14.15, N Latex®-BNII: 12.19), and CSF κFLC/IgG ratio (Freelite®-Optilite: 2.27; N Latex®-BNII: 1.44) differed between the two methods. Sensitivities related to optimal cutoff values were 89.9% (Freelite®-Optilite) versus 94.6% (N Latex®-BNII) for the κFLC index, 91% (Freelite®-Optilite) versus 92.2% (N Latex®-BNII) for the κIgG index, and 81.3% (Freelite®-Optilite) versus 91.4% (N Latex®-BNII) for the CSF κFLC/IgG ratio. However, for iCSF-κFLC, optimal diagnostic cutoff values (0.36 mg/L) and related specificities (81.8%) were identical with a related diagnostic sensitivity of 89.9% for Freelite®-Optilite and 90.5% for N Latex®-BNII. The diagnostic performance of the κFLC index [area under the curve (AUC) Freelite®-Optilite: 0.924; N Latex®-BNII: 0.962] and κIgG index (AUC Freelite®-Optilite: 0.929; N Latex®-BNII: 0.961) was superior compared to CSF oligoclonal bands (AUC: 0.898, sensitivity: 83.8%, specificity: 95.9%). Conclusions: The κFLC index and the κIgG index seem to be excellent markers for identifying MS, irrespective of the method used for κFLC quantification. Based on the AUC, they appear to be the measures of choice. For all measures, optimal cutoff values differed between methods except for iCSF-κFLC. iCSF-κFLC might therefore serve as a method-independent, more cost-efficient, initial screening measure for MS. These findings are particularly relevant for clinical practice given the potential future implementation of intrathecal κFLC synthesis in MS diagnostic criteria and for future multicentre studies pooling data on κFLC measures.

Applications of Artificial Intelligence in Urinalysis: Is the Future Already Here?

BACKGROUND: Artificial intelligence (AI) has emerged as a promising and transformative tool in the field of urinalysis, offering substantial potential for advancements in disease diagnosis and the development of predictive models for monitoring medical treatment responses. CONTENT: Through an extensive examination of relevant literature, this narrative review illustrates the significance and applicability of AI models across the diverse application area of urinalysis. It encompasses automated urine test strip and sediment analysis, urinary tract infection screening, and the interpretation of complex biochemical signatures in urine, including the utilization of cutting-edge techniques such as mass spectrometry and molecular-based profiles. SUMMARY: Retrospective studies consistently demonstrate good performance of AI models in urinalysis, showcasing their potential to revolutionize clinical practice. However, to comprehensively evaluate the real clinical value and efficacy of AI models, large-scale prospective studies are essential. Such studies hold the potential to enhance diagnostic accuracy, improve patient outcomes, and optimize medical treatment strategies. By bridging the gap between research and clinical implementation, AI can reshape the landscape of urinalysis, paving the way for more personalized and effective patient care.

Comparison of coagulation monitoring using ROTEM and Sonoclot devices in cardiac surgery: a single-center prospective observational study.

BACKGROUND: Viscoelastic tests (VETs) are recommended during cardiac surgery to monitor coagulation status and guide transfusion. We compared the results of two VETs, the Sonoclot Analyzer and the ROTEM Sigma. Agreement between viscoelastic tests' subdiagnoses and overall diagnosis severity was assessed. Correlations with conventional coagulation tests (CCT) and the discriminatory potential of numerical VET outputs for transfusion thresholds was determined. METHODS: Single-center, prospective observational study in a tertiary academic center. In fifty adult patients undergoing elective cardiac surgery, parallel Sonoclot, ROTEM and CCT analysis was performed before heparin, or after protamine or coagulation product administration. All patients completed the study, resulting in 139 data points. RESULTS: Agreement on the severity of coagulation disorders was acceptable (83%), but poor (27%) on the differentiation of the underlying causes. Correlations between ROTEM parameters and CCT were good (postprotamine: FIBTEM A5 (r2=0.90 vs. fibrinogen) and EXTEM-FIBTEM A5 difference (r2=0.81 vs. platelet count). Sonoclot correlated less (Clot Rate (r2=0.25 vs. fibrinogen) and Platelet Function (r2=0.43 vs. platelet count). This was reflected in the discriminatory potential of these parameters as found by linear mixed modelling. We suggest clinically useful grey zones for VET cutoff interpretation. CONCLUSIONS: ROTEM and Sonoclot accord well on the detection of severity of coagulation dysfunction, but not on the diagnosis of the underlying cause. ROTEM correlated more closely with CCT then Sonoclot. We propose a testing strategy that could lead to a cost-effective approach to the bleeding cardiac surgery patient.

Direct Oral Anticoagulant removal by a DOAC filter: Impact on lupus anticoagulant testing - Evaluation on spiked and patient samples.

Background: DOAC Filter (DF) is a new device to overcome interference in lupus anticoagulant (LAC) testing by direct oral anticoagulants (DOACs). Objectives: We evaluated DOAC removal from plasma and elimination of DOAC interference in LAC testing by DF, and impact of DF on LAC assays in a representative patient cohort, including a comparison with DOAC-Stop (DS). Methods: Normal pooled plasma (NPP) was spiked with increasing concentrations of apixaban, rivaroxaban, edoxaban, and dabigatran. DOAC and LAC was measured on untreated, DF-treated, and DS-treated spiked samples. Coagulation parameters and thrombin generation were measured on patient samples (n = 20) before and after DF. Patients treated with DOAC, vitamin K antagonist, or heparin and nonanticoagulated patient samples (n = 139) were tested for LAC before and after DF. Results: In spiked NPP, levels were below the lower limit of quantification (LLoQ) after DF/DS treatment for all DOAC concentrations. Following DF, levels were below LLoQ for 53 of 56 DOAC-containing patient samples. Twenty-eight of 33 LAC-positive DOAC-containing samples became negative after filtration, whereas 5 remained LAC-positive (1/5 from a patient with antiphospholipid syndrome [APS]). Four LAC-positive DOAC-containing samples (from patients without APS), became negative after filtration, whereas they remained LAC positive after DS. In the non-DOAC patient groups following DF, LAC changed from positive to negative in 8 (due to a procoagulant effect) and vice versa in 2 cases. Conclusion: DF reduces DOAC interference in LAC testing. As incomplete DOAC removal may occur, DOAC measurements should be performed after filtration. A procoagulant effect after filtration may lead to erroneous LAC results in non-DOAC-containing samples. Therefore, using DF should be restricted to DOAC-containing samples.

Thrombin generation measured by two platforms in patients with a bleeding tendency.

BACKGROUND: Mild to moderate bleeding disorders are a diagnostic challenge. Many patients remain undiagnosed despite thorough and repeated laboratory testing. Thrombin generation (TG) is an overall assay measuring the functionality of the hemostatic system and may be a useful tool in diagnosing patients with bleeding tendency. OBJECTIVES: We examined the added value of TG in patients with mild bleeding tendency with and without diagnosis after classical laboratory testing. Further, we investigated the role of different expressions of results, between-method variation, and reference ranges. METHODS: TG of patients and controls was measured in parallel by two TG platforms (ST Genesia and calibrated automated thrombogram [CAT]). All TG parameters in patient and control groups were compared by statistical analysis (Mann-Whitney U tests) including visual representation with box-and-whisker plots. Results were expressed as normalized ratios (ST Genesia and CAT) or corrected values (ST Genesia). Reference intervals were calculated to which patient results were compared. We studied lot-to-lot reagent variability for both platforms. RESULTS: In 62.7% (ST Genesia) to 69.5% (CAT) of patients undiagnosed with a traditional laboratory work-up, abnormal TG parameters (lag time and endogenous thrombin potential expressed as normalized ratio and/or corrected value) were detected. In the group of previously diagnosed patients, abnormal parameters were found in 58.1% of patients for both TG assays. No relevant lot-to-lot reagent variability was observed. CONCLUSIONS: Adding TG helps with diagnosing patients with mild bleeding disorder. TG seems a promising tool in diagnosis of bleeding tendency, but further evaluation is necessary before application in diagnostic laboratory testing.

Flow cytometric analysis of platelet function to improve the recognition of thrombocytopathy.

INTRODUCTION: Light transmission aggregometry (LTA) is the gold standard for diagnosing bleeding disorders. Although LTA is laborious, requires large volumes of blood and is relatively insensitive to small changes in platelet function, there is still no competing alternative approach to replace LTA for the diagnosis of platelet bleeding disorders. MATERIALS AND METHODS: This study investigates the correlation between flow cytometry-based whole blood platelet activation test (WB-PACT) and LTA and whether WB-PACT is of additional value for the identification of bleeding disorders. In total, 161 patients with suspected bleeding diathesis were tested. RESULTS: A correlation of 0.41 between LTA and WB-PACT was found, and there was agreement between tests in 62% of cases (κ = 0.23). The WB-PACT is of additional value to LTA to detect platelet function disorders (PFD) as 10 patients with elevated bleeding score (BS) were detected with WB-PACT, 4 with LTA and 7 patients were positive with both tests. Interestingly, in contrast to LTA, WB-PACT has an additional option to detect VWF disfunctions. CONCLUSION: WB-PACT may have added value for the routine diagnostic work-up in patients who need to have platelet function tested.

Direct oral anticoagulant adsorption: Impact on lupus anticoagulant testing-Review of the literature and evaluation on spiked and patient samples.

BACKGROUND: Direct oral anticoagulants (DOACs) interfere with lupus anticoagulant (LAC) testing. DOAC-Stop (D-S) represents a preanalytical strategy to cope with this issue. OBJECTIVES: To assess D-S's ability to remove DOACs from plasma and overcome DOAC interference in LAC assays and to evaluate D-S's applicability in a representative patient cohort with routine LAC request. METHODS: Apixaban (30-933 ng/mL), edoxaban (31-1060 ng/mL), rivaroxaban (35-1020 ng/mL), and dabigatran (20-360 ng/mL) were spiked to normal plasma. Aliquots were treated with D-S or untreated before DOAC and LAC testing. Patient samples containing DOAC (n = 43), vitamin K antagonists (n = 25), heparins (n = 21), or no anticoagulants (n = 63) were tested for LAC before and after D-S. RESULTS: Spiking experiments revealed false-positive LAC from low concentrations of DOACs except for apixaban. Following D-S, DOAC levels were below lower limits of quantification, except for apixaban at the highest concentration, and no false-positive LAC was obtained. DOAC levels were below lower limits of quantification after D-S in 39/43 DOAC-containing patient samples. For 23/29 LAC-positive DOAC-containing samples, LAC tests became negative after D-S, whereas 3/6 samples remaining positive were from patients with (high probability for) antiphospholipid syndrome. In the non-DOAC-treated groups, LAC changed from positive to negative in 10 and vice versa in 2 cases. CONCLUSIONS: D-S limits DOAC interference in LAC assays. DOAC concentration measurement should be performed in D-S treated samples because incomplete removal may occur. Applying D-S to vitamin K antagonist-containing, heparin-containing, or not-anticoagulated samples may lead to erroneous LAC results. Therefore, D-S should only be used in plasma from DOAC-treated patients.

The effect of unfractionated heparin, enoxaparin, and danaparoid on lupus anticoagulant testing: Can activated carbon eliminate false-positive results?

BACKGROUND: Heparins and heparinoids interfere with functional clotting assays used for lupus anticoagulant (LAC) detection. However, current guidelines for LAC testing do not provide clear guidance on this matter. OBJECTIVES: We aimed to assess to effect of unfractionated heparin (UFH), enoxaparin, and danaparoid on LAC assays over broad anti-Xa activity ranges and to evaluate whether activated carbon (AC) is able to neutralize these effects. METHODS: UFH (0.1-3.0 IU/mL), enoxaparin (0.2-2.9 IU/mL), and danaparoid (0.6-2.2 IU/mL) were spiked to normal pooled plasma. AC was added at multiple activity levels. Anti-Xa assays and LAC tests were performed on all samples using Stago analyzers and reagents. RESULTS: Abnormal activated partial thromboplastin time (APTT) screening and mixing tests were obtained at the lowest levels for all compounds. Abnormal APTT confirmation tests were seen from 2.5 and 1.9 anti-Xa IU/mL for enoxaparin and danaparoid, respectively. Abnormal dilute Russell's viper venom test (dRVVT) screening tests were obtained from 1.6, 1.4, and 1.1 anti-Xa IU/mL for UFH, enoxaparin, and danaparoid, respectively. Mixing tests were abnormal from 2.5 and 1.3 anti-Xa IU/mL for enoxaparin and danaparoid, respectively. Abnormal dRVVT confirmation results were seen for danaparoid only from 1.9 anti-Xa IU/mL. AC was unable to neutralize anti-Xa activity in plasma and overcome the effect of the tested anticoagulants on LAC assays but may cause prolongation of APTT clotting times. CONCLUSIONS: UFH, enoxaparin, and danaparoid clearly affected LA tests; however, false-positive LAC conclusions were obtained at supratherapeutic enoxaparin and danaparoid levels only. AC may prolong APTT screen clotting times, requiring 3-step testing to avoid potential misdiagnosis of LAC.

Correction for the Hematocrit Bias in Dried Blood Spot Analysis Using a Nondestructive, Single-Wavelength Reflectance-Based Hematocrit Prediction Method.

The hematocrit (Hct) effect is one of the most important hurdles currently preventing more widespread implementation of quantitative dried blood spot (DBS) analysis in a routine context. Indeed, the Hct may affect both the accuracy of DBS methods as well as the interpretation of DBS-based results. We previously developed a method to determine the Hct of a DBS based on its hemoglobin content using noncontact diffuse reflectance spectroscopy. Despite the ease with which the analysis can be performed (i.e., mere scanning of the DBS) and the good results that were obtained, the method did require a complicated algorithm to derive the total hemoglobin content from the DBS's reflectance spectrum. As the total hemoglobin was calculated as the sum of oxyhemoglobin, methemoglobin, and hemichrome, the three main hemoglobin derivatives formed in DBS upon aging, the reflectance spectrum needed to be unmixed to determine the quantity of each of these derivatives. We now simplified the method by only using the reflectance at a single wavelength, located at a quasi-isosbestic point in the reflectance curve. At this wavelength, assuming 1-to-1 stoichiometry of the aging reaction, the reflectance is insensitive to the hemoglobin degradation and only scales with the total amount of hemoglobin and, hence, the Hct. This simplified method was successfully validated. At each quality control level as well as at the limits of quantitation (i.e., 0.20 and 0.67) bias, intra- and interday imprecision were within 10%. Method reproducibility was excellent based on incurred sample reanalysis and surpassed the reproducibility of the original method. Furthermore, the influence of the volume spotted, the measurement location within the spot, as well as storage time and temperature were evaluated, showing no relevant impact of these parameters. Application to 233 patient samples revealed a good correlation between the Hct determined on whole blood and the predicted Hct determined on venous DBS. The bias obtained with Bland and Altman analysis was -0.015 and the limits of agreement were -0.061 and 0.031, indicating that the simplified, noncontact Hct prediction method even outperforms the original method. In addition, using caffeine as a model compound, it was demonstrated that this simplified Hct prediction method can effectively be used to implement a Hct-dependent correction factor to DBS-based results to alleviate the Hct bias.

Alternative Sampling Strategies for Cytochrome P450 Phenotyping.

Interindividual variability in the expression and function of drug metabolizing cytochrome P (CYP) 450 enzymes, determined by a combination of genetic, non-genetic and environmental parameters, is a major source of variable drug response. Phenotyping by administration of a selective enzyme substrate, followed by the determination of a specific phenotyping metric, is an appropriate approach to assess the in vivo activity of CYP450 enzymes as it takes into account all influencing factors. A phenotyping protocol should be as simple and convenient as possible. Typically, phenotyping metrics are determined in traditional matrices, such as blood, plasma or urine. Several sampling strategies have been proposed as an alternative for these traditional sampling techniques. In this review, we provide a comprehensive overview of available methods using dried blood spots (DBS), hair, oral fluid, exhaled breath and sweat for in vivo CYP450 phenotyping. We discuss the relation between phenotyping metrics measured in these samples and those in conventional matrices, along with the advantages and limitations of the alternative sampling techniques. Reliable phenotyping procedures for several clinically relevant CYP450 enzymes, including CYP1A2, CYP2C19 and CYP2D6, are currently available for oral fluid, breath or DBS, while additional studies are needed for other CYP450 isoforms, such as CYP3A4. The role of hair analysis for this purpose remains to be established. Being non- or minimally invasive, these sampling strategies provide convenient and patient-friendly alternatives for classical phenotyping procedures, which may contribute to the implementation of CYP450 phenotyping in clinical practice.

An optimized and validated SPE-LC-MS/MS method for the determination of caffeine and paraxanthine in hair.

Caffeine is the probe drug of choice to assess the phenotype of the drug metabolizing enzyme CYP1A2. Typically, molar concentration ratios of paraxanthine, caffeine's major metabolite, to its precursor are determined in plasma following administration of a caffeine test dose. The aim of this study was to develop and validate an LC-MS/MS method for the determination of caffeine and paraxanthine in hair. The different steps of a hair extraction procedure were thoroughly optimized. Following a three-step decontamination procedure, caffeine and paraxanthine were extracted from 20 mg of ground hair using a solution of protease type VIII in Tris buffer (pH 7.5). Resulting hair extracts were cleaned up on Strata-X™ SPE cartridges. All samples were analyzed on a Waters Acquity UPLC® system coupled to an AB SCIEX API 4000™ triple quadrupole mass spectrometer. The final method was fully validated based on international guidelines. Linear calibration lines for caffeine and paraxanthine ranged from 20 to 500 pg/mg. Precision (%RSD) and accuracy (%bias) were below 12% and 7%, respectively. The isotopically labeled internal standards compensated for the ion suppression observed for both compounds. Relative matrix effects were below 15%RSD. The recovery of the sample preparation procedure was high (>85%) and reproducible. Caffeine and paraxanthine were stable in hair for at least 644 days. The effect of the hair decontamination procedure was evaluated as well. Finally, the applicability of the developed procedure was demonstrated by determining caffeine and paraxanthine concentrations in hair samples of ten healthy volunteers. The optimized and validated method for determination of caffeine and paraxanthine in hair proved to be reliable and may serve to evaluate the potential of hair analysis for CYP1A2 phenotyping.

Does volumetric absorptive microsampling eliminate the hematocrit bias for caffeine and paraxanthine in dried blood samples? A comparative study.

Volumetric absorptive microsampling (VAMS) is a novel sampling technique that allows the straightforward collection of an accurate volume of blood (approximately 10µL) from a drop or pool of blood by dipping an absorbent polymeric tip into it. The resulting blood microsample is dried and analyzed as a whole. The aim of this study was to evaluate the potential of VAMS to overcome the hematocrit bias, an important issue in the analysis of dried blood microsamples. An LC-MS/MS method for analysis of the model compounds caffeine and paraxanthine in VAMS samples was fully validated and fulfilled all pre-established criteria. In conjunction with previously validated procedures for dried blood spots (DBS) and blood, this allowed us to set up a meticulous comparative study in which both compounds were determined in over 80 corresponding VAMS, DBS and liquid whole blood samples. These originated from authentic human patient samples, covering a wide hematocrit range (0.21-0.50). By calculating the differences with reference whole blood concentrations, we found that analyte concentrations in VAMS samples were not affected by a bias that changed over the evaluated hematocrit range, in contrast to DBS results. However, VAMS concentrations tend to overestimate whole blood concentrations, as a consistent positive bias was observed. A different behavior of VAMS samples prepared from incurred and spiked blood, combined with a somewhat reduced recovery of caffeine and paraxanthine from VAMS tips at high hematocrit values, an effect that was not observed for DBS using a very similar extraction procedure, was found to be at the basis of the observed VAMS-whole blood deviations. Based on this study, being the first in which the validity and robustness of VAMS is evaluated by analyzing incurred human samples, it can be concluded that VAMS effectively assists in eliminating the effect of hematocrit.

Paraxanthine/Caffeine Concentration Ratios in Hair: An Alternative for Plasma-Based Phenotyping of Cytochrome P450 1A2?

BACKGROUND AND OBJECTIVE: Although metabolite-to-parent drug concentration ratios in hair have been suggested as a possible tool to study the metabolism of drugs in a non-invasive way, no studies are available that evaluated this in a systematic way. Cytochrome P450 (CYP) 1A2 is a drug-metabolizing enzyme characterized by large inter-individual differences in its activity. The standard approach for CYP1A2 phenotyping is to determine the paraxanthine/caffeine ratio in plasma at a fixed timepoint after intake of a dose of the CYP1A2 substrate caffeine. The aim of this study was to evaluate whether paraxanthine/caffeine ratios measured in hair samples reflect the plasma-based CYP1A2 phenotype. METHODS: Caffeine and paraxanthine concentrations were measured in proximal 3 cm segments of hair samples from 60 healthy volunteers and resulting paraxanthine/caffeine ratios were correlated with CYP1A2 phenotyping indices in plasma. RESULTS: Paraxanthine/caffeine ratios in hair ranged from 0.12 to 0.93 (median 0.41); corresponding ratios in plasma ranged from 0.09 to 0.95 (median 0.40). A statistically significant correlation was found between ratios in hair and plasma (r = 0.41, p = 0.0011). However, large deviations between ratios in both matrices were found in individual cases. Although the influence of several factors on paraxanthine/caffeine ratios and hair-plasma deviations was investigated, no evident factors explaining the observed variability could be identified. CONCLUSION: The variability between hair and plasma ratios complicates the interpretation of hair paraxanthine/caffeine ratios on an individual basis and, therefore, compromises their practical usefulness as alternative CYP1A2 phenotyping matrix.

Spot them in the spot: analysis of abused substances using dried blood spots.

Dried blood spot (DBS) sampling and DBS analysis have increasingly received attention during recent years. Furthermore, a substantial number of DBS methods has recently become available in clinical, forensic and occupational toxicology. In this review, we provide an overview of the different DBS-based methods that have been developed for detecting (markers of) abused substances. These include both legal and illegal drugs belonging to different categories, including cannabinoids, cocaine and metabolites, opioids, benzodiazepines and Z-drugs, amphetamines and analogs, gamma-hydroxybutyric acid, ketamine and novel psychoactive substances such as cathinones. Markers of ethanol consumption and tobacco use are also covered in this review. Since the majority of published methods has shown promising results overall, an interesting role for DBS analysis in diverse toxicological applications can be envisaged. For the distinct applications, we discuss the specific potential and benefits of DBS, the associated limitations and challenges, as well as recent developments and future perspectives.

Potassium-based algorithm allows correction for the hematocrit bias in quantitative analysis of caffeine and its major metabolite in dried blood spots.

Although dried blood spot (DBS) sampling is increasingly receiving interest as a potential alternative to traditional blood sampling, the impact of hematocrit (Hct) on DBS results is limiting its final breakthrough in routine bioanalysis. To predict the Hct of a given DBS, potassium (K(+)) proved to be a reliable marker. The aim of this study was to evaluate whether application of an algorithm, based upon predicted Hct or K(+) concentrations as such, allowed correction for the Hct bias. Using validated LC-MS/MS methods, caffeine, chosen as a model compound, was determined in whole blood and corresponding DBS samples with a broad Hct range (0.18-0.47). A reference subset (n = 50) was used to generate an algorithm based on K(+) concentrations in DBS. Application of the developed algorithm on an independent test set (n = 50) alleviated the assay bias, especially at lower Hct values. Before correction, differences between DBS and whole blood concentrations ranged from -29.1 to 21.1%. The mean difference, as obtained by Bland-Altman comparison, was -6.6% (95% confidence interval (CI), -9.7 to -3.4%). After application of the algorithm, differences between corrected and whole blood concentrations lay between -19.9 and 13.9% with a mean difference of -2.1% (95% CI, -4.5 to 0.3%). The same algorithm was applied to a separate compound, paraxanthine, which was determined in 103 samples (Hct range, 0.17-0.47), yielding similar results. In conclusion, a K(+)-based algorithm allows correction for the Hct bias in the quantitative analysis of caffeine and its metabolite paraxanthine.