Citations

Abstract Background Plasma p-tau217 assays are clinically valuable tools for evaluation of Alzheimer’s Disease (AD) pathology. Discrepancy between plasma p-tau217 and cerebrospinal fluid (CSF) biomarkers (Aß42/40 and p-tau181/Aß42), however, may present an interpretive challenge for clinicians. Two plasma p-tau217 assays (Roche Diagnostics and Fujirebio Lumipulse) were evaluated to assess the frequency of discordant results versus CSF AD biomarkers. Methods The study cohort consisted of 65 paired CSF and EDTA plasma residual patient samples with plasma p-tau217 and CSF AD biomarkers ordered 0.028, CSF Aß42/40?<0.073, Elecsys p-tau217 >0.249 (Youden), and Lumipulse p-tau217 >0.229 (Youden). Creatinine eGFR (mL/min/1.73m2) was calculated using the CKD-EPI 2021 equation. CSF assays were either used individually or considered concordant and used as the reference standard for amyloid positivity. The McNemar test was performed to determine if differences in CSF concordance with plasma assays were statistically significant. Results CSF p-tau181/Aß42?and CSF Aß42/40 were concordant in 91% of cases, whereas plasma p-tau217 assays were concordant with each other in 76% of cases. Overall discordance between plasma Lumipulse p-tau217 and CSF classification was 22% for p-tau181/Aß42?and 25% for Aß42/40. Discordance between plasma Elecsys p-tau217 and the CSF classification was 34% for p-tau181/Aß42?and 31% for Aß42/40. No statistically significant difference in concordance between the CSF and plasma biomarkers were observed between either of the two CSF assays versus the Elecsys plasma p-tau 217 (p=0.69) or for Lumipulse plasma p-tau217(p=0.69). When considering samples in which both CSF assays were concordant (N=59), there were 22% FP and 8% FN results against plasma Elecsys p-tau217 and 12% FP and 8% FN against Lumipulse plasma p-tau217 (see Table 1). Chronic kidney disease (CKD) was examined as a potential contributor to discordance between plasma and CSF determinations. Of the 59 CSF-concordant samples, 31% (4/13) of the false positive plasma p-tau217 results had an eGFR of <60 for the Elecsys p-tau217 assay, and 29% (2/7) had an eGFR of <60 for the Lumipulse p-tau217 assay. Comparatively, in the remainder of the cohort (TP, TN, and FN), 13% (6/46) and 15% (8/52) of samples for the Elecsys and Lumipulse p-tau217 assays, respectively, had an eGFR <60. Conclusion CSF AD biomarkers demonstrated high concordance (91%), whereas plasma p tau217 assays were concordant in only 76% of cases. Both p-tau217 plasma assays produced FP and FN results, with the Roche Elecsys assay exhibiting a higher frequency of false positives (22%) compared with the Lumipulse assay (12%). In both plasma assays, a portion of FP results occurred in individuals with reduced kidney function (eGFR <60), indicating that CKD may contribute to elevated plasma p-tau217 in some, but not all cases of discrepant CSF and plasma p-tau217 results. These findings highlight the importance of clinical context interpretation of elevated plasma p-tau217 results.

Abstract Background LabOps is an internally developed workflow tracking system at Mayo Clinic that records time-stamped specimen handoffs and process steps from extraction through real-time PCR result reporting. Implemented in 2021, LabOps was designed to improve workflow efficiency and reduce turnaround time (TAT) by enabling shared use of patient nucleic acid eluates across multiple targeted, laboratory-developed test (LDT), PCR assays. The objective of this study was to evaluate PCR workflows using LabOps analytics to identify process inefficiencies and implement targeted operational improvements to reduce TAT without increasing staffing. Methods From March 1, 2023, to November 20, 2024, time-stamped workflow data were extracted from LabOps for 40 LDT PCR assays comprising 11,250 individual PCR assay batches ranging from 1 to 94 individual patient samples each. Tracked workflow steps included specimen receipt, pre-extraction setup, extraction completion, PCR batch assembly (up to 94 reactions plus controls per plate), run completion on the Roche LC480 instrument, data analysis, and result report verification. Average time to completion for each step was calculated across assays and specimen types. Line graphs were generated to visualize step-specific TAT trends. Based on workflow analysis, targeted operational interventions—including modification of test run cadence—were implemented. TAT was compared before and after intervention. Results Pre-intervention total TAT averaged 41 hours (range: 29–78 hours) from specimen receipt to result release. Two primary contributors to delay were identified: (1) tissue processing and (2) PCR batch assembly cadence. Tissue processing required an average of 24 hours (range: 15–29 hours), significantly increasing TAT for assays accepting tissue specimens. Approximately 90% of assays were assembled once daily despite available capacity for additional reactions per run. An intervention was implemented for 29 PCR assays (excluding those requiring specialized extraction or amplification protocols) to increase run cadence to twice daily at 12-hour intervals. This change reduced the average batch assembly step TAT from 17 hours (with a range of 16-18 hours) to 9 hours (with a range of 8–10 hours), representing approximately a 50% reduction. Figure 1 demonstrates the reduction in TAT for the Epstein Barr Virus (EBV) PCR test pre- and post-implementation of the cadence change. The modified cadence also distributed workload more evenly between day and evening shifts. Conclusion Data-driven workflow optimization using LabOps analytics enabled a substantial reduction in PCR turnaround time without additional staffing. Increasing run cadence to twice daily reduced batch assembly TAT by approximately 50%. Implementation of an automated, continuous TAT monitoring system provides a scalable framework to proactively adjust workflows in response to increased testing demand, seasonal surges, or future outbreak scenarios.

Abstract Background Benzodiazepines work as mild central nervous system depressants and are among the most prescribed drugs for the treatment of depression, anxiety, sleep disorders and alcohol withdrawal symptoms. Their pharmacological effect is mediated through activation of gamma-amino butyric acid (GABA)-mediated inhibitory neurotransmission in the brain. Although this class of drugs possesses a relatively low potential for addiction and side effects, individual metabolic differences such as impaired organ function, age-related or genetically determined drug metabolism challenges necessitate therapeutic drug monitoring. This study focuses on developing and validating a clinical assay for extraction and quantification of seven benzodiazepines and/or their metabolites in human serum. Methods Calibrators and quality controls (QC) were prepared by spiking benzodiazepines (chlordiazepoxide, oxazepam, nordiazepam, temazepam, diazepam, clobazam, n-desmethylclobazam; Cerilliant, USA) into human serum. Samples (50 µl) were extracted by protein precipitation with 90:10 acetonitrile:methanol (v/v) containing stable isotope labeled internal standards (chlordiazepoxide-d5, oxazepam-d5, nordiazepam-d5, temazepam-d5, diazepam-d5, clobazam-13C6, n-desmethylclobazam-13C6; Cerilliant, USA) in a 96-well plate. Following vortexing and centrifugation, supernatant was diluted with water. The extraction method was configured for automation on a Hamilton liquid pipetting system. Extracted samples were injected onto a LC-MS/MS (Sciex Triple Quad™ 6500) system. LC separation utilized a Restek Ultra Biphenyl 3 µm, 50x3.0 mm column with a 1.8 min method followed by tandem mass spec detection with data analysis performed using Sciex MultiQuant™ software. Results Assay validation was performed across 20 extracted runs on multiple days, three levels of QC and study unknowns. Intra- and inter-run imprecision for the QC concentrations was <10% for all benzodiazepines. Linearity was confirmed by linear regression analysis (r² > 0.98) across the analytical measuring interval (10-2000 ng/mL for most analytes, 10-3000 ng/ml for chlordiazepoxide and 100-10,000 for n-desmethylclobazam) with the limit of detection (LOD) determined to be <15% of the lower limit of quantitation (LLoQ). The signal-to-noise ratio for the LLoQ chromatogram was above minimum quantitation requirement of 10. The sample recovery was excellent with minimal matrix suppression, and no carryover was seen for all analytes. Accuracy was assessed by comparing measured concentrations of spiked samples with expected values or results obtained from an external laboratory. Regression slopes demonstrated minimal proportional bias, with all values within 1±0.1 and r²>0.98 with the % agreement within 80-120%. Conclusion An efficient clinical assay was developed for confirmatory analysis of select benzodiazepines and their metabolites in human serum using a small sample volume. The sample extraction was simple and well suited for automation platforms such as Hamilton pipetting systems. Protein precipitation was employed as a rapid and efficient extraction approach, offering significant advantages over more labor intensive techniques such as liquid–liquid or solid phase extraction. In addition, a short LC-MS/MS analysis method allows maximizing the use of instrumentation in a high-volume laboratory. All together the new assay provides adequate sensitivity, specificity, accuracy and throughput.

Abstract Aims Serial phosphatidylethanol (PEth) testing monitors adherence to substance abstinence policies, including in transplant candidates and workplace monitoring programs, as PEth forms exclusively after ethanol (EtOH) exposure and is not produced endogenously. However, abstinent individuals may generate low but detectable PEth concentrations following repeated exposure to non-consumptive ethanol-containing-household products. PEth concentrations between 10-20 ng/mL are interpreted as “light/no drinking,” yet even these values can carry significant financial, occupational, and clinical consequences regardless of the ethanol source. This study evaluates whether moderate to excessive real-world use of common household ethanol-containing products influences PEth concentrations and aims to better define the upper limits of incidental, non-consumptive ethanol exposure. Methods: Adults older than 18 years (n = 40) with baseline PEth concentrations less than 10 ng/mL were assigned to one of four exposure groups: Group A – hand-sanitizer (10 times/day; 1 mL/use), Group B – mouthwash (4 times/day; 20 mL for 30 seconds/use), Group C – both, or Group D – control (no ethanol-containing products). Participants received a standardized 14-day supply of materials, (56 pre-measured 20 mL vials of mouthwash (21.6% EtOH) and/or a 12 oz bottle of hand-sanitizer (70% EtOH)). Instructions required spacing product use >= 15 minutes apart. Before each blood collection, the venipuncture site was disinfected using 70%-isopropanol wipes. Whole blood (EDTA) samples were collected after 7 days and 14 days of assigned product use. PEth was quantified in-house using a validated LC-MS/MS method. Empty mouthwash vials and remaining hand-sanitizer containers were returned at follow-up visits to verify adherence. Results: At the time of interim analysis, each study group was at least 50 percent enrolled. One participant was withdrawn due to confirmed alcohol consumption during the study period. All remaining participants returned empty mouthwash vials and/or used hand sanitizer containers, confirming adherence to assigned exposure regimens. After 7 days of product use, PEth concentrations in Groups A, B, and C remained below the lower limit of quantitation (LLOQ, less than 10 ng/mL). After 14 days, PEth concentrations in all experimental groups continued to remain less than 10 ng/mL. The control group also remained below the LLOQ throughout the study period. No participant produced PEth concentrations approaching the detection threshold; the highest observed value was 3 ng/mL in one participant in the combined exposure group after 7 days. Conclusions: Despite repeated and extensive use of common ethanol-containing products under controlled conditions approximating high-frequency real-world behavior, PEth concentrations remained undetectable in all participants. These findings indicate that measurable PEth concentrations greater than 10ng/mL are highly unlikely to result from incidental dermal or oral exposure, even when multiple ethanol-containing products are used concurrently. Furthermore, isopropanol-based disinfecting wipes applied prior to phlebotomy did not influence measured PEth concentrations. Collectively, these data support the reliability of PEth as a biomarker of alcohol consumption in high-stakes settings. At the commonly applied clinical cutoff for moderate drinking (20ng/mL), incidental exposure is even less likely to contribute meaningfully to PEth positivity.

Abstract Background Viscoelastic Testing (VET) can provide information about the status of coagulation factors, platelet count, platelet function, and fibrinolysis in fresh whole blood and is used in various critical care settings. Our current VET method, Haemonetics TEG 5000 (5000) (Boston, MA) is being discontinued in the next few years. As potential replacements, we evaluated 3 cartridge-based VET analyzers: Werfen ROTEM Sigma Thromboelastometry System (Sigma) (Bedford, MA), Haemonetics TEG 6S Hemostasis System (6S), and Hemosonics Quantra Hemostasis Analyzer (Quantra) (Durham, NC). These cartridge-based VET analyzers can be used at the point of care or in the laboratory. For our evaluation, we focused on 1 cartridge type per method: ROTEM sigma complete + hep on Sigma, Global Hemostasis-HN on 6S, and QPlus on Quantra. Methods We collected fresh Na Citrate tubes (N=69) from laboratory volunteers. Working in sets of 3, tubes were analyzed native and spiked with drugs (heparin, Eptifibatide, Argatroban, Bivalirudin, tPA) intended to affect clot initiation. We calculated the percentage of clot initiation times (R parameter on 6S, CT parameter on Sigma and Quantra) outside each platform’s reportable range. We timed a subset of samples to determine average time to clot initiation and clot strength (A20 parameter on Sigma, MA on 6S and CS by Quantra). In addition to freshly collected samples, we analyzed residual clinical Na Citrate tubes (N=25) from operating room patients. We compared error rates between the 3 methods. A small group of laboratory staff provided user feedback on each analyzer. Results Percentage of results outside reportable clot initiation range/max times (6, 17, and 8 minutes for Sigma CT, 6S R, and Quantra CT) was 23% (16/69) on Sigma, 38% (26/69) on 6S and 13% (9/69) on Quantra. Average time to first result and clot strength for Sigma (N=11) was 9.1 and 21.7 minutes, 6S 8.9 and 29.1 minutes, and Quantra 6.5 and 11.3 minutes. Overall error rates were 3% (3/101) for both 6S and Sigma and 8% (8/104) for Quantra. On Sigma, the 3 errors indicated partial sample aspiration; recollection would be required. The errors on 6s occurred while inserting the cartridge and required a system reboot to resolve. On Quantra, all 8 samples with errors gave partial results but would have required recollection to obtain complete results. Laboratory staff appreciated the ease of sample loading on Sigma and Quantra, though were concerned that loading the whole 2.7 mL Na Citrate tube could lead to an increase in redraws for errors or sample issues. Sample loading by pipette on 6S may increase the potential for exposures, but staff liked that the smaller sample volume required for the 6S would allow use of the smaller volume 1.8 mL Na Citrate tube. They found the speed to results on the Quantra to be a potential advantage. Conclusion Analytical performance of all 3 platforms was relatively similar and could provide the needed coagulation measures to clinical providers. Time to clot initiation and clot strength, reportable range, error rates and user feedback are important considerations when choosing a VET platform.

Abstract Background Clinical laboratories manage a dynamic and ever-expanding collection of standard operating procedures (SOPs) to support testing, quality assurance, and regulatory compliance. We created an SOP Assistant to help staff rapidly access current information, often spread across multiple documents, using retrieval-augmented generation (RAG) and long-context analysis to ensure responses remain grounded in verified SOP content. The objective of this study was to develop and validate the accuracy and relevance of SOP Assistant responses to ensure they remained grounded in verified SOP content. Methods The SOP Assistant was built on Google Cloud Platform Vertex AI RAG Engine, indexing SOPs spanning 10 laboratory departments and over 1,300 documents. SOPs were chunked and embedded into a vector database using the text-embedding-005 model and Gemini 2.5 Flash parser to give context-preserved chunks. Responses were streamed using a Gradio-based chat interface, containerized with Python 3.11 slim and deployed on Cloud Run. Two query modes were created: "Quick Lookup," which retrieved and analyzed the top 15 relevant document chunks via RAG with re-ranking, and "Deep Analysis," which retrieved the top 5 SOPs and prompted users to select up to three for comprehensive evaluation by long-context analysis. Final responses were generated with Gemini 2.5 Flash at temperature 0.1 to prioritize deterministic, factually grounded output. We developed 100 queries targeting specific challenge types: direct fact retrieval (n=30), conditional logic (n=30), sequential procedural information (n=20), negative/exclusionary (n=10), and vague/disambiguation (n=10). Queries spanned >20 SOPs covering analytes such as potassium, AST, troponin, HDL cholesterol, direct bilirubin, CBCs, and heparin anti-Xa, and workflow SOPs covering instrument maintenance and specimen handling. Responses were graded using a 5-point Likert scale for accuracy and relevance. Accuracy scores of 5 required complete factual concordance with SOPs, including units, values, and conditional logic. Relevance scores of 5 indicated precise, non-extraneous answers. Hallucination count and severity were also recorded. Results Average accuracy and relevance scores were calculated for each query type: direct fact retrieval (4.8/4.7), conditional logic (4.9/4.8), sequential procedural information (4.5/4.8), negative/exclusionary (5.0/4.7), and vague/disambiguation (5.0/5.0). One severe hallucination was noted in a direct fact retrieval query using Quick Lookup mode related to uric acid. Deep Analysis mode did not produce hallucinations. Conclusion Validation through targeted query assessment revealed gaps that drove iterative improvements to the SOP Assistant. RAG-only Quick Lookup produced minor hallucinations when queries spanned multiple SOPs covering similar analytes across different work areas, while Deep Analysis long-context mode generated no hallucinations on complex procedural queries. These findings motivated mandatory inline source citations to support clinical accuracy verification and suggest an optimized architecture in which RAG serves as a rapid document identification layer with final response generation delegated entirely to long-context analysis of complete source documents. This approach leverages the complementary strengths of each method while minimizing the hallucination risk inherent to chunk-based retrieval alone. Validation also uncovered minor SOP wording inconsistencies, suggesting additional utility for this tool as a quality assurance mechanism.

Abstract Background Our institution has a cTnT biomarker panel consisting of an initial cTnT, followed by a 2-hour and if necessary, a 6-hour collection. Accurate cTnT results are important for rapid diagnosis of acute coronary syndrome and to avoid prolonged stays in an emergency room. We compared cTnT results from Roche (Roche, Indianapolis, IN) cobas e 801 (801), e 601 (601), and two e 402 (402) instruments. Methods Comparisons (N=91) were completed using cTnT results from a single 601 instrument as the reference and comparing to the results from two 402 analyzers. Additional comparisons (N=45) were completed using cTnT results from an 801 as reference compared to results from the 402 instruments. Residual plasma separator tubes (PST) from clinically ordered cTnT testing were used. To ensure sample quality all plasma samples were aliquoted and recentrifuged before testing on each analyzer. Analysis was completed on all analyzers within 1 hour of the first comparison measurement. Acceptance limits were set as Passing-Bablok slope 1.00 ±0.10 and bias ±5 ng/L (for cTnT values <100 ng/L) or ±5% (for cTnT values =100 ng/L) when compared to reference 601 or 801. Student’s T-Test was performed to assess statistical significance between the observed mean bias between Pure instruments compared to the reference instrument, significance was defined as p =0.05. Results When compared to 601, results from the first 402 instrument demonstrated a slope of 1.11 with 54% (49/91) of samples within our acceptance criteria and a mean difference of 115 ng/L (Figure 1). Results from the second 402 instrument had a slope of 1.11 with 53% (48/91) within acceptance limits with a mean difference of 121 ng/L (p = 0.53, no difference between 402 instruments). For samples with cTnT <100 ng/L, 88% (36/41) and 88% (36/41) were within acceptance. In cTnT samples =100 ng/L, the mean percent bias was observed as 12% on both 402 analyzers. When using results from the 801 instrument as reference, results from the first 402 instrument had a slope of 0.954 with 77% (33/43) within acceptable limits and a mean difference of -61 ng/L. Results from the second 402 instrument had a slope of 0.978 with 79% (34/43) within acceptable limits and a mean difference of -39 ng/L (p = 0.33, no difference between 402 instruments). In samples with cTnT <100, comparisons to the 801 showed 82% (14/17) and 94% (16/17) within acceptance criteria and a mean difference of 2 ng/L on each 402 instrument. For cTnT samples with concentrations =100 ng/L, the mean percent bias was -2% on both 402 instruments. Conclusion We observed proportional bias on the 402 instruments compared to both 601 and 801 instruments, with large and clinically significant differences particularly when comparing to the 601 at higher cTnT concentrations. To avoid a delta being caused by observed systematic bias we will complete all cTnT testing within a panel on one platform.

Abstract Background Quantitative M-protein Study, Serum (QMPSS), a laboratory developed test, is used to identify and quantify monoclonal immunoglobulins by MALDI-TOF MS MASS-FIX. Monoclonal IgM proteins with glycosylated light chains (LC) are rare (5% of total M-proteins identified). In a given month approximately 3 cases of the 71 reported IgM monoclonal proteins with glycosylated light chains present with indeterminate kappa or lambda LC. The structure of the intact IgM monoclonal protein may restrict access to the Ig-LC constant region resulting in inefficient isolation. The laboratory uses immunofixation electrophoresis (IFE, Sebia) to confirm the Ig-LC isotype in this scenario. This study evaluates a reduction step prior to LC isolation to enhance the detection of the IgM isotype. Methods For QMPSS analysis, serum is added to five immunoaffinity matrices that target IgG, IgA, IgM, Ig-K, and Ig-L (CaptureSelect). After a binding phase, the serum/bead mixtures are washed to remove unbound protein. Finally, Igs are eluted, and the disulfide bonds are reduced, resulting in purified Ig-components. This step is followed by MALDI-TOF MS (Bruker) analysis. M-protein isotypes are determined by pairing each monoclonal protein’s unique LC mass detected in both heavy chain and LC Ig isolates. Twenty de-identified, residual waste serum samples with IgM-M-proteins without clear LC isotype were collected over several months. The proof-of-concept method modification studied to decrease reliance on IFE included use of 2.5 mcL of 0.5M dithiothreitol (DTT) added to kappa and lambda affinity matrix beads prior to sample addition followed by the QMPSS pre-analytical processing. Results For the 20 IgM glycosylated M-proteins, M-spike concentrations ranged from 0.026 to 0.506 g/dL. Using QMPSS and IFE lab routine protocol, the IgM samples were reported as glyco MK (n=17), biclonal MK+ML (n=1), and 2 samples were reported as negative. 16 of the 20 samples used in the study had IgM kappa monoclonal proteins that were identifiable on IFE. 13 of these IgM kappa monoclonal proteins were also identifiable by QMPSS-DTT. Additionally, 1 IgM kappa monoclonal protein was identifiable on QMPSS-DTT but was reported as equivocal on IFE. Of the 6 samples that the DTT method could not resolve, IFE was positive for 3. As shown by the 3 samples in which the light chain was identified only by IFE, the need for IFE as a confirmatory assay is not fully eliminated. The 3 false negative samples had M-quant values of 0.036g/dL, 0.043g/dL, and 0.117g/dL. More extensive testing would be required to determine the cause and prevalence of the observed discrepancy between IFE and QMPSS-DTT. Conclusion The use of the alternative QMPSS-DTT method shows potential to improve the capabilities of the QMPSS assay to identifying IgM monoclonal proteins with glycosylated light chains and would effectively remove the need for IFE in 70% of cases, improving test turnaround time. Further experimentation and optimization could decrease the number of discrepancies.

Abstract Background Albumin measurement in pericardial fluid is frequently requested in our reference laboratory practice. The serum to ascites albumin gradient (SAAG) is used to distinguish between transudative and exudative effusions. However, growing evidence suggests pericardial albumin is less useful in this scenario. Our objective was to validate analytical accuracy of albumin measurement in pericardial fluid and compare frequency of transudate classifications. Methods Albumin concentrations were measured using a Roche Cobas c501, (Roche, Indianapolis, IN). Within-laboratory coefficient of variation (CVWL) and repeatability (CVr) were determined by repeated measurement of pooled clinical residual pericardial fluid. Linearity was assessed by mixing pericardial fluid pools with high and low albumin concentrations. Dilutional recovery from clinically ordered pericardial fluid albumin analyses (n=772) were used to determine analytical accuracy. SAAG transudate diagnostic threshold of 1.1 g/dL (peritoneal) or 1.2 g/dL (pleural) was evaluated in clinical data from patients with paired serum and body fluid albumin measured within a 2-day interval (n=3,031). Results Albumin imprecision in pericardial fluid was assessed at 1.4 g/dL (CVWL = 6.3%, CVr = 5.7%) and 3.1 g/dL (CVWL = 2.2%, CVr = 2.2%). Mixing studies confirmed linearity between 0.5-5.6 g/dL (slope 1.02, R2 = 0.999). Dilutional recovery was performed in 772 clinical orders spanning 0.9 g/dL to 4.8 g/dL, the Passing-Bablok fit of expected versus measured values had slope = 1.09 g/dL and intercept = -0.2 g/dL. In the clinical data, serum albumin was similar for patients in all cases. Pericardial albumin [median(IQR) 2.2(1.0-2.6) g/dL] was significantly higher compared to peritoneal [1.0(0.5-1.7) g/dL; p<0.001] or pleural fluid [1.6(1.0-2.2) g/dL; p<0.001], leading to significantly lower SAAG calculation for pericardial [1.3(0.5-1.9) g/dL] versus peritoneal [1.9(1.4 – 2.4); p<0.001] or pleural fluid [1.5(1.0-2.1); p<0.001]. Applying the SAAG threshold of 1.2 g/dL classified 84% of peritoneal fluids and 58% of pericardial fluids as transudative effusions. The Light’s criteria threshold of 1.2 g/dL led to 65% of pleural fluids and 55% of pericardial fluids being classified as transudative. Conclusion These data show that albumin can be precisely measured in pericardial fluid within a linear range of 0.5-5.6 g/dL. In clinical cases of evaluating SAAG ratio in body fluid samples, albumin concentrations are generally higher in pericardial fluid compared to peritoneal or pleural fluids. This increased pericardial albumin concentration leads to lower SAAG calculations. Application of typical transudate versus exudate thresholds in pericardial fluids leads to more exudative determinations. This data suggests that pericardial fluid may warrant a unique diagnostic threshold to be determined in further studies.

Abstract Background Blood gas (BG) analyzers are essential point-of-care and central laboratory instruments used to assess acid–base status, ventilation, oxygenation, and electrolyte balance in critically ill patients. Cartridge-based systems are increasingly utilized due to simplified maintenance, integrated quality control, and rapid turnaround times. However, inter-instrument variability and method-specific bias remain important considerations, particularly when transitioning between platforms or comparing results with core laboratory analyzers. We evaluated two cartridge-based analyzers—Werfen GEM Premier 7000 (GEM 7000; Bedford, MA) and Siemens RAPIDPoint 500e (RAPIDPoint 500e; Malvern, PA)—compared with our current Radiometer ABL90 FLEX PLUS (ABL90; Brea, CA). Additionally, preanalytical factors—especially hemolysis—can significantly affect potassium and ionized calcium (iCa). Unlike central chemistry analyzers, most BG systems do not assess hemolysis in whole blood. GEM 7000 uniquely incorporates onboard semi-quantitative hemolysis detection. Methods Residual lithium-heparin whole blood syringes (N=120–130) were analyzed on all three platforms within 15 minutes. For analytes with established laboratory methods, paired specimens (N=31–33) were tested on Sysmex XN-1000 (hemoglobin [Hb]; whole blood) and Roche Cobas c501 (sodium [Na], potassium [K], glucose [Glu]; plasma). Acceptability was defined using CLIA Total Allowable Error (TEa) or internally established criteria where CLIA limits were unavailable. Hemolysis index (HI) comparisons (N=76) were performed between GEM 7000 (whole blood) and c501 (plasma following centrifugation). Selected samples were spiked with hemolysate to broaden the analytical range. Agreement was assessed using regression and categorical concordance. Results Compared with ABL90, GEM 7000 demonstrated =90% agreement within CLIA or internal TEa for pH, partial pressure of oxygen (pO2), partial pressure of carbon dioxide (pCO2), Hb, fractional carboxyhemoglobin (FCO2Hb), fractional methemoglobin (FMetHb), iCa, and Glu, but lower agreement for fractional oxyhemoglobin (FO2Hb), Na, K, and Lac. RAPIDPoint 500e achieved =90% agreement for pH, pO2, FMetHb, but showed reduced agreement for pCO2, FO2Hb, FCO2Hb, and Lac. (Table 1) Against primary laboratory methods, clinically relevant variability was observed. For Hb, percent within acceptance was 12.5% (ABL90), 62.5% (GEM 7000), and 28.1% (RAPIDPoint 500e). For Na, agreement was highest for ABL90 (96.8%) and lowest for GEM 7000 (64.5%). For K, GEM 7000 demonstrated strongest agreement (86.1%), whereas ABL90 showed the lowest (36.1%). HI correlation between GEM 7000 and c501 was strong (slope 1.005; mean difference 2), with 93.3% of samples within acceptance limits and 96.0% categorical agreement. Conclusion Both candidate analyzers demonstrated acceptable performance for core BG parameters; however, meaningful inter-platform variability was observed for co-oximetry, and Lac. These findings indicate that BG and core laboratory results are not universally interchangeable, and platform-specific bias may influence clinical interpretation, particularly for Hb and K. Institutions transitioning analyzers should consider method validation, clinician education, and potential adjustment of reference or critical value thresholds. The GEM 7000’s integrated hemolysis detection showed strong agreement with a laboratory-based plasma method, supporting its utility for identifying preanalytical interference at the point of testing. Real-time hemolysis assessment may reduce reporting of spurious results, limit unnecessary redraws, and enhance patient safety.