Citations

Importance Although the Martin-Hopkins equation is widely validated and used to estimate low-density lipoprotein cholesterol (LDL-C) for clinical care, a simplified equation could streamline implementation and enable broader adoption. Objective To develop a simplified machine learning–based alternative to the Martin-Hopkins equation using multivariate adaptive regression splines (LDL-C–MH-MARS) and compare its performance with the Friedewald (LDL-C–F), Sampson–National Institutes of Health (LDL-C–S), Modified Sampson (LDL-C–MS), and Martin-Hopkins (LDL-C–MH) equations. Design, Setting, and Participants The Very Large Database of Lipids is a cross-sectional database of clinical lipid measurements from a population-representative convenience sample of adult and pediatric patients. Lipid measurements were performed by Vertical Auto Profile ultracentrifugation between October 1, 2015, and June 30, 2019. The study included patients with complete lipid panel data who were randomly assigned to a training set and a test set. External validation against preparative ultracentrifugation-based LDL-C measurements was performed in reference laboratory datasets from the Mayo Clinic (broad LDL-C concentrations) and the Further Cardiovascular Outcomes Research With Proprotein Convertase Subtilisin/Kexin Type 9 (PCSK9) Inhibition in Subjects with Elevated Risk (FOURIER) trial (observations from patients with low LDL-C concentrations while taking evolocumab). Study data were analyzed from January to October 2025. Exposures LDL-C–F, LDL-C–S, LDL-C–MS, LDL-C–MH, and LDL-C–MH-MARS equations. Main Outcomes and Measures Accuracy of estimated vs ultracentrifugation-measured LDL-C concentration based on bias, root mean square error (RMSE), and concordance by guideline-based categories. Results The study included 4 939 528 patients (mean [SD] age, 56 [16] years; 2 635 486 female [53%]) with complete lipid panel data who were randomly assigned to a training set (n = 3 292 889) and a test set (n = 1 646 639). LDL-C–MH-MARS equation demonstrated a very low median (IQR) bias of −0.1 (−2.1 to 1.8) mg/dL (to convert to millimoles per liter, multiply by 0.0259), comparable with LDL-C–MH. The median (IQR) difference between the MARS and original Martin-Hopkins equations (ie, LDL-C–MH-MARS – LDL-C–MH) was −0.5 (−1.2 to 0) mg/dL, further suggesting comparability between the 2 methods. The RMSE was smallest for LDL-C–MH-MARS (4.7 mg/dL) and LDL-C–MH (4.9 mg/dL), followed by LDL-C–S (5.8 mg/dL), LDL-C-MS (6.0 mg/dL), and LDL-C–F (7.2 mg/dL). The proportion of patients correctly classified according to clinical categories was nearly identical for LDL-C–MH-MARS (89.7%) and LDL-C–MH (89.6%) but lower for LDL-C–S (86.3%), LDL-C–MS (84.7%), and LDL-C–F (83.1%). The LDL-C–MH-MARS, LDL-C–MH, LDL-C–MS, LDL-C–S, and LDL-C–F equations underestimated LDL-C concentration in 16%, 17%, 28%, 39%, and 60%, respectively, when classifying LDL-C less than 70 mg/dL in patients with triglyceride concentrations of 200 to 399 mg/dL (to convert to millimoles per liter, multiply by 0.0113). The patterns of results were similar in external validation datasets, with the LDL-C–MH-MARS and LDL-C–MH equations demonstrating the highest accuracy. Conclusions and Relevance Study results suggest that the new machine learning–based LDL-C equation provided results comparable with those of the original Martin-Hopkins method while simplifying implementation to a single equation.

Abstract Background Capillary self-collection (CSC) devices allow patients to collect blood samples at home, potentially reducing outpatient phlebotomy visits. This study aimed to (a) assess the patient experience with multiple commercially available CSC devices; (b) determine which laboratory tests are commonly ordered together, which could reduce the need for appointments; and (c) evaluate the analytical performance of these tests using CSC samples. Methods User experience for 3 CSC devices was evaluated. Clinical feasibility was determined by comparing test results in paired sera collected by venipuncture (VP) and CSC devices. VP samples were centrifuged within 2 h and tested immediately. CSC sera were centrifuged and tested both immediately and after delayed processing to simulate shipping temperature extremes (−20°C or 40°C). Basic metabolic panel, lipid panel, thyroid function cascade, and prostate-specific antigen were evaluated. Differences between VP and CSC collections were characterized according to their statistical and clinical differences. Results Patients reported CSC devices were easy to use and painless. Clinically significant differences between VP and CSC sera processed immediately were limited to potassium and bicarbonate. Following delayed processing of CSC serum samples (48 h at room temperature), clinically significant differences in potassium, bicarbonate, and glucose were observed. Frozen samples could not be analyzed. A 48 h delay at 40°C caused clinically significant differences in all analytes except creatinine, lipid panel, prostate-specific antigen, and the thyroid cascade. Conclusion Potassium, bicarbonate, and glucose were not stable at room temperature in CSC sera. CSC serum specimens may be a viable option but require analyte-specific evaluation and consideration of transportation conditions.

Abstract Background Dipeptidyl peptidase 3 (DPP3) is a peptidase released from dying cells. It cleaves proteins in the renin–angiotensin pathway, which can result in hemodynamic instability. At elevated concentrations DPP3 is associated with worse outcomes, particularly in patients with shock. Herein we describe the assay performance of a DPP3 assay (4TEEN4 Pharmaceuticals GmbH) in human plasma. Methods DPP3 concentration was measured using the DPP3 immunoluminometric assay (4TEEN4 Pharmaceuticals GmbH) and the signal was read using a luminometer (Berthold Centro LB963). Analytical performance was established for precision, linearity, accuracy, detection limit, analytical specificity, reference interval, kit lot-to-lot comparison, specimen type, and sample stability. Results Limit of detection was verified at 1.6 ng/mL in EDTA plasma with a coefficient of variation (CV) of <10%. Precision studies revealed a CV ≤ 6% at 28.5 ng/mL and 59.4 ng/mL and comparability with a manufacturer performed assay was demonstrated between 7.7 and 195.2 ng/mL. An upper 97.5% limit of 22 ng/mL without age or sex associations was verified in healthy donors. The assay was not susceptible to interference from lipemia or bilirubin. However, measured DPP3 concentrations increased linearly with increasing hemolysis. DPP3 concentrations are stable in EDTA plasma for up to 24 h and at least 11 months when stored ambient or at −80°C, respectively. Conclusions DPP3 can be measured precisely in EDTA plasma using the immunoluminometric DPP3 assay. Given the potential clinical use of DPP3 in critical care patients, caution should be taken to avoid inducing pre-analytical hemolysis during sample collection.

Abstract Background Insulin resistance leads to changes in lipid and lipoprotein metabolism prior to developing prediabetes, including an increase in the concentration of large very-low-density lipoprotein particles (LVLDL-p), the mean size of VLDL (VLDL-s), and the concentration of small low-density lipoprotein particles (sLDL-p). Conversely, there is a decrease in the concentration of large high-density lipoprotein particles (LHDL-p), mean size of HDL (HDL-s), and mean size of LDL (LDL-s). Clinically validated studies show that proton nuclear magnetic resonance (NMR) can measure these six parameters and derive a lipoprotein insulin resistance (LP-IR) score, ranging from 0 to 100, with higher scores indicating greater risk for insulin resistance. The reference method for calculating LP-IR relies on a proprietary NMR spectrometer, spectral deconvolution method, and algorithm. This study seeks to determine if lipoprotein subfraction parameters measured from a commercially available NMR system can produce LP-IR scores comparable to the reference method. Methods Serum lipoprotein NMR spectra were collected using a Bruker Ascend 600 MHz magnet, Bruker Avance NEO 500 spectrometer, and the Numares AXINON lipoFIT System to report the six LP-IR score parameters between January 2023 and January 2025 (Cohort A, n = 2,474). During the same period, LP-IR score and parameters were collected using the Vantera Clinical Analyzer (400 MHz) with the LipoProfile LP4 method (Cohort B, n = 2,959). Linear regression of decile-matched data for each parameter yielded adjustment equations to align lipoFIT values with the LipoProfile method. LP-IR scores were then calculated for both adjusted and unadjusted values using a previously published LP-IR scoring algorithm. Finally, LP-IR scores were compared for sera with measures from both instruments was (Cohort C, n = 108). Results The median concentrations of sLDL-p, LDL-s, VLDL-s, HDL-S matched between Cohort A and B (<1% difference). Significant differences were noted for LVLDL-p and LHDL-p, with percent differences of 21.5% and 23.5%, respectively. Despite closely matching medians, the distributions varied greatly, with a mean interquartile range (IQR) percent difference across measures of 31.2% (Table 1). After distribution matching, all parameter median and mean IQR percent differences fell to <5% and <10%, respectively, between Cohort A and Cohort B. In Cohort C the LP-IR score matched well (y=1.1x-15, r=0.92; 95%CI 0.88, 0.94). Following distribution matching adjustment of lipoFIT values for Cohort C, new calculated LP-IR scores matched with reduced bias (y=0.94x-1.2) while correlation was slightly reduced (r=0.88 [0.83, 0.92]). By comparison, adjustments made via scaling lipoFIT values proportionally by the ratio of the medians between Cohort A and Cohort B yielded calculated LP-IR scores in Cohort C that maintained a similar bias prior to adjustments (y=1.2x-21), highlighting the ability of distribution matching to reduce bias. Conclusion Adjustments to measured lipoprotein parameters from a commercially available NMR spectrometer can be achieved via distribution matching. These adjustments greatly reduce the bias between a commercially available method and algorithm compared to the reference method, allowing for wider adoption of LP-IR score reporting.

Abstract Background Direct measurement of low-density lipoprotein cholesterol (LDL-C) is widely used and recommended by professional society guidelines despite its potential limitations in patients with hypertriglyceridemia and low LDL-C. This study evaluated the performance of 3 direct LDL-C (LDL-CD) assays, 2 modern LDL-C calculation methods [LDL-C Martin (LDL-CM), LDL-C modified Sampson (LDL-CS)] and the conventional Friedewald (LDL-CF) method against the reference method, beta-quantification (LDL-CBQ). Methods A total of 181 remnant sera from patients with standard lipid panel orders or from patients with LDL-CBQ orders with triglycerides (TG) ≥ 400 mg/dL (4.5 mmol/L), or with TG ≥ 150 mg/dL (1.69 mmol/L) and LDL-C < 70 mg/dL (1.8 mmol/L) were included. LDL-CD and lipid panel data were gathered from Abbott Alinity, Roche Cobas, and Siemens Atellica platforms. Results LDL-CD among the 3 platforms showed a median CV of 11.2%. In patients with TG <400 mg/dL, LDL-CM and LDL-CS demonstrated less bias and less misclassification at the clinical decision LDL-C levels than LDL-CF or LDL-CD. In the 400 to 800 mg/dL (9.0 mmol/L) TG group, LDL-CS was superior to LDL-CD or LDL-CM in accuracy. When TG is ≥ 800 mg/dL, LDL-CD (Roche) showed substantial bias from LDL-CBQ while LDL-CS (Roche) showed smaller but significant bias. Conclusions In summary, LDL-CD or LDL-CF showed little advantage over the 2 modern LDL-C calculation methods. LDL-CS showed the best overall correlation with LDL-CBQ and therefore is recommended to replace LDL-CF and potentially LDL-CD when making clinical decisions in patients with low LDL-C and hypertriglyceridemia.

Abstract Background Low-density lipoprotein cholesterol (LDL-C) is directly associated with coronary artery disease (CAD) risk. Subfractionation of LDL enables differentiation between large-buoyant LDL (>20.5 nm) and small-dense LDL (≤20.5 nm). Small-dense LDL reportedly increases CAD risk, as do LDL-C and LDL particle (LDL-P) concentrations. Nuclear magnetic resonance spectroscopy (NMR) reports LDL-C, LDL-P, and LDL size (LDL-s). We investigated associations between these outputs and their agreement on CAD risk information. Methods Associations between LDL-P, LDL-C, and LDL-s measured by NMR were evaluated in serum from clinically ordered samples (n = 26 710), and a subset of patients with CAD evaluation from coronary angiography (n = 356). Correlations were determined using Spearman ρ, and clinical classifications were compared using the following thresholds for increased risk: LDL-C > 160 mg/dL, LDL-P > 1600 nmol/L, and LDL-S < 20.5 nm. Results The large laboratory NMR data set showed LDL-C was highly correlated with LDL-P (ρ = 0.87), and moderately with LDL-s (ρ = 0.51). Correlation between LDL-P and LDL-s was weakest (ρ = 0.21). When comparing pairwise high- and low-risk laboratory values, concordant values were observed in 99.8%, 43%, and 25% of cases for LDL-P/LDL-C, LDL-s/LDL-P, and LDL-C/LDL-s, respectively. In patients with angiography and NMR results, CAD was diagnosed in 40% (6/15) of patients with concordant high-risk LDL-C and small-dense LDL-s in the smaller patient cohort, but only 19% (6/31) of CAD-positive patients with high-risk LDL-C had small-dense LDL-s. Conclusions LDL-s and LDL-C are frequently discordant at established LDL-C risk cutoffs. CAD diagnosis was found in similar numbers regardless of LDL-s phenotype.