Comparability of venous and capillary blood measurements for a host-protein test differentiating bacterial from viral infections

Abstract Background MeMed BV (MMBV) is a host-protein test for discriminating between bacterial and viral infections. It was analytically and clinically validated in serum and whole blood samples. Here we investigated the comparability of venous versus capillary blood measurements of the MeMed BV score and of the composite TRAIL, IP-10 and CRP biomarkers. Methods A prospective cohort study enrolled adult patients presenting with suspected acute infection between Oct-2024 till Jan-2025 at three medical centers. Eligibility criteria were according to the MeMed BV’s instructions for use. Paired venous and capillary blood specimens were collected from each patient. Each sample (50 µL) was measured on MeMed Key® (MeMed, US) with a next generation MeMed BV cartridge. Passing-Bablok regression was used to evaluate the relationship between venous and capillary measurements for each biomarker and the MeMed BV score. For the MeMed BV score, the first accuracy requirement was that the slope should fall in the range 0.9-1.1 and the intercept in the range of −5 to 5. The second accuracy requirement was that the 95% confidence intervals for the bias at all score bin cutoffs should fall within the interval [−12.5,12.5]. The clinical acceptance criterion for the score was that less than 5% of the paired sample measurements should fall into non-adjacent score bins. Results The study population comprised 58 patients with median age 58.5 (interquartile range: 36.25-75.75). Venous versus capillary blood measurements for TRAIL, IP-10 and CRP demonstrated high correlation with coefficients of 0.98, 0.98 and 0.99 respectively. For the MeMed BV score, the slope in the Passing-Bablok regression analysis was 1.00 (95% confidence interval, CI: 0.99 – 1.00) and the intercept 0.00 (0.00 – 0.08), satisfying the first accuracy requirement. The estimate of bias at each cutoff satisfied the second accuracy requirement: score 10, −0.57 (95%CI: −3.00 – 0.56); score 35, −0.47 (−2.10 – 0.38); score 65, −0.34 (−1.28 – 0.22); and score 90, −0.24 (−0.76 – 0.44). None of the paired samples generated MeMed BV scores falling into non-adjacent bins, satisfying the clinical criterion. Conclusion TRAIL, IP-10, CRP and MeMed BV score measurements are comparable in venous and capillary blood when performed using the next generation MeMed BV cartridge and MeMed Key. The present findings can serve as the basis for expanding MeMed BV’s use to include capillary blood specimens, potentially widening its clinical utility.