Field evaluation of a prototype paper-based point-of-care fingerstick transaminase test.

Publication Type:

Journal Article

Source:

PLoS One, Volume 8, Issue 9, p.e75616 (2013)

Keywords:

Alanine Transaminase, Blood Chemical Analysis, Developing Countries, Drug Monitoring, Drug-Induced Liver Injury, Humans, Liver Function Tests, Microfluidics, Observer Variation, Paper, Point-of-Care Systems, Vietnam

Abstract:

<p>Monitoring for drug-induced liver injury (DILI) via serial transaminase measurements in patients on potentially hepatotoxic medications (e.g., for HIV and tuberculosis) is routine in resource-rich nations, but often unavailable in resource-limited settings. Towards enabling universal access to affordable point-of-care (POC) screening for DILI, we have performed the first field evaluation of a paper-based, microfluidic fingerstick test for rapid, semi-quantitative, visual measurement of blood alanine aminotransferase (ALT). Our objectives were to assess operational feasibility, inter-operator variability, lot variability, device failure rate, and accuracy, to inform device modification for further field testing. The paper-based ALT test was performed at POC on fingerstick samples from 600 outpatients receiving HIV treatment in Vietnam. Results, read independently by two clinic nurses, were compared with gold-standard automated (Roche Cobas) results from venipuncture samples obtained in parallel. Two device lots were used sequentially. We demonstrated high inter-operator agreement, with 96.3% (95% C.I., 94.3-97.7%) agreement in placing visual results into clinically-defined "bins" (<3x, 3-5x, and >5x upper limit of normal), >90% agreement in validity determination, and intraclass correlation coefficient of 0.89 (95% C.I., 0.87-0.91). Lot variability was observed in % invalids due to hemolysis (21.1% for Lot 1, 1.6% for Lot 2) and correlated with lots of incorporated plasma separation membranes. Invalid rates <1% were observed for all other device controls. Overall bin placement accuracy for the two readers was 84% (84.3%/83.6%). Our findings of extremely high inter-operator agreement for visual reading-obtained in a target clinical environment, as performed by local practitioners-indicate that the device operation and reading process is feasible and reproducible. Bin placement accuracy and lot-to-lot variability data identified specific targets for device optimization and material quality control. This is the first field study performed with a patterned paper-based microfluidic device and opens the door to development of similar assays for other important analytes.</p>