Repeatability and reproducibility of a handheld quantitative G6PD diagnostic

Abstract

Background: The introduction of novel short course treatment regimens for the radical cure of Plasmodium vivax requires reliable point-of-care diagnosis that can identify glucose-6-phosphate dehydrogenase (G6PD) deficient individuals. While deficient males can be identified using a qualitative diagnostic test, the genetic make-up of females requires a quantitative measurement. SD Biosensor (Republic of Korea) has developed a handheld quantitative G6PD diagnostic (STANDARD G6PD test), that has approximately 90% accuracy in field studies for identifying individuals with intermediate or severe deficiency. The device can only be considered for routine care if precision of the assay is high.

Methods and findings: Commercial lyophilised controls (ACS Analytics, USA) with high, intermediate, and low G6PD activities were assessed 20 times on 10 Biosensor devices and compared to spectrophotometry (Pointe Scientific, USA). Each device was then dispatched to one of 10 different laboratories with a standard set of the controls. Each control was tested 40 times at each laboratory by a single user and compared to spectrophotometry results. When tested at one site, the mean coefficient of variation (CV) was 0.111, 0.172 and 0.260 for high, intermediate, and low controls across all devices respectively; combined G6PD Biosensor readings correlated well with spectrophotometry (rs = 0.859, p<0.001). When tested in different laboratories, correlation was lower (rs = 0.604, p<0.001) and G6PD activity determined by Biosensor for the low and intermediate controls overlapped. The use of lyophilised human blood samples rather than fresh blood may have affected these findings. Biosensor G6PD readings between sites did not differ significantly (p = 0.436), whereas spectrophotometry readings differed markedly between sites (p<0.001).

Conclusions: Repeatability and inter-laboratory reproducibility of the Biosensor were good; though the device did not reliably discriminate between intermediate and low G6PD activities of the lyophilized specimens. Clinical studies are now required to assess the devices performance in practice.

Fig 1. Schematic overview of the study design.

CV: coefficient of variation. (Images reproduced with permission from SD Biosensor and Hemocue).

Fig 2. G6PD activity measured by each Biosensor device and spectrophotometry.

Red circled bars = reference method (Hb normalized spectrophotometry result), Blue shade = recommended range for ACS Low controls, Red shade = recommended range for ACS Intermediate controls, Green shade = recommended range for ACS High controls, Blue dotted line = median activity of ACS Low controls across all devices excluding spectrophotometry, Red dotted line = median activity of ACS Intermediate controls across all devices excluding spectrophotometry, Green dotted line = median activity of ACS High controls across all devices excluding spectrophotometry, dots represent outliers.

Fig 3. Hb levels measured by each Biosensor device and Hemocue.

Red circled bars = reference method (Hemocue), Blue line = recommended point estimate for ACS Low controls, Red line = recommended point estimate for ACS Intermediate controls, Green line = recommended point estimate for ACS High controls, Blue dotted line = median Hb reading for Low controls, Red dotted line = median Hb reading for Intermediate controls, Green dotted line = median Hb reading for High controls, dots represent outliers.

Fig 4

Phase A: G6PD activity measured by Biosensor and spectrophotometry. Red circled results are influential points generated by spectrophotometry; Left: Red horizontal lines indicate upper and lower end of 95% limit of agreement, black dotted line indicates mean difference. Right: horizontal dotted lines indicate the overlap between Low and Intermediate readings by Biosensor.

Fig 5

Box and whisker plot of G6PD activity / control by Biosensor (A) and spectrophotometry (B) across Phase B sites. Blue shade = recommended range for ACS Low controls, Red shade = recommended range for ACS Intermediate controls, Green shade = recommended range for ACS High controls, Blue dotted line = median activity of ACS Low controls across all devices, Red dotted line = median activity of ACS Intermediate controls across all devices, Green dotted line = median activity of ACS High controls across all devices, dots represent outliers.