Evaluation of a point-of-care haemozoin assay (Gazelle device) for rapid detection of Plasmodium knowlesi malaria

Abstract

Plasmodium knowlesi is the major cause of zoonotic malaria in Southeast Asia. Rapid and accurate diagnosis enables effective clinical management. A novel malaria diagnostic tool, Gazelle (Hemex Health, USA) detects haemozoin, a by-product of haem metabolism found in all Plasmodium infections. A pilot phase refined the Gazelle haemozoin identification algorithm, with the algorithm then tested against reference PCR in a larger cohort of patients with P. knowlesi mono-infections and febrile malaria-negative controls. Limit-of-detection analysis was conducted on a subset of P. knowlesi samples serially diluted with non-infected whole blood. The pilot phase of 40 P. knowlesi samples demonstrated 92.5% test sensitivity. P. knowlesi-infected patients (n = 203) and febrile controls (n = 44) were subsequently enrolled. Sensitivity and specificity of the Gazelle against reference PCR were 94.6% (95% CI 90.5-97.3%) and 100% (95% CI 92.0-100%) respectively. Positive and negative predictive values were 100% and 98.8%, respectively. In those tested before antimalarial treatment (n = 143), test sensitivity was 96.5% (95% CI 92.0-98.9%). Sensitivity for samples with ≤ 200 parasites/µL (n = 26) was 84.6% (95% CI 65.1-95.6%), with the lowest parasitaemia detected at 18/µL. Limit-of-detection (n = 20) was 33 parasites/µL (95% CI 16-65%). The Gazelle device has the potential for rapid, sensitive detection of P. knowlesi infections in endemic areas.

Figure 1

Study enrolment flow chart and number of samples tested.

Figure 2

(a) Parasite count distribution of P. knowlesi clinical samples and corresponding test outcomes, grouped according to antimalarial treatment status at sampling point, further sub-grouped by pilot and validation testing phases; horizontal dotted line denotes 200 parasites/µL cutoff. (b) Sensitivity of the Gazelle device accounting for all clinical samples and subset of samples with parasite counts less than 200 parasites/µL, grouped according to antimalarial treatment status at sampling point, further sub-grouped by pilot and validation testing phases; vertical bars represent 95% confidence intervals.

Figure 3

Limit of detection (LOD) of the Gazelle device using individual patient’s parasite counts, grouped according to antimalarial treatment status at sampling point. Horizontal dotted line denotes LOD geometric mean (gmean) was calculated to be 33 parasites/µL.