The results of this study show that the XN-31 can be used in clinical practice as a fast and easy screening assay for malaria that provides reliable qualitative and quantitative results. Therefore, XN-31 can easily be integrated in regular 24/7 patient care diagnostics settings in non-endemic counties and can provide all required information to clinicians to timely start proper treatment. Compared to other screening assays, the XN-31 is the only test that can provide rapid results (< 1 min) to determine the presence of Plasmodium-infected RBC with a detection limit equivalent to thick blood film examination [5] in combination with a Plasmodium species differentiation and an accurate parasitaemia quantification. The XN-31 can accurately determine the parasitaemia in samples with low numbers of infected erythrocytes, as the limit of detection and quantification was determined to be 5.9 and 19 infected RBC per µL, respectively. Thereby the detection limit of the XN-31 is equivalent to thick and thin blood examination which has on average a detection limit of ~ 10 parasites per µL [5].
Although there were no false positive or false negative results for samples of patient suspected for malaria in this study, it is known that submicroscopic malaria exists [20]. In these cases the number of infected erythrocytes is below the detection limit of thick blood film examination, and thus also below the detection limit of the XN-31. Therefore, false negative results can occur, but most submicroscopic malaria cases are asymptomatic and occur in patients from endemic areas with extensive immunity against malaria or in patients infected with a benign, non-falciparum, Plasmodium species that will present with a typical and characteristic fever pattern returning every 48 or 72 h [21,22,23]. Hence, for patients whom malaria is clinically suspected, but a negative result is obtained, regardless of which test is used, repeat testing should be undertaken periodically. In specific cases further examinations by more sensitive methods can be indicated for patients for which a negative result by the XN-31 has been generated. In this analytical performance evaluation an LoD of ~ 6 MI-RBC/µL was achieved, which is significantly lower than the 20 MI-RBC/µL cut-off set for qualitative judgment of MI-RBC present or absent. The threshold for defining a sample as positive could therefore possibly be adjusted by the manufacturer.
Next to the hypothetical possibility of false negative results due to patients with very low parasitaemia, this study demonstrated that indeterminate and false positive MI-RBC results can occur as well. Examination of a large panel of blood samples of patients with a variety of RBC abnormalities demonstrated not only a high frequency of indeterminate results (~ 50%), but also false positive results (~ 5%). These false positive MI-RBC results were predominantly encountered for blood samples from patients with sickle cell disease (HbSS and HbSβ0) and patients with increased numbers of nucleated erythrocytes and/or reticulocytes. Conditions of stressed erythropoiesis such as may occur in thalassaemia or other haemoglobinopathies are mentioned in the instructions for use as potentially giving an erroneous MI-RBC positive result. Analysis of the scattergrams demonstrated substantial differences between the true positive and false positive MI-RBC samples and, therefore, future refinement of the automatic interpretation script of the XN-31 for the obtained scattergrams should result in improved performance of the XN-31. Thus, caution is required for patients with abnormal blood cell morphology and review of the scattergram is advised when authorizing results. It should however be noted that in this study the number of samples measured was deliberately enriched with those expected to cause interferences and that the actual occurrence of such issues may be substantially lower in the routine setting where only samples from patients suspected to have malaria will be measured on XN-31.
In order to better determine what could produce indeterminate and false positive MI-RBC results by the XN-31, abnormal RBC samples were examined. For detection of MI-RBC, the software determines the number of events in the M-gating area as shown in Fig. 4. If the number of detected events exceeds a certain threshold an indeterminate or positive result will be generated according to clustering patterns defined by algorithms. When interfering cell types are present that produce a scattergram with a distinct cluster of particles in the M-gating area, the algorithm will override the presence of generalized background scatter, producing a false positive MI-RBC result. This also means that when 20 or more parasites/μL are present, but no distinct cluster can be detected due to interference, an indeterminate result will be produced that is marked with an abnormal scattergram flag. Examination of the panel of abnormal RBC samples showed that a number of diseases and conditions frequently occurred in both the false positive and indeterminate MI-RBC results group; beta thalassaemia major, leukaemia, lymphoma, premature newborns, and sickle cell disease. The observation occurring most frequently for the false positive samples was sickle cell disease in crisis. Although there were no sickle cell patients included in the study population for suspected Plasmodium infection, it can be hypothesized that a patient with sickle cell disease and malaria should get a valid MI-RBC present result (cluster detected) although the actual parasitaemia value would be overestimated. Premature newborns, beta thalassaemia and haemochromatosis are associated with stressed or disturbed erythropoiesis as well and we speculate that triggering of indeterminate and false positive results is highly correlated with diseases and conditions associated with the presence of immature cells in the erythrocyte lineage or with severely abnormal RBC morphology.
Although in this study the population of patients suspected for a Plasmodium infection, none of the mentioned diseases were present, it is very well possible as RBC abnormalities occur relatively frequently in the population in malaria endemic areas [24]. Some of these abnormalities can even cause mortality in malaria patients making it even more important to understand exactly which RBC abnormalities cause indeterminate or positive result on the XN-31. Mortality in sickle cell patients with malaria is a problem in endemic countries that have a high prevalence of sickle cell disease. More than 80% of people that have sickle cell disease live in sub-Saharan Africa where most Plasmodium deaths occur [25]. In one study from Cameroon, it was found that 2 out of every 10 sickle cell patients who died had malaria [26]. To date, there have been preliminary studies done how interferences affect the results of the XN-31, but there is some discrepancy in the results and what conditions can trigger an abnormal scattergram on the XN-31 [14, 15]. For the investigation of RBC abnormalities, abnormal samples were deliberately collected to determine the effect on the XN-31 result. This does not necessarily represent the frequency of false positive results in endemic regions where thalassaemia and sickle cell disease are more prevalent as there have been studies performed previously on the XN-30 in regions such as Burkina Faso that showed no false positive results [16]. However, it is important to augment this data with further studies in order to determine whether the XN-31 can properly detect Plasmodium infected RBC in patients with sickle cell disease and other diseases that significantly affect red blood cell morphology and/or erythropoiesis.
When an indeterminate result occurs, it is clear that additional examinations by other methods are required to confirm whether Plasmodium infected RBC are present or not to prevent the reporting of a false positive MI-RBC result, which could lead to misdiagnosis. To mitigate the possibility of having a false positive MI-RBC result with potentially serious consequences, it is recommended to evaluate the haemocytometric scattergrams in order to determine whether or not there is distinct cluster formation of particles within the M-gating area where parasites are detected, as seen in a true positive Plasmodium sample. In case of abnormal RBC morphology and a positive malaria result, Plasmodium infected RBC should be confirmed by microscopic examination of thick and/or thin blood films. However, the scattergrams of true positive and false positive MI-RBC samples are different and therefore further refinement of XN-31 gating and interpretation algorithms should be able to increase the specificity.
Finally, there are some limitations of this study. Firstly, the number of included patients for diagnostic accuracy was smaller than anticipated due to travel restrictions during the COVID-19 outbreak. Secondly, the LoD and LoQ studies were performed using P. falciparum parasites cultured in vitro in RBC as there were no patient samples with a parasitaemia high enough to dilute serially across the linear range. Although the use of in vitro cultures on the XN-31 is not approved in the specifications, the results show no interferences were present if the dilution series is prepared by dilution of P. falciparum parasites cultured in RBC in vitro in freshly collected blood of a healthy donor. Using this method accurate results were obtained for the LoB, LoD, and LoQ.