Clinical diagnostic evaluation of HRP2 and pLDH-based rapid diagnostic tests for malaria in an area receiving seasonal malaria chemoprevention in Niger

Rapid diagnostic tests (RDT) for malaria have become the in-the-field diagnostic standard in most countries in Africa. Current malaria RDTs are immunochromatographic tests that detect the presence of circulating parasite antigens. The two most commonly targeted antigens are histidine-rich protein 2 (HRP2), which is specific to Plasmodium falciparum, and Plasmodium lactate dehydrogenase (pLDH), which is present in all Plasmodium species; currently available tests detect one or both antigens. HRP2-based tests have been preferred in areas where P. falciparum is predominant, due to a higher reported sensitivity [1]. There is also some evidence that they are more heat-stable than pLDH-based tests [2]. In vivo, the two antigens are cleared with different speeds, which appears to affect their specificities (notably, their ability to detect current infection) depending on the context. An evaluation in both high- and low-transmission areas in Uganda showed that median time for an HRP2 test to become negative after an effective treatment was 35–42 days, but that the median time to become negative for a pLDH test was only 2 days [3]. This likely explained the higher specificity of the pLDH test (94%) compared to HRP2 test (80%) in the high-transmission area, while specificities were similarly high (~ 99%) in the low-transmission area. The higher specificity of pLDH tests was also seen in hospitalized children in Burkina Faso [4]. These findings have important implications in high-transmission environments, where interpretation of HRP2 tests can be difficult in the peak malaria season, leading to false positive results and hence unnecessary treatments. In part because of these results, a pLDH-based malaria RDT was introduced by the medical humanitarian organization Médecins Sans Frontières in several health structures it supports in the Magaria District of Niger. In all other health centres of the District, and throughout the rest of Niger, HRP2-based tests are used by the Ministry of Public Health and its partners.

The CareStart™ Malaria PAN (pLDH) Ag RDT (Reference RMNM-02571, hereafter referred to as CareStart pLDH(pan) is among the highest-scoring RDTs in the most recent WHO-sponsored evaluation of RDTs [5]. These evaluations do not formally estimate the sensitivity and specificity of the RDTs, but instead use a geographically diverse set of blood samples with known levels of parasitaemia to calculate a ‘Panel Detection Score’ (PDS). In the most recent round, the SD BIOLINE Malaria Antigen (Reference 05FK50, hereafter referred to as SD Bioline HRP2) test had a PDS > 90% for the lowest level parasitaemia (200 parasites/µl), and a false positivity rate < 2%, both of which comfortably meet WHO thresholds for performance. At the lowest level parasitaemia, the PDS for the Carestart pLDH(pan) was slightly lower at 84–86%, but still comfortably met the thresholds, and the false-positivity rate was 0. However, the testing is performed in ideal conditions, so the conclusions are not always transferrable to field conditions. In the recent study in Uganda, sensitivity of the Carestart pLDH (pan) test was estimated at 96.1% (95% CI 92.9–98.1) and 94.7% (95% CI 91.2–97.0) in areas of low and high malaria transmission, respectively; its specificity was estimated at 99.8% (99.6–99.9) and 93.9% (89.6–96.8), respectively [3]. In the same study, the positive predictive values of the test were above 95% in both high and low-transmission environments.

A formal clinical diagnostic evaluation of the newly introduced pLDH-based RDT, in parallel with the previously used HRP2-based RDT, was performed in field conditions in rural Niger. Off-site microscopy by at least two blinded microscopists was the gold standard comparator.

The specificity and PPV of the pLDH RDT were low, much lower than previously described, and very similar to the HRP2 test, which was unexpected considering the different performance and characteristics of these types of tests reported previously [1, 3‐5, 9]. This result is both difficult to explain and potentially problematic as the pLDH test is scaled up across different high-transmission environments.

The prevalence of parasitaemia during the high transmission season was slightly lower than expected (46% vs the 60% expected among febrile children presenting to a health centre), which may be due to the protective effect of SMC. Previous evaluations of the pLDH and HRP2 tests have not been performed in areas receiving SMC. Nonetheless, it is unclear how SMC would affect the performance of an RDT. One possibility would be that SMC could lead to high levels of sub-microscopic parasitaemia, some of which would be detected by the rapid tests but not by microscopy, a phenomenon described elsewhere outside the setting of SMC [10]. This could potentially explain why false positive results during the high season were identical with the HRP2 and pLDH tests. However, it could be expected that the proportion of sub-microscopic infection would change over the course of an SMC distribution, as the protective effect of the drug combination wanes. Thus, the fact that the characteristics remained similar over the course of an SMC distribution cycle would weaken this hypothesis. Although the specificity of the pLDH test improved during the low season, the positive predictive value was still poor. Importantly, the specificity did not vary by production lot.

The second surprising result of this study is the poor sensitivity of the pLDH (91.0%, 95% CI 85.3–95.0) and particularly HRP2 (85.8%, 95% CI 79.3–90.9) tests during the low malaria transmission season. These sensitivities are lower than would be expected [3‐5], even given the overall lower parasite burden during the low transmission season. Indeed, even when restricting the analysis to specimens with parasitaemia > 200 parasites/µl, sensitivities were still low, particularly for the HRP2 test, given that rapid tests based on this antigen typically have sensitivities which approach 100% at these levels of parasitaemia [3, 11, 12]. Even more surprising is that the sensitivity of the HRP2 test was lower than that of the pLDH test, which goes against previous experience that HRP2 tests are more sensitive than pLDH tests [3, 12].

The photographs of RDT cassettes provide a valuable record of what tests look like when performed on ‘real-life’ samples, and are an added value for monitoring, especially given the surprising results obtained. The intensity of bands should not be correlated with severity of infection, but nonetheless, in this evaluation, interesting patterns emerged. The visual appearance of the pLDH test is different from the HRP2 test, with a fainter, slightly hazier line than the HRP2 test. While this does not call into question the accuracy of the test, the experience in the field was that this was less desirable for staff accustomed to the clearer lines of the HRP2 test.

Another hypothesis for the discrepant results between the two tests in the low season would be the presence of HRP2 deletions in the parasite genome, which in turn cause HRP2-based RDTs to be negative even in the presence of Plasmodium [13]. This has been recognized as a problem in some parts of Africa, including Kenya and Eritrea [14, 15]. A nationwide survey in the Democratic Republic of Congo suggested that 6.4% of infections had had HRP2 deletions, and that there was significant spatial clustering [16]. The absence of such discrepancies (pLDH+/HRP2−) in the high transmission season in Magaria suggests that the HRP2 deletions are probably not widespread in the area, but it should be monitored.

One of the limitations of this study is that information about recent use of artemisinin combination therapy (ACT) was not collected. One hypothesis to explain the high false-positivity rate would be that persistent antigenaemia remained high after effective ACT treatment, but given the characteristics of the time to become negative of these tests, all of these false positives would have had to present to the clinic very quickly after finishing their ACT, which seems improbable. The RDTs were stored at ambient temperatures for relatively short periods of time, and those temperatures may have surpassed recommended storage temperatures, so degradation might have been of concern. On the other hand, this mimics usual storage condition for the vast majority of RDTs used in routine clinical care in the Sahel.

PCR was not performed on these samples, which may be a limitation of this study. PCR comes with certain advantages, including that it relies on a standardized process, with less room for human error in interpretation of a result than microscopy, and also that it has a higher ability to detect low-level parasitaemia [17, 18]. For these reasons, PCR and ultra-sensitive RDTs have become standard in malaria elimination settings [19]. On the other hand, in a setting such as Niger during the high transmission season, one could question the clinical significance of a PCR assay that detected a very low-level sub-microscopic parasitaemia that would less likely be the cause of a febrile child’s acute illness [20, 21]. Therefore, in this setting, where the goal was to differentiate clinical malaria from non-malaria, PCR was not chosen as a comparator, relying instead on microscopy as a gold standard, which was believed to be a better indicator of clinical illness in this setting. Microscopy, like PCR and other diagnostic tools, is operator-dependent, but given the results here, the usual lower limits of detection for expert microscopists (between 50 and 100 parasites/µl) seem to have been applicable in this case. Theoretically, it is possible that another antigen-detection method such as ELISA would have been a more appropriate comparator for antigen-based RDTs given that correlations between antigen concentration and parasite level is different for HRP2 and pLDH [22, 23]. However, malaria antigen-specific ELISAs are not routinely available and the study site was not equipped for sample storage for this purpose. More importantly, using ELISA would not have met with the objective of assessing the clinical performance of the RDTs in routine conditions.

In summary, this clinical diagnostic evaluation was undertaken to support the routine use of the pLDH RDTs, but these results showed that the tests had similar performance to an HRP2 test, particularly regarding specificity during the high transmission season. This likely led to a significant amount of over-treatment of malaria, though in this setting, over-treatment is preferable to under-treatment. Nonetheless, as the pLDH test continues to be introduced into new settings, many questions remain. It should be monitored in areas with and without SMC, and future evaluations should include additional PCR analyses, possibly including consideration of HRP2 deletions.