Background
Rapid diagnostic tests (RDTs) have become the focal point of malaria control. The central role of RDTs is the result of a recent paradigm shift in malaria case management, based on the World Health Organization (WHO) 2010 recommendation that all persons thought to have malaria should have their diagnosis confirmed by microscopy or an RDT before treatment [
1]. However, the value of a “test before you treat” policy depends on accurate diagnosis. False-negative tests may delay the provision of life-saving treatment for individual patients and may simultaneously increase the number of persons capable of infecting mosquitoes in the community.
Although microscopy has been used most commonly to detect malaria parasites, it requires equipment, reagents and skilled microscopists [
2]. Thus, in parts of sub-Saharan Africa where microscopy is inaccessible or of low quality, RDTs have become the primary tool for the parasitologic diagnosis or confirmation of malaria [
3,
4]. Since 2005, the proportion of suspected malaria cases examined using a diagnostic test (microscopy or RDT) in sub-Saharan Africa rose from 36% in 2005 to 41% in 2010 and 65% in 2014. In 2014, RDTs accounted for 71% of the diagnostic tests performed in sub-Saharan Africa [
3]. Given the central role RDTs now play in determining whether persons with symptoms have malaria and should be treated, it is increasingly important to understand the factors that influence their performance.
RDTs are immunochromatographic tests which detect proteins released from parasitized red blood cells. Most of the RDTs used currently to diagnose
P. falciparum infections target HRP2 [
5]. HRP2-based RDTs are specific for
P. falciparum because
P. falciparum is the only human parasite that produces HRP2 [
3]. In contrast, RDTs targeting pan-lactate dehydrogenase (pLDH) or aldolase can detect all the
Plasmodium species that infect humans, although they are reported to be less sensitive than HRP2-based tests, especially with low parasite densities [
6,
7]. In regions where
P. falciparum predominates and non-falciparum infections occur as mixed infections with
P. falciparum, including most of sub-Saharan Africa, WHO generally recommends HRP2-based RDTs. In contrast, pLDH and aldolase-based RDTs are recommended in areas where non-falciparum infections predominate. Currently, WHO suggests restricting combined HRP2/pLDH RDTs to regions where
P. falciparum and non-falciparum infections occur as single-species infections [
5].
Although the sensitivity of HRP2-based RDTs has been reported as >90% for
P. falciparum, several groups have reported decreases in the sensitivity of HRP2-based RDTs after decreases in the intensity of transmission [
7‐
9]. For example, in Zanzibar, the sensitivity of HRP2-based RDTs in relation to thick smears fell from 93 to 79% as the percent of malaria-attributable fever episodes in the population decreased from 30 to <3% [
9].
Plasmodium falciparum parasites without the central repeat region of the
hrp2 gene may cause false-negative RDTs [
10,
11] because they fail to produce the HRP2 target molecule for HRP2-based RDTs. Isolates with
hrp2-negative
P. falciparum have now been identified in the blood of infected human subjects in South America, Asia and Africa [
10‐
13]. However, despite the diagnostic threat and malaria control concerns posed by parasites without
hrp2, there is a paucity of data on the frequency of those parasites and the factors driving (responsible for) their selection.
Preliminary studies from Mali have found seasonal fluctuations in the prevalence of false-negative RDTs and suggest the peak prevalence of
hrp2-negative isolates is at the end of the dry season [
8]. However, it is not clear whether the seasonal variation in RDT sensitivity for
P. falciparum infection observed in Mali occurs in East Africa or elsewhere. There is also a need to determine whether
hrp2-negative parasites can be identified using pLDH-based RDTs.
To address these knowledge gaps, this study compared the sensitivities of HRP2- and pLDH-based RDTs at sites with varying intensities of malaria transmission in Rwanda to determine whether deletions of hrp2 were responsible for false-negative HRP2-based RDTs.
Discussion
In sub-Saharan Africa, HRP2 RDTs are the test used most commonly for parasitologic confirmation of malaria before treatment [
5]. However, several reports have noted significant declines in the sensitivity of HRP2 RDTs after declines in the intensity of transmission [
7‐
9]. In Mali, preliminary studies found seasonal declines in RDT sensitivity were associated with peaks in the prevalence of
hrp2-negative
P. falciparum isolates at the end of the dry season [
8]. It is not clear if the association between declining RDT sensitivity and increasing prevalence of
hrp2-negative isolates observed in Mali occurs elsewhere.
Thus, this study compared the sensitivity of HRP2 RDTs at 3 sites with varying intensities of transmission in Rwanda to determine whether deletions of hrp2 were responsible for false-negative HRP2-based RDTs. RDT performance was examined in relation to microscopy which was considered the gold standard. Samples with false-negative HRP2 RDTs (positive smear, negative RDT) were re-examined using PCR to test for the hrp2 gene.
Consistent with previous reports, this study found that HRP2-based RDTs were more sensitive than pLDH-based RDTs, although less specific [
6,
7,
17]. However, when the HRP2 and pLDH RDTs were considered together, sensitivity increased slightly, without a decline in specificity. Using both RDTs, sensitivity was 91.8% and specificity was 85.7%.
Notably, the sensitivity of HRP2-based RDTs varied across the study sites and there was a decrease in the sensitivity of the HRP2 RDT after a fall in malaria transmission. In Rukara, monthly estimates of HRP2 RDT sensitivity ranged from 90 to 97%. Conversely, in Kibirizi, the sensitivity of the HRP2 RDT declined from 88 to 67% after two rounds of IRS as slide positivity rate for symptomatic patients fell from 46 to 3%.
Although IRS appeared to reduce the incidence of malaria in Kibirizi, a recent study in an area of The Gambia with high LLIN coverage found no additional benefit from adding IRS. Of note, in The Gambia, ≥93% LLIN coverage was achieved for all sleeping spaces and IRS coverage was 83–86%. [
21]. In contrast, the 2014–2015 Rwandan DHS found that IRS achieved >98% coverage of targeted areas while only 81% of households had at least one LLIN and 43% of households had an LLIN for every two persons [
16].
Potential explanations for false-negative HRP2 RDTs and the decline of RDT sensitivity in Kibirizi include: loss (deletion) of the
hrp2 gene and low parasite densities [
6,
10,
11].
Parasites lacking the hrp2 gene are a potential source of false-negative HRP2 RDTs. The hrp2 gene is absent in P. falciparum isolates with hrp2 gene deletions and in non-falciparum Plasmodium parasites. PCR analysis was used to identify isolates without the hrp2 gene and to confirm the presence P. falciparum DNA. Of 138 P. falciparum infections with false-negative HRP2 RDTs, 32 were negative by PCR for hrp2 (consistent with deletion of the hrp2 gene).
Plasmodium falciparum isolates lacking the
hrp2 gene appear to be a significant source of false-negative RDTs in Rwanda. However, in this study, most
P. falciparum isolates lacking the
hrp2 gene were detected by the pLDH-based RDT and improved malaria control was not associated with an increased frequency of false-negative RDTs due to
hrp2-negative
P. falciparum isolates (Fig.
5).
For the majority (106/138 = 77%) of
P. falciparum samples with false-negative HRP2 RDT results, PCR for
hrp2 was positive.
P. falciparum isolates containing the
hrp2 gene may produce a false-negative RDT if the parasite density is below the threshold for RDT detection. Although this study lacks quantitative data on parasite density, a positive pLDH RDT may provide a crude estimate of the parasite density (≥200–1000) parasites per microlitre [
22]. Because HRP2-based RDTs are more sensitive than pLDH-based RDTs at low parasite densities, a positive pLDH RDT suggests the parasite density was at or above the threshold for HRP2 RDT detection [
6,
7,
23,
24]. Conversely, microscopy positive/pLDH negative
P. falciparum samples may reflect low density infections.
In this study,
P. falciparum infections with parasite densities below the threshold for detection may be responsible for many of the false-negative RDTs. Of the 106
P. falciparum isolates with false-negative HRP2 RDTs and the
hrp2 gene (confirmed by PCR), most (77%) were negative by pLDH RDT. Additionally, in Kibirizi, the proportion of microscopy positive/pLDH RDT negative samples increased as the slide positivity rate fell (Fig.
6). The proportion of microscopy positive/pLDH negative samples rose from 13.9% (95% CI 12.7–16.0) during April to August 2014 to 38.6% (95% CI 32.1–45.6) during December 2014 to April 2015. This increase in the proportion of microscopy positive/pLDH negative samples may reflect an increase in the proportion of low density infections. Thus, the pLDH RDT data suggest that a decline in parasite density may have contributed to the decrease in HRP2 RDT sensitivity as malaria control improved in Kibirizi.
Other potential causes for false-negative RDTs which were not examined in this study include partial deletions of the
hrp2 gene, prozone effects due to excess antigen, sequence variability of
P. falciparum hrp2 and circulating antibodies to HRP2 which have been reported to interfere with RDT detection of HRP2 [
11,
25‐
28]. While the primers used in this study amplified only exon 2 of the
hrp2 gene, the
hrp2 gene is also known to have chromosomal breaking points outside exon 2 [
28].
There have been several reports of prozone-like effects with HRP2-based RDTs in patients with hyperparasitaemia. Although the mechanism of prozone-like effects for antigen detection tests is not well defined, one plausible explanation is that the amount of antigen may exceed the binding capacity of the dye-labelled antibodies used for antigen detection. In this situation, unlabelled target antigen reaches the test strip and saturates the binding capacity of the capture antibodies affixed to the test strip. As a result, antigen captured by dye-labelled antibodies may be unable to bind to the test strip to form a visible band [
29]. However, false-negative HRP2 test lines attributed to the prozone effect have been described only in samples with ≥288,000 parasites/µL [
30,
31]. While this study lacks data on parasite density, results of previous studies suggest that hyperparasitaemia is unlikely to have been a significant cause of false-negative RDTs in this study [
31].
Of the 343 samples with false-negative HRP2 RDTs that were tested by PCR, 21 were negative by PCR for both
hrp2 and 18S rRNA of the four
Plasmodium parasites known to cause human infection. Sub-optimal PCR sensitivity may have occurred as a result of inadequate DNA sample, degradation of DNA sample, presence of PCR inhibitors and deletion or mutation of the targeted DNA [
20]. Other possible explanations for these discrepancies include false-positive microscopy results and pLDH RDT cross-reactivity with other infectious agents, such as African trypanosomes [
32].
Importantly, there is potential for confusion about the RDT used in this study because Premier Medical Corporation Ltd. submitted two different products with the name “First Response
® Malaria Ag. pLDH/HRP2 Combo Card Test” to WHO for testing. The RDT used in this study, catalogue number I16FRC, was tested in rounds 1, 2 and 5. However, a different product was tested in round 6 (catalogue number PI16FRC). Product I16FRC did not meet WHO recommended procurement criteria during round 5 of WHO RDT lot testing because the panel detection score (PDS) for
P. vivax at 200 parasites per microlitre was 74.5 (below the WHO criterion of ≥75). In contrast, product I16FRC had a satisfactory PDS score of 85.0 for
P. falciparum (please note that PDS is not equivalent to sensitivity) [
6]. These data are available at:
http://www.rdt-interactive-guide.org/ [
33]. Because
P. falciparum is the predominant species in Rwanda, the marginally low sensitivity of this RDT for
P. vivax would not be expected to have a significant impact of the finding of this study [
5].
Finally, the authors recognize the lack of testing for additional single-copy genes is a theoretical limitation because the reported sensitivity of PCR is greater for multi-copy genes (18S rRNA) than single copy genes (
hrp2) (1 vs 10–100 parasite per microlitre) [
34]. However, several factors suggest the parasite densities of the
P. falciparum samples that were negative by PCR for
hrp2 were above the threshold for detection by PCR for single-copy genes: all patients had symptoms consistent with malaria infection, DNA was extracted within 3 months of sample collection and most
P. falciparum isolates without the
hrp2 gene were detected by the pLDH RDT. Most symptomatic individuals in malaria-endemic areas have parasite densities ≥1000 parasites per microlitre [
35] and prompt extraction of DNA limits the time for DNA degradation which may disproportionately reduce the sensitivity of PCR for single-copy genes. Additionally, based on the results from round 5 of WHO RDT quality testing, a positive pLDH RDT suggests the parasite density was above the threshold for detection by PCR for single-copy genes (≥200 parasites per microlitre). The pLDH RDT of the First Response
® Malaria pLDH/HRP2 Combo test (catalogue number I16FRC) had a sensitivity of 31% for wild-type (clinical)
P. falciparum smear-positive samples with 200 parasites per microlitre. In contrast, with parasite densities of 2000 per microlitre, the sensitivity of the pLDH RDT was 100% [
22].
Ultimately, the factors driving the decline in RDT sensitivity as malaria control improves are not clear. If parasite density declines as malaria control improves, the decrease in RDT sensitivity could be driven in part by an increase in the number of infections with parasite densities below the RDT threshold for detection. In addition, there are concerns that
hrp2-negative parasites may have an increased impact on RDT performance as malaria control improves [
10,
11]. Conversely, in high transmission settings,
hrp2 negative parasites may have less impact on RDT sensitivity because individuals are commonly infected with more than one
P. falciparum parasite strain (genotype) and the RDT will yield a true-positive result if any one parasite (genotype) is
hrp2-positive [
10]. However, the number of parasite genotypes infecting individual subjects, the multiplicity of infection (MOI), declines as malaria control improves. Conversely, individuals infected by only a single parasite genotype negative for the
hrp2 gene will produce false-negative results when tested with an HRP2-based RDT [
10]. In the present study, the decrease in HRP2 RDT sensitivity as malaria control improved was not associated with an increased frequency of
hrp2-negative
P. falciparum isolates. However, it may have been associated with an increase in the frequency of low density infections.
Authors’ contributions
CTK, NU, SR, CK and DK contributed substantially to the conception and design of this study. CTK and NU coordinated and supervised field work. CTK, EM and JPH supervised and performed the laboratory analysis. CTK and DK performed the data analysis and prepared the first draft of the manuscript. All authors read and approved the final manuscript.