There is a growing interest in using saliva as an alternative diagnostic medium to blood because of its relative ease of collection and reduced biohazard. This proof-of-concept study was designed to validate the diagnostic potential of saliva for malaria by quantifying levels of Pf HRP2 in clinical samples.
HRP2 as a biomarker in blood and saliva
Pf HRP2 was selected as the target biomarker based on its characterization and precedented use in commercial RDTs.
Pf HRP2 is a multiplet (
M
r
50 to 85 kDA [
12,
13]), but antigenically invariant, water-soluble protein that mediates the formation of haemozoin [
14].
Pf HRP2 is secreted by parasites at all stages, exported through the membrane of the infected red blood cell and then fully released into the blood upon schizont rupture [
15].
Since sequestration of parasitized erythrocytes in the host vasculature is a characteristic feature of
P. falciparum pathology [
16], microscopic measurement alone of parasitaemia in peripheral blood could be an inaccurate indicator of the parasite biomass. Methods to measure circulating
Pf HRP2 were developed to improve estimates of the total parasite burden in cases of extensive sequestration [
17]. Qualitative detection of
Pf HRP2 has become a viable alternative to microscopy to diagnose malaria in remote areas. It is an effective clinical indicator of the severity of past and present parasitaemia. However, the utility of
Pf HRP2 in monitoring patient response to anti-malarial therapy is limited by the persistence of reactive circulating antigen for several weeks post-treatment [
18].
Quantitative studies have measured the levels of
Pf HRP2 in blood and established its diagnostic significance. Parra
et al first identified
Pf HRP2 in the plasma of infected individuals [
19]. Other studies reported average levels of
Pf HRP2 in the range of 1.012 fg per parasite [
20] and 8.53 fg per infected RBC [
21] in culture medium, and 0.57 to 1.11
μ g/mL in plasma [
17]. Generally,
Pf HRP2 is present at higher levels in whole blood than plasma [
22,
23] and is released in larger quantities than pLDH [
21]. The above studies by Desakorn
et al and Kifude
et al also reported a correlation between parasite density and plasma levels of parasite antigens.
In the same way, simultaneous measurement of the parasite density and the concentrations of
Pf HRP2 in plasma and in saliva could reveal a correspondence between the salivary proteome and systemic parasitaemia. Such a study would need to control for the presence of circulating complexed antigen (e.g., with neutralizing IgA) and stage-dependent secretion of
Pf HRP2 [
15]. Furthermore, if
Pf HRP2 persists in the blood of patients following an active infection, it is also likely that the marker persists in the saliva. These studies are the focus of future work. The correlation between parasite density and salivary
Pf HRP2 in the present work was not evaluated due to the size of the study cohort (the power of the correlation coefficient was 0.053).
Negative-control samples were not collected from the endemic area due to the possible persistence of Pf HRP2 in patients with prior exposure but no active infection. Without complete patient history, one could not discern a patient with persistent antigen from a prior infection from one who had never been infected. The presence of persistent Pf HRP2 in the negative controls would have artificially raised the background signal. Since the study measured Plasmodium protein and not host response antibody, it was deemed acceptable to recruit negative controls from a non-endemic population.
The results of the ELISA should be interpreted in light of several factors that may complicate true reconciliation of the assay responses to recombinant
Pf HRP2 and the antigen found in clinical samples. The primary structure of
Pf HRP2 contains numerous repeated sequences and is, therefore, thought to present multiple epitopes for antibody binding [
24]. The degree of multivalence could vary with the size of
Pf HRP2, which differs among strains. While multivalence enhances the detection signal in immunoassays, the final interpretation is complicated by the genetic diversity of the antigen. Secondly, reports of cross-reactivity [
13] suggest that some epitopes on
Pf HRP2 are also present on the highly homologous
Pf HRP3. Thus it is possible that the total assay response also included other parasite histidine-rich proteins present in saliva. Finally, as discussed above, the saliva of semi-immune individuals could contain a mixture of free and antibody-bound antigen. Only the free fraction of
Pf HRP2 would yield ELISA signal.
Saliva collection
Serum molecules can reach saliva through the gingival crevicular fluid and via mechanisms of intracellular and extracellular transport. The transport of a protein into saliva depends on its molecular mass, solubility, ionization [
4], and the salivary pH. Therefore, different molecules can experience varying degrees of dilution during transfer from plasma to saliva. While the precise route followed by
Pf HRP2 is not known, it most likely enters the saliva duct by pericellular ultrafiltration from the surrounding vasculature. Further investigation into the mechanisms may aid optimization of sample collection.
For analysis by ELISA, robust protocols for the collection and stable storage of saliva samples are important to minimize sample degradation. At room temperature, breakdown of salivary proteins occurs within 30 min of collection [
25]. For longer procedures, protein degradation can be mitigated by processing at 4 °C and adding protease inhibitors. Protein degradation may explain why Gbotosho
et al observed decreased sensitivity for antigen detection in saliva samples that were stored overnight [
10].
In the present study, since -80 °C storage was not available in the field, all samples were stored at -20 °C and used within 14 days. The single freeze-thaw cycle was used to denature mucins and improved their separation by centrifugation [
26]. The addition of Tween 20 surfactant to the saliva reduced non-specific binding in the immunoassay.
Complex sample preparation and handling are not amenable to a low-cost rapid test. However, it is expected that short (i.e., under 30 min) analyses of fresh samples would largely circumvent problems of degradation. The removal of mucins could be accomplished by extracting the saliva from a sponge collector [
27]. The integration of such sample preparation would further enable simple processing for saliva rapid tests.
Enzyme-linked immunosorbent assay
Whereas diagnostic development requires absolute quantitation of salivary antigens, previous field studies have only reported qualitative detection using commercial tests designed for higher levels of antigen in blood or plasma [
9,
10,
17,
20]. Rapid diagnostic tests that rely on the accumulation of gold particles in lateral-flow strips do not achieve a sufficiently low limit of detection for use with saliva samples. Wilson
et al drew similar conclusions about colorimetric microplate assay kits, i.e., Malaria Ag CELISA, which has reported LODs of 1.5 to 3.91 ng/ml [
15,
20]. By comparison, an assay suitable for saliva requires a greater signal-to-noise ratio, a lower detection range, and mitigation of matrix effects.
To meet these requirements, this study developed a more sensitive custom chemiluminescent [
28] ELISA for
Pf HRP2 (Figure
2). The pair of antibodies was pre-validated by the vendor for sandwich ELISA. Amplification of the signal was achieved using biotinylated detector antibody with the strong tetravalent binding of the streptavidin-enzyme conjugate. The resulting readout yielded a high signal-to-noise ratio (i.e., compared to colorimetric assays) so that the effect size,
d, between positive and control populations was very large (
d=7). This allowed significant conclusions to be drawn despite the limited number of subjects.
It was important to match the detection range of the ELISA with respect to clinical analyte levels. In the absence of prior reports about the physiological levels of Pf HRP2 in saliva, it was inferred that salivary concentrations of Pf HRP2 would be in the range of 101 to 102 pg/mL based on typical plasma:saliva ratios of protein. Agitation of the samples during incubation provided a useful degree of freedom in the field to tune the detection for clinical samples. The predicted levels of salivary Pf HRP2 were indeed supported by the results of the ELISA. Since the enrollment process of this study favoured symptomatic subjects with relatively high parasite densities, further investigation is required to determine the lowest parasite density that yields detectable salivary Pf HRP2.
The numerous components of saliva matrix aside from the analyte can have a considerable impact on the performance of the ELISA. Collective matrix effects of binding proteins, drugs, degrading enzymes, and heterophilic antibodies, etc. can differ between binding systems [
29]. A common approach to mitigate the matrix effects of saliva is to dilute the sample in a more tractable buffer [
30] and measure it against calibration samples prepared in the same buffer. The dilution of saliva samples with PBS was evaluated, but the minimum required dilution was deemed unsuitable for the detection limits that were required for this investigation. Due to the variation of recovery rates in saliva relative to buffer, it was decided to prepare calibration standards in pooled saliva from malaria-negative donors. Calibration standards were included on each microtitre plate to account for inter-plate variation.
The protease inhibitors added to saliva block the activity of serine proteases, tyrosine phosphatases, and alkaline phosphatases. Their collective reactivity significantly reduced the signal and background of the detection curve (Figure
3). Clearly, the assay performance needs to be carefully assessed in the presence of inhibitors or any other additives involved in the collection procedure.
Design guidelines for saliva immunodiagnostics
Saliva has garnered significant attention as a diagnostic medium for systemic disease, and is particularly attractive as a low-cost, non-invasive approach to meet health needs in developing countries [
3,
31]. As the number of investigations towards such applications is expected to increase, guidelines are offered for the early-stage development of saliva diagnostics.
When selecting a biomarker of systemic disease, one should begin with a short list of those whose detection is well-characterized in blood or its derivative components. Proteins transferred from blood to saliva may be diluted by up to 100,000×, but the dilution factor is not constant for all analytes [
32]. Thus, the physiological range of the target biomarker in saliva should be determined. For pilot studies, physiological levels may not be known
a priori, and available assay kits for serum measurements could lack the sensitivity to detect the biomarker in saliva. Such cases may require the design of a new assay with a suitable detection range.
The matrix effect of saliva on the assay response should be characterized. If the assay is sufficiently sensitive, then the saliva could be diluted. An alternative strategy is to reduce the sample viscosity by removing mucous with mechanical filtration, or chemical digestion by an
in-vitro mucolytic agent (e.g., N-Acetyl Cysteine) [
33]. Non-specific binding can be mitigated by the addition of detergent or a competitive binding molecule. When undiluted saliva is assayed, it would also be useful to prepare calibration standards in a matrix that yields a consistent recovery rate.
The authors further recommend that the collection of oral fluid should be detailed because this can significantly affect the composition of the sample. For example, gingival cervicular fluid differs markedly from saliva, which can differ yet depending on whether a specific gland was targeted and whether the collection was stimulated or resting. Where possible, fresh saliva should be used and kept on ice after centrifugation. If analysis is to be done at a later date, the samples should be refrigerated and stabilized with appropriate inhibitors.