Background
Plasmodium vivax remains a major cause of malaria [
1,
2], with its importance, relative to
P. falciparum, increasing as case management and conventional vector-control measures such as long-lasting insecticidal bednets reduce transmission to low levels [
3]. This difficulty in achieving pre-elimination levels of
P. vivax transmission is largely attributed to the parasite’s particular biology, with long-lasting liver-stage hypnozoites causing blood-stage relapses weeks to years following the initial infection. These relapses are thought to account for up to 80% of all detected
P. vivax bloodstage infections [
4,
5] and contribute to sustained malaria transmission. These hypnozoites form a hidden parasite reservoir that needs to be targeted in order to effectively reduce the
vivax malaria burden.
The current mainstay of
P. vivax control is treatment with 8-aminoquinolines such as primaquine or tafenoquine, the only drugs effective at clearing hypnozoites. However, their administration requires careful medical assessment due to the risk of potentially inducing severe hemolysis in patients with glucose-6-phosphate dehydrogenase deficiency (G6PDd), an X-linked genetic disorder affecting ~5% of the population in malaria-endemic regions with geographical variations [
6‐
8]. G6PDd may be diagnosed either by point-of-care rapid diagnostic tests or using quantitative laboratory assays. Primaquine efficacy is affected by dosage and patient adherence. The duration and dose of a primaquine regimen vary greatly in both duration of treatment (7 or 14 days) and total dose (3.5–7mg/kg) [
9]. Poor adherence is regularly reported [
10] which may substantially undermine the efficacy of radical cure of
vivax malaria, resulting in the development of short-course high-dose primaquine regimens [
11]. In recent clinical trials, tafenoquine has shown promising results with comparable efficacy to low-dose primaquine (3.5mg/kg total dose) in preventing relapses with only a single dose [
12]. While a patient’s G6PDd status affects eligibility for 8-aminoquinline treatment, the efficacy of these drugs can also be altered by insufficient dosing or low CYP2D6 metabolization, for which treatment failures have been observed previously [
13,
14].
For the treatment of individuals with symptomatic
P. vivax infection, the risk of prescribing 8-aminoquinolines needs to be balanced against the benefit to the patient, accounting for the high probability that they carry hypnozoites and will experience future relapses with associated chronic morbidity and sustain onward. For public health interventions where entire populations are targeted for treatment with 8-aminoquinolines, the balance between benefit and risk is different, given that a large proportion of individuals will not be hypnozoite carriers. One strategy is mass drug administration (MDA) where all individuals are offered treatment, subject only to some basic eligibility criteria (e.g., age, pregnancy status). MDA with 8-aminoquinolines has been predicted to be highly effective in aiding
P. vivax control and elimination [
4,
15]. However, given the high prevalence of G6PDd in most malaria endemic settings, MDA with 8-aminoquinolines without prior G6PD testing would expose many individuals (including those without hypnozoites) to potentially dangerous drugs and is thus considered unsafe in most remaining
P. vivax endemic areas. As a consequence, WHO does not currently recommend the use of MDA with 8-aminoquinolines for
P. vivax malaria [
16]. An alternative to MDA is mass screen and treat (MSAT) where individuals are first tested for blood-stage parasites with a rapid diagnostic test (RDT) or light microscopy, and only individuals with detectable blood-stage parasites are treated. MSAT with blood-stage drugs and primaquine has been shown not to have an effect on
P. vivax transmission [
17]. This is thought to be because a high proportion of
P. vivax blood-stage infections and all
P. vivax liver-stage infections are not detectable by RDTs or light microscopy.
The existing options for population-level interventions that are based on treatment with 8-aminoquinolines either lack impact (MSAT) or expose populations to a high degree of risk (MDA). By measuring antibodies to a panel of
P. vivax antigens, it is possible to identify individuals with recent blood-stage infections who are likely to carry hypnozoites and target them for treatment with 8-aminoquinolines [
18].
P. vivax serological test and treat (
PvSeroTAT), where people are screened by a serological test and, if classified as exposed, treated with a combination blood- and liver-stage treatment could thus provide a potentially safe and efficacious alternative for population-level treatment (see malaria serology use case 5 in Greenhouse
et al. [
19]).
Future programmatic implementation of
PvSeroTAT will depend on having effective, field-deployable
P. vivax sero-diagnostic tests (SDTs). As with any diagnostic test, a
P. vivax SDT will have to trade off sensitivity and specificity, and thus, the ability to detect all true positives requiring treatment while limiting overtreatment. The current diagnostic performance of these serological markers is approximately 80% sensitivity and 80% specificity [
18]. Here, we model a wider range of test performance values (50–100%) and explore the relationship between the
P. vivax SDT performance and
PvSeroTAT public health impact as well as overtreatment in order to inform
P. vivax SDT target product profiles (TPPs). The mathematical model of
vivax transmission from White
et al. [
20,
21] was adapted to implement
PvSeroTAT campaigns at the population level. We explored the range of possible sensitivities and specificities of this serological diagnostic tool and evaluated the corresponding drop in
P. vivax prevalence after evaluating different
PvSeroTAT scenarios. These were compared to those potentially achievable in MDA or MSAT campaigns.
Discussion
Similar to previous observations from systematic reviews of field studies, our models predict that MDA programs have a strong but transient effect on
vivax malaria transmission [
23] that is estimated to be higher in low transmission settings [
24,
25], but result in a large proportion of people receiving unnecessary treatments. In contrast with MDA, MSAT would result in poor impact in both transmission settings, but no overtreatment. This confirms the finding from a clinical trial in Indonesia [
17] that treating only blood-stage
vivax cases was not enough to lower the overall transmission.
PvSeroTAT is a newly proposed public health intervention where community members are screened using an SDT and those classified as currently and previously exposed to
P. vivax infections treated with anti-blood- and liver-stage therapy [
18,
19].
PvSeroTAT can bridge the gap between MDA and MSAT by achieving an impact on transmission that is close to that of MDA while dramatically reducing the level of overtreatment. Using a test with performance comparable to the currently available research assay (i.e., 80% sensitivity and 80% specificity [
18]),
PvSeroTAT is predicted to achieve approximately 80–85% of the impact of MDA while reducing the number of people over-treated by 80%. The modeling of different combinations of SDT diagnostic performance showed that under both high and low transmission setting and all implementation scenarios the impact of 1 to 3 rounds of
PvSeroTAT impact is directly dependent on the sensitivity of the sero-diagnostic test (Table
2 & Additional File
2: Table S2, Fig.
3). The degree of overtreatment was however related entirely to test specificity.
This has important implications for
P. vivax SDT development. Trade-offs between sensitivity and specificity during SDT development results in related trade-offs between public health impact and the rate of overtreatment. Yet, we found that multiple rounds of interventions with a lower sensitivity test may achieve the same impact as fewer rounds with a more sensitive one. For example, a public health campaign consisting of three rounds of
PvSeroTAT 6 months apart, at 70% sensitivity and 70% specificity would yield a public health impact comparable to that of two rounds at 90% sensitivity and 90% specificity (Tables
2 and 3). However, three rounds at 70/70 will result in more than four times as many people being unnecessarily treated compared to two rounds at 90/90. On the other hand, three rounds of
PvSeroTAT at 80% sensitivity and 80% specificity diagnostic performance is predicted to result in a similar impact to that achieved with two MDA rounds but accompanied by a 3-fold reduction in overtreatment.
As with other public health interventions, the impact is predicted to decrease as the implementation scenario moves from “best-case” to “real-life”. While results from the ideal scenario provide a higher boundary of treatment efficacy (i.e., universal eligibility for hypnozoiticidal drug treatment), the differences in impact observed between the “high efficacy” and the “real-life” scenario are attributable to decreased efficacy and adherence to primaquine regimens. This highlights the importance of strong locally led programmatic implementation and the need for effective communication and engagement with treated populations. Shorter treatment with primaquine at increased dosage, such as that used in the IMPROV study [
11] or a single dose of tafenoquine may also provide increased adherence. Both the use of a
P. vivax SDT with higher sensitivity or an increase in the number of
PvSeroTAT rounds delivered can at least partially counteract the impact decays inherent in “real-life” programmatic implementation. In a recent Phase III clinical trial, tafenoquine demonstrated non-inferiority with respect to low-dose Primaquine regimen to prevent hypnozoite-caused relapses. Therefore, although we did not model explicitly Tafenoquine due to it not being readily available in many endemic regions, we expect its efficacy to be similar to what was observed in our “high efficacy” scenario.
A key output from our model was that
PvSeroTAT alone was not enough to reach elimination either in low or moderate transmission settings. Rather than modeling elimination scenarios, we set out to estimate the expected magnitude of reduction in PCR prevalence upon implementing a population-level public health campaign. We propose
PvSeroTAT to be used as a tool with high potential for temporarily shrinking the human parasite reservoir, jointly with tools such as vector control or other measures targeting residual pockets of transmission (e.g., reactive case-detection, RCD) and vector populations. In low-transmission settings, some countries have started to implement RCD to focus interventions around newly identified index cases. A notable example is the 1–3–7 surveillance and response strategy developed in China with the aim to report new cases within 1 day, investigate these within 3 days, and take action within 7 days [
26‐
28]. As part of these actions,
PvSeroTAT could be used to improve upon microscopy-based detection methods (i.e., MSAT) upon investigating possible asymptomatic infections linked to the index case. A limitation of our model lies in the absence of geographically-explicit enhanced actions to be taken in the vicinity of infections treated by the routine case-management health system.
In our modeling approach, overtreatment was defined as administering hypnozoiticidal drugs to an individual without a blood-stage infection in the previous 9 months. Most
P. vivax relapses are expected to occur within 9 months of the primary infection [
29]. Overtreatment is therefore the result of false positive diagnostics (i.e., blood-stage infection happened more than 9 months previously). A different definition of overtreatment, based on presence or absence of hypnozoites, would result in somewhat different estimates of overtreatment as the antibody levels may indicate a positive diagnostic signal for individuals who had a blood-stage infection in the last 9 months but do not harbor hypnozoites anymore.
A limitation of our analysis is that we do not account for potential variation between geographic regions and across transmission intensities, factors which are known to affect diagnostic accuracy [
30]. Another limitation of this analysis is that we did not account for the effects of targeting diagnostics and treatment strategies at high-risk populations. For example, in Cambodia, malaria transmission was found to be mostly associated with occupational activities such as forestry or living close to forested areas [
31]. The effectiveness of
PvSeroTAT interventions could be optimized by targeting people who are at increased risk of malaria infection, for example when transmission is mostly occupational. Furthermore, malaria transmission fluctuates seasonally and we did not set out to investigate the timing of interventions. Further studies will assess the impact of single or multiple rounds happening either during low or peak transmission season, so that public health campaigns can be enforced at the time where they will most reduce
P. vivax transmission.
Conclusion
Current antibody panels allow for detecting individuals who are at a high risk of
P. vivax relapse with a diagnostic performance of approximately 80% sensitivity and 80% specificity. We predict that
PvSeroTAT campaigns may achieve an impact similar to that of MDAs, with the benefit of massively reducing overexposure to primaquine, thus lowering the risks of G6PDd-induced haemolysis. To achieve effective
PvSeroTAT implementation, the development of
P. vivax SDTs should aim for both high performance and easy field-use thus facilitating deployment of multiple
PvSeroTAT rounds. Further cost-impact modeling will be required to determine if a more expensive but higher performing test may have a better cost-effectiveness than a lower performing test that requires an additional
PvSeroTAT round to achieve the same public health impact. We suggest that with a field-deployable test achieving a diagnostic performance of the current research assay (i.e., 80% sensitivity and 80% sensitivity [
18],
PvSeroTAT would provide sufficient public-health impact for programmatic implementation in countries that have already achieved a reduction in transmission levels to low-to-moderate levels (2–10%
P. vivax prevalence by qPCR), where highest
PvSeroTAT impact is expected. The target product profiles informing
P. vivax SDT development should reflect the trade-offs between impact, overtreatment, and ease of programmatic implementation that are identified here.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.