Background

Malaria continues to be a global scourge of humans. According to the World Health Organization (WHO), the latest estimate of malaria burden was 229 million cases in 2019, of which 409,000 died [1]. A mosquito-borne disease, malaria is caused by the infection of protozoan parasites of the genus Plasmodium. The vast majority of clinical incidences of malaria are due to Plasmodium falciparum and Plasmodium vivax. In the last two decades, the global malaria reduction effort has led to a significant decrease in the case number, but the progress appeared to have slowed down in the last few years [1]. Because malaria is vector-borne, disease control and elimination strategies rely heavily on vector control measures, such as insecticide-treated nets and indoor residual spraying. Additional means to reduce transmissions such as prompt treatment and reporting, reactive surveillance, and use of the 8-aminoquinoline drug primaquine to inactivate gametocytes have also been adopted by various countries [1,2,3].

Most P. vivax infections in endemic areas are asymptomatic, even in low transmission settings approaching elimination [4,5,6,7,8,9]. Because asymptomatic carriers can transmit the disease [10,11,12,13,14,15,16], they represent a challenge for malaria elimination. At present, the extent to which P. vivax carriers contribute to transmission has not been quantified in most settings. Although asymptomatic carriers are much more prevalent than patients, they are less infectious to the mosquitoes due to the lower parasite density in blood. The relative contributions of asymptomatic carriers and patients in a population to malaria transmission depend on the relative prevalence, infectivity, and contact frequency with the mosquito vectors. Few studies have attempted to assess the relative infectivity of asymptomatic carriers and patients [12,13,14,15,16,17]. Recent studies from Ethiopia and Amazonia predicted that asymptomatic carriers contribute approximately 30–80% of transmission [14, 16]. In mainland Southeast Asia, it was recently found that human asymptomatic infections and mosquito infections arose closely in time after mass drug administration (MDA) of dihydroartemisinin-piperaquine with low dose primaquine, suggesting that asymptomatic carriers be the key drivers of transmission [18].

There remains a gap in understanding of how well P. vivax transmits from humans to mosquitoes. Although many mosquito feeding assays have been performed and together provided a valuable dataset of P. vivax infectivity [10,11,12,13,14,15, 19,20,21,22,23,24,25,26,27], the majority of data were obtained at the clinical level of parasite densities. Far fewer data have been acquired at low-density levels typical of asymptomatic carriers [28]. This study aims to fill this gap by determining the quantitative relationship between P. vivax parasite density and mosquito infectivity over a broad range of parasitaemia. Serial dilution of patient blood was used to generate parasite densities at different levels. The membrane feeding assay (MFA) was used with laboratory-reared Anopheles dirus, a major vector in Southeast Asia, to measure P. vivax transmission.

Methods

Blood collection from Plasmodium vivax patients

Plasmodium vivax-infected blood was collected from symptomatic patients who attended malaria clinics in Tak and Ubon Ratchathani provinces of Thailand. The study was approved by the Ethics Committee of Faculty of Tropical Medicine, Mahidol University (MUTM 201 L-040-05). Enrollment was limited to malaria patients who were ≥  18 years old. From each patient, venous blood (20 ml) was collected in a heparinized tube. Blood was immediately placed in a warm box (37 ℃ ± 5 ℃) and transported to a field laboratory or the Mahidol Vivax Research Unit (MVRU) in Bangkok for MFA.

Blood preparation and membrane feeding assay

For every P. vivax-infected blood sample, P. vivax mono-species infection was confirmed by nested PCR as previously described [29, 30]. Blood was washed with pre-warmed serum-free RPMI 1640 medium and centrifuged at 800 × g for 10 min at 37 °C. Packed infected blood was resuspended with an equal volume of naïve human AB serum to 50% haematocrit (hct). The resuspended blood was subjected to 2-fold serial dilution with warmed 50% hct-O cells in human AB serum, generating 13 different levels of parasitaemia. The final volume for each dilution was 600 µl. The diluted blood was kept at 37 °C until MFA. A thick-film slide was prepared at each parasite concentration for microscopic examination by dotting 1 µL of blood at per spot, five spots per side. The remaining blood was used to feed 100 female An. dirus mosquitoes using MFA.

Light microscopic examination of blood smears

Thick blood smears were stained with 10% Giemsa solution for 10 min, rinsed with running tap water, and air dried. Total parasites per 1 µl spot were counted once under a light microscope (LM) for all original and serially diluted blood samples fed to the mosquitoes. Counts were made separately for asexual and sexual stages.

Nested PCR analysis

To confirm P. vivax mono-infection, nested PCR was performed as previously described [29, 30], using genus and species-specific primers against the 18 S ribosomal RNA genes. Briefly, DNA was extracted by QIAamp® DNA Mini Kit (Qiagen, Germany) according to the manufacturer’s instructions. The first round of PCR was performed using the purified DNA as the template with genus-specific primers. The PCR product was then used as the DNA template for a second PCR reaction which used the same forward primer with a species-specific reverse primer. The second PCR was performed in 5 separate reactions, each detecting P. falciparum, P. vivax, Plasmodium malariae, Plasmodium ovale, or Plasmodium knowlesi DNA. Water was used as the template for the negative controls. The PCR products were visualized on agarose gel after electrophoresis.

Mosquito rearing

A colony of An. dirus was maintained in the insectary of the MVRU. The colony was established in 2011 from the original colony obtained from the Armed Force Research Institute of Medical Sciences, Bangkok, Thailand, as previously described [25]. Briefly, the mosquitoes were reared at 26–27 °C (± 1 °C), 70–80% (± 10%) relative humidity, and with a 12 h day/night cycle. For MFA, female mosquito cartons were placed inside an insulated plastic cooler and ground transported from Bangkok to the field sites. For MFA, female mosquitoes (only 5 to 7 day-old mosquitoes) were used after 6 h sugar starvation. Approximately 100 starved mosquitoes were used for each diluted blood sample.

Membrane feeding assay (MFA)

To feed 100 female mosquitoes, 400 µL of each diluted blood was added to a water-jacketed glass membrane feeder covered with a Baudruche membrane and maintained at 37 °C with a circulating-water system to prevent the transition of gametocytes to gametes. Mosquitoes were allowed to feed for 30 min. The engorged mosquitoes were maintained with 10% sucrose solution. All engorged-mosquito midguts were dissected on day 7 post feeding, stained with 0.05% mercurochrome, and examined under an LM. Two parameters were determined for each feeding experiment: (i) mosquito infection rate (percent oocyst-positive mosquitoes in all mosquitoes dissected) and (ii) oocyst intensity (the mean oocyst count in all mosquito dissected).

Data analysis

When available, parasite and gametocyte densities used in the analysis were direct microscopic counts. At low densities, when no parasite was detected under the microscope, the densities were imputed based on the known fold-dilution of the original samples. All imputed values were less than 5 parasites (or gametocytes) per µl. To determine the parasite (or gametocyte) densities yielding 10% and 50% oocyst prevalence, the rising phase of each dose response (from 0% up to 85% infection rate) was fit to the Hill’s equation using Quest Graph IC50 Calculator [31] with two free parameters, the Hill coefficient and the half-maximal concentration; the minimum and maximum infection rates were set to 0% and 100%, respectively.

Results

General characteristics of P. vivax isolates

Eight P. vivax-infected blood samples were obtained from different malaria patients who visited a health facility in Thailand in 2016–2017. All infections were confirmed to be mono-species by nested PCR [30]. The infection characteristics are summarized in Table 1. The geometric mean parasite density was 1328 (CI95 593–2973) parasites/µL. The geometric mean gametocyte density was 153 (CI95 92–254) gametocytes/µL. The mean female/male ratio was 4.4 (CI95 1.6–7.2).

Table 1 Characteristics of P. vivax isolates used in membrane feeding assays

Relationship between parasite density and mosquito infection

Twelve parasite densities were obtained for each parasite isolate in addition to the original parasite density. The lowest data point represents a 4096-fold reduction of parasite density relative to the original. The parasite density of each diluted blood sample was re-evaluated by LM to ensure that no gross error was introduced by serial dilution. Figure 1 demonstrates the consistency between the LM counts and the calculated parasite densities based on known dilution factors. Actual counts were used in all analysis, except when no parasite was detected, in which case the imputed parasite density value was used.

Fig. 1
figure 1

Plasmodium vivax densities in serially diluted blood samples. Shown are the relationships between actual LM counts and calculated parasite densities based on the dilution factors for a total blood-stage parasites, b total gametocytes, c female gametocytes and d male gametocytes. For each plot, data from all 8 cases were combined. Only data with calculated densities above 3 per microliter were included. Trendlines represent the best linear fits with zero crossing on the log-log scale

In all cases, the MFA showed that both the mosquito infection rate and the oocyst intensity declined as the infected blood was diluted (Fig. 2). In 5/8 cases, a complete or nearly complete dose response was observed. In these cases, the infection rate was over 80% at the highest parasite density (no dilution) and decreased to 0% after the blood was sufficiently diluted. In the three other cases, the infection rate at the highest parasite density was 37–58% resulting in a partial dose response. The lowest infective parasite density (IPL), i.e. the parasite density causing infection in at least one mosquito, ranged from 2 to 129 parasites/µl. The lowest infective gametocyte density (IGL) ranged from 0.2 to 5 gametocytes/µl (Table 2).

Fig. 2
figure 2

Mosquito infection as a function of parasite or gametocyte densities. a mosquito infection rate vs. total parasitaemia, b the oocyst intensity vs. total parasitaemia, c mosquito infection rate vs. total gametocytaemia and d the oocyst intensity vs. total gametocytaemia

Table 2 Lowest infective parasite and gametocyte densities

The dose response of the mosquito infection rate was further analyzed by non-linear regression. For each parasite isolate, the characteristic parasitaemia resulting in 10% and 50% mosquito infection, referred to as IP10 and IP50, respectively, were determined by fitting the dose response to Hill’s equation and solving the best-fit equation for the 10% and 50% infection rates respectively. The infective gametocytaemias (IG10 and IG50) were similarly determined. Table 3 displays these parameters for each parasite isolates. The geometric means of IP10 and IG10 were 33 (CI95% 9–120) parasites/µl and 4 (CI95% 1–17) gametocytes/ µl, respectively. The geometric means of IP50 and IG50 were 146 (CI95%  36–586) parasites/µl and 13 (CI95%  3–49) gametocytes/µl, respectively. Thus, on average, 33 parasites/µl (or 4 gametocytes/µl) was required to achieve 10% infection in An. dirus. Similarly, 146 parasites/µl (or 13 gametocytes/µl) was required to infect 50% of the mosquitoes.

Table 3 Characteristic parasitaemia and gametocytaemia

Relationship between the oocyst density and mosquito infection rate

The data from the feeding experiments of the 8 isolates demonstrated a well-defined and robust relationship between the mosquito infection rate and the oocyst intensity (Fig. 3). The trend (filled symbols) increased monotonically and plateaued near the 100% infection rate. This relationship is nearly identical to the previously published one (open symbol) [25] despite the two studies being conducted several years apart. Therefore, the relationship between the mosquito infection rate and the oocyst intensity is robust; one parameter can reasonably predict the other.

Fig. 3
figure 3

Relationship between oocyst density and mosquito infection rate. Black circles represent values from individual feeding experiments in this study. White circles represent values from individual feeding experiments with AB serum replacement from a previous study [25]. The solid red line is the best fit by Hill’s equation

Discussion

The relationship between P. vivax parasite density in blood and mosquito infectivity is important for estimating the potential of each infected individual to transmission. Several studies have examined this relationship [12, 14, 20, 21, 23,24,25], but the results are variable, presumably reflecting biological differences between different mosquito species and/or parasite isolates as well as methodological differences in the feeding assays [28]. For example, in a study from Ethiopia using Anopheles arabiensis and Anopheles pharoensis [23], the infection rate by MFA was found to increase with gametocyte density and plateaued at about 30–40%. In the same study, the gametocyte density of 50–100 gametocytes/µl was needed to attain the sizable infection rate of 20%. A separate study using An. arabiensis [24] also found that similarly high gametocyte densities were needed to achieve this level of infection. In contrast, in a study using An. dirus [25], only a few P. vivax gametocytes/µl were sufficient to result in a 20% infection rate. In this study, MFAs were performed using serially diluted patient blood with the specific aim of generating a dataset with a balanced parasite density distribution. It was found that the lowest parasite density that gave rise to mosquito infection ranged from 2 to 129 parasites/µl, or from undetectable to 5 gametocytes/µl. While these density values provide a glimpse of transmission potential at low parasite densities, they have limited utility. This is because the values of IPL and IGL depend on the absolute number of mosquitoes examined for infection. As more mosquitoes are examined, IPL and IGL will tend to be lower due to the increased oocyst detection sensitivity. Because of this, it is preferable to characterize the mosquito infectivity by characteristic concentrations like IP10 or IG10. In this study, the geometric mean of IP10 was 33 (CI95% 9–120) parasites/µl and the geometric mean of IG10 was 4 (CI95% 1–17) gametocytes/µl. The 95% confidence interval was broad, reflecting high variation across different feeding experiments. Of note, the value of IP10 is similar to the value predicted by the trendline in a previous study [25], which used 94 independent P. vivax isolates, each at a single parasite density. Thus, the serial dilution MFA provided comparable mosquito infection data with far fewer parasite isolates.

The value of IP10 of 33 parasites/µl suggests that many asymptomatic P. vivax carriers have non-negligible potential to transmit. According to the estimated density distribution of P. vivax in Southeast Asia [32], 22% of asymptomatic carriers are predicted to have parasite density higher than 33 parasites/µl. With the following set of simple assumptions: (a) An. dirus as the sole vector, (b) asymptomatic carriers representing 95% of all P. vivax infections at a given time, (c) 22% of carriers transmitting at the 10% infection rate (or higher), (d) carriers receiving twice more mosquito bites than patients based on the relative outdoor/indoor biting rates [33], and (e) all P. vivax patients transmitting at the average 80% infection rate [25], the contribution of asymptomatic carriers to onward transmission would be similar to that of patients, if not higher. Thus, in Southeast Asia, asymptomatic carriers are likely a critical infectious reservoir. Consistently, in a previous MDA study, it was found that P. vivax asymptomatic infections in humans and wild mosquitoes arose closely in time after the blood stage clearance by the MDA, suggesting a close relationship between the two [18].

This study also has a few limitations. Using serum replacement and serial dilution in MFA imposed two unnatural conditions that may impact the interpretation of the findings. Firstly, MFA may not fully reflect natural transmission through skin feeding. Although studies have suggested that MFA is an acceptable surrogate for natural feeding [19, 34, 35], these studies were conducted at high parasitaemia using patient blood. There is currently no evidence for, or against, the use of MFA for determining transmission at very low P. vivax density. It is plausible that gametocytes are sequestered in the skin to promote their uptake during blood feeding [36, 37], in which case the true potential of P. vivax transmission would have been underestimated by MFA. On the other hand, other unknown factors may also impede transmission from the skin in which case MFA would have overestimated the transmission. Secondly, serially diluting blood in naive AB serum with O red cells removed the immunological factors in the original blood which can influence transmission. Plasma components such as naturally acquired antibodies have been shown to modulate parasite transmission, either by inhibiting it [38, 39] or promoting it [40]. Other soluble mediators including cytokines could also interfere [35, 41,42,43]. With serum replacement, the experiments in this study did not capture these effects. However, in transmission settings where malaria has become low for several years, the effect of antibodies is likely to be small. This is because naturally acquired transmission-blocking immunity had a short duration, declining significantly within months without boosting [39, 44]. Consistent with this, the infectivity dose response in a previous study conducted in a low transmission setting of Thailand was not significantly affected by plasma replacement [25].

Conclusion

The present study quantified the sexual transmission of P. vivax at different parasite dilutions, encompassing the parasitaemia range found in asymptomatic blood-stage parasite carriers. The findings provide critical information for estimating the contribution of asymptomatic infections in malaria transmission. The data suggest that the asymptomatic reservoir is an important source of transmission.