Anti-Pfs25 Human Plasma Reduces Transmission of Plasmodium falciparum Isolates That Have Diverse Genetic Backgrounds

Abstract

Pfs25 is a leading candidate for a malaria transmission-blocking vaccine whose potential has been demonstrated in a phase 1 trial with recombinant Pfs25 formulated with Montanide ISA51. Because of limited sequence polymorphism, the anti-Pfs25 antibodies induced by this vaccine are likely to have transmission-blocking or -reducing activity against most, if not all, field isolates. To test this hypothesis, we evaluated transmission-blocking activities by membrane feeding assay of anti-Pfs25 plasma from the Pfs25/ISA51 phase 1 trial against Plasmodium falciparum parasites from patients in two different geographical regions of the world, Thailand and Burkina Faso. In parallel, parasite isolates from these patients were sequenced for the Pfs25 gene and genotyped for seven microsatellites. The results indicate that despite different genetic backgrounds among parasite isolates, the Pfs25 sequences are highly conserved, with a single nonsynonymous nucleotide polymorphism detected in 1 of 41 patients in Thailand and Burkina Faso. The anti-Pfs25 immune plasma had significantly higher transmission-reducing activity against parasite isolates from the two geographical regions than the nonimmune controls (P < 0.0001).

INTRODUCTION

Despite decades of effort battling malaria, the disease is still a major cause of morbidity and mortality, mainly due to Plasmodium falciparum. Transmission-blocking vaccines (TBVs) that target sexual-stage parasite development within the mosquito midgut are an integral part of the malaria control and elimination plan currently under development (1, 2). Among multiple TBV targets, Pfs25, an antigen expressed on the surface of P. falciparum zygotes and ookinetes in mosquito midguts, is a leading vaccine candidate with which there is substantial evidence of induction of transmission-blocking activity (TBA) or transmission-reducing activity (TRA). A phase 1 trial conducted on U.S. adults testing a recombinant Pfs25 formulated with Montanide ISA51 demonstrated transmission-reducing activities in a standard membrane feeding assay (SMFA) using gametocyte cultures from the NF54 P. falciparum isolate (3).

While Pfs25 gene transcripts were detected in blood-stage gametocytes in the human host, the bulk of the Pfs25 expression commences only after fertilization in the mosquito host (4, 5). Due to the presence of little or no expression of the Pfs25 protein in the human host, it is unlikely to be naturally targeted by adaptive immunity. This is consistent with the absence of detectable anti-Pfs25 antibodies in human sera from areas where P. falciparum is highly endemic (6, 7), suggesting that the protein is not subject to selective pressure to evade the human immune system by antigenic variation and that the Pfs25 gene is likely highly conserved. Indeed, a survey of Pfs25 gene sequences from 9 parasite isolates from various geographical locations and 20 patients in Papua New Guinea revealed only one nonsynonymous mutation (8, 9). Recent large-scale analysis of P. falciparum diversity in natural infections by deep sequencing showed only one synonymous mutation in the Pfs25 gene (10). It is therefore hypothesized that a recombinant Pfs25 vaccine based on the Pfs25 sequence in P. falciparum clone 3D7 of isolate NF54 will induce strain-transcending immunity against field isolates. In this study, we evaluated the transmission-blocking and -reducing activities of human plasma from the Pfs25/Montanide ISA51 trial by a direct membrane feeding assay (DMFA). The assay uses P. falciparum gametocyte-containing blood obtained from malaria patients from two distant countries where the parasite is endemic, Thailand and Burkina Faso, in two distinct local mosquito vectors, Anopheles dirus and Anopheles gambiae, respectively. Our results showed that the anti-Pfs25 human plasma reduces transmission of P. falciparum isolates with diverse genetic backgrounds.

MATERIALS AND METHODS

Processing of plasma.

The anti-Pfs25 plasma used in this study was collected by plasmapheresis of a volunteer from the phase 1 trial of Pfs25/Montanide ISA51, under NIAID IRB-approved protocol 05-I-0118 (3). All subjects provided written informed consent before participation. Plasma from a naïve volunteer (Interstate Blood Bank) was used as a control. Both anti-Pfs25 and naïve plasma samples were heat inactivated prior to processing. In order to avoid clotting due to ABO blood type mismatch, we preadsorbed the plasma with whole AB⁺ type blood (Interstate Blood Bank, Memphis, TN) to remove anti-A and anti-B antibodies, as described previously (11). Briefly, the whole AB⁺ type blood was washed to remove white cells and platelets. The AB⁺ red blood cell pellet was mixed at room temperature with the anti-Pfs25 immune plasma or the naïve-control plasma by gentle rotation for 20 min and was then centrifuged at 2,500 × g for 10 min at 4°C. Supernatant was collected and filtered through a 0.22-μm filter, aliquoted, and stored at −80°C for feeding assays. Using a standard enzyme-linked immunosorbent assay (ELISA) as previously described (3), the anti-Pfs25 titer of this plasma was set to 3,200 ELISA units. This plasma had been tested by SMFA in the United States and had demonstrated 68 to 83% TRA (3).

DMFA.

DMFAs were conducted separately at laboratories in Thailand and Burkina Faso. Both laboratories used their own local laboratory-reared mosquitoes exposed to gametocyte-infected blood from malaria patients in Burkina Faso and Thailand. The gametocyte-infected blood samples from malaria patients were collected under protocols approved by the institutional ethical committee in Burkina Faso (003-2009/CE-CM) and from the Ministry of Public Health in Thailand (under protocol WRAIR 1308). Prior to testing of the immune plasma at the two sites, the standard operating protocols (SOPs) used by both laboratories were reviewed and harmonized to minimize assay variation. The only differences in the SOPs used by the two sites, in addition to the source of gametocytes, were the mosquito species, the volume and the hematocrit of the feeding mixture in the feeder, and the types of membrane used for feeding.

In Thailand, infections were carried out using a colony of the local vector Anopheles dirus and gametocyte carriers were identified among adult patients of the Malaria Clinic, Mae Sot, Tak Province. Gametocyte density in donor carriers was determined by a microscopy read of 500 leukocytes on a thick blood smear. In Burkina Faso, an Anopheles gambiae (M molecular form) colony, established from the local natural vector population, was used. P. falciparum gametocyte carriers were selected among children screened in villages around Bobo Dioulasso. Gametocyte density in donor carriers was determined by a microscopy read of 1,000 leukocytes on a thick blood smear.

The gametocyte-infected blood was collected by venous blood draw and washed with incomplete RPMI medium. The blood was mixed with aliquots of anti-Pfs25 or heat-inactivated naïve control plasma. Taking into account the usual observed hematocrit in patients from the two study sites, the blood cell-to-plasma mixing ratios were 150 μl:180 μl in Thailand and 175 μl:180 μl in Burkina Faso. The mixture was incubated at room temperature for 15 min and transferred in glass feeders covered with Parafilm (in Burkina Faso) or Baudruche membrane (in Thailand) and maintained at 37°C using a water jacket circulation system. Three feeders were used per blood donor-plasma combination in order to avoid a potential feeder effect. Fifty 2- to 3-day-old female A. gambiae or 100 5- to 7-day-old A. dirus mosquitoes were placed in each membrane feeder and were allowed to feed for 30 min, after which all unengorged mosquitoes were removed. Mosquitoes were provided a 10% sucrose solution daily until they were dissected for oocyst counting at 7 days (for A. gambiae) or 9 to 11 days (for A. dirus) after blood feeding.

Transmission-blocking and -reducing activities of the anti-Pfs25 plasma.

We considered two types of activity of the anti-Pfs25 plasma, TBA and TRA. The TBA measures the blocking of infection prevalence and is calculated as follows: TBA = 100 × [1 − (prevalence of infection for mosquitoes fed with immune plasma/prevalence of infection for control group)]. The TRA measures the reduction of oocyst density in infected mosquitoes and is calculated as follows: TRA = 100 × [1 − (mean oocyst number with immune plasma/mean oocyst number in control group)]. We tested the TBA and TRA of the anti-Pfs25 plasma against parasite isolates from 5 gametocyte carriers from Thailand and 7 from Burkina Faso.

Statistical methods.

To evaluate whether we used comparable parasite exposures for mosquito vectors in Thailand and Burkina Faso, we used a Wilcoxon-Mann-Whitney (WMW) test for testing gametocyte density between countries and the t distribution on the log-transformed values for confidence intervals on the geometric mean. To compare gametocyte density to mean oocyst count, proportion of infected mosquitoes, or TRA, we used Spearman correlations with confidence intervals (CI) calculated using rank transform methods.

We modeled the log of the mean oocyst count (for TRA) or the log of the infection prevalence (for TBA) as normally distributed. This assumption is motivated by the central limit theorem, which states that because each statistic comes from a mean (either the mean of the oocyst counts or the mean of the indicators of infection), the normality assumption is reasonable even if the individual mosquito oocyst counts follow a more complicated model such the zero-truncated negative binomial model (12). The log of the mean can be treated as normally distributed using the delta method (13). For a single donor, we estimated the log of the ratio (test over control) of the oocyst means (for TRA) or prevalences (for TBA), using Welch's two-sample t test and the associated confidence intervals. In this case, the observations are the log means for the 6 feeders (3 with tested sera and 3 for control). This method automatically accounts for the feeder-to-feeder variability. The estimates and confidence limits are then transformed to TBA or TRA by the formula 100[1 − exp(logR)], where logR is the estimate (or confidence limit) of the log ratio from the t test. For testing the country effects, we used meta-regression methods (14, 15). Meta-regression methods are typically used for meta-analyses where there are systematic effects of interest between studies. Here each donor acts like a study and the systematic effect is the country effect. We used the mean and variance of the logR effect for each donor from the t test methods just described and allowed for donor variability through a random donor effect. We estimated the ratio of the effect from Burkina Faso over the effect from Thailand, where the effect is the test-over-control ratio for each country (estimated by the exponential of the average logR effect per country). Thus, confidence intervals on that ratio account for both the feeder variability and the donor variability.

Pfs25 sequencing and microsatellite genotyping.

Blood from each gametocyte carrier was spotted on filter paper, dried, wrapped in aluminum foil, and stored at ambient temperature. Blood samples from an additional 29 P. falciparum carriers in Burkina Faso were added for Pfs25 sequencing. Genomic DNA was extracted from each gametocyte carrier blood sample using CTAB extraction buffer (16). The Pfs25 gene was amplified by using forward PCR primer PF1 (5′-CTTTGTTTTCTTCAATTTATTC-3′) and reverse primer PR1 (5′-TCATGGTATTTTTTTTTGTC-3′) upstream of the Pfs25 gene start codon and downstream of the Pfs25 stop codon, respectively. Both strands were sequenced using the BigDye Terminator, version 3.1, cycle sequencing kit (Applied Biosystems) and an Applied Biosystems sequencer. Sequences were assembled and verified using SeqScape (Applied Biosystems). The Pfs25 sequence from the NF54 P. falciparum isolate (8) was used as a reference sequence.

Microsatellites were genotyped for P. falciparum in blood from each gametocyte carrier using fluorescence-labeled forward primers. The markers used in this study (B5M5, C14M17, C13M63, B5M124, BM17, C1M67, and C1M4) were described previously (17, 18). Fragment analyses were conducted with a sequencer (Applied Biosystems), and alleles were sized relative to an internal size standard using GENEMAPPER 4.0 (Applied Biosystems).

RESULTS

Transmission-blocking and -reducing activities of the anti-Pfs25 plasma on P. falciparum gametocyte isolates in Thailand and Burkina Faso.

The DMFAs were conducted by feeding vector mosquitoes on blood from P. falciparum gametocyte carriers. The geometric mean of gametocyte density of carriers in Burkina Faso was lower (133/μl [95% CI, 78 to 227/μl]) than that of Thai carriers (409/μl [95% CI, 62 to 2,706/μl]), although there was no significant difference between the 2 countries (P = 0.146 by WMW test). To evaluate the effect of gametocyte density on baseline mosquito infectivity, we calculated the Spearman correlation (r_s) between gametocyte density and the proportion of infected mosquitoes or the mean oocyst density in mosquitoes from feeding cups using control plasma. We found no significant correlation between proportion of infected mosquitoes and gametocyte density in the blood meal (r_s = 0.46; 95% CI, −0.16, 0.82; P = 0.137), but there appears to be a strong positive correlation between the mean oocyst density in mosquitoes and gametocyte density (r_s = 0.88; 95% CI, 0.62, 0.97; P = 0.0002).

We observed significant reductions in oocyst density in 10 of 12 feeding assays for mosquitoes fed with gametocyte-containing blood mixed with anti-Pfs25 plasma compared to mosquitoes fed on the same infectious blood mixed with control sera (Table 1 and Fig. 1). The gametocyte density in the blood meal, however, did not strongly impact the TRA (r_s = −0.18; 95% CI, −0.68, 0.44; P = 0.58), as the mean oocyst density in the controls was used to standardize the TRA. The activity may also be measured by TBA: inhibition of mosquito infection (prevalence). We observed a significant correlation between TRA and TBA by anti-Pfs25 plasma (r_s = 0.76; 95% CI, 0.32, 0.93; P = 0.0045). However, there were fewer significant TBA events, with only 7 of 12 cases significantly blocking infection (Table 1 and Fig. 2). This lack of significance may be due to the high oocyst density in the infection control typically observed in these assays (19) (Fig. 2).

Table 1.

Transmission-blocking and -reducing activities in DMFAs conducted in Thailand and Burkina Faso^a

Study site	Gametocyte carrier	No. of mosquitoes infected/no. dissected		% Reduction in prevalencec (95% CI)	P valuec	Oocyst density mean; median (IQRb)		% Reduction in oocyst density (95% CI)c	P valuec
		Control plasma	Test plasma			Control plasma	Test plasma
Thailand	Thai 1	96/120	17/120	82.5 (72.1, 89.0)	0.003	20.1; 9 (1–30)	0.7; 0 (0–0)	96.9 (88.9, 99.2)	0.0021
	Thai 2	52/60	34/60	34.8 (14.5, 50.3)	0.013	144.1; 140 (96–236)	12.9; 2 (0–14)	91.8 (66.5, 98.0)	0.0097
	Thai 3	74/120	17/120	77.4 (59.2, 87.5)	0.003	2.6; 2 (0–4)	0.3; 0 (0–0)	91.2 (56.9, 98.2)	0.0190
	Thai 4	51/60	48/60	5.6 (−20.5, 26.0)	0.512	149.3; 112 (8–246)	97.2; 58 (5–166)	33.8 (−67.1, 73.8)	0.2827
	Thai 5	79/84	63/66	−1.1 (−15.1, 11.2)	0.772	222.9; 162 (49–345)	70.0; 53 (22–120)	71.0 (−21.7, 93.1)	0.0743
Burkina Faso	BF 1	53/63	53/73	14.9 (−7.6, 32.7)	0.125	33.4; 28 (6–49)	6.3; 4 (0–9)	80.6 (58.5, 91.0)	0.0059
	BF 2	45/47	93/109	10.5 (−10.4, 27.5)	0.181	86.4; 88 (24–133)	22.0; 16 (5–33)	73.6 (63.6, 80.9)	0.0003
	BF 3	80/93	60/96	27.1 (5.2,43.9)	0.034	13.7; 13 (3–23)	3.9; 2 (0–7)	72.0 (63.9, 78.2)	0.0002
	BF 4	42/68	19/61	48.5 (1.2, 73.2)	0.047	11.4; 3 (0–16)	2.0; 0 (0–2)	80.0 (29.2, 94.4)	0.0251
	BF 5	64/82	67/104	19.2 (−10.4, 40.8)	0.099	54.9; 22 (2–92)	6.8; 2 (0–7)	87.1 (76.1, 93.1)	0.0015
	BF 6	60/75	36/86	48.7 (31.9, 61.3)	0.003	11.1; 5 (1–14)	1.6; 0 (0–2)	84.6 (66.2, 93.0)	0.0050
	BF 7	110/132	65/130	40.1 (25.8, 51.6)	0.003	14.0; 13 (2–21)	2.7; 1 (0–4)	80.9 (72.4, 86.7)	0.0003

^aThe DMFAs were conducted using A. dirus and A. gambiae mosquitoes in Thailand and Burkina Faso, respectively.

^bInterquartile range (IQR) is the difference between the third and the first quartiles.

^cPercent reduction was calculated by comparing the infection prevalence or the oocyst count in mosquitoes fed with anti-Pfs25 test plasma and a nonimmune control plasma after combining information from feeders using log transformations. For details on the estimation and calculation of the 95% confidence intervals and P values, see the statistical methods section.

Fig 1.

TRA measured as percent reduction on mean oocyst density. For each subject, we give the estimate and the 95% confidence interval. The meta-analysis columns represent the overall estimate from that country with the 95% confidence interval calculated by meta-regression.

Fig 2.

TBA measured as percent reduction on infection prevalence. For each subject, we give the estimate and the 95% confidence interval. The meta-analysis columns represent the overall estimate from that country with the 95% confidence interval calculated by meta-regression.

Meta-regression analysis was conducted to determine whether TBAs and TRAs, as measured by inhibition of infection prevalence and oocyst density, respectively, determined in Thailand and Burkina Faso are comparable. The average TBAs were 51.2% (95% CI, 15.8%, 71.8%) for Thailand and 31.0% (95% CI, −9.4%, 56.5%) for Burkina Faso, whereas the average TRAs were estimated as 86.1% (95% CI, 68.6%, 93.8%) for Thailand and 80.3% (95% CI, 64.1%, 89.2%) for Burkina Faso. Percent inhibition was calculated as 100 × (1 − R), with R being the ratio of test to control. The ratio of the ratios (RR) was used to compare the two countries ((RR = R (Burkina Faso)/R (Thailand)). A value of RR close to 1 would mean that the two countries have comparable effects. We estimate the RR_TBA to be 1.41 (95% CI, 0.69, 2.89) and the RR_TRA to be 1.41 (95% CI, 0.51, 3.88). Despite the higher-than-one values for both RR_TBA and RR_TRA, we find no significant differences between the two sites, and the confidence intervals of RR_TBA and RR_TRA rule out extreme differences.

The meta-regression methodology allows us to adjust for effects of gametocyte density on RRs by adding a log gametocyte density term to the linear meta-regression model. The adjusted RR_TRA (RR_TRA[adj] = 1.87; 95% CI, 0.57, 6.13; P = 0.26) shows no significant country effect, whereas the gametocyte-adjusted RR_TBA (RR_TBA[adj] = 2.21; 95% CI, 1.16, 4.20; P = 0.021) does show a significant country effect, indicating that DMFA parameters other than gametocyte density may contribute to the TBA readout.

Pfs25 protein is highly conserved among field isolates with independent genetic backgrounds.

The recombinant Pfs25 in the Pfs25/ISA51 vaccine was manufactured based on the sequence of isolate NF54 of P. falciparum clone 3D7. We genotyped parasites from gametocyte carriers involved in the study by using 7 microsatellite markers and sequenced the Pfs25 genes in these parasites and additional samples.

Genotyping was conducted using DNA from human blood samples containing both sexual and asexual blood-stage parasites. The number of alleles detected by microsatellite markers allowed us to estimate the genetic diversity among the parasite population in the patients. Parasite isolates from 4 of 5 Thai carriers revealed only one allele for each of the 7 microsatellite markers, whereas 2 alleles were detected for 1 microsatellite in parasites from the fifth carrier (Table 2). The microsatellite haplotypes identified in the individual Thai carriers were different, indicating that while carriers from Thailand were infected with one or two parasites, each was infected with a different isolate. In Burkina Faso, however, parasite genetic diversity was much greater. The 7 Burkina Faso gametocyte carriers showed evidence of multiple infections, with 2 to 9 different haplotypes detected per donor; at least 42 distinct genotypes or clones were present within this group (Table 2).

Table 2.

P. falciparum microsatellite allele sizes in blood samples from gametocyte carriers

Study site	Gametocyte carrier	Allele size (base pairs) of microsatellite markerc							Haplotype no.a
		B5M5	C14M17	C13M63	B5M124	BM17	C1M67	C1M4
Thailand	Thai 1	221	174	192	220	131	200	189	1
	Thai 2	226	203	202	189	131	209	214	1
	Thai 3	214	186	201	177	127	207	200	1
	Thai 4	NDb	196	381	152	132	205	184	1
	Thai 5	230	167/163	240	171	128	199	205	2
Burkina Faso	BF 1	192/204/207 211/224/230	180/192/202	191/212/217 226/253	159/171/200/211/220 226/239/247/261	131/136 138 151	189/193/205 210/212	178/185/189 200/206	9
	BF 2	183/192/207 211/217/230	180/184/194 205/215/226	197/208/215	183/200/207/213/220 232/247	131/142 144	183/189/195 199	186/189/204 209	7
	BF 3	211/217/220 227/237	182/194/202	203/223	177/189/200/211/232 247/260	131/138 144	183/187/199 214	190/192/200 214	7
	BF 4	198/204/217 220/230	178/186/190 207/215	199/203/216 223/272	194/207/213/220/232	138/142 147	183/193/205	180/185/189 193/200	5
	BF 5	207/224	228	219	200/211	131/143	185	194/200	2
	BF 6	211/220/224 233	180/192	219	194/200/213/220/226	131/138	189/208	184/200	5
	BF 7	192/204/211 217/220/224	180/192/207 213	206/215	177/183/188/194/211 239/260	130/144	183/189/201 203/210/216	178/193/206	7

^aNumber of detected parasite haplotypes in the human blood donor, according to the maximal number of alleles observed.

^bND, not detected.

^cMultiple alleles in mixed infection are separated by slashes.

Despite the highly divergent genetic backgrounds of parasites among the carriers from the 2 countries, Pfs25 sequences revealed a very low level of diversity. Among the Thai isolates, one nonsynonymous mutation was observed compared to the previously published NF54 isolate (8). This mutation observed in the blood donor identified as Thai 3 has previously been detected in other geographical areas (http://plasmodb.org/) and results in a glycine-to-alanine substitution. No synonymous polymorphisms were detected among the Thai isolates. In Burkina Faso, no polymorphism, synonymous or nonsynonymous, compared to the NF54 isolate was detected in the Pfs25 gene among all 7 gametocyte carriers and 29 additional parasite carriers.

Interestingly, the Pfs25 gene sequence of the donor with the lowest TBA (Thai 4) was identical to that in NF54. Conversely, the antisera of the donor with the Pfs25 gene substitution for alanine (Thai 3) had highly significant TBA (P_TBA = 0.003 and P_TRA = 0.019). These results from our study suggest that sequence polymorphism may have little impact on efficacy of recombinant Pfs25 vaccines based on sequences in the NF54 isolate.

DISCUSSION

Previously, Arakawa and colleagues demonstrated effective transmission-blocking activity against field isolates using anti-Pfs25 sera raised in rodents (20). Similarly, in a pilot study, we observed that pooled rabbit sera from animals immunized with Pfs25/Montanide ISA51 inhibited parasites from a Thai malaria patient with a single Pfs25 haplotype identical to the 3D7 strain (data not shown). This article is the first report that human anti-Pfs25 sera can confer transmission-blocking activity in a DMFA against field parasites responsible for local transmission. It supports our hypothesis that Pfs25 vaccines based on the NF54 isolate sequence can induce strain-transcending immunity. The limited polymorphism of Pfs25 may be central to strain-transcending immunity and may facilitate development of a Pfs25-based transmission-blocking vaccine. We recognize that this proof of concept was based on a single human serum with a relatively high anti-Pfs25 titer and that the observation needs to be confirmed with immune sera from future vaccine trials.

To minimize assay variability between two sites, we harmonized the DMFA SOPs for blood processing, feeder setup, feed duration, and temperature, while gametocyte carrier recruitment (patients in Thai clinics versus asymptomatic Burkina Faso volunteers), hematocrit in blood meal, and mosquito species were adapted to local conditions. Our results demonstrated similar anti-Pfs25 serum transmission-reducing and/or transmission-blocking activity against local mosquito vectors in both sites.

We observed strong efficacy of the anti-Pfs25 plasma in reducing oocyst intensity (mean TRAs were above 80% in both Thailand and Burkina Faso) and a more limited blocking activity (mean TBAs were 31% and 51.2% in Thailand and Burkina Faso, respectively). The same plasma had been tested in SMFA in the United States and had demonstrated 68 to 83% TRA (3). Further analyses of the same data sets revealed lower TBAs (0 to 11%) in multiple assays, with mean oocyst densities of 5.1 to 52.1 in control mosquitoes fed with an autologous preimmune serum (Y. Wu, unpublished data). The TRA and TBA results of DMFAs in Burkina Faso and Thailand seem comparable to those of SMFAs in the United States. We therefore observed similar efficacies of the anti-Pfs25 plasma against cultured and field parasites.

It was anticipated that the anti-Pfs25 plasma would show a lower effect on mosquito infection prevalence (TBA) than on oocyst density (TRA) in these assays. As recently highlighted by Churcher and colleagues, variation in oocyst density associated with little or no change in prevalence occurs at high levels of infection, with high infection prevalence and high oocyst density. This is due to the relationship between prevalence and oocyst density, in which infection prevalence reaches a plateau close to 100% when oocyst density is high (19). Consequently, TBA is more dependent on control infection intensity than TRA, and a higher TBA may be observed when mosquitoes are fed on blood meals with lower infectious gametocyte densities. In the present study, oocyst densities in the control groups were high as defined by Churcher and colleagues (19), probably as a result of selecting volunteer blood donors with higher gametocyte densities, whereas in natural settings, wild mosquito vectors may be infected by gametocyte carriers at a submicroscopic detection level (21). Actual exposure of mosquitoes to P. falciparum in natural settings remains poorly documented (22, 23). Studies to characterize the parasite intensity in naturally infected malaria vectors and subsequent transmissibility to human populations could potentially fill a major gap of knowledge for evaluating TBV candidates at relevant infection levels.