Estimating the Prevalence of Asymptomatic Malaria Parasite Carriage in Southern Ghana: Utility of Molecular and Serological Diagnostic Tools

Background. Asymptomatic malaria infections can serve as potential reservoirs for malaria transmission. These infections range from microscopic to submicroscopic densities, making an accurate estimation of asymptomatic parasite carriage highly dependent on the sensitivity of the tool used for the diagnosis. This study sought to evaluate the sensitivities of a variety of molecular and serological diagnostic tool at determining the prevalence of asymptomatic Plasmodium falciparum parasite infections in two communities of varying malaria parasite prevalence. Methods. Whole blood from 194 afebrile participants aged between 6- and 70-years old living in a high (Obom) and a low (Asutsuare) malaria transmission setting of Ghana was used in this study. Thick and thin blood smears, an HRP2-based malaria rapid diagnostic test (RDT) and lter paper dried blood spots (DBS) were prepared from each blood sample. Genomic DNA was extracted from the remaining blood and used in Plasmodium specic photo-induced electron transfer polymerase chain reaction (PET-PCR) and Nested PCR, whilst the HRP2 antigen content of the DBS was estimated using a bead immunoassay. Comparison of prevalence as determined by each method was performed. at 37.8% and Nested PCR, HRP2 bead and microscopy Parasite at 50.1%, 11.2%, and Nested PCR, HRP2 bead PET-PCR, RDT and microscopy respectively.


Background
Asymptomatic parasite carriage in Plasmodium falciparum infections is a well-known phenomenon [1]. Previously, it was assumed that residents of high transmission areas were at a greater risk of harboring asymptomatic (subclinical) infections as a result of acquired immunity to clinical manifestation of the infection developed over repeated exposures [1,2]. However, recent studies conducted in lowtransmission areas of malaria endemic countries, especially in Africa have identi ed a high prevalence of asymptomatic P. falciparum carriers [3]. Asymptomatic Plasmodium carriage in low transmission settings has been suggested to be responsible for 20-50% of all malaria transmission in those settings [4].
Recent estimates of high asymptomatic parasite carriage in low transmission settings could be due to the sensitivity of the parasite detection tools used, where highly sensitive molecular tools increase parasite prevalence estimates [5]. Light microscopy, the gold standard for laboratory con rmation of malaria [6] has a sensitivity of detection ranging from 30-50 parasites/µl of blood according to Gilles 1993 [7] or 50-500 parasites/µl according to Moody et al [6]. In addition to having low sensitivity, microscopy is dependent on the quality of reagents and techniques used in preparing and staining the smear [8] as well as the expertise of the microscopist who examines the smear [9]. These limitations and the di culty of deploying microscopy to all testing sites have led to the expansion of tools used in malaria diagnosis and detection of infection to include tools such as the rapid diagnostic tests (RDTs), with a sensitivity of ~ 100 parasites/µl [6,9] and molecular tools such as polymerase chain reaction (PCR), with a sensitivity of about 2-5 parasites/µl of blood for Nested PCR [10] and 0.01 to 1 parasite/µl of blood for real time PCR [11].
Although the main rationale to improve malaria diagnostic tools is to ensure prompt and accurate parasite detection and treatment of clinical cases, the new diagnostic tools are frequently used by Malaria Control Programs to assess parasite carriage in population surveys [5,12,13].
Malaria RDTs are predominantly based on the histidine-rich protein (HRP2) and or lactate dehydrogenase (LDH) antigens, and despite RDTs having a similar sensitivity to microscopy [14][15][16], their ease of use and fast turnaround time have made them a preferred diagnostic tool [17,18]. The most commonly used malaria RDTs are the HRP2-based tests, because of abundant production of the protein by the parasite and its enhanced sensitivity compared to LDH [19,20]. A major limitation of RDTs is the fact that they are not quantitative [21]. Additional limitations of HRP2-based RDTs include a high false positive rate due to HRP2 antigen persisting in the blood for up to four weeks after the clearance of an active infection [22] and increasing reports of false negative results due to parasites not producing HRP2 as a result of pfhrp2 gene deletions [23].
A recently developed tool for detecting parasite antigen is a sensitive HRP2 bead assay, which can simultaneously measure multiple parasite antigens including HRP2, LDH and aldolase. The HRP2 bead assay has a limit of detection of 0.24 pg/mL, 1.43 pg/mL or 71.9 pg/mL for three unique forms of HRP2 antigens (Type A, B, and C, respectively) that are captured by the beads [24]. The main disadvantage of the HRP2 bead assay is that it cannot be used as a point of care test [24][25][26].
Molecular diagnosis of malaria largely comprises the use of a wide variety of polymerase chain reaction (PCR) platforms to detect parasite nucleic acids. A photoelectron induced transfer PCR (PET-PCR), has a limit of detection of 3.2, 5.8, 3.5 and 5 parasites/µl for P. falciparum, P. ovale, P. malariae and P. vivax respectively, and the possibility of duplex detection of both P. falciparum and another human Plasmodium species in a single reaction is presently available [27]. PET-PCR has also been optimized for use in detecting asymptomatic malaria parasite carriers in large community surveys [24]. Although molecular tools are more sensitive than microscopy and RDTs, they are not suitable for point of care diagnosis as they are time-consuming and require expensive specialized equipment and reagents as well as highly-skilled personnel to run them [10].
This study evaluated the sensitivities of a variety of malaria parasite detection tools; microscopy, HRP2based malaria RDT, HRP2 bead assay, PET-PCR and Nested PCR in determining the prevalence of asymptomatic P. falciparum parasite carriage amongst participants from two communities with varying malaria parasite prevalence in southern Ghana.

Ethical consideration
Ethical approval for the study was obtained from the Institutional Review Board of the Noguchi Memorial Institute for Medical Research, Ghana (Study number 089/14-15). Written informed consent, assent and parental consent (for children) were obtained from all study participants.

Study site and population
This pilot study used consecutive sampling to select 194 participants from a larger cross-sectional study conducted in Obom and Asutusare during the off-peak malaria season (February 2016) [28]. Participants from the larger study were aged between 6 and 70 years old and selected based on the absence of any sign or symptom suggestive of malaria. Obom is a high malaria parasite prevalence setting in the Ga South municipality of Greater Accra Region of Ghana ( Figure 1) with a microscopy estimated parasite prevalence of 35% in 2014 [5,12] and 41.8% in 2019 [29]. Asutsuare is a low malaria parasite prevalence setting in the Shai Osudoku District of the Greater Accra Region of Ghana. Microscopy estimates of parasite prevalence in Asutsuare was 8.9% in 2009 [30] and 3.6% in 2016 [31]. According to the WHO, an annual parasites prevalence of 1-10 % is considered as low and ≥ 35 % considered as high [32].

Sample collection and processing
Prior to sample collection, the axillary temperature of each participant was measured using a digital thermometer. Venous blood (5 ml) was collected from each volunteer into EDTA vacutainer® blood collection tubes (BD, New Jersey, USA). An aliquot of the blood was used to prepare thick and thin blood smears for microscopy. The blood smears were air dried, xed (thin lm only) and stained with Giemsa following the WHO standard protocol [8,33]. The slides were observed at 100X magni cation under a light microscope by two microscopists working independently. A sample was scored as negative for malaria if no parasite was seen after observing 200 elds and scored positive if parasites were observed.
Parasite density (PD/μl) was determined as the number of malaria parasite observed per 200 white blood cells (WBCs) X 40, with the assumption that 1 μl of blood contains 8,000 WBCs [34].
Additionally, 5 μl of the blood was used for P. falciparum diagnosis using the Malaria Pf (HRP2) Ag RDT Multi Kit (Access Bio Inc, New Jersey, USA), following the manufacturer's instructions.
Four, 50 μl drops of blood sample were spotted on Whatman #3 lter paper (GE Life sciences, USA). The lter paper blood spots were individually air dried and stored at room temperature in a sealed plastic bag containing a desiccant. The remaining blood from each volunteer was separated into plasma and packed blood cells, which were subsequently stored frozen at -20 o C until required. All samples from the eld were subsequently transported to the Immunology Department of the NMIMR for further processing and analysis. An aliquot of the whole blood was sent to the CDC (USA) for analysis. DNA extraction DNA for the Nested PCR was extracted at the NMIMR from two 3 mm disks punched out of the DBS using the Chelex extraction method as previously described [35]. Whereas DNA for the PET-PCR was extracted at the CDC from 200 μl of packed blood cell pellets using the QIAamp DNA Mini Kits (Qiagen, USA) according to the manufacturers protocol. The DNA extracted from both procedures was either stored at 4 °C for immediate use or stored at -20 °C for later use.

Nested PCR
The Nested PCR ampli cation of the P. falciparum 18S rRNA gene was adapted from Singh et al. [36] with slight modi cation as previously reported [12]. Brie y, 200 nM dNTPs, 2 mM MgCl 2 , 133 nM each of forward (rPLU6) and reverse (rPLU5) primers (Additional le Table s1) and 1 U OneTaq DNA polymerase (NEB, UK) was used to amplify the 18S rRNA gene from 5 μl (~20 ng) of DNA in the primary PCR. The secondary PCR was performed using similar concentrations of reagents as in the primary reaction mix; however, rFal1 (forward) and rFal2 (reverse) primers were used to amplify 1 μl of the primary product. The cycling parameters for the primary and secondary reactions are listed in Additional le Table s1. Genomic DNA from the 3D7 strain of P. falciparum (MRA 102G) was used as the positive control sample and a distilled water (no template) served as the negative control sample. Positive and negative control samples were included in each PCR reaction set up. The ampli ed PCR products were separated alongside a 100 bp ladder (New England Biolabs, UK) on a 2% agarose gel stained with Ethidium bromide. The gels were subsequently viewed under ultra violet light using the FUSION-FX7 advanced (Vilber Lourmat, Germany) chemiluminescence documentation system. All PCR assays were performed using the Eppendorf Mastercycler Nexus PCR Cycler (Eppendorf, UK).

PET-PCR
The multiplex PET-PCR assay was performed as previously described [27]. Brie y, the ampli cation of Plasmodium genus was performed in a 20 μl reaction containing 2 μl (~20 ng) of each DNA template, TaqMan Environmental buffer 2.0 (Applied BioSystems, USA), 125 nM each of forward and reverse primers (Additional le Table s1) except for the P. falciparum HEX-labeled primer which was used at a 62.5 nM. The cycling parameters used were an initial denaturation at 95 °C for 10 min, followed by 45 cycles of denaturation at 95 °C for 10 sec, annealing at 60 °C for 40 sec. Genomic DNA from the 3D7 strain of P. falciparum (CDC, USA) was used as positive control. All assays were performed in duplicate and using the Agilent Mx3005pro thermal cycler (Agilent technologies, USA).

HRP2 bead assay
The HRP2 concentrations (pg/ml) of each sample was determined using an HRP2 bead assay previously described by Rogier et al. [24]. Brie y, a 6 mm disc was punched out of the dried blood spot (DBS) and incubated over night in 200 μl of buffer (0.05% phosphate-buffered saline, Tween 20, 0.5% bovine serum albumin, 0.1% casein, 0.5% polyvinlyalcohol, 0.5% polyvinylpyrrolidine, 0.05% NaN 3 , and 3 μg/mL E. coli extract to prevent non-speci c binding). The test samples, buffer blank and negative controls (O+ Packed red blood cells that tested negative by two RDT kits and microscopy) were added in duplicate on each plate. Following the assay incubation steps, 100 μl PBS was added to each well and was incubated at room temperature with shaking for 1 min. The plate was subsequently read on a Luminex-200 machine (Luminex Corporation, USA) with a target of 50 beads per region.

Data analysis
All samples that yielded visible fragments after agarose gel electrophoresis or positive CT values after real time PCR analysis were classi ed as positive for the particular PCR reaction. The CT cut off for the PET-PCR was set at 40. The mean threshold uorescence intensity (MFI) positivity cut off value for the HRP2 bead assay was the log normal of the background signal (bg) + 3 SD of the MFI value for the malaria-negative population. IBM SPSS version 20 was used to generate the descriptive statistics including median and to compare median age, hemoglobin and temperature between the two sites. Graph Pad Prism version 7 was used to determine Pearson Chi-Square for sex and parasite prevalence estimated by RDT, microscopy, Nested PCR and HRP2 bead assay, Mann-Whitney test for age and Cohen's kappa test was used to determine the level of agreement between parasite prevalence estimates determined by two different tests (RDT, microscopy, Nested PCR and HRP2 bead assay). Statistical signi cance was set as P ≤ 0.05, unless otherwise stated. Kappa values of < 0 are classi ed as no agreement (disagreement), 0.0 -0.20 are classi ed as poor agreement; 0.21 -0.40 are classi ed as fair agreement; 0.41 -0.60 are classi ed as moderate agreement and values of 0.61 -0.80 classi ed as substantial agreement and 0.81-1.0 as an almost perfect agreement [37].

Demographics
Of the 194 participants, 105 (54.1%) were residents of Obom, a high parasite prevalence area and 89 (45.9%) were residents of Asutsuare, a low parasite prevalence area. There was no signi cant difference (p= 0.652) in the distribution of males between the two study sites (53% in Asutsuare and 49% in Obom) (Table 1) or in terms of age (p= 0.109). The median (IQR) age of participants from Obom was 14 (12 -24.3) years and the median (IQR) age in Asutsuare was 16 (13 -25.8). Estimation of parasite prevalence and density by microscopy A total of 19.1% (20/105) and 2.2% (2/89) of the samples were identi ed containing P. falciparum by microscopy in the high (Obom) and low (Asutsuare) transmission sites respectively ( Figure 2 and Table  1). One of the samples from Obom contained a mixture of P. falciparum and P. malariae (however, this was not con rmed by PCR). A higher number of P. falciparum parasite carriers were detected in the high parasite prevalence setting (Obom) relative to the low parasite prevalence setting of (Asutsuare) (Pearson Chi-Square, p= 0.0002) ( Table 1). Parasite density estimated per microlitre (PD/µl) blood from Obom ranged between 32 and 5080 with a median (IQR) of 180 (80-405), whilst in Asutsuare, both samples that tested positive by microscopy had a parasite density of 40 ( Figure 3).

Estimation of parasite prevalence based on antigen detection
RDT positivity rates The HRP2-RDT identi ed a total of 37.8% (39/101) of the samples collected from the high transmission area as positive. RDT results were not available for 4 samples from the high parasite prevalence area.
None of the samples from the low parasite prevalence setting of Asutsuare tested positive by the HRP2 RDT ( Figure 2 and Table 1).

HRP2 Bead detection of Plasmodium antigen levels
Detection of the P. falciparum HRP2 antigen was signi cantly higher in Obom (61.9%) when compared to Asutsuare ( Figure 4B).
There were 8 samples from Obom that tested positive for P. falciparum by all the ve methods tested ( Figure S1), whilst 8 samples were negative by the same ve methods ( Figure S1). In the low transmission setting, no sample was identi ed as positive by all the methods ( Figure S1), whilst 35 samples were identi ed as negative by all the ve tests.
Comparison among sensitive detection methods by areas In the high transmission setting, parasite prevalence estimated by Nested PCR was signi cantly higher than that estimated by PET-PCR and the HRP2 bead assay (Pearson Chi square=13.06 and 6.76 respectively, p< 0.001 for both), but parasite prevalence estimated by the HRP2 bead assay and PET-PCR were similar (Pearson Chi square=31.89 and p> 0.05) (Table S3).
In the low transmission setting, parasite prevalence estimated by the HRP2 bead assay was signi cantly higher than that recorded by PET-PCR (Fisher's Exact Test p< 0.000) (Table S3) and the difference between parasite prevalence estimated by both Nested PCR and PET-PCR on the one hand and Nested PCR and the HRP2 bead assay on the other were similar (Fisher's Exact Test p=1.000 and 0.156 respectively).

Comparison between antigen detection methods by areas
The HRP2 bead assay detection of HRP2 antigen levels identi ed a signi cantly higher number of positive samples compared to the HRP2 based RDT in the high malaria transmission setting (Pearson Chi-Square=17.22, p< 0.001) (Figure 2 and 5B). Comparisons could not be made in the low transmission site, as no sample tested positive by HRP2 RDT ( Figure 5D, Table 2 and Table S3).

Comparison between nucleic acid detection methods by areas
Nested PCR identi ed a signi cantly higher number of positive samples compared to PET-PCR in both the high transmission setting, Obom (Pearson Chi-Square=13.06, p< 0.001) (Figure 2, 5A and 5C) and the low transmission setting -Asutsuare. In comparing diagnostic methods that measure similar parasite features, HRP2 antigen (RDT and the HRP2 bead assay) and parasite DNA (Nested PCR and PET-PCR), fair and signi cant agreements were observed only for the samples collected from the high transmission setting (Obom) ( Table 2). A crosstabulation analysis between PET-PCR and the HRP2 bead assay found that the two methods agreed substantially and signi cantly (Cohen kappa value = 67.9%, p= 0.004).

Agreement between diagnostic tests
Microscopy is generally referred to as the gold standard diagnostic test for malaria. When results from the microscopy read out by the microscopists used in this study was set as the gold standard (reference test) (Table 3), the level of agreement between microscopy and the PET-PCR and the HRP2 bead assay tests in Obom was poor, with a fair agreement observed between results obtained by microscopy and RDT. In Asutsuare, the interrater agreement between microscopy and both PET-PCR and Nested PCR was poor but the agreement between microscopy and the HRP2 bead assay was fair. All the poor agreements were not signi cant, whilst the fair agreements were signi cant. There was no agreement between microscopy and nPCR in both Obom and Asutsuare (Table 3).
When Cohen's kappa analysis (Table 3) was repeated with Nested PCR set as the gold standard (reference), there was a poor agreement between Nested PCR and RDT but fair agreement between Nested PCR and PET-PCR and the HRP2 bead assay in Obom, whilst in Asutsuare, all the agreements were poor. All the agreements in Obom were signi cant whilst those in Asustuare were not signi cant. There was no agreement between the results obtained by Nested PCR that was compared to microscopy in Obom (Table 3).

Discussion
This study independently utilized ve different diagnostic tools, PET-PCR, a HRP2 bead assay in addition to commonly used HRP2 RDT, microscopy and Nested PCR to determine the prevalence of asymptomatic P. falciparum parasite carriage in two communities with varied malaria parasite prevalence in southern Ghana. Asymptomatic malaria infections are usually characterized by low and submicroscopic parasite densities [38] and depending on the transmission intensity of the area can contain lower than 100 parasites per microlitre [39,40]. Relying solely on microscopy for detecting malaria infections containing such low parasite densities will likely result in missing many infections. Although microscopy, RDTs and nPCR are routinely used for malaria diagnosis in Ghana, PET-PCR and the HRP2 bead assay that are known to be more sensitive than microscopy at detecting low density parasitaemia [24,41] are rarely used. The sensitivities of various combinations of commonly used malaria diagnostic tools have been compared in different malaria endemic countries, including Ghana [5], none of the studies conducted in Ghana have compared the performance of PET-PCR and an HRP2 bead assay to microscopy, an HRP2 based RDT and nPCR at determining malaria parasite prevalence in different settings of Ghana. This study was conducted to evaluate the sensitivities of especially PET-PCR and the HRP2 bead assay as effective tools to detect asymptomatic malaria parasite carriage in settings of varying parasite prevalence in Ghana.
In this study, microscopy and HRP2-based RDT, the most commonly used malaria diagnostic tests in community surveillance studies in malaria endemic countries [42] produced the lowest estimates of asymptomatic parasite carriage in both the high and low malaria parasite prevalence settings. This was not surprising as the parasite densities of infections in samples from even the high parasite prevalence setting was very low. Asymptomatic infections are noted to contain low (submicroscopic) parasite densities [43], below the limit of detection of both microscopy and RDT kits [42].
In this study, parasite prevalence detected by the RDT was higher than the microscopy in the high transmission setting, but a reverse trend was observed in the low transmission setting. One likely reason for these results could be that the HRP2 antigen concentrations measured in samples from the high parasite prevalence setting were often higher and can be detected by the RDT than in the low parasite prevalence setting where it is below the detection limit of the RDT [51,52]. Higher levels of HRP2 antigen could also result from longer duration of antigen persistence in the high parasite prevalence setting due to more frequent infection. This would account for the higher positivity rates detected compared to microscopy in Obom but not in the low parasite prevalence setting (Asutsuare). Persistence of the HRP2 antigen after the clearance of infecting parasites is a well-known phenomenon [44,45]. Consequently, HRP2 based malaria RDT kits may test positive for HRP2 antigens in the absence of an active infection.
Additionally, as demonstrated in the study sites described here, parasite densities in low transmission settings are generally low and likely to be below the limit of detection of the RDT and microscopy especially in the off peak season [31].
Although the parasite prevalence estimated by PET-PCR, nPCR and the HRP2 bead assay were similar in the high prevalence setting, the level of agreement among the three tests was low. This observation may be due to differences in limits of detection, assay targets and other fundamental differences between the methods. Persistence of HRP2 antigen for up to four weeks following a resolved P. falciparum infection can result in false positive HRP2 bead assay results, whilst parasites with deletions in the hrp2 gene (not tested in this study) can cause false negative tests [46][47][48]. Nested PCR protocols generally have much higher numbers of ampli cation cycles compared to real time PCR protocols including PET-PCR and as such are likely to detect and amplify lower template concentrations than real time PCR. Nested PCR has previously been found to be more sensitive than PET PCR [49,50]. However, the increased number of steps involved in nPCR make it more tedious and prone to contaminations and other operator errors that can increase the number of false negative as well as false positive test results compared with real time PCR processes.
When the results obtained from PET-PCR and nPCR, both DNA-detecting, were compared to the results from the HRP2 bead assay, there was a much higher level of agreement between PET-PCR and the antigen-detecting assay. A possible explanation for this could be that PET-PCR and the HRP2 bead assay have a similar parasitaemia threshold of approximately two parasites per microliter [24], which is higher than that of nPCR. However, both the HRP2 bead assay and PET-PCR are quantitative, require fewer processing steps, and are faster processes than nPCR.

Limitations
The different diagnostic tests used in this study detect different parasite components and also have varying limits of detection. The samples used in this study were collected during the off-peak malaria season where parasite densities are generally low and thus would require diagnostic tests with low limit of detection and high sensitivity to detect.

Conclusion
Nested PCR exhibited the highest sensitivity by identifying the highest prevalence of asymptomatic P. falciparum in both the high and low parasite prevalence setting. However, parasite prevalence estimated by the HRP2 bead assay and PET-PCR had the highest level of inter-rater agreement relative to all the other tools tested and have the advantage of requiring fewer processing steps and producing quantitative results relative to Nested PCR. These advantages make PET-PCR and the HRP2 bead assay very useful tools for detecting and estimating malaria parasite density especially amongst asymptomatic individuals during community surveys. as well as the parents or guardians of participants who were minors before they were enrolled onto the study.

Consent for publication
Not applicable Availability of data and materials All data generated or analysed during this study are included in this published article. Figure 1 Map of Ghana highlighting the study sites. The study sites, Obom and Asutsuare are represented by green circles on the map. The map was created for this study by Awiah Dzantor Selorm, ACECoR, University of Cape Coast, using shape les from the Survey Department of the Ghana Statistical Services and ArcMap GIS v10.5.

Figure 2
Parasite prevalence determined by the different diagnostic tests. Microscopy and RTD detected the least number of P. falciparum parasite carriers in both the high and low parasite prevalence setting. Whilst parasite prevalence determined by Nested PCR, PET PCR and the HRP2 bead assay were similar in the high parasite prevalence setting, they were different in the low parasite prevalence setting Figure 3 Parasite density determined by different tools. The median (IQR) parasite density of samples that tested positive for P. falciparum by microscopy (a) and PET-PCR (c) as well as the median (IQR) HRP2 antigen content of the samples estimated using the bead assay (b) from each site. Signi cant differences were observed in values obtained using microscopy (a) and the bead assay (c) but not by PET-PCR (b) when samples from Obom were compared to those from Asutusare.

Figure 4
Comparisons of PET-PCR, nPCR and the HRP2 bead assay detection tools. A Venn diagram illustrating the number of positive parasites by the three methods, [A] High transmission site, the three methods identi ed 42 samples as positive for the parasites, 7 positives between nPCR vs HRP2 bead assay, and also between PET-PCR vs HRP2 bead assay, and 6 positives between PET-PCR vs N-PCR.
[B] Low transmission, the three methods identi ed 3 samples as positive for the parasites, 4 positives between nPCR vs HRP2 bead assay, and 1 positive between PET-PCR vs HRP2 bead assay, and no positives between PET-PCR vs nPCR