qPCR in a suitcase for rapid Plasmodium falciparum and Plasmodium vivax surveillance in Ethiopia

Many Plasmodium spp. infections, both in clinical and asymptomatic patients, are below the limit of detection of light microscopy or rapid diagnostic test (RDT). Molecular diagnosis by qPCR can be valuable for surveillance, but is often hampered by absence of laboratory capacity in endemic countries. To overcome this limitation, we optimized and tested a mobile qPCR laboratory for molecular diagnosis in Ziway, Ethiopia, where transmission intensity is low. Protocols were optimized to achieve high throughput and minimize costs and weight for easy transport. 899 samples from febrile patients and 1021 samples from asymptomatic individuals were screened by local microscopy, RDT, and qPCR within a period of six weeks. 34/52 clinical Plasmodium falciparum infections were missed by microscopy and RDT. Only 4 asymptomatic infections were detected. No hrp2 deletions were observed among 25 samples typed, but 19/24 samples carried hrp3 deletions. The majority (25/41) of Plasmodium vivax infections (1371 samples screened) were found among asymptomatic individuals. All asymptomatic P. vivax infections were negative by microscopy and RDT. In conclusion, the mobile laboratory described here can identify hidden parasite reservoirs within a short period of time, and thus inform malaria control activities.


Introduction
Malaria remains a major public health threat in many countries in the tropics and subtropics. After a decade of progress with a pronounced reduction of the number of clinical cases and deaths, progress has stalled in recent years. In 2020, over 240 million cases and 600,000 deaths were recorded [1].
Accurate and fast diagnosis and treatment are key aspects of malaria control. In most malariaendemic countries, diagnosis by light microscopy is routinely conducted at health centers and hospitals. The sensitivity and specificity of local microscopy depends greatly on the training of local microscopists [2], and field microscopy can be substantially less sensitive than expert microscopy [3]. As an alternative, rapid diagnostic tests (RDTs) have become increasingly common. RDTs are lateral flow devices that detect parasite-specific proteins through immunohistochemistry. RDTs require less training, and results are obtained within 10 minutes. They are thus used by small health posts with no microscopy infrastructure and by health workers conducting household visits and diagnosis, e.g. in the frame of reactive case detection activities [4]. Sensitivity of RDTs can be impaired by incorrect storage and handling, wrong interpretation of results, or deletion of the gene coding for Histidine-Rich Protein 2 (HRP2), which is detected by most RDTs for P. falciparum [5]. False-positive results can be caused by non-malarial infections [6]. Light microscopy and RDT have a limit of detection of approximately 50-100 parasites per uL of blood [7]. A large number of clinical infections remain below this density [8,9]. Further, in all transmission settings, a proportion of infections remain asymptomatic, and many of them are subpatent [10,11]. Asymptomatic infections and low-density clinical infections escaping routine diagnosis conducted at health centers among febrile patients sustain transmission and present a major challenge to control [12][13][14][15].
Molecular diagnosis by PCR or other nucleic acid amplification tests are required to assess the quality of local diagnosis, to determine the true number of infections among febrile patients, and to understand population parasite prevalence in asymptomatic individuals. Rapid, sensitive screening might also be required to coordinate the response to outbreaks, for example to decide where intensified vector control is warranted because of a large asymptomatic reservoir. Molecular surveillance is often complicated by the absence of laboratory infrastructure and lack of skilled personnel in malaria endemic sites. Shipment of samples to reference laboratories can be complicated and time consuming. Molecular screening is thus seldom applied to select control strategies tailored to local conditions, or in response to outbreaks.
In order to speed up time to result and enable in-country scientists and control programs to process samples, efforts are increasingly being made to bring laboratory capacity to field sites [16][17][18]. Numerous devices and protocols for molecular screening for pathogens are being developed and trialed. Often, these assays rely on custom-built devices [19]. Throughput of commercially available platforms is often low [20][21][22][23][24]. In addition, the need for highthroughput, mobile DNA extraction platforms is not addressed.
For this study, a mobile qPCR lab was trialed for malaria surveillance in a low transmission site in Ethiopia. All equipment and consumables needed are commercially available and fit in suitcases for transport on airplanes. Up to two 96-well plates can be processed in a day, at a cost of approximately USD 2.5 per sample for DNA extraction and P. falciparum and P. vivax qPCR. Within a brief period of 2 months, nearly 2000 samples from febrile cases and asymptomatic individuals were screened using highly sensitive qPCR.

Ethical approval
Informed written consent was obtained prior to sample collection from each study participant or, in the case of minors, from their parent or legal guardian. The study protocol was approved by the University of Notre Dame IRB (#19-03-5201), Trinity College Dublin, Addis Ababa University, and the National Research Ethics Review Committee at Ministry of Science and Higher education (MoSHE).

Study site
In Ethiopia, P. falciparum and P. vivax are endemic. Malaria transmission ranges from very high in the tropical lowlands along the borders with Sudan and South Sudan to low and sporadic in the highlands [25]. In 2019, over 900,000 confirmed cases were reported. This represents a pronounced reduction compared to 2013, where the number of cases peaked at 2.6 million, but only a moderate reduction compared to 2010, with 1.2 million confirmed cases [26]. Diagnosis is provided at over 20,000 health centers across the country. Larger health centers perform diagnosis by microscopy, while RDTs are used by smaller health posts. In addition, over 70,000 health extension workers visit households and provide basic medical services, including malaria diagnosis by RDT [27].
This study was conducted in Ziway, Oromia region. Transmission intensity of P. falciparum and P. vivax is low. Samples for the current study were collected in the low transmission season in June and July 2019. Clinical samples were collected from individuals presenting with febrile illness to Batu and Dembel Health Centers. Cross-sectional surveys were conducted in 3 rural kebele (the lowest administrative units in Ethiopia), Bochessa, Dodicha, and Golba, which are under Adami Tulu Jiddo Kombolcha district administration.

Sample and data collection
For the clinical samples, patients with suspected malaria infection were invited to join the study and provide an additional blood sample for diagnosis. A brief questionnaire was completed including age, sex, and kebele of residence of the patient. For community samples, a convenience sampling strategy was applied. The study team visited the villages, approached households, and asked all household members who were present to provide a sample. 100-200 μL blood were collected by finger prick into EDTA tubes. Blood samples were stored on ice packs in Styrofoam boxes until bringing them to the lab each evening, where they were stored at -20˚C. RDT positive individuals among the community samples were referred to their health center for further diagnosis and treatment.

Diagnosis by microscopy, RDT, and qPCR
Samples were collected by finger prick into EDTA tubes. All samples were screened by RDT (AccessBio CareStart Pf(HRP2)/Pv(LDH) combo) upon collection, and by local microscopy. For microscopy, WHO protocols were followed. 100 fields were assessed before declaring a sample negative.
DNA extraction was done using the Macherey-Nagel NucleoMag kit according to manufacturer's instructions, with the following modification (Box 1xy): As the kit is optimized for extraction from 200 uL blood, but DNA was extracted from only 100 uL of blood, the volume of all reagents was reduced by 50%. Thus, per kit 8x 96 samples could be extracted, further reducing the amount of materials required and cost per sample. As proposed by the manufacturer as option, after the ethanol wash-step, beads were air-dried for 15 minutes instead of using buffer MBL-4. Some of the volumes of buffers were slightly modified to be able to complete all steps with a 30-300 μL multichannel pipette (S1 Protocols). In a recent side-by-side comparison, the extraction kit used yielded significantly more DNA than a spin-column based kit [28].
qPCR was done in a total volume of 12 μL, including 4 μL DNA, corresponding to 4 μL blood. For P. falciparum qPCR the varATS assay was used. This assay targets a multicopy gene that is present in 10-20 copies per parasite [29]. Using the extraction method we chose for this study, the limit of detection of this assay is 0.3 parasites/μL blood [28]. P. vivax qPCR was done using the cox1 assay. This assay targets a mitochondrial gene that is present in approximately 10 copies per parasite [30]. Detailed qPCR protocols are given in S1 Protocols. Due to a manufacturing problem with the P. vivax probe (low yield), only a random subset of samples was screened by qPCR (653/1021 asymptomatic and 718/899 clinical samples). Infection prevalence and test positivity rate was compared among three age groups of 0<5 years, 5<15 years, and �15 years, between males and females, and between kebele (community sampling only) using Pearson's Chi-square test.
HRP2-based RDTs can also detect the HRP3 protein, though sensitivity is lower [31]. Deletions of the hrp2 gene result in false-negative RDTs in low-medium density infections. Deletions of hrp2 and hrp3 result in negative HRP2-based RDTs irrespective of parasite density [32,33]. P. falciparum positive samples were typed for hrp2/3 deletions by droplet digital PCR (ddPCR). In this assay, either hrp2 or hrp3 is multiplexed with a control gene, serine-tRNA ligase. Both targets are amplified with very high sensitivity and specificity, thus providing highly accurate data on deletion status [34]. For ddPCR, samples were shipped to the University of Notre Dame.

Mobile laboratory
The mobile DNA extraction and qPCR systems were established in a makeshift laboratory on the compound of Addis Ababa University in Ziway. It consisted of a basic shed with two simple tables, and thus is representative for many locations with no laboratory infrastructure. All equipment and consumables required for this study are commercially available and given in Table 1. All protocols were optimized to achieve high throughput, i.e. work in 96-well format for extraction and 48-well format for the qPCR, while maintaining a low weight of the instruments required. Most importantly, the need for low weight instruments precluded the use of a centrifuge as used for common spin-column DNA extraction protocols. Instead, a protocol based on magnetic beads was used.
The main equipment required include the MIC qPCR system (including laptop computer), one plate shaker (for binding of DNA to the beads, wash steps, and DNA elution), one magnetic block (to bind the beads and remove the supernatant), two 12-channel pipettes (1-10 μL, 30-300 μL), and one set of single channel pipettes (1-20 μL, 20-200 μL, 100-1000 μL) (Fig 1). The total weight of all equipment was 7.2 kg, and it fits into one carry-on bag for air transport. The cost for all instruments totals approximately USD16,900 (not including the laptop computer). Costs are as follows: qPCR instrument: USD13,000, plate shaker: USD300, magnetic block: USD700, 2 x multichannel pipette: $1000 per pipette, 3 x single channel pipette: USD300 per pipette. The weight of all consumables for 8 plates (8x96 samples) was approximately 9.3 kg and thus can be easily transported by air as check-in luggage ( Table 1). The cost of consumables including extraction kit, pipettes tips and other plasticware, and qPCR reagents was approximately USD 2.5 per sample.
The following items were purchased from local pharmacies: 99% Ethanol, nitrile gloves, lancets for blood sample collection, and microscopy slides. Further, a -20˚C freezer was purchased locally for storage of reagents and samples. Plastic buckets were purchased as waste bins. All extractions were done at room temperature. No water bath or incubator was needed.

P. falciparum screening
Among 899 samples collected from febrile patients presenting to clinics, 55 (5.8%) were positive by qPCR. By local microscopy, only 13/52 qPCR positive samples were correctly diagnosed. One of the samples positive by qPCR for P. falciparum was misdiagnosed by microscopy as P. vivax. RDT was moderately more sensitive, with 18/52 qPCR positive samples detected by RDT. A total of six samples were positive by microscopy and/or RDT, but not confirmed by qPCR (Fig 2). They remained negative when the qPCR was repeated. Mix up at the health center during sample and data collection might have occurred. Demographic and qPCR data are given in Table 2. No significant difference in qPCR positivity by age group or sex was observed.
P. falciparum prevalence among asymptomatic individuals was very low with only 4/ 1021 (0.4%) individuals positive by qPCR. None of them were positive by microscopy or RDT. One individual tested positive by RDT, but the infection was not confirmed by qPCR. The four qPCR positive individuals were 4, 14, 15, and 40 years old; two were male and two were female. Three of those positive were from Golba, and one from Dodicha ( Table 2). 25 P. falciparum positive samples were successfully typed for hrp2 deletion, and 24 samples for hrp3 deletion. No hrp2 deletions were observed, but 19/24 samples lacked the hrp3 gene.

P. vivax screening
Among 718 samples collected from febrile patients presenting to clinics and screened for P. vivax by qPCR, 12 (1.7%) were positive (Fig 2). Seven out of these twelve samples were also positive by microscopy and RDT. The twelve individuals that were positive by qPCR were 3 to 27 years old; with 10/12 being 15 years and older (Table 2). Four more samples were positive by microscopy and RDT, but not confirmed by qPCR. As for qPCR-negative/microscopy or RDT positive P. falciparum samples, they remained negative when the qPCR was repeated. P. vivax prevalence among asymptomatic individuals was 3.8% (25/653). None of them was positive by microscopy or RDT (Fig 2). The age of positive individuals ranged from 1.5 to 60 years. No significant difference in prevalence rate among age groups was observed ( Table 2). Prevalence differed significantly among kebele (P = 0.003). It was highest in Dodicha at 6.3% (21/335), and lower in Golba at 2.5% (3/122) and Bochessa at 0.5% (1/196) ( Table 2).

Discussion
Molecular screening for infections below the limit of detection of microscopy or RDT is a key component of molecular malaria surveillance, and often the first step for subsequent studies, such as parasite genotyping to quantify drug resistance or establish transmission networks. Lack of adequate laboratory infrastructure is a problem in many endemic countries. In this study, using a mobile qPCR setup, high quality data on P. falciparum and P. vivax infection status with a limit of detection of <1 parasite μL blood was obtained from almost 2000 samples within a period of a few weeks. A short turnaround time of a few weeks is required in order to integrate molecular surveillance into control activities, for example to determine the extent of the asymptomatic reservoir during an outbreak [35]. The protocol tested is fully based on commercially available instruments and reagents, and offers high throughput with DNA extraction done in 96-well format, and qPCR run in 48-well format. A pair of trained laboratory technicians can process two 96-well plates within a day, and thus screen approximately 180 samples plus controls. The two-month study period included laboratory setup (e.g. cleaning, and procurement of freezer), visits to health centers and communities before sampling began, and sample collection. The limiting factor in the present study was the number of individuals that could be sampled per day. The cost of all consumables for DNA extraction, and separate P. falciparum and P. vivax qPCR, is approximately 2.5 USD per sample. The protocol used is suited for extraction of any DNA or RNA from blood and thus can be applied for molecular diagnosis of any blood-borne pathogen. The protocol requires multiple pipetting steps and thus molecular laboratory skills are needed. Likewise, knowledge is required to interpret qPCR data. Training of malaria control program personnel will be crucial in order to integrate qPCR data into routine surveillance activities. Based on training experiences in multiple malaria endemic countries, individuals with no prior laboratory experience can learn the skills to conduct DNA extractions and qPCR within two weeks (C. Koepfli, unpublished).
The makeshift laboratory presented multiple challenges. No running water was available, and power supply was unreliable with regular power cuts lasting several hours. As a result, the PCR was often run in the hotel, which had a back-up generator. For future surveillance by control programs, use of a generator to power the mobile lab is recommended. Of note, the DNA extraction can be done without the plate shaker, thus not requiring any power. Mixing steps can be done by pipetting. This protocol requires substantially more tips. The main risk of high-throughput manual DNA extraction in 96-well format is cross-contamination. This risk does not differ in a field laboratory compared to the same protocol being used in a reference laboratory, and can be minimized by proper training of personnel. The challenges of the mobile lab were offset by the rapid availability of data. This was highlighted by the extended period of time required to obtain permit to ship samples to the US for hrp2 and hrp3 deletion typing.
This study revealed crucial reservoirs for transmission not identified by current control. Two thirds (34/52) of P. falciparum infections detected by qPCR in febrile patients were missed by microscopy and RDT. These untreated infections likely contribute to transmission for an extended period of time [36]. More sensitive diagnostic tools at health centers would be expected to reduce transmission. In contrast, very few infections were detected among asymptomatic individuals. A contrasting pattern was observed for P. vivax. Most infections were detected among asymptomatic individuals, and fewer among febrile patients. Asymptomatic P. vivax infections clustered mostly on one kebele. Possibly, many of the asymptomatic infections were relapses. Yet, while asymptomatic, they can still contribute to transmission [37]. This study corroborated high rates of subpatent P. falciparum and P. vivax infections in Ethiopia [38,39]. Of note, transmission intensity in the Ziway region has declined drastically since 2005/2006, when a prevalence by microscopy of 16-19% was recorded [40].
In conclusion, this proof of concept study showed that actionable data on subpatent P. falciparum and P. vivax infections can be obtained in a short period of time using a mobile qPCR lab. Molecular screening has identified a gap in the sensitivity for diagnosis of clinical P. falciparum cases and a substantial asymptomatic P. vivax reservoir, which was mostly concentrated in one village. P. falciparum control should focus on more sensitive diagnosis in health centers, e.g. though the introduction of novel, ultra-sensitive rapid diagnostic tests [41]. P. vivax control also needs to focus on prevention of onward transmission from the asymptomatic reservoir.