Development and validation of a sample-to-result real time multiplex RT-PCR test for human African trypanosomiasis and arboviral febrile illnesses in the Democratic Republic of the Congo

Abstract

Background

Human African trypanosomiasis (HAT) is a fatal vector-borne disease caused by Trypanosoma brucei. Although HAT incidence has declined, meeting WHO’s elimination targets remain difficult, particularly due to limited diagnostic sensitivity for low-parasite load-infections. Arboviruses such as dengue (DENV 1–4), chikungunya (CHIKV), and yellow fever (YFV) virus, present with nonspecific febrile symptoms similar to HAT and are underdiagnosed in Sub-Saharan Africa. Due to this overlap in symptoms and a suspected geographical overlap of vectors and pathogens in the Democratic Republic of the Congo (DRC) the pathogens were combined in a multiplex-PCR panel. Sample-to-result platforms (S2R) can reduce hands-on time and infrastructure requirements, making them ideal for peripheral laboratories. We developed a multiplex real-time RT-PCR assay on the ARIES^® platform, for simultaneous detection of HAT, DENV, CHIKV and YFV, showing how automated, closed-cartridge PCR can simplify testing.

Methods

A technical validation and retrospective sample testing (n = 242) were performed at the Institute of Tropical Medicine (ITM). Field validation took place in the DRC with retrospective samples from a CHIKV outbreak (n = 121) in Institut National pour la Recherche Biomédicale (INRB) Kinshasa and 52 prospective whole blood samples from acute febrile patients in Centre de Recherche en Santé de Kimpese (CRSK) in Kimpese.

Results

The assay showed a slight loss of sensitivity, evidenced in the technical validation, and the non-detection of some T.b.gambiense and retrospective arboviral samples at ITM with low pathogen loads. CHIKV samples tested in Kinshasa showed a sensitivity of 89.4%. Although all samples tested in Kimpese were negative for the pathogens, it demonstrated how just a few days of training and the simplified workflows of a S2R-platform can enable robust diagnostics under challenging conditions.

Conclusion

Ensuring rapid, sensitive molecular diagnostics in resource-limited settings is critical for eliminating HAT and strengthening surveillance of emerging arboviruses. Despite the recent discontinuation of ARIES^®, our findings confirm the feasibility and reliability of detecting diverse pathogens with minimal laboratory resources. The assay aligns with WHO and FIND target-product profiles, highlighting its relevance for neglected diseases in low-resource settings. These results emphasize the ongoing need for open, flexible S2R platforms to support disease surveillance and outbreak preparedness.

Clinical trial

Clinicaltrials.gov NCT04760678, registered on February 17th, 2021.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-026-12550-0.

Introduction

Human African trypanosomiasis (HAT), or sleeping sickness, is a vector-borne parasitic disease affecting rural communities in Sub-Saharan Africa and is fatal if left untreated [1]. The disease is caused by the Trypanosoma brucei (T.b.) parasite, with T.b. gambiense causing a chronic form of the disease in western and central Africa, and T.b. rhodesiense leading to the acute form of sleeping sickness mostly found in eastern and southern Africa. Over 90% of the reported cases are caused by T.b. gambiense, with more than 80% of cases between 2013 and 2022 occurring in the Democratic Republic of the Congo (DRC) [1, 2].

As a vector-borne disease, HAT is transmitted by the bite of infected tsetse flies (Glossina genus), other transmission forms such as mother-to-child, prick accidents or sexual transmission can occur but remain rare [2].

Clinically, HAT advances in two stages. The first stage, or haemolymphatic stage, is characterized by non-specific symptoms such as fever, headache, general malaise, weakness, fatigue, myalgia and arthralgia, which are very similar to acute infections with dengue, chikungunya and yellow fever viruses. In the first stage the trypanosomes are found in blood, lymph and subcutaneous tissues. If untreated, the parasites eventually cross the blood-brain barrier leading to the second stage, or meningoencephalitic stage, causing the more typical neurologic symptoms, where the disease then can be diagnosed in the cerebrospinal fluid (CSF) [2].

Major progress in HAT control has led to obtaining the WHO’s initial goal of eliminating gambiense HAT (gHAT) as a public health problem by 2020 (defined as less than 1 new case per 10,000 population in at least 90% of foci, and less than 2,000 cases globally [3]). The next WHO goal sets out a far more ambitious target with elimination of transmission by 2030 (defined as zero reported cases) [3, 4]. In 2023, the actual number of reported gHAT cases in endemic countries was 675 [5]. Although numbers of reported cases are low, there is concern about underreporting since most cases occur in remote, rural areas [6].

HAT diagnostics and case detection rely on active case detection by mobile teams in populations at risk and passive case detection of symptomatic patients consulting health facilities. Diagnosis of HAT is a two-step process using antibody detection as screening tool and microscopic confirmation through parasite detection in the lymph, blood or CSF [7]. However, for HAT elimination of transmission a very sensitive and specific diagnostic tool is needed, as parasitological confirmation can be challenging in patients with low parasite burdens. The risk of missing these cases could sustain or reintroduce transmission in endemic foci [8], hence impeding the intended elimination of transmission by 2030 [9]. Due to toxic side effects of the treatment, most national policies require the visualization of parasites prior to gHAT treatment. Consequently, the current diagnostic algorithm’s sensitivity is insufficient for achieving the new goals. The expected advent of acoziborole, a single-dose oral treatment, effective in both disease stages with a good safety profile, should offer the potential of treatment of seropositive suspects without requiring parasitological confirmation [10]. Post hoc diagnostic confirmation would still be necessary to determine whether T.b. gambiense is circulating by distinguishing true cases from false positives, which is crucial for surveillance. However, this confirmation could be conducted later in a reference laboratory. Also, to prevent disease resurgence, post-elimination surveillance is expected to play a major role in the elimination process [11] and molecular techniques confirming the presence of the parasite would be a gamechanger [10].

Acute febrile illnesses are widely underdiagnosed in Sub-Saharan Africa [12, 13]. Whereas sleeping sickness is epidemiologically well documented in DRC and on the path to elimination, the exact geographical distribution of arthropod-borne viruses, such as dengue virus (DENV), chikungunya virus (CHIKV) and yellow fever virus (YFV) in DRC remains poorly defined. CHIKV and YFV are considered endemic in DRC, primarily causing sporadic outbreaks [12]. However, the true burden and circulation of these viruses remain unclear. In recent years, arboviruses have been expanding globally, posing an increasing public health threat [14, 15]. There is an urgent need to map the presence and spread of arboviruses in Sub-Saharan Africa in order to improve the management of clinical fever cases and ensure appropriate diagnosis and treatment of non-malaria febrile illnesses [16].

We developed a multiplex real-time RT (reverse transcriptase)-PCR assay, HAT-ARBO, for the simultaneous detection of HAT, CHIKV, DENV serotypes 1–4 and YFV on the ARIES^® platform. The combination of these pathogens in a multiplex-PCR panel was based on overlapping clinical symptomatology, as they all cause acute undifferentiated febrile illnesses (i.e. fever, headaches, general malaise, myalgia) [1, 2, 17–19]. Although treatment and management vary, these symptoms can be a common entry point for medical attention. Furthermore, although the vectors differ, there is a geographical overlap of the vector presence and a suspicion of epidemiological overlap of the diseases associated with these pathogens in the DRC (Fig. 1). With samples for HAT surveillance taken all over the country this could help to map the presence of arboviruses in DRC.

Fig. 1.

A) Showing the presence of the tsetse fly by the number of species found in Africa, in white = no species found, in blue = more than one species found [20]. HAT cases are shown in the black circle for T.b. gambiense and in a black triangle for T.b. rhodesiense cases the years 2021 and 2022 [4]. B) Showing the suitability for Aedes aegypti in Africa in a binary mode 0 (white = not suitable) and 1 (blue = suitable) surface, using an average suitability value from all observed detections [21]. The occurrence data for DENV, CHIKV and ZIKV spans the years of 1927 to March 2024, assembled by Lim et al. [22]

Molecular diagnostic tests, such as real-time PCR, are widely recognized for their high sensitivity and specificity in detecting pathogens [23]. However, implementing these tests at the point-of-care (POC) presents several challenges, particularly concerning sample handling and processing. To achieve the desired sensitivity, effective nucleic acid extraction from clinical samples is crucial, requiring precise equipment and controlled laboratory environments to prevent contamination and ensure result integrity. In POC settings, maintaining such stringent conditions can be difficult [24, 25]. Additionally, sample transport from POC sites to centralized laboratories poses further challenges. Sample degradation during transit, the need for cold chain logistics, can all adversely affect the specimen quality. To address these challenges, secondary peripheral laboratories near patient populations could reduce transport distance, ensuring faster diagnostic and improving patient management. However, setting up molecular diagnostics in such facilities, requires investment in infrastructure, equipment and trained personnel to ensure that tests remain accuracy and reliable.

The ARIES^® system, developed by Luminex (now Diasorin) provides a comprehensive solution for implementing molecular diagnostics in peripheral laboratories. This sample-to-result (S2R) platform integrates nucleic acid extraction, purification, amplification, and detection via real-time PCR within a closed cartridge system with a capability of running 6 cartridges at a time. The multiplexing capability and built-in sample process control (SPC) enhance reliability, while the fully automated workflow reduces hands-on time and the contamination risk [25, 26].

To evaluate the performance of the HAT-ARBO PCR on ARIES^®, a technical validation was first conducted at the Institute of Tropical Medicine (ITM) in Antwerp. This was followed by feasibility testing in the Democratic Republic of the Congo (DRC), with implementation in both the national reference laboratory in Kinshasa and a peripheral laboratory at the district hospital of Kimpese.

Methods

Ethical approval

Ethical approval was obtained by the Institutional Review Board of the ITM (IRB reference number 1449/20), the Ethics Committee (EC) of the University Hospital in Antwerp (Belgian registration number: B3002021000025) and the EC of the Université Protestante du Congo (UPC) for the use of clinical samples collected within the FIKI² study (Clinicaltrials.gov NCT04760678). All participants gave their written informed consent. The study was conducted according to the principles stated in the Declaration of Helsinki, and to all applicable local and international regulations. Only stored samples with authorization for secondary usage were used in this study and the authorization from the local Ethical Committee was obtained. The viral RNA used for these experiments was extracted from viral culture supernatant obtained from virus isolations from returning travellers, or in the case of CHIKV, from a previous study in Aruba [27], which were completely anonymized and cannot be linked to the original donor.

The use of rodents for the expansion of T.b. gambiense LiTat 1.3 was approved by the Animal Ethics Committee (AEC) of the Institute of Tropical Medicine (ITM) under protocol TU2021-1.

HAT-ARBO PCR: Primers and probes in silico analysis

An in silico validation of the primers and probes used for HAT-ARBO PCR was done using the OligoAnalyzer™ tool of Integrated DNA Technologies (IDT) to detect potential secondairy structures or dimer formation [28]. All primers and probes were BLASTed [29] against all target viruses used in the PCR (CHIKV, DENV, YFV) and other related arboviruses (i.e. Tick-borne encephalitis virus, Mayaro virus, O’nyong’nyong virus, West Nile virus, Semliki Forest virus). The Trypanosoma primers and probes were BLASTed against target (evansi, brucei) and off-target Trypanosoma (vivax and cruzi), Leishmania and Plasmodium spp.

Complete genomes for CHIKV, DENV1-4 and YFV were downloaded from the VIPR databank, now known as BV-BRC [30] [accessed June 2019]. The primers and probe binding regions were aligned to detect potential mutations across different strains. If mismatches occurred, the primers and probes were modified accordingly. The dataset included 644 genomes for CHIKV, 1973 for DENV1, 1544 for DENV2, 940 for DENV3, 246 for DENV4 and 157 for YFV. An overview of the final optimized primers and probes and their sequences, concentrations and references can be found in Table 1.

Table 1.

Table with all used primers and probes with their sequences, concentrations and references

PCR	Oligo	Sequence (5’ → 3’)	Conc.	Ref.
Tryp	18S2 For	CCAATCGGACGCTCTCTTT	0.3µM	[42]
	18S2 Rev	GTGGAGGCGTTGGTTCTAAT	0.3µM
	18S2 Probe	/56-FAM/TTGTGTTTA/ZEN/CGCACTTGTCGTGGC/3IABkFQ/	0.1µM
CHIKV	ChikSII	CCGACTCAACCATCCTGGAT	0.4µM	[39]
	ChikAsII	GCAGACGCAGTGGTACTTCCT	0.4µM
	ChikSI	TGATCCCGACTCAACCATCCT	0.4µM
	ChikAsI	GGCAAACGCAGTGGTACTTCCT	0.4µM
	Chik Probe	/5HEX/TCCGACATC/ZEN/ATCCTCCTTGCTGGC/3IABkFQ/	0.2µM
YFV	YFV For	CTAATTGAGGTGYATTGGTCTG	0.4µM	[40, 41] adapted
	YFV Rev	CTGCTAATCGCTCAAMGAACG	0.4µM
	YFV Probe	/5Cy5/ATCGAGTTG/TAO/CTAGGCAATAAACAC/3IAbRQSp/	0.2µM
DENV1	D1 For	CAAAAGGAAGTCGTGCAATA	0.2µM	[45]
	D1 Rev	CTGAGTGAATTCTCTCTACTGAACC	0.2µM
	D1 Probe	/5TEX615/CATGTGGTTGGGAGCACGC/3IAbRQSp/	0.1µM
DENV2	D2 For	CAGGCTATGGCACYGTCACGAT	0.2µM	[45]
	D2 Rev	CCATYTGCAGCARCACCATCTC	0.2µM
	D2 Probe	/5TEX615/CTCTCCRAGAACGGGCCTCGACTTCAA/3IAbRQSp/	0.1µM
DENV3	D3 For	GGACTGGACACACGCACCCA	0.2µM	[45]
	D3 Rev	CATGTCTCTACCTTCTCGACTTGYCT	0.2µM
	D3 Probe	/5TEX615/ACCTGGATGTCGGCTGAAGGAGCTTG/3IAbRQSp/	0.1µM
Pan-DENV	pDENV For	GCATATTGACGCTGGGARAG	0.4µM	[43, 44] adapted
	pDENV Rev1	CTGTGCCTGGAATGATGCTG	0.4µM
	pDENV Rev2	CGCTCTGTGCCTGGATTGAT	0.4µM
	pDENV Probe	/5TEX615/CCA + GA + GAT + CCT + GCT/3IAbRQSp/	0.2µM
SPC	MHV For	CGGTGTTAGCGGTTTTGCTG	0.2µM	[46]
	MHV Rev	GCCACTCGGTTTGTTTGAGGGCA	0.2µM
	MHV Probe	/56-TAMN/TATGTGAAGTCCAAGGTCGGAAATTACCGACTG/3IAbRQSp/	0.1µM

Manual extraction and real-time PCR

Nucleic acid extraction

RNA was extracted to detect 18S ribosomal RNA from T. brucei and the genomic single stranded positive RNA of arboviruses, all of which require an RNA-based extraction approach for optimal detection. A 140µL sample (viral supernatant, spiked EDTA-blood or serum) was processed using the QIAamp^® Viral RNA Mini kit (Qiagen) according to the manufacturer’s protocol. Elution was done in two steps with 40µL of elution buffer, including a 5 min incubation time to maximize RNA recovery between each elution. If the initial sample volume was below the recommended 140µL, nuclease-free water was added to meet the required input volume. The extracted nucleic acids were stored at -80 °C.

Bio-Rad CFX96 PCR

One-step real-time RT-PCR was performed on the Bio-Rad CFX96 using the iTaq™ Universal Probes One-Step Kit (Bio-Rad). The reaction was conducted in a total volume of 25µL with 5µL template and 20µL master mix, which included the corresponding concentrations of primers and probes for the different assays (see Table 1), 12.5µL of iTaq™ Universal Buffer and 0.5µL of iScript RT enzyme. Cycling conditions consisted of four steps: 600s at 50 °C (1), 300s at 95 °C (2), 10s at 95 °C (3) and 30s at 60 °C (4), with steps 3 and 4 repeated 50 times.

Data analysis was performed using the single threshold setting for Cq determination, and baseline subtraction with curve fitting was used as configured by the CFX96 data analysis program.

ARIES^® molecular diagnostic platform

Sample pretreatment optimization

Several pretreatment conditions in different sample types were examined to optimize the extraction in the ARIES^® cartridges. Different volumes were tested for whole blood (WB), serum samples and buffy coat (BC) samples, which were spiked with target pathogens (ZIKV, YFV, or T.b. gambiense parasites) and subjected to different lysis and incubation protocols. The different protocols varied in their inclusion of Proteinase K (30µL), carrier RNA (4µL), Qiagen AVL lysis buffer (300µL) and in the case of BC dilution with PBS. To evaluate sample stabilization, WB was mixed 1:1 with DNA/RNA Shield (Zymo Research) in different volumes and processed under the same conditions as above.

These optimization experiments established baseline parameters for maximizing nucleic acid recovery from diverse sample types, for more detailed information refer to the supplementary data.

Final ARIES^® pre-treatment protocol

Following optimization, the definitive and final pre-treatment protocols were established for whole blood, serum, buffy coat, and blood preserved in DNA/RNA shield to be used for the validation and field implementation. For whole blood or serum, 100µL of sample was mixed with 300µL of lysis buffer (AVL-Buffer, Qiagen), 30µL of proteinase K (Qiagen), followed by incubation for 10 min at 56 °C to ensure lysis. Buffy coat samples were first diluted 1:5, by mixing 20µL of PBS with 80µL of buffy coat, after which 100µL was processed according to the pre-treatment protocol described above for whole blood and serum. Blood samples were mixed with DNA/RNA Shield (Zymo Research) in 1:1 ratio, where 150 µl of the mixture was combined with 250µL of lysis buffer (AVL-Buffer, Qiagen) and 25µL of proteinase K (Qiagen), followed by incubated for 10 min at 56 °C. After pretreatment, 400µL of each processed sample was transferred to the ARIES PCR cartridges for automated extraction and amplification.

ARIES^® amplification

The reaction mixture for PCR was prepared with primers and probes used in the corresponding concentrations (see Table 1), for a final reaction volume of 50µL in the ExoReady Mix tube, and 5µL of this mix was pipetted on top of the wax sheet of the ExoReady Mix tube. The ExoReady Mix tube was clicked on the cartridge, with the wax sheet melting when amplification starts. The following PCR program was used: 420s at 50 °C (1), 120s at 95 °C (2), 10s at 95 °C (3) and 30s at 60 °C (4). Steps 3 and 4 were repeated 45 times.

A fixed threshold method for all targets was implemented by a service engineer from Luminex/Diasorin after PCR validation. This was determined empirically, by comparing high signal known negatives and low signal known positives (controls) and adjusting the threshold to avoid false positive and false negative results. The thresholds were optimized based on the fluorescent dyes brightness and the sensitivity of the ARIES detection channels. The final cut-off values were set to 50,000 RFU (relative fluorescence units) for HAT (6FAM Channel) and CHIKV (AP525 Channel) and to 25,000 RFU for DENV (AP593 Channel), YFV (AP 662 Channel) and SPC (AP559 Channel).

HAT-ARBO validation on Bio-Rad CFX96 and ARIES^®

A basic validation for semi-quantitative real-time multiplex RT-PCR was done following the MIQE-guidelines. Performance characteristics such as standard curves with PCR efficiency and R²-value, the precision (repeatability and reproducibility), analytical sensitivity or limit of detection (LoD) and analytical specificity were established [31]. In a perfect PCR, the amplicons theoretically double at each cycle leading to a slope of -3.32 and a PCR efficiency of 100%. A real-time PCR efficiency between 90 and 110% and R² > 0.98 is generally considered acceptable [32–34]. When comparing multiplex-PCRs with simplex-PCRs a difference in Cq-value of less than 1 is ideal [35].

Standard curves on the Bio-Rad CFX96

To establish standard curves, serial tenfold dilutions (10^− 1 to 10^− 7) from extracted viral RNA or DNA were prepared in nuclease-free water. PCR efficiency, R²-value, slope and y-intercept were automatically calculated by the Bio-Rad CFX Manager™ Software (Version 3.1). Standard curves were generated separately for simplex and multiplex PCR reactions.

Simplex vs. multiplex PCR

For the ARIES^® system, 1mL of nuclease-free water was spiked with extracted RNA from viral supernatants or DNA from Trypanosoma to achieve a final dilution of 1:10,000. Each sample was split in 5 equal aliquots (200µL each) and tested using both simplex and multiplex-PCRs. All tests were performed in duplicate and repeated three times. The difference in Cq-values between the simplex- and multiplex-PCR was calculated as described above.

Analytical sensitivity

The analytical sensitivity or limit of detection (LoD) is the lowest number of copies that can still be detected with a certain level of confidence (mostly 95%) [31, 34]. A probit regression was done in R Studio to determine the 95% detection limit, by analyzing the detection rates, representing the proportion of replicates detected at each concentration level.

WB samples spiked with parasites or viruses were used. T.b. gambiense (LiTat 1.3) was diluted to different parasite/mL concentrations in WB. CHIKV, DENV and YFV were titrated to TCID₅₀ and further diluted in WB. From the TCID₅₀ the PFU/mL was calculated with the formula: 1 TCID₅₀ = 0.7 PFU/mL [36]. All samples underwent pretreatment per the established protocols.

Analytical specificity

Reference virus strains were obtained from BEI Resources and from the virus repository of the Virology Unit at ITM to assess the analytical specificity or cross-reactivity. A total of 18 reference strains were tested: CHIKV (n = 11), DENV1 (n = 1), DENV2 (n = 1), DENV3 (n = 2), DENV4 (n = 2) and YFV (n = 1). Each reference virus strain (2 µL) was spiked in 200µL of nuclease-free water and run on the ARIES platform.

To assess cross-reactivity, the following arboviruses and related pathogens were tested: Mayaro virus (MAYV, n = 2), Tickborne encephalitis virus (TBEV, n = 3) (Hypr and Neudörfl strains), Japanese encephalitis virus (JEV, n = 1), O’nyong-‘nyong virus (ONNV, n = 1), West Nile fever virus (WNV, n = 2), Usutu virus (USUV, n = 1), Russian spring-summer encephalitis virus (RSSV, n = 1), St. Louis encephalitis virus (SLEV, n = 1), Toscana virus (TOSV, n = 1), Sandfly Fever Sicilian and Naples virus (SFS/NV, n = 1), Ross River virus (RRV, n = 1), Sindbis Virus (SINV, n = 1), Venezuelan equine encephalitis virus (VEEV, n = 2), Eastern equine encephalitis virus (EEEV, n = 1), Western equine encephalitis virus (WEEV, n = 1), Rift Valley Fever virus (RVFV, n = 1), Zika virus (ZIKV, n = 4). Furthermore, other Kinetoplastids, such as Leishmania chagasi (ECCHa7) and Leishmania DNA were tested to ensure no cross-reactivity with Trypanosoma species.

For more detailed information on the strains used refer to the Supplementary Table 2.

Repeatability/reproducibility

Repeatability or intra-assay variance refers to the agreement of samples analyzed in the same assay run [31]; whereas reproducibility or inter-assay variance refers to the agreement of the same sample analyzed in a different assay run or by a different person, with different lots of PCR kits or in a different laboratory [34]. Acceptable precision is defined as SD lower than 0.5 [35].

To assess repeatability, six ARIES^® cartridges were loaded with 200µL of nuclease free-water spiked with trypanosome DNA at a concentration of 1 fg/µL and with a 1:10,000 dilution of DENV4-RNA. All samples were processed simultaneously to evaluate inter-run variability and Cq-values were analyzed to determine consistency across runs. For reproducibility testing, EDTA blood was spiked with 100 cells/mL T.b. gambiense parasites (LiTat 1.3) or with a 1:1,000 dilution of YFV from viral supernatant (stored at -80 °C). The EDTA blood was divided in 100µL aliquots, stored at -20 °C and tested six times over 14 days. For both repeatability and reproducibility, the mean Cq-values and SD were calculated to determine inter-run and inter-day precision.

External quality control (EQC) samples

To evaluate the performance and robustness of the assay, external quality control samples from the Quality Center for Molecular Diagnostics (QCMD) were tested on ARIES^®. Each EQC sample (100µL) was processed using the established pre-treatment protocol. This panel included samples for CHIKV (n = 6), DENV1 (n = 2), DENV2 (n = 2), DENV3 (n = 3), DENV4 (n = 2) and negative samples (n = 2). No YFV EQC samples were available for testing.

Clinical samples

The study design was a diagnostic performance study with retrospective samples at the ITM, Antwerp in Belgium and at the INRB, Kinshasa in DRC and prospective samples at the CRSK, Kimpese in DRC. Retrospective samples were selected based on their availability in the biobanks and ethical permission for secondary use of samples. Selection prioritized samples that were well-characterized with known Cq values, covering a range of viral loads and clinical presentations in the case of HAT, to ensure representation of different infection stages. Especially for the retrospective samples a selection bias is possible, since their selection was based on availability and not randomness.

The prospective samples were collected consecutively from patients following inclusion and exclusion criteria set by the FIKI²-study (Clinicaltrials.gov NCT04760678). The eligibility criteria to be enrolled in the study were, fever (24 h prior to preesentation, defined as: axillary temperature > 37.5 °C, or tympanic (or oral or rectal) temperature > 38 °C), children over 2 years and willing and able to provide written informed consent (including assent for minors older than 12y). Exclusion criteria were age less than 2 years old, hospitalization in the last 14 days or refusal to provide written consent.

Retrospective samples tested at ITM, Antwerp, Belgium

The diagnostic assay was validated using negative and positive retrospective samples. Ten whole blood samples from the Red Cross Belgium served as non-endemic controls, while 118 whole blood samples from the DRC were included as endemic controls, of which six samples were positive for malaria. Trypanosoma-positive samples and controls, as buffy coats, were obtained from the WHO HAT Specimen Bank. These samples included 20 endemic negative controls, 20 T.b. gambiense-positive samples from the first stage, 20 T.b. gambiense positives from the second stage and eight T.b. rhodesiense-positive samples. The original Cq-values of these samples were unknown, and it was not further specified whether the positive cases were detected in blood, CSF or lymph node aspirates. To supplement the endemic negative controls, 14 additional Trypanosoma negative samples were obtained through the ITM’s clinical laboratory.

For arbovirus-positive samples, retrospective serum samples from patients attending the ITM’s outpatient clinic included DENV1 (n = 1), DENV2 (n = 2) and DENV3 (n = 3). Furthermore, arbovirus-positive samples from studies were tested, including CHIKV positive sera (n = 9) from a study in Aruba [27] and dengue positive samples (n = 12) from a study in DRC [12], including eight DENV1, two DENV2 and two undetermined dengue-positive cases (Table 4).

Table 4.

Number, sample type and references for retrospective clinical samples used for the ARIES validation at ITM

	+	-	Total	Sample type	Reference
DENV (DRC)	12	117	129	Serum	[12]
CHIKV (Aruba)	13	0	13	Serum	[27]
ZIKV, CHIKV, DENV (travelers)	5	4	9	WB	Samples obtained from patients at Central Clinical reference laboratory, ITM Antwerp
Tryp. Neg.	0	14	14	WB, spinal fluid
Non-endemic healthy	0	10	10	WB	Healthy blood donations Red Cross Belgium
Tryp. Pos. (WHO CC)	48	20	68	Buffy Coat	INSTITUT PASTEUR - ICAReB
	77	165	242

Field implementation in the DRC

To evaluate the field implementation, three ARIES^® machines were installed in the DRC. Two of these machines were installed at the Institut National pour la Recherche Biomédicale (INRB) in Kinshasa, while the third was placed at the Centre de Recherche en Santé de Kimpese (CRSK) in Kimpese. At INRB Kinshasa, two laboratories were dedicated for analysis. One air-conditioned laboratory housed the two ARIES machines, while the second lab was used for sample preparation.

All necessary reagents were imported to DRC and local personnel in both Kinshasa and Kimpese were trained for testing and analysis.

Retrospective sample testing in INRB, Kinshasa, DRC

Retrospective serum samples from a 2019 CHIKV outbreak in DRC were re-tested at INRB, Kinshasa. These samples, which included 85 CHIKV positives and 36 negative controls [37] were processed following the established pre-treatment protocol for ARIES^® testing.

Prospective sample testing in CRSK, Kimpese, DRC

The Centre de Recherche en Santé de Kimpese (CRSK), is a secondary health resource center for the health zone of Kimpese, embedded within the IME Hospital (Institút Médical Evangélique Kimpese), which provides medical care for approximately 40,000 inhabitants. The site was selected due to its existing research infrastructure, stable power supply and available laboratory personnel. At the time of selection, it lacked molecular diagnostic capacity, making it an ideal site for evaluating the feasibility of decentralized molecular testing.

The face-to-face training was for three days and was followed up remotely via the Iristick smart glasses (https://iristick.com/), which was a very useful tool to guide the laboratory technicians after not performing ARIES^® testing for a few weeks.

As the smart glasses have a camera, speaker and microphone, they can be operated handsfree and leave the technician to perform the experiment with remote advice from the expert at ITM, Antwerp [38]. An air condition was installed to be run during the use of ARIES^® and temperature and humidity logs were kept during the study period. Over a period of 9 months, 52 whole blood samples were collected from acute fever patients participating in the FIKI²-study and stored in DNA/RNA Shield. Additional sample testing was planned; however, due to unforeseen circumstances, the CRSK laboratory was forced to close, leading to the premature termination of this part of the study.

Results

Primers and probes in silico analysis

The primers and probes for CHIKV [39] and YFV (adapted from [40, 41]) have been validated extensively for diagnostic purposes within the National Reference Laboratory for Arboviruses at ITM, Antwerp. The 18S2 RT-qPCR assay for HAT detection, designed in-house, demonstrated high sensitivity and specificity and in molecular reference testing for T.b. gambiense [42]. For the pan-DENV-PCR strategy, a variety of published primers and probes were evaluated for their sensitivity detecting all four DENV serotypes. An adapted combination of two published primers and probes [43, 44], together with the primers and probes for the serotype-specific DENV1-3 PCR were identified as the most sensitive combination for detection of all four dengue virus serotypes [45].

Little heterodimerization was found with the OligoAnalyzer™ Tool (IDT website) [28]. All primers and probes were run through BLAST [29] to confirm target specificity. All PCRs had very low E-values (E-02 until E-05) for their corresponding target sequences and high E-values for sequences from similar and unrelated viruses (TBEV, MAYV, ONNV, WNV, SVF) and parasites (Plasmodium vivax/falciparum, Leishmania, T. cruzi, T. vivax). Alignments against publicly available genomes from BV-BRC [30] confirmed no significant off-target binding. The sequences of all primers and probes are listed in Table 1 and alignments of the primers and probes with the reference genomes for the different arbovirus strains in Supplementary Table 1.

Optimization of nucleic acid extraction for ARIES^®

Sample volume

We first compared two different sample volumes (100µL and 200µL) of nuclease-free water, serum and whole blood as input for ARIES^®. Increasing the sample volume to 200µL led to higher Cq-values for the SPC, in both serum (Δ Cq of 2.9) and whole blood (Δ Cq of 5.5), whereas no difference was observed for the nuclease-free water samples (Fig. 2). This indicates a less efficient extraction of whole blood or serum with higher sample volumes. For the target (ZIKV), while the Cq-value was significantly higher for the 100µl water sample, it remains nearly the same for both volumes of serum and whole blood. This suggests that despite increased target availability, larger sample volumes resulted in less efficient nucleic acid extractions. We therefore selected a sample volume of 100µL for all subsequent experiments.

Fig. 2.

Sample volume optimization. For arbovirus (ZIKV) target in water, serum and whole blood (WB). The sample processing control (SPC) is marked with a green circle and the target (arbovirus ZIKV) with a blue triangle. H2O/S = nuclease free water run with the serum experiment (n = 1); serum 100 = 100µL, serum 200 = 200µL (n = 5). H2O/WB = nuclease free water run with the WB experiment (n = 2); WB 100 = 100µL, WB 200 = 200µL (n = 6). The serum and WB experiments were run independently

Pre-treatment conditions

We next examined whether adding a commercial lysis buffer alone or with Proteinase K, carrier RNA or both could enhance nucleic acid recovery (Fig. 3). For whole blood spiked with T.b. gambiense or YFV, combining lysis buffer with Proteinase K lowered Cq values by up to 2 compared to lysis buffer alone. Addition of carrier RNA to Proteinase K only provided small additional improvements (< 1 Cq value), and carrier RNA alone showed a higher Cq value for Trypanosoma and only slightly lower Cq value for YFV of 0.3. For serum, the impact was less pronounced, but Proteinase K generally improved detection by nearly 2 Cq values for both the SPC as the target. Consequently, we standardized pre-treatment with Proteinase K for both whole blood and serum in lysis buffer to allow conformity and simplicity when the protocols were applied in Kinshasa and Kimpese.

Fig. 3.

A) Pre-treatment optimization for whole blood. B) Pre-treatment optimization for serum. LB = lysis buffer, ProtK = proteinase K, carrRNA = carrier RNA. All done in 2 repetitions, except WB + LB + carrRNA only done once, YFV WB + LB + ProtK + carrRNA one repetition excluded since SPC was too high. Untreated serum was repeated three times

DNA/RNA shield

We tested the compatibility of DNA/RNA Shield [47], a solution to stabilize whole blood samples at ambient temperatures, with ARIES extraction. To assess its performance, 75µL and 150µL of shielded blood were compared, each spiked with 10 and 100 parasites/mL. Lower volumes (75µL of blood with 75µL of shield) resulted in better sensitivity (Fig. 4). Overall, a 1:1 ratio of sample to DNA/RNA shield, followed by lysis buffer and Proteinase K treatment, provided optimal extraction conditions.

Fig. 4.

DNA/RNA Shield optimization. (A) Volume optimization for T.b. gambiense target in whole blood. The SPC is marked with a green circle, and the target (T.b. gambiense) with a blue triangle. T.b. gambiense parasites were in different concentrations of 10 parasites/mL and 100 parasites/mL. (B) Pre-treatment optimization for the use of DNA/RNA Shield with the target of T.b. gambiense and YFV. SPC = sample processing control (green circle), the target were T.b. gambiense parasites and YFV (blue triangle). WB = whole blood, SH = DNA/RNA Shield, LB = lysis buffer, ProtK = Proteinase K

Buffy coat

The optimal extraction condition for BC, as shown in Fig. 5, was a 1:5 dilution (20µL of PBS with 80µL of BC) which resulted in minimal increase in Cq values relative to whole blood or serum.

Fig. 5.

Volume optimization for the use of buffy coat (BC) as sample. SPC = sample processing control (green circle), the target were T.b. gambiense parasites (blue triangle). WB = whole blood, BC = buffy coat, PBS = Phosphate-buffered saline

Assay performance parameters on Bio-Rad CFX96 and ARIES^®

As summarized in Table 2, all simplex- and multiplex-PCR reactions fell within the acceptable ranges of efficiency and R²- values.

Table 2.

Results of technical validation with standard curves and limit of detection

	Bio-Rad CFX96				ARIES
	Efficiency	Slope	R2	y-intercept	M-S 10 − 4	95% LoD
Tryp S	97.9%	-3.37	0.998	36.234	1.4	5.22 p/mL
Tryp M	97.9%	-3.37	0.999	36.933
CHIKV S	100.0%	-3.32	0.999	47.168	0.2	4.67 PFU/mL
CHIKV M	94.2%	-3.47	0.995	40.562
YFV S	96.0%	-3.42	0.998	39.632	0.4	589.88 PFU/mL
YFV M	101.8%	-3.47	0.997	38.717
DENV1 S	96.9%	-3.40	0.999	42.035	1.3
DENV1 M	92.3%	-3.52	0.993	42.592
DENV2 S	96.7%	-3.40	0.997	42.067	1.2	37.13 PFU/mL
DENV2 M	100.3%	-3.32	0.998	41.068
DENV3 S	105.7%	-3.19	0.998	39.818	0.7
DENV3 M	103.2%	-3.25	0.996	39.963
DENV4 S	100.6%	-3.31	0.975	40.673	1.5
DENV4 M	98.5%	-3.36	0.985	40.663
Optimum	90–100%	-3.32	> 0,98	similar	< 1

M = multiplex PCR; S = simplex PCR; c/mL = copies/mL; M-S = Cq value from simplex PCR subtracted from multiplex PCR; LoD = limit of detection; 10 − 4 = 4th 1:10 dilution

Comparing the multiplex-PCR to the simplex-PCR, 4 Cq values were slightly above 1, but under 1.5, indicating minimal loss when multiplexing.

The LOD for Trypanosoma was 5.22 cells/mL. For the arboviruses, the CHIKV PCR was the most sensitive assay with 4.67 PFU/mL in, the LOD of DENV2 was 37.13 PFU/ml and of YFV 589.88 PFU/ml (see Fig. 6).

Fig. 6.

A) The limit of detection for the different detection methods of T.b. gambiense were calculated as median from 2–4 references for each method [48–51]. B) The viral load in patients was calculated as median from references, for CHIKV from 5 references [52–56], for DENV from 3 references [57–59] and for YFV only one reference was available [60]. The lowest and highest value were the median of the lowest and highest value given in the references. The limit of detection for each pathogen was determined with Probit regression

All of the 17 arbovirus strains (1 YFV, 11 CHIKV, 1 DENV1, 1 DENV2, 2 DENV3 and 2 DENV4) and none of the related viruses or parasites were detected. For a more detailed overview over the viruses used please refer to Supplementary Table 2.

As shown, in Table 3, the standard deviation for intra-run repeatability remained below 1 Cq. Inter-assay reproducibility was equally robust, except for the standard deviation in the reproducibility of the Trypanosoma-PCR where the SPC had a Cq difference of 1.5., still near the manufacturer’s acceptable threshold (< 1.2) [26] (Table 3).

Table 3.

Repeatability and reproducibility results for HAT-Arbo multiplex on the ARIES

	Repeatability	Target	SPC		Reproducibility	Target	SPC
T.b.gamb.	Mean	28.63	31.32	T.b.gamb.	Mean	26.72	31.23
	SD	0.53	0.55		SD	1.11	1.50
CHIKV	Mean	33.52	33.10	YFV	Mean	26.08	31.62
	SD	0.64	0.61		SD	0.85	0.85

SPC = sample processing control

External Quality Control (EQC) samples

We tested six CHIKV and eleven DENV samples from the QCMD panel. Out of the six CHIKV samples, five were positive for CHIKV and one was negative for CHIKV but included YFV 17D, DENV1 and West Nile virus NY99. From the eleven DENV samples, nine were positive for DENV (n = 2 DENV1, n = 2 DENV2, n = 3 DENV3, n = 2 DENV4), one was negative and one was negative for DENV, but included JEV, West Nile virus NY99, TBEV, YFV and Zika virus.

Samples with enough sample volume were retested with the reference method at the reference lab (CHIKV = 3; DENV = 10). Out of the 13 retested samples 4 samples tested negative that were previously still detected at Cq values higher than 33 with testing done in previous years.

All samples that tested negative in the retesting, were confirmed negative by ARIES^®. One sample of DENV1 tested positive in the reference laboratory testing and negative with ARIES^®, however it was on the limit of detection with Cq value of 35.

With the exception of one sample, all samples were detected with ARIES^® with similar Cq values than those detected with the reference method resulting in a high Cohen’s kappa of 0.881 compared to ITM Central Clinical reference laboratory results retested in at the same time as the ARIES.

Clinical performance on retrospective samples at ITM

A total of 245 retrospective samples were tested with the ARIES^® assay at ITM. For more detailed information about the provenance of these samples refer to Table 4.

All 165 negative samples were negative with ARIES^®. Out of the 13 tested CHIKV samples, 4 were not detected. However, the original Cq values of these samples in previous testing were high with 36.8, 34.2, 32.9 and 38.9. Seven samples were retested with the CHIKV simplex PCR after a new RNA extraction with QIAamp^® Viral RNA Mini kit (Qiagen) and this revealed an average loss of 2.8 Cq values, suggesting sample degradation over time. When comparing these latter results with ARIES^® results the average Cq value was 0.4 better with ARIES^® than the reference testing. This is in agreement with the results obtained comparing spiked CHIKV-RNA in simplex and multiplex-PCR with a difference of only 0.2 Cq values.

For DENV, four out of fifteen samples were not detected by the ARIES^® assay. When tested with reference testing these four samples had high Cq values of 33.6, 35.1, 35.8 and 38.9. Two samples tested positive with ARIES^® but were not tested for DENV in the original study. When comparing original Cq values to renewed reference testing, an average loss of 1.6 Cq values was observed, likely as a result of RNA degradation. The difference between the Cq values of ARIES^® and the renewed reference testing showed better Cq values of 0.3 on average in favor of ARIES^®.

For Trypanosoma, all 8 T.b. rhodesiense were detected and out of 40 T.b. gambiense samples, 32 were detected. Out of the eight samples not detected, three had a negative SPC. When separated by stage, three were not detected in stage 1 samples and five in stage 2 samples. As the Cq values are unknown for these samples, the concentration of trypanosomes in these samples may have been under the limit of detection of the PCR. The results of all retrospectively tested samples can be found in Table 5.

Table 5.

Results of the retrospective sample testing at ITM with the ARIES

Positives	+ Reference	+ ARIES
CHIKV	13	9
DENV	17	13
T.b.gambiense	40	32
T.b.rhodesiense	8	8
Negatives	- Reference	- ARIES
DRC: DENV	117	117
ITM: neg. For Typ.	14	14
Non-endemic healthy	10	10
WHO: Tryp	20	20

Field implementation in the DRC

Retrospective sample testing in INRB, Kinshasa, DRC

Among 85 confirmed CHIKV positives from a previous outbreak in 2019 [37], ARIES^® detected 76 (89.4%). When comparing Cq values to original reference testing, ARIES^® results showed an average increase of 2.9 Cq values, suggesting a mild loss of RNA integrity over time. The 9 samples that tested negative had an average Cq value of 34, indicating low viral loads, probably as a result of RNA degradation during prolonged storage.

Prospective sample testing in CRSK, Kimpese, DRC

Fifty-two acute-febrile whole-blood samples collected in DNA/RNA shield (FIKI² study) were tested using the HAT-ARBO PCR, with none testing positive for HAT, CHIKV, DENV or YFV. The average SPC Cq value was 32.2 ± 1.7, comparable to those observed at ITM Antwerp with 31.3 ± 2.6 (n = 130), confirming successful extraction and amplification in the peripheral laboratory setting in Kimpese.

Four samples had a negative SPC, suggesting possible PCR inhibition or extraction failure. One sample failed due to an expired ExoReady mix tube, while another was extremely viscous, likely affecting pipetting and extraction efficiency. Despite these minor challenges, the successful processing of febrile illness samples at CRSK, Kimpese, supports the feasibility of decentralized molecular testing for HAT and arbovirus surveillance in field settings.

Since the machine settings are only in English and most technicians do not speak English, a detailed standard operating procedure (SOP) was written in French, including pictures of the steps of handling the ARIES machine. Short laboratory instructions were put up, where technicians could see them easily during the workflow. Iristick smart glasses (https://iristick.com/) were used to follow the workflow remotely during the first experiments performed by the technician after the training. This helped to clarify any doubts and build confidence in the workflow. A close communication was established for trouble shooting and possible questions.

Discussion

There is an urgent need for affordable, easy and rapid molecular diagnostics in Low and Middle-Income Countries (LMIC), that can be used outside reference laboratories [61, 62]. S2R-platforms, exemplified by GeneXpert (Cepheid), which has been implemented widely in peripheral laboratories, have transformed the diagnosis of Tuberculosis, HIV, Ebola and more recently COVID-19 [25, 63, 64]. However, many commercial S2R solutions are not specifically designed for epidemiological needs, infrastructure constraints or range of pathogens endemic in LMIC [61, 65]. The ARIES^® molecular platform is an open system in which end-users can implement PCR tests of their own choice allowing flexibility in terms of pathogen targets [65]. Unfortunately, this platform was discontinued by Diasorin in 2024.

To address both HAT and emerging arboviral infections in single diagnostic workflow, we developed a real-time multiplex RT-PCR assay (HAT-ARBO) on the ARIES^® platform, targeting Trypanosoma brucei, alongside dengue virus, chikungunya virus and yellow fever virus. Because these pathogens present overlapping clinical features, a combined assay can optimize resources, turnaround time and simplify testing algorithms [14]. Applying a multiplex-PCR can theoretically lead to a reduced sensitivity due to primer-dimers formations or competing amplification reactions [31]. We found that on the ARIES^® platform, the difference between simplex and multiplex-PCR sometimes exceeded 1 Cq value, likely reflecting the additional variability generated by the automated extraction and cartridge-to-cartridge differences [26]. Despite this, and when applied to retrospective clinical samples, ARIES^® performance paralleled that of manual extraction plus simplex PCR, occasionally even performing better, except in samples with low pathogen load.

In HAT infections, residual DNA can persist even after successful treatment, so RNA based detection is considered instrumental for confirming active infection [66–68]. Our 18S rRNA based real-time RT-PCR aligns with this approach. Under the ARIES^® protocol, which processes just 100 µl of whole blood per run, we observed an LoD of 5.2 live parasites/mL, well below the < 10 parasites/mL benchmark set by WHO [69]. This highlights the critical role of physically capturing parasites within a limited input volume. Similar volume constraints apply across different test chemistries (NASBA, LAMP, CRISPR-based SHERLOCK or SL-RNA) each reporting LoDs in the 10 to 100 parasites/mL range [49, 66, 70, 71]. Although alternate sample matrices (e.g., serum, lymph, CSF, PaxGene tubes, dried blood spots, serum) and extraction methods can shift detection thresholds, our data confirms that even a relatively small input volume can still yield high analytical sensitivity. This is especially relevant when using DNA/RNA shield for blood stabilization, as it further dilutes the blood sample. More importantly, as shown in Fig. 6A, ARIES^® outperforms the mini anion-exchange centrifugation technique (mAECT), which typically detects 10–50 parasites/mL from 500µL of whole blood [48, 50], further underscoring the substantial sensitivity of the HAT-ARBO PCR in a limited sample volume.

Nevertheless, clinical sensitivity with the WHO HAT Specimen Bank samples proved much lower for T.b. gambiense than for T.b. rhodesiense, for which all samples were detected. This discrepancy mirrors results from other RT-qPCR, NASBA [72], LAMP [73, 74] and CRISPR-based assays [71], all showing reduced sensitivity for T.b. gambiense compared to T.b. rhodesiense. A key factor may be the significantly lower parasite load in T.b. gambiense infections, often below 100 parasites/mL, whereas T.b. rhodesiense can exceed 10,000 parasites/mL [70, 72, 75]. Moreover, T.b. gambiense may be detected in lymph node aspirates or in cerebrospinal fluid for second-stage disease, without necessarily being mAECT-positive in blood. Recent studies have shown that purifying even 2.5mL of stabilized blood sometimes failed to confirm T.b. gambiense RNA in stage 2 patients [76, 77]. This lower diagnostic sensitivity could present a problem, as for the elimination phase of HAT it is very important to also detect low parasite loads. However, with the LoD in spiked samples being lower than current diagnostic methods, even with these false negatives the HAT-ARBO still presents a good diagnostic test. One reason could be the compromised sample integrity by prolonged storage (> 10 years) at -80 °C without preservatives [78]. Another reason could be the sample type. The specimens from the WHO HAT Specimen Bank are obtained from buffy coat, a frequent storage medium for Trypanosoma samples as it leads to a higher sensitivity mAECT [79], microscopy [80] and higher concentration of nucleic acids [81, 82]. Although higher nucleic acid concentrations would be expected to improve PCR sensitivity [71, 83], this is not always the case due to the increased viscosity of buffy coat, which complicates pipetting. Furthermore, extraction from buffy coat samples often introduces more contaminants and may require a buffy coat-optimized extraction procedure [82, 84]. This difficulty pipetting was also observed in our experiments and by the negative SPC, indicating a failed extraction.

Crucially, however, the HAT-ARBO PCR retained 100% clinical specificity in this collection consistent with a previous report that demonstrated high specificity of this 18S rRNA assay on blood in DNA/RNA shield [42].

The LoD testing of the multiplex-PCR for the arboviral targets on the ARIES^® platform was evaluated using virus titrated to PFU/mL. As shown in Fig. 6B, these LoDs remain below the typical viremia levels reported in patients. The LoDs were also comparable to the ones reported in the literature when determined on the real-time PCR machine (Biorad CFX96) in copies/ml [39–41, 43–45] (see Supplementary Fig. 1). Compared to another commercial S2R-platform, the bioMérieux BioFire Global Fever Panel, the ARIES^® LoD values were similar or lower, indicating comparable or improved sensitivity depending on the target pathogen [49].

One advantage of PCR compared to other diagnostic methods is its high sensitivity and specificity [23, 25, 34, 70]. Flaviviruses are so closely related that serological methods suffer from extensive cross-reactivity [85, 86]. A sensitivity of 95%, with 90% being minimally acceptable and a specificity of 99% were set in the target product profiles (TPPs) by WHO for HAT and by FIND for YFV [69, 87]. As already evidenced by the in silico analysis and the analytic specificity testing with related pathogens, the specificity testing had good results with no amplification of related pathogens, in accordance with the TPP. Due to the low sample numbers available for most pathogens (except CHIKV retrospective testing in Kinshasa), sensitivity could not be calculated. Some samples could not be detected with the ARIES, probably due to the age of the retrospective samples. Degradation of RNA is expected to increase with storage time. The implementation in Kinshasa with testing retrospective CHIKV samples with more samples (n = 85) showed a sensitivity of almost 90%. Samples with a Cq value over 33 in the simplex-PCR were not detected, therefore the ARIES^® assay shows decreased sensitivity at the lower end of the viral concentrations. Usually patients with arboviral disease show a high viremia (see Fig. 6B) but this viremia is also quite short (5–7 days after symptom onset) [88, 89]. Since most patients seek medical attention only when symptoms persist, molecular detection methods may miss infections in those patients attending medical care towards the end of the viremia. One limitation of the study was the lacking availability of YFV samples to be tested.

In 2023, WHO published three TPPs for the detection of gambiense HAT in different settings, with two applicable for our assay [69, 90], while for the arboviruses only YFV has a published TPP [87]. Although our assay predates the WHO’s TPPs for gHAT, it fulfils most of the requested criteria.

The extraction and PCR on the ARIES^® platform takes 1 h 52 min total run time, even with a 30–40 min pre-treatment, it is still faster than manual extraction and PCR, and significantly faster than the 48 h ceiling stated by the TPPs [69, 87]. Automated qualitative result interpretation (including PDF output) is also available. Only three days of training were needed for laboratory technicians with moderate experience, aligning with the TPPs targets of under seven days [69] and in line with similar findings for other S2R-platforms [25, 63, 65, 91–94]. Although the ARIES^® platform is validated for use of swabs and clear liquids, a sample extraction optimization for whole blood with a sample volume of 75–100µL was optimized to meet the TPP requests on venous blood for HAT and arboviruses [69, 87]. Adding DNA/RNA Shield improved sample preservation (up to 2 weeks at 37 °C or nearly 2 months at 4 °C (unpublished data)). Since samples are often collected in remote areas, it is an important minimal acceptable requirement in the TPPs [69]. The preparation of the sample extraction optimization is a simple procedure involving 4 pipetting steps and an incubation for 10 min, not exceeding the minimally acceptable steps in the TPP. Since the cartridges are individually packed, single sample testing is possible without waste of unused reagents. In addition, the ARIES^® platform functions in an environment of 10–30 °C and 40–70% of humidity (ARIES^®: 15–35 °C and 20–80% humidity) [26, 69, 87]. The shelf life of the cartridges and ExoReady mix tubes is 1 year and 9 months, respectively, and storage temperature is 15–30 °C, fulfilling the minimum requirement of the TPP of WHO and FIND [69, 87]. No contamination was observed, although sample preparation and the ARIES machine were in the same room at CRSK Kimpese. Even though the S2R ARIES^® technology is designed to lower the threshold for having molecular diagnostics accessible in LMIC settings, it still does pose several limitations such as the language setting or service. The only available system language was English, whereas not all lab-technicians trained were fluent in English. Normally, the ARIES^® platform would be installed by a Luminex Field engineer, however DRC was not covered by Luminex/Diasorin’s service and the ARIES^® was installed without running the initial tests and calibrations [26]. The recent discontinuation of the ARIES platform shows the weakness of this application due to the dependence on one company. Even with an open system, end users remain dependent on the company for the production of consumables, system installation and maintenance, and—if the desired PCR or sample type is not part of the company’s portfolio—for performing costly and time-consuming validations. The PCR-assay design presented in this manuscript can be transferred to other S2R-platforms with an open system or to a conventional real-time PCR machine. The SARS-CoV-2 pandemic has shown the feasibility of installing molecular diagnostic laboratories quickly with a great success over a range of low-resource settings.

The main limitations of this study were the limited availability of samples and the geographic representation. For some pathogens, only a small number of samples were available, most notably no samples with co-infections, YFV samples and a limited number of DENV and T.b. rhodesiense-positive samples. Retrospective samples at ITM came from travelers, with infections from all around the world, not necessarily from DRC. Only CHIKV samples were available from DRC. The Trypanosoma samples were provided by WHO and their provenance was unknown.

Ultimately, the ARIES^® platform and the HAT ARBO PCR were implemented successfully in the national reference laboratory in Kinshasa and in a rural hospital in Kimpese. This demonstrates both the system’s versatility and the feasibility of deploying automated molecular diagnostics in peripheral settings lacking a dedicated molecular biology unit. Despite the discontinuation of ARIES^®, this study highlights the need, feasibility and practicality of context-adapted, S2R molecular diagnostics in low-resource environments.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AEC

Animal ethics committee

BC

Buffy coat

c/ml

Copies/ml

carrRNA

Carrier RNA

CHIKV

Chikungunya virus

CRSK

Centre de Recherche en Santé de Kimpese

CSF

Cerebrospinal fluid

DENV

Dengue virus

DRC

Democratic Republic of the Congo

EC

Ethics committee

EEEV

Eastern equine encephalitis virus

EQC

External quality control

FIND

Foundation for Innovative New Diagnostics

gHAT

Gambiense HAT

HAT

Human African trypanosomiasis

INRB

Institut National pour la Recherche Biomédicale

IRB

Institutional review board

ITM

Institute of Tropical Medicine

JEV

Japanese encephalitis virus

LB

Lysis buffer

LMIC

Low and middle-income countries

LOD

Limit of detection

M

Multiplex-PCR

mAECT

mini Anion-Exchange Centrifugation Technique

MAYV

Mayaro virus

M-S

Cq value from simplex PCR subtracted from multiplex PCR

ONNV

O’nyong-‘nyong virus

PBS

Phosphate-buffered saline

POC

Point-of-care

protK

Proteinase K

PSG

Phosphate-buffered saline-glucose

QCMD

Quality Center for Molecular Diagnostics

RFU

Relative fluorescence units

RRV

Ross river virus

RSSV

Russian spring-summer encephalitis virus

RT-PCR

Reverse-transcription polymerase chain reaction

RVFV

Rift valley fever virus

S

Simplex-PCR

S2R

Sample-to-result

SD

Standard deviation

SFS/NV

Sandfly fever sicilian and naples virus

SH

DNA/RNA shield

SINV

Sindbis virus

SLEV

St. Louis encephalitis virus

SOP

Standard operating procedure

SPC

Sample process control

T.b.

Trypanosoma brucei

TBEV

Tickborne encephalitis virus

TOSV

Toscana virus

TPPs

Target product profiles

UPC

Université Protestante du Congo

USUV

Usutu virus

VEEV

Venezuelan equine encephalitis virus

WB

Whole blood

WEEV

Western equine encephalitis virus

WHO

World Health Organization

WNV

West Nile fever virus

YFV

Yellow fever virus

ZIKV

Zika virus

Author contributions

A.H. was responsible for the conceptualization of the project, the data analysis and preparation of all figures, methodology, supervision and writing of the original draft of the manuscript; S.R. contributed with the methodology, investigation, writing and revision of the manuscript; S.M., M.K., M.M.K. were responsible for the investigation in INRB and CRSK; N.V.R. contributed to the methodology and the writing and revision of the manuscript; L.C. conceived the idea and supervised the validation at ITM; R.I.L. contributed to the conceptualization of the project, project administration and reviewed the manuscript; M.V.E. provided resources for the validation at ITM and reviewed the manuscript; O.K. and D.P. provided resources and supervision for the investigation in CRSK; P.V. contributed to the conceptualization of the project, the funding acquisition, the writing and the revision of the manuscript; S.A-M. provided resources and supervision for the investigation at INRB and reviewed the manuscript; K.K.A. was responsible for the conceptualization of the project, the funding acquisition, supervision, writing of the original draft and reviewing the manuscript.

Funding

This research was funded by the Belgian Directorate General for Development Cooperation and Humanitarian Aid (DGD).

Data availability

Additional data is available on request.

Declarations

Ethics approval and consent to participate

Ethical approval was obtained by the Institutional Review Board of the ITM (IRB reference number 1449/20), the Ethics Committee (EC) of the University Hospital in Antwerp (Belgian registration number: B3002021000025) and the EC of the Université Protestante du Congo (UPC) for the use of clinical samples collected within the FIKI² study (Clinicaltrials.gov NCT04760678). All participants gave their written informed consent. The study was conducted according to the principles stated in the Declaration of Helsinki, and to all applicable local and international regulations. Only stored samples with authorization for secondary usage were used in this study and the authorization from the local Ethical Committee was obtained. The viral RNA used for these experiments was extracted from viral culture supernatant obtained from virus isolations from returning travellers, or in the case of CHIKV, from a previous study in Aruba [27], which were completely anonymized and cannot be linked to the original donor. The use of rodents for the expansion of T.b. gambiense LiTat 1.3 was approved by the Animal Ethics Committee (AEC) of the Institute of Tropical Medicine (ITM) under protocol TU2021-1.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.