Utility Of POC Xpert HIV-1 Tests For Detection-Quantification Of Complex HIV Recombinants Using Dried Blood Spots From Kinshasa, D. R. Congo

Point-of-Care (POC) molecular assays improve HIV infant diagnosis and viral load (VL) quantification in resource-limited settings. We evaluated POC performance in Kinshasa (Democratic Republic of Congo), with high diversity of HIV-1 recombinants. In 2016, 160 dried blood samples (DBS) were collected from 85 children (60 HIV−, 18 HIV+, 7 HIV-exposed) and 75 HIV+ adults (65 treated, 10 naive) at Monkole Hospital (Kinshasa). We compared viraemia with Cepheid-POC-Xpert-HIV-1VL and the non-POC-COBAS®AmpliPrep/COBAS®TaqMan®HIV-1-Testv2 in all HIV+, carrying 72.4%/7.2% HIV-1 unique/complex recombinant forms (URF/CRF). HIV-1 infection was confirmed in 14 HIV+ children by Cepheid-POC-Xpert-HIV-1Qual and in 70 HIV+ adults by both Xpert-VL and Roche-VL, identifying 8 false HIV+ diagnosis performed in DRC (4 adults, 4 children). HIV-1 was detected in 95.2% and 97.6% of 84 HIV+ samples by Xpert-VL and Roche-VL, respectively. Most (92.9%) HIV+ children presented detectable viraemia by both VL assays and 74.3% or 72.8% of 70 HIV+ adults by Xpert or Roche, respectively. Both VL assays presented high correlation (R2 = 0.89), but showing clinical relevant ≥0.5 log VL differences in 15.4% of 78 cases with VL within quantification range by both assays. This is the first study confirming the utility of Xpert HIV-1 tests for detection-quantification of complex recombinants currently circulating in Kinshasa.

The access to routine molecular tools for early infant HIV-1 diagnosis (EID) and viral load (VL) quantification in children and adults is required for an early antiretroviral treatment failure identification and the prompt linkage to care. It can reduce HIV-associated mortality and morbidity in infected populations [1][2][3] . However, most of 37 million HIV-infected individuals live in resource-limited countries with a high number of different circulating HIV-1 variants, high rates of infection and no or limited access to routine HIV monitoring 4,5 . These settings have insufficient access to laboratory facilities, cold-chain management shortcomings, and difficulties for plasma collection and sample transportation 5,6 .
Until now, conventional molecular tests for EID and VL needed long procedures conducted in specialized and centralized laboratory settings requiring substantial infrastructure and training, needing turnaround times of several weeks or months 7,8 . This could increase the risk of loss to clinical follow-up of patients, thus having a negative impact in the HIV treatment cascade 9,10 .
To improve the linkage to care of HIV-exposed and infected subjects, some new easy to perform molecular assays for EID and VL quantification have been developed: point-of-care or POC assays. They have emerged as 1 HIV-1 Molecular Epidemiology Laboratory, Microbiology and Parasitology Department, University Hospital Ramón y Cajal-IRYCIS and CIBEREsp-RITIP, Madrid, 28034, Spain. 2 Monkole Hospital, Kinshasa, Democratic Republic of the Congo. 3 Microbiology Department, Clínica Universidad de Navarra, Navarra Institute for Health Research (IdiSNA), Institute of Tropical Health, Universidad de Navarra (ISTUN), Pamplona, 31008, Spain. 4  HIV diagnosis and viraemia quantification. HIV diagnosis was firstly performed in DRC using rapid serological tests: Determine ™ HIV-1/2 Ag/Ab (Alere), Double-Check Gold HIV 1&2 (Orgenics) and Uni-Gold HIV (Trinity Biotech) from 18-months old and by Biomerieux 4th generation immunoassay VIDAS ® HIV Duo Ultra or exceptionally by molecular Abbott real-time HIV-1 Qualitative in infants under 18-months old. In Madrid, Spain, HIV serological status in the 85 children was confirmed with BioRad Geenius TM HIV-1/2 confirmatory assay using one DBS dot per patient, as previously reported 33 . All HIV seropositive and undetermined pediatric DBS by Geenius were then tested by POC Cepheid Xpert Qual (Xpert Qual), which provides a binary "detected"/"not detected" result 34 . In Navarra, Spain, HIV serostatus was confirmed in all adults by two 4th generation immunoassays: Elecsys ® HIV combi PT (Roche) and VIDAS ® HIV Duo Quick (bioMerieux).
HIV-1 viremia was quantified using Cepheid Xpert HIV-1 VL (Xpert VL) 35 and COBAS ® AmpliPrep/ COBAS ® TaqMan ® HIV-1 Test v2.0 (Roche VL) 36 in all HIV+ DBS, both techniques based on real time amplification of HIV genome. All assays were performed using one dot eluted in Xpert Qualitative buffer for Xpert assays or Roche SPEX buffer for Roche-VL as lysis buffer to elute the DBS dots, according to manufacturer's instructions. GeneXpert ® Instrument automates and integrates specimen preparation, HIV-1 total nucleic acids (viral RNA and proviral DNA) extraction and amplification, and detection of the target sequence in specimens using real-time reverse transcriptase PCR (RT-PCR). The systems require the use of single-use disposable GeneXpert ® cartridges that hold all the necessary RT-PCR reagents and host the RT-PCR processes.
For statistical analysis of VL data, any viraemia values reported by the system as <40 cp/ml (by Xpert VL) or <20 cp/ml (by Roche VL), lower limit of detection of each assay, were reported as 39 cp/ml or 19 cp/ml, respectively, being considered detected but not quantifiable. We identified treated subjects under therapeutic failure when they present HIV-1 viraemias of 1,000 cp/ml or higher, clinical treatment failure threshold using DBS 37 . Both HIV-1 VL assays were based on real time PCR, providing an assay-specific cycle threshold (Ct), which inversely correlates with the starting concentration of the viral genome in the infected specimen. Ct values were recorded following DBS VL quantification by both Xpert VL and Roche VL platforms using one DBS dot in each sample.
We provided the number of HIV-1 RNA copies per dot and per plasma milliliter after considering patient's hematocrit assuming 39% hematocrit for children, 42% for women and 47% for men, according to previous studies 38,39 . This lead to plasma volumes of 42.7 µl, 40.6 µl and 37.1 µl, respectively, in 70 µl blood collected per dot. The main features of the three molecular HIV assays used in the study are described in Table 1.

HIV-1 variant characterization.
For HIV-1 variant characterization, RNA was extracted from 2 DBS dots using the NucliSENS easyMAG automated platform (BioMerieux) or manual High-Pure Viral Nucleic Acid (Roche) kit. Viral RNA was amplified in the HIV-1 pol coding region by RT-PCR and nested-PCR using primers designed by WHO 40 as previously described 41 Table 2. The 160 study subjects were mainly seropositive by rapid serological tests in DRC (58.75%), female (57.5%) and antiretroviral experienced (46.2%). The mean age for HIV diagnosis in DRC was 8.1 (SD 5.38) years old in children and 40.4 (SD 12.13) in adults. The mean age at DBS collection was 9.8 (SD 5.11) years old in children and 46.5 (SD 12.27) in adults. All but one children were born in Kinshasa, 8.2% were orphaned and 6 presented HIV/ Mycobacterium tuberculosis confection. Most adults were female (70.7%) Among the 92 children and adults diagnosed as HIV positive in the local laboratory in Congo by rapid tests and 80.4% were antiretroviral (ARV) experienced at sampling. The remaining were ART-naive or with unknown treatment data. NRTI and NNRTI were the most used ARVs in the study cohort, Zidovudine + Lamivudine + Nevirapine being prescribed in 60 treated patients (13 children and 47 adults), mainly as first line therapy (50 cases). Only 8 patients received protease inhibitors (PIs) based treatment with Lopinavir/Ritonavir, while integrase inhibitor use was absent among study subjects. (Table 2).
HIV-1 infection was confirmed in 16.5% of 85 children by Xpert Qual and in 93.3% of 75 HIV+ adults diagnosed in DRC by both Xpert VL and Roche VL. However, we identified false positive HIV diagnosis in DRC after following rapid serology testing algorithm in 4 adults (range 23.8-28.5 years age) and in 4 children (range 5.3-13.1 years age), and 5 of them (2 adults, 3 children) were under unnecessary ART for a mean time of 3.9 years (Table 3) Roche, respectively. Two specimens not detected by Roche were detected by Xpert (<40 cp/ml), while 4 specimens below limit of detection of Xpert could be only detected by Roche, with lower detection limit (<20 cp/ml) ( Table 4).
High correlation among Xpert and Roche VL assays using DBS. Viral Load results within the quantification range of both assays were available for 78 (92.8%) of 84 HIV+ patients. The POC Cepheid Xpert HIV-1 VL assay showed excellent agreement (ICC = 1) with Roche VL for HIV-RNA quantification. A high and significant correlation was observed among both VL assays (R 2 = 0.89, P < 0.001), as shows the estimated regression line (Fig. 1). However, Ct values for HIV-1 quantification were VL assay dependent when quantifying the 78 DBS detected by both VL assays.
The similarities between both VL assays were evaluated by the Bland-Altman plot method (Fig. 2). HIV-1 VL overestimation by one of the two assays (Xpert VL or Roche VL) was observed in all but one specimen among the 78 DBS, although the difference was below a clinically relevant threshold of 0.5 log 10 cp/ml in most cases (84.6%). The POC Xpert HIV-1 VL assay tended to overestimate HIV-1 VL in 69.2% samples, and the non-POC Roche VL in 29.5% specimens (Fig. 3, Table 4). The overall mean difference in the HIV-1 RNA values obtained by Xpert VL assay and Roche VL was 0.30 log 10 cp/dot (95% CI: 0.26 to 0.35 log 10 cp/dot) (P < 0.001). However, clinical relevant differences (≥0.5 log VL) ranging from −0.55 to 1.07 were observed in 12 (15.4%) of 78 DBS specimens with VL above detection limit by both assays (Figs 2 and 3), differing across samples and assays. Eleven cases corresponded to Xpert VL use, while only one to Roche VL testing (Table 4).
The impact of each complex recombinant on VL quantification was unclear (

Discussion
POC test use can improve the clinical management of HIV-infected infants and adults and reduce the delay in diagnosis and in ART failure identification 14 . Early infant diagnosis is a WHO priority 14,48 , since it allows early ART to be established and reduces irreversible damage to central nervous and immune systems, viral reservoirs, as well as HIV transmission and morbidity/mortality associated with HIV-1 infection 49 . A correct early HIV diagnosis is also essential, since false positive HIV tests might result in unnecessary antiretroviral treatment and psychological distress in falsely diagnosed individuals and families 50 . The use of DBS has been proposed as an alternative sample to plasma/serum, easier to be collected, stored and shipped, very convenient in limited resource countries 15,33,37,51 . The aim of this study was to evaluate the performance of two POC HIV-1 assays (Xpert-VL and Xpert Qual) for HIV detection and/or quantification using DBS in the DRC, a country with a high HIV-1 diversity including a high prevalence of complex recombinants, mainly URFs.
Although POC molecular testing prevents inappropriate HIV serological diagnosis and is cost-effective 52 , it is not yet globally adopted in all EID or adult programs 48 . In the DRC, the National Program of fight against HIV-AIDS (PNLS) recommends performing EID 0-2 days after birth, but POC assays for EID have not been implemented yet within the clinical routine of HIV-exposed infants.
We have evaluated the clinical impact of the lack of routine EID molecular testing in HIV-exposed newborns shortly after birth and of confirmatory serological testing in older children and adults in a cohort in Kinshasa. We have found false HIV diagnoses among 5% of the study participants that lead to unnecessary ART in five HIV uninfected subjects. The high prevalence of false positive diagnosis among HIV-exposed infants less than 18 years old could be explained by the long delay in PCR results from a centralized national laboratory, which were only available at the clinical center in Kinshasa 6 months after the original HIV serological test. Wrong diagnoses may also be a consequence of the absence of a confirmatory molecular test with a second new sample as recommended by WHO 14 for infants. In older children and adults, false HIV diagnosis can be due to the local absence of confirmatory serological analysis and the exclusive use of serological rapid testing for HIV diagnosis. Although rapid immunochromatographic test for HIV are recommended in low income countries 53 , the low HIV prevalence among the general population in the DRC (0.7%) 4 may be associated with a lower positive predictive value for these methods. In addition, rapid HIV testing is not appropriate for acute infection diagnosis 53 . Another aspect that could influence a misdiagnosis is the fact that a subjective reading of rapid HIV tests may speed up the (2019) 9:5679 | https://doi.org/10.1038/s41598-019-41963-y www.nature.com/scientificreports www.nature.com/scientificreports/ communication of false positive results. The high mean age at HIV diagnosis in DRC in children (8.1 years) from the study cohort would suggest HIV diagnosis delay, since most of them acquired HIV infection by vertical route according to clinical reports.
Results also show that Xpert and Roche molecular tests were superior to fourth generation serological screening assays to identify HIV infection in HIV-exposed children older than 18 months and adults. Moreover, Xpert resolved HIV status and rule out HIV infection in DBS from 10 individuals with undetermined results by Geenius confirmatory assay, in agreement with recent studies 54 . It could be due to the fact that Xpert HIV-1 Qual can detect HIV-1 infections up to 7-10 days before seroconversion, an average of nine days earlier than a panel of HIV-1 antibody tests and five days earlier than a panel of HIV-1 antigen p24 tests 34 . Xpert ® HIV-1 Qual may play a role in the diagnosis of HIV, either in EID, as a confirmatory test after antibody-based testing, or for the detection of acute HIV infection in antibody negative patients recently infected 55 .
POC Xpert VL and non-POC VL Roche assays were also suitable in quantifying VL using DBS in the DRC. A high percentage of subjects under ART in the DRC showed VL levels above 1,000 cp/ml by Xpert (92.7%) or Roche (63.8%) (Table 1), the VL threshold defined by WHO to confirm virological failure in low-and middle-income countries and when using DBS in adults and children 37 . This is the optimal threshold for presenting the lowest percentage of misclassification compared with higher thresholds 51 . In fact, most patients in the study had a clinical suspicion of virological failure. However, the prevalence of ART failure was overestimated using Roche VL and this threshold, resulting less specific than other VL assays using DBS, as described in previous studies 51,56 .
Despite the good correlation in VL quantification using Xpert VL with the no-POC Roche VL assay (R 2 = 0.89), still 15.4% of 78 samples with VL within the quantification range of both assays presented clinically significant VL differences above 0.5 log 10 cp/ml, according to previous studies 16  www.nature.com/scientificreports www.nature.com/scientificreports/ expanded use of VL in the DRC for an early detection of virological failure as well as the use of the same VL technique for each patient during ART monitoring to reduce potential assay-associated viraemia overestimations, which could be interpreted as virological failure events. This could reduce unnecessary ART regimen switches in these patients, favoring an early clinical response by reinforcing adherence or changing ART regimen if resistant variants are detected before clinical symptoms associated with treatment failure appear. The achievement of the 90-90-90 UNAIDS objectives depend on HIV monitoring, otherwise a future epidemic of HIV resistant strains may occur and delay these objectives in Sub-Saharan Africa 58 . The finding of assay dependent Ct values for HIV-1 quantification reinforces the risk of establishing a standard Ct cutoff as accurate threshold value to differentiate virological failures in subjects under ART.
The continuous evolution of HIV can hinder diagnosis and complicate clinical practice 59 . Thus, one of the main challenges for molecular diagnostic and VL assays is to detect and/or quantify different HIV-1 variants correctly. According to the manufacturer's information, the Xpert ® HIV-1 Qual assay has been validated for specimens including groups N, O and M (9 subtypes and recombinants A/E, A/B and AG/GH) 60 and the Xpert VL for groups N, O and M (9 subtypes, CRF01_AE, CRF02_AG, and CRF03_AB) 61 . However, most 98 HIV-1 CRF 43 and complex unique recombinant forms (URF) have not been validated yet. Roche VL was evaluated by analysis of HIV-1 group O and group M subtypes A through H from cell culture origin 36,62 , although it was also able to quantify a number of CRFs 16 . However, none of them has been evaluated across a large panel of URF variants, as we reported. We demonstrate that POC-Xpert assays and VL and Roche-VL can successfully detect and quantify complex recombinants in pol.  www.nature.com/scientificreports www.nature.com/scientificreports/ We also provide new data related to the HIV molecular epidemiology in Kinshasa, reporting an extremely high rate of unique inter-subtype recombinants in recently infected populations, although URF prevalence could be underestimated since HIV-1 variants were characterized considering pol gene but not the complete genome. Other authors have also highlighted the extreme diversity of HIV strains circulating in the country with a high presence of URFs and different CRFs, together with a low presence of subtype C 28,31 .
Some studies have reported clinically-significant differences (>0.5 log) in VL quantification across techniques with the same non-B strains 23,63-65 . These results would suggest different rates of detection across assays due to genetic variability in the HIV-1 gene region targeted by the assays in incorrectly quantified samples, as previously demonstrated for a CRF02_AG variant 63 75 has already been used in the DRC. The National HIV/AIDS Program in the DRC plans to use the existing Xpert machines for HIV EID testing and VL monitoring in order to improve the clinical care of HIV-infected children and adults. Our results demonstrate that Xpert performance is adequate for testing HIV-1 variants currently circulating in the DRC. Thus, due to simplicity, rapid results and good performance, POC Xpert HIV-1 can be useful in the decentralization of EID and VL monitoring from specialized laboratories in the DRC to local hospitals and clinics within their routine clinical care. This will help to reach the ambitious 90-90-90 goals in the country. We also confirmed that DBS could be a suitable sample for Xpert use in the DRC, requiring a minimum volume of blood, favoring molecular testing in infants and low-weight children. In addition, DBS samples offer additional advantages, they are not considered biohazardous once dried and are not as time and temperature sensitive as plasma specimens 37 .
An important limitation of this study is that we did not compare DBS to plasma due to the lack of paired plasma/DBS specimens collected for each subject in the study population. Moreover, due to the design of the study and sample size, we could not determine the statistical power of possible performance differences across assays in each specific HIV-1 variant. Finally, we could not explore the effect of DBS lysis buffer in viraemia quantification, which could influence DBS VL results according to previous reports 76 . The main strength of our study is that it shows the first results confirming the utility of POC Xpert HIV-1 tests and Roche VL platform for early HIV-1 diagnosis and for VL quantification of complex recombinants (mainly URF) currently circulating in Kinshasa, the epicenter of HIV-1 group M epidemic and where a large number of complex recombinants cocirculate. To our knowledge, there are no previous studies that have included a large panel of different HIV-1 complex recombinants characterized by phylogenetic analysis during Xpert-POC HIV-1 evaluation for EID and VL. We also report some of the current limitations of HIV diagnosis and monitoring in DRC. Since Xpert assays and DBS use can improve early diagnosis in HIV-exposed infants and early detection of ART failures in countries with complex HIV-1 recombinants and limited infrastructures, as in the DRC, our results could have a direct clinical impact in global HIV diagnosis and monitoring to reach early the 90-90-90 objectives.