Short Communication: Validation of the Asante HIV-1 Rapid Recency Assay for Detection of Recent HIV-1 Infections in Uganda

Abstract

Point of care rapid recency testing for HIV-1 may be a cost-effective tool to identify recently infected individuals for incidence estimation, and focused HIV prevention through intensified contact tracing. We validated the Asante™ HIV-1 rapid recency^® assay for use in Uganda. Archived specimens (serum/plasma), collected from longitudinally observed HIV-1 recently and long-term infected participants, were tested with the Asante HIV-1 rapid recency assay per manufacturer's instructions. Previously identified antiretroviral therapy (ART)-naive samples with known seroconversions within 6 months of follow-up were tested in independent laboratories: the Rakai Health Sciences Program (RHSP) and the Uganda Virus Research Institute HIV Reference Laboratory (UVRI-HRL). In addition, samples from participants who seroconverted within 6–18 months and samples from individuals with chronic HIV-1 infection of at least 18 months duration were classified into three categories: ART naive, ART exposed with suppressed viral loads, and ART exposed with detectable viremia. Of the 85 samples seroconverting in ≤6 months, 27 and 42 samples were identified as “recent” by the Asante HIV-1 rapid recency test at the RHSP laboratory and UVRI-HRL, corresponding to sensitivities of 32% and 49%, respectively. There was 72% agreement between the laboratories (Cohen's kappa = 0.481, 95% CI = 0.317–0.646, p < .0001). Specificity was 100% (200/200) among chronically infected ART-naive samples. The Asante HIV-1 rapid recency assay had low sensitivity for detection of recent HIV-1 infections in Uganda, with substantial interlaboratory variability due to differential interpretation of the test strip bands. Specificity was excellent. Assessment of assay performance in other settings is needed to guide decisions on test utility.

Introduction

Determination of HIV-1 incidence is important for epidemic monitoring, identification of individuals at high risk of transmission, and for measuring the impact of intervention programs.^1,2 The gold standard for incidence assessment is based on prospective cohort studies¹ that are expensive, time consuming, and findings may not be generalizable to the general population due to potential selection bias, loss to follow-up, and behavioral modification due to repeated testing and counseling.³ Alternatively, incidence may be estimated from tests for recent infections (TRIs), which differentiate recent from long-term infections using cross-sectional testing of biomarkers associated with different durations of infection from samples collected at a single time point.⁴ TRIs have evolved and the Limiting Antigen Avidity (LAg-Avidity) Assay ⁵ is commercially and widely available.

The Asante™ HIV-1 rapid recency^® assay (Sedia Biosciences, Portland, OR) is a LAg-Avidity assay that has been modified to a rapid lateral flow format.⁶ According to the manufacturer, the test can identify individuals recently infected with HIV-1 defined as <180 days from seroconversion within 20 minutes.⁷ The President's Emergency Plan for AIDS Relief and the Ministry of Health of Uganda adopted use of the Asante HIV-1 rapid recency assay in a nationwide surveillance program. However, there was no validation of the test in Uganda. We validated the diagnostic performance of the Asante assay among recent seroconverters and chronically infected antiretroviral therapy (ART)-naive samples in Uganda.^8,9

Materials and Methods

Ethical approval

The study was approved by the Uganda Virus Research Institute (UVRI)'s Research Ethics Committee (REF: GC/127/20/01/758) and Uganda National Council for Science and Technology (REF: HS534ES). Samples used in the validation were only those from participants in longitudinal studies that had provided prior written informed consent for use of their archived samples in future research studies.

Study samples

The study used archived samples from longitudinally followed participants at two study sites: the Rakai Health Sciences Program (RHSP) and the Medical Research Council (MRC)-Masaka. The RHSP participants had confirmed seroconversions ≤6 months and the MRC participants were from a high-risk cohort with follow-up visits every 3 months. A total of 203 samples from seroconverters within ≤18 months and 600 chronically infected samples were tested. Seroconverter samples were from participants who had seroconverted within ≤6 months (n = 85), within 6–12 months (n = 35), and within 12–18 months (n = 83). Chronically infected samples were collected from participants with HIV-1 infection of at least 18 months duration and included ART naive (n = 200), ART exposed with detectable viral loads (i.e., ≥1000 copies/mL, n = 200), and ART exposed with suppressed viral loads (i.e., <1000 copies/mL, n = 200). We also tested 100 HIV-negative samples to evaluate the specificity of the assay (Table 1).

Table 1.

Samples Used in the Validation

Category	Subcategory	No.
Seroconversion samples	Within ≤6 months	85
	Within 6–12 months	35
	Within 12–18 months	83
Chronically infected samples (collected at least 18 months after seroconversion)	ART naive	200
	ART exposed with detectable viral loads	200
	ART exposed with suppressed viral loads	200
HIV negative samples	HIV negative	100

Laboratory testing

The 85 HIV-1 seroconverter samples ≤6 months were tested at both the RHSP laboratory and the UVRI HIV Reference Laboratory (HRL). Chronically infected HIV-1 samples were only tested at the RHSP laboratory. Two RHSP technicians independently retested samples with discordant results between the RHSP and UVRI-HRL. Owing to limited residual volumes, repeat tester 1 tested 23 out of the 24 discordant samples, whereas repeat tester 2 tested 20 out of the 24 discordant samples. Laboratory technicians performed testing with all results read visually per the manufacturer's instructions. The Asante recency rapid test strip returns three possible results: recent HIV-1 infection (<180 days post-HIV seroconversion), long-term HIV-1 infection (>180 days post-HIV seroconversion) and a negative HIV result. Samples from HIV-1–infected individuals with a HIV-negative result on the Asante assay were retested using the Uganda national HIV testing algorithm (Alere Determine™ HIV–1/2, HIV 1/2 Stat-Pak, SD-Bioline HIV 1/2).¹⁰ All seronegative samples were excluded from the analysis.

Data analysis

Assay sensitivity for detection of recent HIV-1 infections was estimated for each laboratory. Sensitivity estimates excluded HIV-1–negative samples on the recency assay. Percentage agreement between the two testing laboratories and Cohen's kappa statistic were computed. Percentage agreement was also estimated between repeat testers and for discordant results. Asante assay detection of HIV-negative samples and specificities for long-term infections were also determined.

Results

Sensitivity for infection and recency

The assay verified HIV-1 infection with a sensitivity of 99.2% (803/809). Of the 85 samples from seroconverters within ≤6 months, 27 and 42 samples were accurately identified as “recent” by the Asante test when tested at RHSP and UVRI-HRL laboratories, corresponding to sensitivities of 31.8% (95% CI = 22.3–42.9) and 49.4% (95% CI = 38.5–60.4), respectively (Table 2).

Table 2.

Results from Testing of the ≤6 Months Seroconverter Samples

Asante test strip result	Frequency (%)		p a
	RHSP	UVRI
Recent (sensitivity)	27/85 (31.8)	42/85 (49.4)	.048
Long-term infected	55/85 (64.7)	39/85 (45.9)
Negative	3/85 (3.5)	4/85 (4.7)
Total	85	85

^a Fisher exact p-value for comparison of recency results between laboratories.

RHSP, Rakai Health Sciences Program; UVRI, Uganda Virus Research Institute.

Agreement between and within laboratories (for recency)

There was 72% agreement between the two laboratories with a Cohen's kappa = 0.481 (95% CI = 0.317–0.646, p < .0001). There were 24 (28%) samples with discordant results between the RHSP and UVRI-HRL. There was an 80% agreement (20/24) between the two repeat testers of the discordant samples between the RHSP and UVRI-HRL laboratories with kappa = 0.575 (95% CI = 0.258–0.891, p = .00038). There was 65% and 75% agreement with the initial RHSP laboratory results for repeat testers 1 (n = 23) and 2 (n = 20), respectively, and 35% and 25% agreement with UVRI-HRL results by the same two testers.

Specificity for HIV-1–negative samples and long-term infection diagnosis

All the 100 longitudinally observed HIV-1–negative samples tested negative on the Asante HIV-1 rapid recency assay. We assessed assay performance on other seroconverter samples (between 6 and 18 months duration) with 11.4% and 16.9% classified as recent for 6–12- and 12–18-month categories, respectively. Specificity and sensitivity were not determined for these categories due to inability to resolve actual duration of HIV-1 infection (Table 3). Recency testing of HIV-1 chronically infected (≥18 months duration) ART-naive samples resulted in a specificity of 100% (95% CI = 97.7–100), whereas specificities of 98.5% (95% CI = 95.3–99.6) and 87.4% (95% CI = 81.4–91.2) were estimated for chronically infected samples with detectable (i.e., ≥1000 copies/mL, range 1005–1,937,225 copies/mL) and suppressed viral loads (i.e., <1000 copies/mL), respectively (Table 4). After correcting for viral suppression as specified in the kit insert, specificity for long-term infection among those with a suppressed viral load increased from 87.0% to 99.5%.

Table 3.

Results from Testing of ART-Naive Samples That Seroconverted Within 6–18 Months

Asante test strip result	Frequency (%)a		p
	6–12 months	12–18 months
Recent	4/35 (11.4)	14/83 (16.9)	.532
Long term	30/35 (85.7)	68/83 (81.9)
Negative	1/35 (2.9)	1/83 (1.2)
Total	35	83

^a Specificity/sensitivity not determined due to inability to resolve the actual duration of the infection at the sampling date, negative sample retested on HIV rapid diagnostic test and confirmed positive.

ART, antiretroviral therapy.

Table 4.

Results from Testing of Samples from Chronically Infected Individuals

Asante test strip result	Frequency (%)
	ART naive	ART exposed (detectable VL)	ART exposed (suppressed VL)
Recent	0/200 (0.0)	3/200 (1.5)	25/200 (12.5)a
Long term (specificity)	200/200 (100)	197/200 (98.5)	174/200 (87.0)
Negative	0/200 (0.0)	0/200 (0.0)	1/200 (0.0)
Total	200	200	200

^a As per manufacturer's kit insert, all recent samples must be assessed for HIV-1 viral load, and if suppressed, reclassified as long term and this correction improves specificity to 99.5.

VL, HIV viral load.

Of the 803 HIV-1–positive samples (both seroconverter and chronic), 6 (0.7%) tested negative on the Asante HIV-1 rapid recency assay (Tables 2–4).

Discussion

The Asante HIV-1 rapid recency assay is a rapid lateral flow format assay incorporating a recombinant immunodominant gp41 region common to HIV-1 group M that detects recent HIV-1 infections within the preceding 180 days. This assay has been promoted in many Sub-Saharan African countries to identify new HIV-1 infections for targeted prevention and treatment, and to estimate HIV-1 incidence.⁷ We assessed the performance of this assay on samples from individuals with known duration of HIV-1 infection from Uganda.

The Asante recency assay had a low sensitivity of between 30% and 50% for identifying samples from individuals infected <6 months. These results support the data presented by the independent evaluation by the Consortium for the Evaluation and Performance of HIV Incidence Assays, which demonstrated that the mean duration of recent infection was 65 days, not 180 days as stated in the test kit package insert.¹¹ We found substantial interlaboratory variability in test strip scoring, which is also likely to be a problem in field settings with suboptimal lighting that may further limit the programmatic utility of the test. The assay manufacturer recommends use of a strip reader for improved accuracy. However, current programmatic rollout in Uganda does not use strip readers as they were unavailable, so it is not known whether this would have improved test performance. Studies validating performance of the assay with a strip reader are needed. Given the low sensitivity for recent infections, this assay using visual assessment of bands may be of limited utility for HIV-1 recency screening and cross-sectional incidence estimation in this or similar settings, but may be of some value in identifying recent infections within high incidence subgroups or populations. Also, given the high cost of the optical readers ($2000–3000), routine use may not be cost effective in resource-limited settings. Improvements in band intensity may improve inter-reader variability.

However, the assay accurately categorized all HIV-1–negative samples, consistent with estimated specificity of 100%, from a field validation in Vietnam¹² with an HIV-1 clade A/E (CRF01-AE) epidemic.

Conclusion

In this validation study, the Asante HIV-1 rapid recency assay had a low to moderate sensitivity for HIV-1 recency detection <6 months in longitudinally characterized seroconverter samples in Uganda. Test performance varied substantially between two independent laboratories with test strips read visually. Further validations in diverse settings with different HIV-1 clades and incorporation of optic readers may be useful in determining the utility of this assay for detection of recent HIV-1 infection.

Authors' Contributions

Laboratory testing was carried out by C.S., A.N.K., J.E., R.M.G., S.N.K., and D.S. MRC cohort follow-up was taken care of F.M.K. and P.K. Data analysis was done by A.N. and A.J.A. Article preparation was taken care of C.S., R.M.G., G.K., D.S., G.N., T.L., E.N.K., J.B.W., J.K., P.K., A.N.K., D.S., C.W., M.G., M.W., R.G., L.W.C., O.L., S.J.R., and T.C.Q.

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This study was supported, in part, by the Division of Intramural Research, National Institute of Allergy and Infectious Diseases (NIAID), National Institutes of Health (NIH), and grant R01-095068 of NIAID, NIH.