Correlation of Prospective and Cross-Sectional Measures of HIV Type 1 Incidence in a Higher-Risk Cohort in Ho Chi Minh City,Vietnam

Abstract

The primary aim of this study was to estimate HIV incidence within a high-risk population in Ho Chi Minh City (HCMC), Vietnam using both cross-sectional and prospective methodologies. A secondary aim was to develop a local correction factor for the BED and avidity index incidence assays. The research study design consisted of three phases: (1) cross-sectional, (2) prospective, and (3) BED false recent (BED FR). A total of 1619 high-risk, sexually active individuals were enrolled in the cross-sectional phase and 355 of the opiate-negative, HIV-negative women were subsequently enrolled in the prospective phase. Four-hundred and three men and women with known HIV infection duration of greater than 12 months were enrolled in the BED FR phase. The HIV prevalence for all participants in the cross-sectional phase was 15.8%. HIV incidence in the cross-sectional group was estimated using the BED IgG capture assay and AxSYM avidity index assay for recent HIV infection and incidence within the prospective cohort was determined by observations of HIV seroconversion. HIV incidence in opiate-negative women was estimated using the BED assay to be 0.8% unadjusted and 0.5% after applying the locally derived BED false recent rate of 1.7%; no seroconversions were observed in the prospective cohort. We also screened the cross-sectional samples for evidence of acute infection using nucleic acid testing, 4th generation HIV EIA, and SMARTube coupled with Genscreen and Determine diagnostic tests; no confirmed acute infections were identified by any method. HIV incidence within this opiate-negative study population was low and incidence estimates from the two methods compared favorably with each other. Incidence estimates and false recent rates using the AxSYM assay were higher: AI FRR of 2.7% and adjusted incidence of 1.7% per year (95% CI, 0.6, 2.8). By comparison, both HIV prevalence and incidence estimates for the opiate-positive group were higher.

Introduction

HIV/AIDS continues to be a significant global health problem with major implications for the health, economics, and political infrastructure of communities around the world. Recent prevention studies involving male circumcision, the use of preexposure prophylaxis with antiretrovirals, and the early treatment of HIV-positive partners to reduce transmission in serodiscordant couples have all shown significant promise.^1

–5 However, the evaluation of existing and new prevention methods requires additional research sites with populations at high risk for HIV infection and accurate methods for estimating the rate of new HIV infection.

Prospective follow-up of a cohort of HIV-negative persons has frequently been deemed the “gold standard” for determination of HIV incidence rates. This approach is, however, time-consuming, expensive, and logistically challenging in resource-limited settings. Furthermore, this type of study has inherent biases such that the enrollment of persons into a cohort study with repeat testing may lead to behavior changes that result in a lower observed HIV incidence than in the source population of interest. To address these challenges, laboratory testing methods have been developed to measure HIV incidence using specimens from cross-sectional sample sets.^6,7 These incidence assays and testing algorithms rely on the ability to distinguish recently and chronically infected individuals using biomarkers such as antibody levels and maturation as indicators of recent HIV infection.

Currently, the HIV BED incidence assay is the only commercial kit specifically available for incidence estimation. The BED assay has been used to estimate national HIV incidence, monitor trends, and identify risk factors in populations by applying the assay to specimens derived from cross-sectional, population-based surveys, integrated behavioral and biologic surveys, and sentinel surveillance activities.^8

–13 Evidence has shown, however, that when applied to cross-sectional settings, particularly in areas with moderate to high HIV prevalence, the BED may misclassify a proportion of long-term infections as recent, thus resulting in an overestimation of population HIV incidence. Samples from participants with known long standing HIV infection that are misclassified as recent in incidence assays are termed “false recent.” Several factors that influence the host's antibody or immune response to the infection have been identified that contribute to this misclassification rate. Foremost among these factors are antiretroviral treatment (ART) and the development of AIDS, as both conditions result in a decreased antibody response. Similarly, elite suppressors with consistently low antibody titers, or persons with long-term infection and very low CD4 cell counts who experience a decline in antibody as the body's response becomes more compromised by disease, may be misclassified as recent on the BED assay.^14,15

Recruitment and screening strategies for studies applying cross-sectional incidence methods have thus been adapted to exclude participants on ART and those with known long-term infection in this type of analysis. Other researchers have also developed mathematical correction factors to help mitigate this error,^16
–18 however, misclassification rates may vary in different populations, place, and time resulting in the need to generate real-time false recent rates within the same population concurrent with incidence estimations. Current HIV incidence research efforts are focused on standardizing approaches to evaluating incidence assays and building consensus on statistical methodologies to interpret the test data, as well as defining false recent rates from different populations and developing new and improved incidence assays or multitest recent infection testing algorithms (RITAs) that have lower false recent rates and are less susceptible to the host factors that impact the performance of the existing incidence assays.

In addition to the BED assay, other assays have been developed that measure a functional parameter, antibody avidity, to distinguish recent from long-term infections. In our study we used the AxSYM avidity index assay^19,20 to estimate HIV incidence within the cross-sectional population and also compared the results of the two assays for incidence estimations as well as the false recent rates in this population for the two assays. We also sought to identify acute infections by testing for p24 Ag (using a 4th generation Ag/Ab combo test), RNA (using pooled sample testing) and inducible antibody (using SMARTube), and to correlate these results with cross-sectional and directly observed incidence.

We report here the HIV incidence values obtained from the prospective and cross-sectional methodologies as well as the derivation of local false recent rates for the BED and AxSYM Avidity Index assays in Vietnam. This study was conducted in Ho Chi Minh City (HCMC), Vietnam as part of a larger initiative that used a standardized incidence protocol to establish local HIV incidence rates to aid in the identification and development of new study sites for HIV prevention research. This approach provides real-time information on HIV incidence and the suitability of the site for future HIV prevention research. In particular, this study aimed to determine the feasibility of the site for future microbicide studies; thus only opiate-negative women were recruited to the prospective arm to control for route of HIV infection. To address concerns of overestimation using cross-sectional methodologies, we collected data on the false recent rate within the local population to calibrate the cross-sectional incidence values. This study also provided the opportunity for study staff to develop their research skills using a preparatory study protocol and gain experience using two different methods for estimating HIV incidence.

Materials and Methods

Study sites

The study was conducted at three sites located in three different districts within HCMC: the Binh Thanh, District 4, and District 8 Community Counseling and Support Centers (CCSC). Each of these CCSCs houses a Voluntary Counseling and Testing (VCT) and HIV Outpatient Clinic (OPC) unit. These sites are part of the VCT/OPC clinic network overseen by the HCMC Provincial Aids Committee (PAC), and were research-naive prior to study initiation. In addition to these clinics, four other OPC clinics (District 3, 5, 7, and Go Vap) were accessed to recruit participants for the false recent (FR) phase of the study.

Ethical considerations

This research was reviewed and approved by the local PAC Ethics Committee (FWA#00003315) and the FHI Protection of Human Subjects Committee (FWA#00000025). All research was conducted per Good Clinical Practice (GCP) standards and International Conference on Harmonization (ICH) guidelines.

Study populations and specimens

Cross-sectional phase

Male and female participants for the cross-sectional phase of the study were recruited by local peer educators as well as by other study participants. In brief, local peer educators recruited a limited number of seed individuals who they judged would meet specific high-risk sexual behavior eligibility criteria. Upon presenting at the clinic, seed individuals and their referrals were screened for eligibility, enrolled in the cross-sectional phase, and offered up to three referral coupons to give to peers and sexual partners who were screened, enrolled, and asked to recruit in an analogous manner. Since the goal was not to reach population equilibrium as is the case with respondent-driven sampling (RDS) but simply to reach HIV-positive individuals, recruitment chains were terminated after they were deemed to be inactive or not connected to any HIV-positive individuals.

Eligibility was determined by age (18–35 years) and elevated risk of sexual acquisition of HIV as defined by having greater than three partners in the past month and greater than three sex acts per week, and/or sex with a man who has sex with men (MSM), an injection drug user (IDU), or a sex worker in the past month. Participants in the cross-sectional phase completed two visits separated by approximately 10–14 days; the first visit involved consent, interview, and specimen collection procedures, while test results were disclosed at the second visit consistent with standard VCT visit schedules and testing practices.

Demographic, risk behavior, and medical history information was collected at the first visit and a blood and urine sample taken. Approximately 20 ml of venous blood was collected from each participant at the first visit, processed for plasma, and aliquots stored frozen at −70°C prior to testing. Urine samples were also collected at the first visit; pregnancy testing and opiate testing were completed on the sample within 24 h. Pregnancy results were returned to the participants at their follow-up visit; opiate testing was for research purposes only and results were not returned to the participants.

Prospective phase

HIV-negative, opiate-negative women identified in the cross-sectional phase were recruited to create a prospective cohort that was followed at monthly intervals for 9–12 months. As one of the primary objectives of the study was to determine the feasibility of the site for a future microbicide study, opiate-positive participants were not enrolled due to the likely nonsexual route of HIV infection in opiate users. Risk behavior, medical history information, as well as blood and urine samples were collected at each visit as described above. As part of the data analysis plan, two interim analyses were conducted using incidence estimations from the BED assay to ascertain the possibility of the study site meeting the site selection criteria of 2% incidence (lower bound of the 90% incidence confidence interval). Due to the low incidence observed in the interim analysis, the prospective phase was closed out early such that all women did not complete all of their scheduled follow-up visits.

For the cross-sectional and prospective phases all HIV testing was conducted in the context of pre-and posttest counseling. Participants received their HIV test results at their follow-up visits; participants with positive pregnancy tests were contacted and asked to return to the clinic to receive their results.

False recent phase

Men and women with documented HIV-positive status of greater than 12 months who were not on ARVs were recruited from Outpatient Clinics (OPCs). Screening informed consent was obtained before accessing medical records to verify the date of the first reported HIV-positive test result. Only participants with a documented HIV-positive test result more than 12 months prior to their visit date were eligible for this phase of the study. The dates and results of their most recent CD4 and viral load tests were also collected from the medical record. Eligibility criteria for this phase was male or female age 18–35 with known HIV status of greater than 12 months and no previous history of ART. BED FR participants completed one study visit in which demographic and behavioral information, medical history, as well as two 10-ml blood samples (EDTA and serum) were collected. Urine samples were not collected in the BED FR phase.

Laboratory methods

HIV diagnostic tests

The HIV testing algorithm approved by the Vietnam MOH consists of two HIV EIA assays followed by one rapid HIV test. Plasma samples from the cross-sectional and prospective phases were tested using this three-step algorithm, which included the following HIV tests: (1) MUREX Ag/Ab COMBINATION EIA, (2) Genscreen HIV 1/2 Enzyme Immunoassay version 2, and (3) Abbott Determine HIV-1/2 Rapid Test. All three diagnostic tests must be positive before a positive diagnosis can be returned to the participant. Note that the standard algorithm includes Murex 3rd generation EIA, Genscreen, and Determine tests run sequentially; we modified the algorithm and substituted the Murex 4th generation assay for the 3rd generation assay. All tests were run according to the manufacturer's kit instructions.

SMARTube

The SMARTube (SMARTbiotech) is not a diagnostic test, but rather is intended to be used as a pretreatment for blood samples to stimulate antibody production in vitro; treated samples are then to be tested using currently licensed HIV antibody-based tests. In this study, a 2-ml whole blood sample (sodium heparin tube) was collected from all cross-sectional participants and at each visit of the prospective participants. One milliliter of whole blood was added to the SMARTube and the sample incubated at 37°C with 5% CO₂ for 3.5 days as per the manufacturer's kit instructions. The samples were then centrifuged and the supernatant (S-plasma) transferred to a cryotube and stored at −70°C prior to use. S-plasma was applied to the Genscreen HIV 1/2 Enzyme Immunoassay version 2 and the Abbott Determine HIV-1/2 Rapid Test and the results compared with unincubated matching samples. Samples that scored positive on the diagnostic test after incubation in the SMARTube but negative prior to incubation were termed “potential acute” and were tested by PCR for the presence of HIV RNA using AMPLICOR HIV-1 Test, version 1.5 (Roche).

BED testing

Cryopreserved plasma samples from the cross-sectional and BED FR phase were tested using the BED-CEIA (Calypte Biomedical Corporation) as per the manufacturer's instructions.²¹ Specimens with a final ODn reading of less than 0.8 were scored as recent infection.

AxSYM Avidity Index

Cryopreserved plasma samples from the BED FR phase and the cross-sectional phase were tested in the AxSym Avidity Index as described by Suligoi et al.^19,20 with the Abbott AxSYM HIV-1/2 gO kit. Specimens with an AI cutoff of ≤0.8 were scored as recent infection.

Urine testing

Urine samples collected from female participants in the cross-sectional and prospective phases were tested using the One Step Pregnancy Test Strip HgG Pregnancy test (Cortez Diagnostics). Samples from the cross-sectional phase of the study were screened for opiates using the MOP One Step Morphine Test Strip rapid test.

Statistical analysis

The HIV incidence rate was calculated directly from the prospective phase, calculated as the number of HIV seroconversions divided by the woman-years of follow-up observation.

The unadjusted HIV incidence estimation by BED in the cross-sectional phase was computed using the formula provided in the assay's package insert as: \documentclass{aastex} \usepackage{amsbsy} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{bm} \usepackage{mathrsfs} \usepackage{pifont} \usepackage{stmaryrd} \usepackage{textcomp} \usepackage{portland, xspace} \usepackage{amsmath, amsxtra} \pagestyle{empty} \DeclareMathSizes {10} {9} {7} {6} \begin{document} \begin{align*}{\rm I} = (365/ w) \times N {\rm inc} / [N {\rm neg} + (365 / w) \times N {\rm inc} / 2]\end{align*}\end{document}

where w is the window period for detecting recent infections (assumed to equal 197 days, regardless of the clade, population, or sensitivity of the HIV screening test algorithm for detection of early seroconversion),²² Ninc is the number of incident (recent) infections identified by BED among participants with a positive rapid test, and Nneg is the number of HIV-negative individuals. In addition, the adjusted HIV incidence estimation using Hargrove adjustment¹⁷ and Welte adjustment¹⁸ were evaluated by accounting for the BED false recent rate. The BED false recent rate (FRR) from participants with established HIV infection was computed as the total number of BED recent participants divided by the total number of established HIV-infected participants. The 95% confidence intervals of prospective phase were computed using exact methods. Confidence intervals for the cross-sectional incidence rates were calculated based on the assumption that the number of incident infections follows a Poisson distribution except for Welte adjustment. The same approaches were applied to calculate the HIV incidence estimation by Avidity Index assay (e.g., cutoff of 0.8 and window period of 180 days and the locally derived FRR for the avidity index assay was used).

Results

Study enrollment

The study enrolled 1619 men and women in the cross-sectional phase. Retention in this phase of the study was high (99%); all but 10 participants returned for their follow-up visit to receive their HIV test results. HIV-negative women without evidence of opiate use (355) from the cross-sectional phase were then recruited into the prospective phase, which entailed 9–12 monthly follow-up visits. As a result of the early close out of this study phase none of the women was able to complete a full 12 months of follow-up. Among those women who had chance to complete the abbreviated study, 88% of them contributed at least 6 months follow-up data, and 95% of them completed at least 3 months. The FR phase enrolled 403 men and women from HIV OPC clinics who were known to have been diagnosed with HIV via a positive HIV test result of greater than 12 months before enrollment in this study.

Demographic features

The demographic profiles of participants in these cross-sectional and prospective phases differed little since the prospective participants were a subset of cross-sectional participants (Table 1). As anticipated, the demographics of the individuals with long-term HIV infection who were enrolled in the FR phase were significantly different from the cross-sectional and prospective phase participants. The goal of the FR phase was to enroll an approximately equal number of men and women, whereas women comprised 81% of those enrolled in the cross-sectional phase and 100% of those in the prospective phase. In both cross-sectional and prospective phases over 50% of the participants self-reported commercial sex work as their primary occupation and up to 83% of the cross-sectional participants reported having sex in exchange for payment within the past month. In contrast, less than 1% of the FR phase participants reported having engaged in sex work. FR participants were also more likely to be married and live with their families and less likely to live alone or with friends. Despite being more likely to report education beyond ninth grade than the cross-sectional and prospective phase participants, FR participants were also significantly more likely to report being unemployed and earning less than 500,000 dong a week, which is the equivalent of $24.

Table 1.

Participant Demographics

	Study phase
	Cross-sectional	Prospective	False recent (FR)
Enrolled population	1619	355	403
Age ^a
<22	340 (21)	75 (21.1)	11 (2.7)
22–24	305 (18.8)	70 (19.7)	42 (10.4)
25–31	691 (42.7)	136 (38.3)	252 (62.6)
>31	283 (17.5)	74 (20.8)	98 (24.3)
Gender ^a,b
Female	1313 (81.1)	355 (100)	194 (48)
Male	304 (18.8)	0 (0.00)	209 (52)
Education ^a
None	105 (6.5)	18 (5.1)	12 (3)
Grade 1–5	571 (35.3)	134 (37.8)	85 (21)
Grade 6–9	686 (42.4)	153 (43.1)	175 (43.6)
Grade 10–12	251 (15.5)	47 (13.2)	114 (28.2)
University/college	6 (0.3)	2 (0.9)	17 (4.2)
Employment ^a
No, unemployed	25 (1.5)	3 (0.8)	87 (21.5)
Yes, part-time	1022 (63.1)	249 (70.1)	139 (34.4)
Yes, full-time	567 (35)	103 (29)	177 (43.8)
Type of work ^a,c
Sex worker	893 (55.3)	198 (55.8)	1 (0.25)
Food service	362 (22.4)	117 (33)	23 (5.7)
Manual labor	154 (9.5)	5 (1.4)	166 (41.1)
Other	206 (12.8)	35 (9.9)	213 (53)
Income (weekly) ^d
<500,000^e	358 (22.1)	60 (17)	284 (70.5)
500,001–1,500,000	920 (56.8)	206 (58)	112 (27.7)
1,500,001–3,200,000	275 (17)	69 (19.4)	4 (1)
>3,200,000	66 (4)	20 (6)	3 (1)
Are you married ^a
No	1039 (64.2)	207 (58.3)	140 (34.6)
Yes	293 (18.1)	65 (18.3)	237 (58.9)
Other	287 (17.7)	83 (23.4)	26 (6.4)
Whom you live with ^a,f
Alone	239 (14.8)	47 (13.2)	12 (3)
Spouse	57 (3.5)	11 (3.1)	74 (18.4)
Family/relatives/ children	834 (51.6)	179 (50.4)	304 (75.9)
Roommates/friend	266 (16.4)	76 (21.4)	5 (1.2)
Partner	218 (13.5)	40 (11.3)	6 (1.5)

Chi-square test of difference between XS and BED FR statistically significant at p<0.05.

Two transgender participants.

Four cross-sectional participants did not answer this question.

In Vietnamese dong.

This is the equivalent of $24 U.S. and below the poverty line, according to the Urban Poverty Assessment conducted by the People's Committee of Ha Noi and HCMC in 2009.

One cross-sectional participant did not answer this question.

HIV prevalence

The overall HIV seroprevalence in the cross-sectional phase was 15.8% (95% CI, 14.1, 17.6; Table 2). Prevalence ratios ranged from 14.9% in Binh Thanh to 17.2% in District 4. Men had significantly higher prevalence rates at all three sites, as did older participants (age ≥25), self-reported sex workers, and those reporting more than three sex partners in the past month (all p-values<0.05). Among opiate-negative female participants in the cross-sectional phase, HIV seroprevalence was 7.5% (95% CI, 6.2, 9.4) compared to the 42.3% (95% CI, 36.1, 48.6) prevalence observed in opiate-positive women. This finding was consistent with the finding that overall HIV prevalence was 5-fold higher among cross-sectional participants who were identified as opiate users either by testing or self-report (43.9%) than among those who were nonopiate users (8.6%) (Table 2).

Table 2.

HIV Prevalence in Cross-Sectional Phase

	Opiates tested OR self-reported ^a
	Positive N (%)	Negative N (%)	Total N (%)
All participants
HIV-tested population,^b N	330	1279	1609
HIV positive, N	145	110	255
Prevalence (95% CI^c)	43.9% (38.6, 49.3)	8.6% (7.1, 10.1)	15.8% (14.1, 17.6)
Female
HIV-tested population,^b N	241	1067	1308
HIV positive, N	102	83	185
Prevalence (95% CI^c)	42.3% (36.1, 48.6)	7.5% (6.2, 9.4)	14.1% (12.3, 16.0)
Male
HIV-tested population,^b N	89	212	301
HIV positive, N	43	27	70
Prevalence (95% CI^c)	48.3% (37.9, 58.7)	12.7% (8.2, 17.2)	23.3% (18.5, 28.0)

Opiate positives were identified by laboratory testing or self-reported opiate use.

Ten subjects without HIV test results.

Approximate confidence interval.

HIV incidence in cross-sectional/prospective phases and acute infection

In 89.5 women-years of follow-up, there were no seroconversions detected in the prospective phase; HIV incidence was zero per 100 PY (95% CI 0–4.1%). The estimated BED incidence rate for opiate-negative women in the cross-sectional phase was 0.5% per year (95% CI 0, 1.3 %) using the Welte formula,¹⁷ which was consistent with the 0.5% (0, 1.0%) using the Hargrove formula; both values were adjusted using the locally derived BED FRR of 1.7% (95% CI 0.7, 3.6%) in the incidence formula (see below). The unadjusted incidence rate among opiate-negative cross-sectional samples was 0.8% per year (95% CI 0.1, 1.5) using the BED assay. In the same opiate-negative samples, the adjusted estimated incidence using the AxSYM avidity index (AI) assay (AI FRR of 2.7%) was 1.7% per year (95% CI 0.4, 3.0) using the Welte formula and 1.7% per year (95% CI 0.6, 2.8%) using the Hargrove formula. The unadjusted incidence was 2.1% per year (95% CI 0.8, 3.4). All samples identified as recent infection were positive on Western blot.

Although the sample size is small, we observed higher incidence estimates in opiate users with both assays. The BED incidence estimate for opiate nonusers was consistently below 1%, whereas the BED estimate among opiate users was 7.7% (2.9, 12) (Fig. 1). This correlation of higher incidence rates among opiate users was also observed using data from the AxSYM avidity index assay (data not shown).

FIG. 1.

HIV incidence in (a) opiate-negative participants and (b) opiate-positive participants using the BED assay.

The study also employed three approaches to identify acute infection: (1) testing for p24 Ag (using a 4th generation Ag/Ab combo test followed by a 3rd generation HIV EIA assay algorithm (Genscreen), (2) analysis of Determine and Genscreen results from SMARTube incubated and unincubated samples, and (3) RNA detection, using pooled sample testing of HIV-negative samples.²³ The algorithmic approach identified 15 potential acute HIV infections that were positive on Murex combo assay and negative on both Genscreen and Determine. As some of the Murex values were near the positive cutoff, the samples were retested in the Murex assay; 6 of the 15 samples did not yield repeat positive results and none of the remaining 9 samples had detectable RNA and hence all were scored as negative (false positive Murex 4th generation test result).

We also evaluated the SMARTube and observed no decrease in sensitivity of either Genscreen or Determine using samples that had been incubated in the SMARTube; of the 229 HIV-positive samples incubated in SMARTube, all incubated samples tested positive on both diagnostic assays. Of the 1206 negative samples, five samples were identified as potential acute infection using the SMARTube approach coupled to the Genscreen diagnostic test. As the majority of these samples had low values near the cutoff, the samples were retested and four of the five samples retested as negative; the one positive sample was also negative by viral load test. Of note, the Genscreen test run using plasma generated a comparable number of false-positive tests. When the Determine rapid test was used as the downstream HIV diagnostic test, no potentially acute samples were identified; all negative plasma samples were also negative after incubation.

There was also no overlap of samples identified as potentially acute between the Murex Ab/Ag and the SMARTube methods; the results most likely indicate either initial low/borderline EIA OD values or false-positive test results for the EIA tests used. To further explore this finding, all negative and potential acute samples identified from the methods described above were tested in pools of 50 samples using the Roche Amplicor 1.5 test (cut off of 400 copies/ml); no positive pools were identified except for the control pool in which a spiked, HIV-positive plasma sample was added (data not shown). The lack of identification of acute infections by any of these methods is consistent with the low incidence estimates in the cross-sectional phase and the lack of seroconversions in the prospective phase.

False recent rates (FRR)

Seven (1.7%) of 403 participants known to be HIV positive for >12 months (range of first positive tests was 1–9 years with a median of 2.5 years) scored as recent in the BED assay. Of the seven participants, four were female and three were male, and two self-reported injection drug use in the past 12 months. All three men had CD4 counts less than 50, whereas all of the women who scored recent had CD4 >250. The range of CD4 counts in this population was 2–1156 cells per μl; 27 participants with CD4 counts below 50 did not score as recent. Using a cutoff of ≤0.8, 11 participants were scored as recent in the AI assay. The AI FRR was 2.7% (95% CI 1.4, 4.8) using plasma samples. Only one sample was identified as recent in both BED and AI assays.

Discussion

Our study provides an example of a direct comparison of the performance of two different HIV incidence assays and the correlation of the incidence results with directly observed incidence in a matched cohort. We found that the BED incidence estimate from the cross-sectional opiate-negative participants (0.5% per year, adjusted) was similar to that found in the prospective study (0% per 100 PY), and the 95% confidence intervals largely overlapped. Incidence estimates on the same samples using the AI assay yielded a higher adjusted incidence rate (1.7% per year) and FRR than the BED. This finding perhaps reflects the difference in subtype coverage for the two assays as the BED assay contains a multisubtype peptide that may provide better coverage than the AxSYM assay. The low incidence observed in both the cross-sectional and prospective phases correlate with the lack of acute infections detected in this population. The Murex 4th Ag/Ab and SMARTube approach for detecting acute infection both yielded some false-positive results; in the latter case the false positives were not observed when the Determine rapid test was used as the diagnostic test illustrating the differences in selection of test kits for identification of acute infection. As no acute infections were identified by screening for HIV RNA, additional investigation is needed to fully evaluate the performance of the SMARTube in a population with higher incidence and acute infection rates.

Our study also illustrates the importance of deriving and using a local FRR that applies to the study population to calibrate the incidence estimate. Current Office of the Global AIDS Coordinator (OGAC) guidance on the use of the BED assay specifies that locally derived correction factors should be used when calculating HIV incidence rates using the assay since FRRs may be impacted by ARV use and other factors such as virus subtype, population variation, and the duration and status of the epidemic.²⁴ While the sample size is small, our BED FRR of 1.7% fell within the general range of an independent study sponsored by CDC in HCMC,²⁵ and was lower than FRRs reported for similar studies in Africa (5–12%). As shown in Fig. 1a, had we used the higher FRRs reported in South African studies (5.57%) in place of the locally derived values we would have generated a negative incidence value and would have underestimated the incidence. We also compared the results from the simplified formulas from Welte et al.¹⁸ to Hargrove et al.¹⁷ and obtained similar results. To improve consistency in methodologies between study sites, guidance documents from OGAC and the WHO recommend using the Welte formula. The AxSYM FRR was higher than that found for the BED assay, and the two assays, with one exception, identified different samples as false recent, suggesting possible use of both assays in a RITA. In our study, we screened out participants who were either on ARVs or who had AIDS (based upon self-report, clinical presentation, and/or medical records) from both the cross-sectional and BED FR phases, which may in part account for the lower rate than that observed in studies that included this population. While small, our dataset is also reflective of other studies that observed a correlation with low CD4 counts and false recent tests in the BED assay.

As topical microbicide studies are focused on preventing HIV acquisition via vaginal sex, our recruitment efforts for the cross-sectional were focused on high-risk women and our prospective cohort excluded female opiate users in whom a likely route of infection was via injection. We observed, however, a 5-fold higher HIV prevalence ratio in participants who tested positive for opiate use or self-reported injection drug use (44.6% vs. 8.7%). The estimated cross-sectional adjusted incidence among opiate users was also much higher than among nonusers (9.7% vs.0.7% per year). These findings support the current regional data indicating that transmission is occurring predominantly through IDU. While a greater percentage of male participants were IDU (29.6%), we found a substantial proportion of our female participants also either tested positive for opiate use or self-report IDU (18.4%). The majority of female participants also reported either commercial sex work or received payment for sex (83.4%) and the HIV prevalence among the female sex workers (FSW) who reported IDU was significantly higher than among those who did not (42.6% vs. 7.9%). We believe that the intersection of IDU and female sex work is cause for concern as it represents a potential cross-over point for the infection to spread more broadly from the IDU population into the general population. Continued surveillance of and additional research among the FSW IDU population in Vietnam are warranted to better refine HIV incidence estimates in this population as this group may represent a suitable population for preexposure prophylaxis studies that use an orally available intervention.

This study has a number of limitations; as mentioned above, since recruitment chains were terminated after they were deemed to be not connected to any HIV-positive individuals, the observed sample became a convenience sample and the HIV prevalence could be overestimated for the study population. Nevertheless, the impact on HIV incidence estimation using BED and AI assays would be minimum (or insignificant) because the incidence estimation is solely dependent on HIV-positive participants. Most notably is the small number of women-years of observation in the prospective cohort due to low retention rate and early study close-out. The low retention rate contrasted sharply with the high (99%) follow-up and retention of participants in the cross-sectional phase. In the cross-sectional phase the follow-up period was 2 weeks or less and participants were motivated to receive their HIV test results. The primary reason given by prospective participants for early discontinuation from the lengthier prospective phase was that the time commitment was difficult to accommodate in participants' schedules. In addition, several participants stated that after having tested HIV negative in the cross-sectional phase they no longer felt the need to return for additional visits. Retraining of staff to stress the importance of returning for follow-up visits resulted in only a slight improvement in retention, indicating that any future studies in this population would require intensified approaches to ensure participant retention. Based upon the findings from the interim and final cross-sectional BED analyses that yielded estimated incidence rates below the threshold established for site selection, and the lack of seroconversions and acute infections, a decision was made by the sponsor to close the study before the full 12 months of follow-up was completed.

In conclusion, reliable and timely information on HIV incidence is essential to the selection of suitable research sites to conduct HIV prevention research. We compared cross-sectional incidence estimations with the directly observed incidence figures to evaluate the utility of using laboratory incidence methodologies as a surrogate measurement for longitudinally derived incidence. Our study demonstrates the potential but also the limitations and challenges for cross-sectional incidence methodologies in combination with other criteria to help guide the selection of future research site.

Footnotes

Acknowledgments

This work is made possible by the generous support of the American people through the U.S. Agency for International Development (USAID). The contents are the responsibility of FHI and do not necessarily reflect the views of USAID or the United States Government. Financial assistance was provided by USAID under the terms of Cooperative Agreement No. GPO-A-00-05-00022-00, the Contraceptive and Reproductive Health Technologies Research and Utilization (CRTU) Program. Additional funding was also provided by the Centers for Disease Control and Prevention under contract no. 200-20074-05314, Task Order #7. For support of the study, we thank the Provincial AIDS Committee, the People's Committee, the Medical Research Center, the Center for Preventative Medicine (CPM) laboratory, the FHI/Vietnam office, and the FHI/Bangkok office. We would also like to acknowledge the contributions of all of the following individuals to the successful completion of this study and the preparation of this manuscript: FHI Vietnam Country Director Steve Mills, the SIDI VN site teams, Nguyen Thi Xuan Hai, Duong Dinh Cong, Vo Thi Xuan Hanh, Nguyen Thi Ngoc Bich, Shelly Fischer, Motiur Rahman, Margaret Farrell-Ross, Sola Park, Thuy Nguyen Xuan, Dao Duc Giang, and all of the SIDI Study Participants from HCMC. We also thank David Sokal and Laneta Dorflinger and Tim Mastro for valuable input and Bharat Parekh and Tamar Jehuda-Cohen for helpful discussions.

Author Disclosure Statement

No competing financial interests exist.

References

Karim

, Karim

et al. on behalf of the CAPRISA 004 Trial Group. Effectiveness and safety of tenofovir gel, an antiretroviral microbicide, for the prevention of HIV infection in women. Science, 2010; 329:1168–1174.

Grant

, Lama

, Anderson

et al. Preexposure chemoprophylaxis for HIV prevention in men who have sex with men. N Engl J Med, 2010; 363:2587–2599.

Auvert

, Taljaard

, Lagarde

et al. Randomized, controlled intervention trial of male circumcision for reduction of HIV infection risk: The ANRS 1265 Trial. PLoS Med, 2005; 2:e298.

Bailey

, Moses

, Parker

et al. Male circumcision for HIV prevention in young men in Kisumu, Kenya: A randomised controlled trial. Lancet, 2007; 369:643–656.

Cohen

, Chen

, McCauley

et al. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med, 2011; 365:493–505.

Busch

, Pilcher

, Mastro

et al. Beyond detuning: 10 years of progress and new challenges in the development and application of assays for HIV incidence estimation. AIDS, 2010; 24:2763–2771.

Mastro

, Kim

, Hallett

et al. Estimating HIV incidence in populations using tests for recent infection: Issues, challenges and the way forward. J HIV/AIDS Surveillance Epidemiol, 2010; 2:7.

Hall

, Song

, Rhodes

et al. Estimation of HIV incidence in the United States. JAMA, 2008; 300:520–529.

Jiang

, Wang

, Ni

et al. HIV-1 incidence estimates using IgG-capture BED-enzyme immunoassay from surveillance sites of injection drug users in three cities of China. AIDS, 2007; 21,Suppl 8:S47–S51.

10.

Saphonn

, Parekh

, Dobbs

et al. Trends of HIV-1 seroincidence among HIV-1 sentinel surveillance groups in Cambodia, 1999–2002. J Acquir Immune Defic Syndr, 2005; 39:587–592.

11.

Wolday

, Meles

, Hailu

et al. Temporal trends in the incidence of HIV infection in antenatal clinic attendees in Addis Ababa, Ethiopia, 1995–2003. J Intern Med, 2007; 261:132–137.

12.

Rehle

, Shisana

, Pillay

et al. National HIV incidence measures—new insights into the South African epidemic. S Afr Med J, 2007; 97:194–199.

13.

Mermin

, Musinguzi

, Opio

et al. Risk factors for recent HIV infection in Uganda. JAMA, 2008; 300:540–549.

14.

Hladik

, Olara

, Mermin

et al. Effect of CD4(+) T cell count, antiretroviral treatment on two serological HIV incidence assays. AIDS Res Hum Retroviruses[Epub ahead of print]10.1089/aid.2010.0347.

15.

Marinda

, Hargrove

, Preiser

et al. Significantly diminished long-term specificity of the BED capture enzyme immunoassay among patients with HIV-1 with very low CD4 counts and those on antiretroviral therapy. J Acquir Immune Defic Syndr, 2010; 53:496–499.

16.

McDougal

, Parekh

, Peterson

et al. Comparison of HIV type 1 incidence observed during longitudinal follow-up with incidence estimated by cross-sectional analysis using the BED capture enzyme immunoassay. AIDS Res Hum Retroviruses, 2006; 22:945–952.

17.

Hargrove

, Humphrey

, Mutasa

et al. Improved HIV-1 incidence estimates using the BED capture enzyme immunoassay. AIDS, 2008; 22:511–518.

18.

Welte

, McWalter

, Barnighausen

. A simplified formula for inferring HIV incidence from cross-sectional surveys using a test for recent infection. AIDS Res Hum Retroviruses, 2009; 25:125–126.

19.

Suligoi

, Galli

, Massi

et al. Precision and accuracy of a procedure for detecting recent human immunodeficiency virus infections by calculating the antibody avidity index by an automated immunoassay-based method. J Clin Microbiol, 2002; 40:4015–4020.

20.

Suligoi

, Butto

, Galli

et al. Detection of recent HIV infections in African individuals infected by HIV-1 non-B subtypes using HIV antibody avidity. J Clin Virol, 2008; 41:288–292.

21.

Parekh

, Kennedy

, Dobbs

et al. Quantitative detection of increasing HIV type 1 antibodies after seroconversion: A simple assay for detecting recent HIV infection and estimating incidence. AIDS Res Hum Retroviruses, 2002; 18:295–307.

22.

Parekh

, Hanson

, Hargrove

et al. Determination of mean recency period for estimation of HIV type 1 incidence with the BED-capture EIA in persons infected with diverse subtypes. AIDS Res Hum Retroviruses, 2011; 27:265–273.

23.

Pilcher

, McPherson

, Leone

et al. Real-time, universal screening for acute HIV infection in a routine HIV counseling and testing population. JAMA, 2002; 288:216–221.

24.

Office of the Global AIDS Coordinator (OGAC): Update on HIV-1 incidence estimation and surveillance in resource-constrained settings using tests for recent infection—statement from the surveillance and survey and the laboratory technical working groups to the office of the global AIDS coordinator September 2010. OGAC: Washington DC, 2010.

25.

Tuan

, Le

, Shah

, Kim

. Validation study of HIV-1 incidence assays in Vietnam. The second Asia Regional HIV Surveillance Meeting, Lessons learned in second generation surveillance: the Asia experience, Ho Chi Minh City, Vietnam, 2010.