A novel serological testing algorithm for detecting recent HIV seroconversion has been developed that distinguishes recent from longer-standing HIV infections using a single serum sample . It is based on the slow antibody increase after HIV infection, which causes samples from recently infected individuals to be non-reactive in a less-sensitive HIV assay, while being positive in a sensitive assay. Its development is a major step forward for monitoring purposes, because it is a simple and low-cost method that can easily be optimized when coupled to existing systems, such as sentinel surveillance at sexually transmitted disease (STD) clinics [2,3].
We used this novel method to investigate trends in HIV incidence from 1991 to 2001 among homosexual men participating in voluntary unlinked HIV prevalence surveys at a large STD clinic in Amsterdam, the Netherlands. Local and international reports have recently demonstrated dramatically increasing rates of gonorrhoea and syphilis, as well as increases in HIV-related sexual risk behaviour among homosexual men [4,5]. As these rising trends might forecast a resurgent HIV epidemic, it is especially now, imperative to monitor HIV incidence in this population.
Materials and methods
The STD clinic of the Municipal Health Service in Amsterdam offers anonymous and free examination for STD. Over the past 5 years 15 000–20 000 new consultations have been provided annually at this outpatient clinic. Since 1991, except for 1993, anonymous and unlinked cross-sectional HIV prevalence surveys have been conducted each spring and autumn . Attendees are eligible for these surveys if they visit the clinic for evaluation of a possible new STD episode during a survey period. In each period, approximately 1000 consecutive clinic attendees are enrolled. Those who consent to participate are interviewed on sociodemographic and sexual characteristics and offered testing for HIV antibodies. All clinic attendees are routinely screened for gonorrhoea, syphilis and C. trachomatis.
For the present study, data were used only from male participants in the 1991–2001 HIV prevalence surveys, who identified themselves as being homosexual (approximately 15% of all participants) and who consented to blood testing for the presence of HIV (96.3% of all homosexual participants), a total of 3090 homosexual men. During the study period, 19 186 consultations were made at the STD clinic by homosexual men. Participants in the surveys did not substantially differ from non-participants with respect to age and nationality.
Conventional analysis of HIV antibodies was performed with a commercially available enzyme immunoassay (Axzym; Abbott Laboratories, North Chicago, IL, USA) and confirmed by Western blot analyses. After this testing, specimens were stored at −20°C. HIV-positive samples from which sufficient material was available for further testing (93.0% of the positive samples) were then tested with a less-sensitive HIV enzyme immunoassay (Organon Teknika Vironostika, Durham, NC, USA) to determine which individuals had been infected recently. Sera from seropositive individuals who were non-reactive on the less-sensitive assay were further tested for the presence of antiretroviral drugs (amprenavir, efavirenz, indinavir, lopinavir, nelfinavir, nevirapine, ritonavir and saquinavir) by a validated liquid chromatographic method with tandem mass spectrometric detection (LC/MS/MS; Sciex API 3000).
A positive culture was used as the diagnostic criterion for gonorrhoea (GC-Lect agar, BBL; Becton Dickinson, Cockeysville, USA). A diagnosis of early syphilis was based on clinical symptoms and a reactive serology. The Treponema pallidum haemagglutination assay (Fujirebo, Tokyo, Japan) was used for syphilis screening. When the assay was positive, the Venereal Disease Research Laboratory test (Venereal Disease Research Laboratory, Wellcome, Dartford, England) and the FTA-absorption test (Trepo-spot IF; Biomerieux, Marcy l'Etoile, France) were performed to confirm results. Patients previously treated for syphilis were considered to have a new syphilis infection only when they had a threefold or more increase in Venereal Disease Research Laboratory titre or when T. pallidum was demonstrated by dark-field microscopy. Cultures for herpes simplex virus and dark-field microscopic examination for T. pallidum were performed only when attendees presented with a genital ulcer on clinical examination. Infection with C. trachomatis was diagnosed on the finding of a positive ligase chain reaction (Abbott Laboratories). Non-specific urethritis or proctitis was diagnosed when more than 10 leukocytes per high-power microscopic field were seen in the absence of intracellulair Gram-negative diplococci.
Individuals who tested reactive with the sensitive HIV assay and non-reactive with the less-sensitive assay are considered to be in the period of early HIV infection [i.e. < 170 days, 95% confidence interval (CI) 162–183 days] , when antibody titre is increasing but has not yet peaked. However, as antibody concentrations may decrease considerably with the use of highly active antiretroviral therapy (HAART) [1,8], individuals in whom antiretroviral drugs were detected in the blood (see Laboratory methods) were considered to use HAART and were regarded as having longer-standing HIV infections. For the purposes of this study, only individuals who tested non-reactive on the less- sensitive assay and did not use HAART were classified as ‘recently infected'. Annual HIV incidence (I) was calculated as being the prevalence of individuals with recent infection (n) among the susceptible population [i.e. HIV-negative plus recently infected men (N)], divided by duration (T) of the mean time between seroconversion on the assays (here 170 days), thus I = (n/N)(365.25/T)(100) . To determine 95% CI for the incidences, we used the Bonferroni principle [1,7].
Material was insufficient for testing with the less-sensitive assay for 7.0% of HIV-infected participants overall: 25.7% participants in 1991, 21.2% in 1992, and 0.9% in 1994–2001. For purposes of comparison, annual HIV incidence was calculated with an adjustment for the proportion not available for testing (Table 1). The prevalence of recent infections was assumed to be similar among HIV-positive individuals who were not tested with the less-sensitive assay, and this adjusted number of recent infections was used to calculate the adjusted HIV incidence.
We used a logistic regression model to investigate trends in (unadjusted) incidence by entering the survey period number as a continuous variable in the model. The probability of recent infection among the susceptible population (n/N in the incidence formula) was the fitted value from the logistic regression. Furthermore, we examined whether trends differed by age groups, by adding an interaction term between calendar time and age. Information on age was regrouped according to the median age of the study population. Levels of statistical significance for time trends were based on the likelihood ratio test and confidence intervals for the odds ratios were based on the wald test. HIV incidence was compared with the rates of rectal gonorrhoea and recent syphilis diagnosed among all homosexual men visiting the STD clinic. Analyses were repeated by excluding 1991 and 1992, because these years had a high percentage of samples unavailable for testing with the less-sensitive assay.
From 1991 to 2001, 19 186 new consultations were made by homosexual men at the Amsterdam STD clinic. A total of 3090 of these homosexual men participated in the semi-annual HIV seroprevalence surveys. These men were predominantly of Dutch nationality (77.1%), and had a median age of 34 years (interquartile range 28–40). Over time, the median age increased from 32 years in 1991–1996 to 35 years in 1997–2001.
HIV prevalence and incidence
Of the 3090 participants, 2636 were HIV seronegative and 454 were seropositive. Sera from 422 of the seropositive men were tested with the less-sensitive assay, and 58 of these tested non-reactive. In 21 men, antiretroviral agents were detected (see Laboratory methods); these men were classified as having non-recent (longer-standing) infection together with the seropositive men who tested reactive on the less-sensitive assay, totalling 385 men with longer-standing infection. The remaining 37 individuals were determined as having recent HIV infection. Most (83.8%) of these recently infected men were unaware of their HIV infection.
The overall HIV prevalence was 14.7% (454/3090) and incidence was 3.0 infections/100 person-years (PY) (95% CI 1.8–4.6) (Table 1). Incidence increased over calendar time (P time-trend 0.024). The odds ratio for each subsequent survey period was 1.07 (95% CI 1.01–1.13). When we excluded the years 1991 and 1992 from analyses, because these years had a high proportion of samples unavailable for testing, the increasing time trend remained clearly present (P time-trend 0.019).
Strikingly, there appeared to be an interaction between calendar time and age on the risk of having a recent HIV infection, although the interaction did not reach statistical significance, P = 0.120. This model, which included both factors and their interaction (model fit P = 0.054), showed that the increase in incidence was strongly present among older (≥ 34 years) men [P time-trend 0.008, for each subsequent survey period the odds ratio was 1.13 (95% CI 1.02–1.24)], but not among younger men [P time-trend 0.482, for each subsequent survey period the odds ratio was 1.03 (95% CI 0.96–1.10)] (Fig. 1). Similar results were obtained when the years 1991 and 1992 were excluded from analyses.
Recent HIV infection in relation to other sexually transmitted diseases
Among the 37 men with recent HIV infection, 26 (70.3%) had a concurrent STD (see Methods). This proportion was 54.5% (6/11) in 1991–1996 and 76.9% (20/26) in 1997–2001. Of the 26, 11 (29.7%) had anogenital gonorrhoea, none had syphilis, 13 (35.1%) had anogenital chlamydia, two (5.4%) had anogenital herpesinfection, 10 (27.0%) had non-specific urethritis, and two (5.4%) had proctitis.
STD were diagnosed in 63.9% (246/385) men with longer-standing HIV infection and among 48.6% (1280/2636) HIV-negative men in the period 1991–2001.
Examining the total of 19 186 new consultations made by homosexual men at the STD clinic from 1991 to 2001, the number of diagnoses of rectal gonorrhoea and recent syphilis increased between 1998 and 2001 by more than threefold and 17-fold for each STD, respectively (Fig. 2).
Using a novel approach for detecting recent infections, we investigated HIV incidence from 1991 to 2001 among homosexual men attending an STD clinic in Amsterdam. We demonstrated that incidence is increasing in this group, a finding that agrees with earlier national and international reports of rising rates of sexual risk behaviour and STD among homosexual men [4,5,9].
Such STD as rectal gonorrhoea and syphilis are useful indicators of HIV-related sexual risk behaviour, and also facilitate HIV transmission and acquisition. It has been postulated that individuals who are in the early phase of their HIV infection contribute considerably to the spread of HIV because HIV-RNA levels, and thus infectiousness, are very high in this phase [10,11]. Our study adds to this hypothesis by showing that a large proportion of men with a recent HIV infection had a co-existing STD, and that men with both conditions reported large numbers of recent sexual partners; factors likely to contribute to the risk of transmission of HIV and other STD.
In contrast to the high and increasing incidence found among STD clinic attendees, the incidence of HIV among homosexual men who participated in the Amsterdam Cohort Study was much lower. From 1997 to 2001, incidence among cohort participants younger than 35 years was more or less stable at 1.1/100 PY (95% CI 0.6–1.8) (unpublished data), whereas in the same time period incidence was 3.6/100 PY (95% CI 1.5–7.4) among young men attending the STD clinic. Such variation illustrates the importance of using different sub-populations when monitoring HIV in a certain population. STD clinic visitors may represent men who more often engage in risky sexual behaviour than cohort participants, and may not necessarily represent the ‘general population'. However, any trends detected in such a high-risk group could be the first indicators of a general change. For this reason, STD clinics are important sites for HIV surveillance. As soon as a valid less-sensitive assay becomes easily available, it should be implemented in the routine of HIV screening, thereby facilitating prevention measures such as partner notification, determining timing of anti-HIV treatment, for identifying patterns and locations of HIV transmission, and investigating the occurrence of treatment resistance.
However, the novel method to detect new HIV infections has two shortcomings that may affect incidence estimates [1,7,8]. First, the less-sensitive assay has unknown validity for non-B subtypes, hampering its immediate use in populations in which these types are frequently found, such as heterosexuals and drug users. However, this problem is unlikely to have affected the HIV incidence estimates in our study, because homosexual men nearly always present with subtype-B HIV infection (E. op de Coul, personal communication, 2000). Second, in someone using HAART or diagnosed with AIDS, HIV antibody titres can be reduced. All individuals in whom antiretroviral drugs were detected, indicative of HAART use, were thus classified as not recently infected. However, some of these individuals may well have been infected recently, and we may thus have slightly underestimated the incidence in recent years, when HAART became widely used. Of individuals with an AIDS diagnosis who test non-reactive on the less-sensitive assay, 5% would not be truly recently infected . Such misclassification may have resulted in a slight overestimation of the incidence in the early years of our study. It has not affected the estimates in recent years, when HAART became widely used, because all individuals who used HAART (including those diagnosed with AIDS) were classified as having longer-standing HIV infection. Despite these uncertainties, it remains likely that HIV incidence is increasing among homosexual STD clinic attendees in Amsterdam.
Our study deals with a small number of outcomes and with findings for age and the interaction between calendar time and age that do not reach statistical significance. We thus have to be careful in interpreting results on the effect of age. Nevertheless, an interesting assumption might be that incidence is increasing especially in older men, and that contrary to previous years, when new infections largely occurred in younger men, incidence is now becoming highest among older men. Likewise, the increase in syphilis over time is only observed among older (P = 0.002) but not younger men (P = 0.983) participating in our study, as was observed among all homosexual attendees of the STD clinic in Amsterdam (unpublished data) and internationally .
The incidence of HIV is increasing among homosexual attendees of an Amsterdam STD clinic. Together with the rising rates of syphilis, gonorrhoea, and related risk behaviour among homosexual men, this finding calls for preventative action, especially for those who have recently been infected.
The authors would like to thank the personnel of the STD clinic for their contribution in data collection; the staff of the Public Health Laboratory of the Municipal Health Service for laboratory support, especially Paul Lakeman, Heidie Schreijer, Marja Pospiech and Judith Merks; Kristel Crommentuyn for laboratory HAART analyses; and Lucy Phillips for editing the final manuscript.
1. Janssen RS, Satten GA, Stramer SL. et al
. New testing strategy to detect early HIV infection for use in incidence estimates and for clinical and prevention purposes. JAMA 1998, 280: 42–48.
2. Gupta SW, Gill N, Graham C, Grant AD, Rogers PA, Murphy G. What a test for recent infection might reveal about HIV incidence in England and Wales. AIDS 2000, 14: 2597–2601.
3. Schwarcz S, Kellogg T, McFarland W. et al
. Differences in the temporal trends of HIV seroincidence and seroprevalence among sexually transmitted disease clinic patients, 1989–1998: application of the serologic testing algorithm for recent HIV seroconversion. Am J Epidemiol 2001, 153: 925–934.
4. Dukers NHTM, Goudsmit J, de Wit JBF, Prins M, Weverling GJ, Coutinho RA. Sexual risk behavior relates to the virological and immunological improvements among homosexual men in HIV infection. AIDS 2001, 15: 369–378.
5. Stolte IG, Dukers NHTM, de Wit JBF, Fennema JSH, Coutinho RA. Increase in sexually transmitted infections among homosexual men in Amsterdam in relation to HAART. Sex Transm Infect 2001, 77: 184–186.
6. Fennema JSA, van Ameijden EJC, Coutinho RA, van Doornum GJJ, Cairo I, van den Hoek JAR. HIV surveillance among sexually transmitted disease clinic attenders in Amsterdam, 1991–1996. AIDS 1998, 12: 931–938.
7. Centers for Disease Control. Guide to calculating HIV incidence using STARHS results from Organon Teknika Corporation (OTC) Vironostika Less Sensitive HIV EIA. Atlanta; April 2001.
8. Notermans DW, de Jong JJ, Goudsmit J. et al
. Potent antiretroviral therapy initiates normalization of hypergammaglobulinemia and a decline in HIV type 1-specific antibody responses. AIDS Res Hum Retroviruses 2001, 20: 1003–1008.
9. Stolte IG, Coutinho RA. Risk behaviour and sexually transmitted diseases are on the rise in gay men, but what is happening with HIV? Curr Opin Infect Dis 2002, 15: 37–41.
10. Yerly S, Vora S, Rizzardi P. et al
. Acute HIV infection: impact on the spread of HIV and transmission of drug resistance. AIDS 2001, 15: 2287–2292.
11. Pilcher CD, Shugars DC, Fiscus SA. et al
. HIV in body fluids during primary HIV infection: implication for pathogenesis, treatment and public health. AIDS 2001, 15: 837–845.
12. Centers for Disease Control and Prevention. Consultation on recent trends in STD and HIV morbidity and risk behaviors among MSM [Meeting report]. Atlanta, Centers for Disease Control and Prevention, 30–31 October 2000.