Pre-AIDS mortality and its association with HIV disease progression in haemophilic men, injecting drug users and homosexual men

Objective: To study pre-AIDS mortality and its association with HIV disease progression in different exposure groups with known intervals of HIV seroconversion.

Design and methods: The type and rate of pre-AIDS deaths were assessed in 111 HIV-infected haemophilic men followed in London, and 118 injecting drug users and 158 homosexual men followed in Amsterdam. In each group, the association between CD4+ T-cell count, HIV RNA and pre-AIDS mortality was studied using proportional hazards analysis.

Cited Here...: By 10 years after seroconversion 7.3% of the haemophilic men had died without AIDS and 38.2% had developed AIDS. These figures were 20.2 and 30.5% for injecting drug users, and 8.0 and 55.0% for homosexual men. The major causes of pre-AIDS mortality appear to differ in the three exposure groups. The risk of pre-AIDS death tended to increase with decreasing CD4 cell count and increasing HIV RNA levels in injecting drug users and homosexual men. In men with haemophilia the associations were less obvious, although the log-transformed CD4 cell count was predictive for pre-AIDS death.

Conclusions: Pre-AIDS deaths occur and are at least partially related to HIV disease progression irrespective of how individuals became infected. Because of the longer life expectancy due to highly active antiretroviral therapy (HAART), pre-AIDS deaths are likely to show a further increase. Methods to incorporate these intermediate outcomes should be considered in the estimation of the size of the HIV epidemic and in the survival analysis of HIV-infected individuals. Prevention and treatment of non-AIDS infections, especially hepatitis C virus infection, and cancers will become increasingly important in HIV-infected individuals. The interaction between these therapies and HAART should be closely monitored.

Introduction

A substantial proportion of HIV-infected injecting drug users (IDU), heterosexually infected patients from developing countries, and haemophilic men die without having satisfied the formal AIDS definition ^[1-4]. Without precise data on these pre-AIDS deaths, it is impossible to accurately estimate the size of the HIV epidemic from seroprevalence and AIDS surveillance data.

In both IDU and haemophilic patients, death rates from non-AIDS causes are higher among HIV-positive individuals than among HIV-negative individuals ^[1,4,5], suggesting an association between pre-AIDS mortality and HIV disease progression. A previous study among a large group of IDU demonstrated that pre-AIDS death rates from natural causes increased with ongoing time since HIV seroconversion and decreasing CD4 cell count ^[2]. However, to our knowledge no evidence is available on the relationship between pre-AIDS deaths and HIV disease progression from other HIV-infected exposure groups with a high likelihood of premature death.

Among those infected with HIV, men with haemophilia and IDU are almost universally co-infected with hepatitis C virus (HCV) ^[6,7]. Although most individuals with acute HCV infection are symptom-free, HCV infection becomes chronic in about 85% and may cause progressive liver disease, cirrhosis or liver cancer in approximately 20% of chronic patients within 20 years of infection ^[8].

Furthermore, it has been shown that HIV infection accelerates progression to hepatic decompensation ^[5]. Because most HIV-infected haemophilic patients became infected with HCV earlier in time than HIV-infected IDU, the impact of pre-AIDS deaths due to HCV infection is likely to be most notable in the former group. Moreover, as the recent introduction of highly active antiretroviral treatment (HAART) delays HIV disease, pre-AIDS deaths from HCV and other causes are likely to take on greater significance. As the latter is also true for risk groups previously shown to be at low risk of pre-AIDS mortality, we assessed the type and rate of pre-AIDS deaths in haemophilic men and IDU with well-estimated dates of HIV seroconversion, as well as in HIV-infected homosexual men in whom pre-AIDS death and HCV infection are uncommon ^[9-12]. We investigated whether the three exposure groups share common causes of pre-AIDS death and whether pre-AIDS mortality is associated with HIV disease progression in each of these groups.

Methods

Study population

The study population comprised patients with haemophilia registered at the Royal Free Hospital in London, and IDU and homosexual men participating in the Amsterdam Cohort Studies on HIV infection and AIDS. The Royal Free Hospital Haemophilia cohort consists of 111 men infected with HIV between 1979 and 1985. All had a sample that gave a positive HIV serology, and 63 men (56.8%) had an earlier sample which tested negative. The first documented HIV seroconversion in a haemophilic patient in the UK was in October 1979 ^[13], and in 48 men in whom no negative test result was available this date was used as the date of the last negative test. By April 1985, all clotting factor concentrates used in the UK were heat-treated and therefore it has been assumed that all seroconversions in the cohort would have taken place by this date. Therefore, in 15 patients with a positive test result after July 1985, the date of first positivity has been set at July 1985. From the Amsterdam cohorts all IDU (n = 118) and homosexual men (n = 158) with a documented seronegative and seropositive HIV test were eligible for this analysis. IDU were defined as individuals who, since 1979, had injected drugs before HIV seroconversion. Eighty-one IDU and 134 homosexual men underwent HIV seroconversion during follow-up in the cohort studies, which started in December 1985 and October 1984, respectively. The remaining 37 IDU and 24 homosexual men entered the cohort studies seropositive but had earlier seronegative blood samples. These samples were obtained for reasons unrelated to HIV disease progression, such as syphilis and hepatitis B testing or participation in a hepatitis B vaccine trial.

The design of these three cohorts has been described in detail elsewhere ^[14-16] and are largely comparable. Briefly, participants were seen every 3-6 months for clinical examination and laboratory review. Information on AIDS diagnosis and cause of death was obtained from review of clinical records or from local and national registers.

Laboratory methods

Sera were tested for antibodies to HIV with a commercial immunoassay and confirmed by immunoblotting. Lymphocyte subsets were determined by flow cytometry. Investigation for T-lymphocyte subsets started in November 1982 in the haemophilia cohort, October 1984 in the cohort of homosexual men and March 1989 in the drug users cohort and subsets were determined at each subsequent study visit.

In the haemophilia cohort, HIV RNA was measured using a reverse-transcriptase polymerase chain reaction assay (PCR; Roche Amplicor HIV-1 Monitor test: v1.0 plus add-in non-B primers, Roche Molecular Systems, Branchburg, New Jersey, USA) according to the manufacturer's instructions. The lower detection limit of this kit was 400 copies/ml. In the Amsterdam cohorts, HIV RNA was quantified by using a nucleic acid sequence-based amplification assay (NASBA HIV-1 RNA QT; Organon Teknika, Boxtel, The Netherlands) performed according to the manufacturer's instructions. The threshold of quantification was 1000 copies/ml. At both sites, HIV-RNA was retrospectively determined in stored serum samples, if available, taken at the first positive visit and at 1-year intervals until the end of the follow-up. Routine prospective testing of HIV RNA started at the end of 1997 in the haemophilic men, February 1997 in the IDU and July 1996 in the homosexual men.

Statistical methods

Product limit-estimates of the cumulative pre-AIDS mortality and AIDS incidence were calculated using a competing risks model based on influenced transition probabilities ^[17]. Follow-up time was calculated from the estimated time of seroconversion through pre-AIDS death, AIDS, loss to follow-up or censor date (i.e. 1 January 1998). The time of seroconversion was the midpoint between the last negative and first positive HIV test for haemophilic and homosexual men. For IDU, the cumulative HIV-seroincidence distribution was additionally taken into account when estimating the expected date of seroconversion. Those who were seropositive at enrolment but who had an earlier blood sample available to determine HIV seroconversion were included in the risk set at the date of study entry (i.e. left truncation). AIDS was diagnosed according to the definition effective in Europe at the time of diagnosis.

Cox proportional hazards analysis, allowing for late entry, was used to study the association between pre-AIDS death and disease progression. We studied the prognostic value of the CD4+ T-cell count, and HIV RNA on the risk of pre-AIDS death. In this analysis, these laboratory markers were treated as time-dependent covariates, and they were evaluated as continuous variables (raw values and logarithmic/square-root transformations) as well as after categorization. CD4 cell counts were categorized as ≥ 500, 200-499 and < 200 × 10⁶ cells/l. RNA levels were divided into three groups: ≤ 3.00, 3.01-5.00, > 5.00 log₁₀ copies/ml. If the interval between two consecutive marker measurements or between the last measurement and the date of end of the follow-up for pre-AIDS mortality exceeded 2 years, additional censoring took place 2 years after this measurement to limit misclassification due to temporal changes in markers. Individuals were re-entered into the risk set when their next measurement became available. Adjustment was made for age at seroconversion, gender (IDU only) and calendar period of follow-up. To correct, at the population level, for the release and use of nucleoside reverse transcriptase inhibitors and prophylaxis for Pneumocystis carinii pneumonia (PCP), the changing AIDS case definition, and the introduction of HAART, respectively, the time-dependent variable calendar period of follow-up was categorized as: before 1990 (reference period), 1991-1993, 1994-1995, 1996 and later. In the Cox analysis, categories of a variable were sometimes combined when no pre-AIDS deaths occurred in one of the different categories.

All analyses were conducted separately for each exposure group. As a previous study demonstrated that pre-AIDS mortality from natural causes but not from overdose and suicide was associated with HIV disease progression ^[2], in IDU we repeated our analysis separately for pre-AIDS deaths from natural causes and those from other causes. A P-value < 0.05 was considered statistically significant.

Results

General characteristics of the 111 haemophilic men, 118 IDU [70 men (59.3%) and 48 women] and 158 homosexual men are shown in Table 1. All haemophilic men seroconverted between 1979 and 1985 and therefore this cohort had a longer follow-up than both Amsterdam cohorts in which seroconversions continue to be observed. The haemophilic men seroconverted at a younger age than IDU. Homosexual men had the oldest age at seroconversion.

Table 1

By 1 January 1998, 57 haemophilic men, 27 IDU and 78 homosexual men had developed AIDS, and 60, 38 and 76 had died, respectively. Most haemophilic men, IDU and homosexual men with AIDS had an opportunistic infection, with PCP the most common infection among haemophilic men (35.1% of the AIDS cases) and homosexual men (26.9%), and oesophageal candida among IDU (44.4%). Kaposi's sarcoma was frequent among homosexual men, whereas it was not seen in the haemophilic men or IDU. Of those not known to have died, 76.5% of haemophilic men, 81.3% of IDU and 81.0% of homosexual men were seen in 1997. In this year, their median CD4 cell counts were 250 [interquartile range (IQR), 55-343], 320 (IQR, 202-428) and 404 × 10⁶ cells/l (IQR, 313-509), respectively. Of the indivdiuals seen in 1997, 23.1% of haemophilic men, 24.2% of IDU and 57.8% of homosexual men received antiretroviral therapy with protease inhibitors (intention to treat), 48.7, 19.7 and 6.3% were treated with non-protease inhibitors and 28.2, 56.1 and 35.9% did not receive antiretroviral treatment, respectively.

Deaths

At the end of the follow-up, there had been 48 deaths among haemophilic men with AIDS, 19 among IDU and 65 among homosexual men (Table 2). Among IDU and homosexual men all these deaths were attributable to AIDS, although one IDU died from AIDS with liver cirrhosis and another from a cause unknown. Among haemophilic men, most deaths were AIDS-related, but three were due to liver failure and one of the deaths attributable to AIDS occurred in a man with liver failure. The median value of the last CD4 cell count measurement in the year before death with AIDS was 10 (IQR, 0-40) for haemophilic men, 20 (IQR, 10-50) for IDU and 50 × 10⁶ cells/l (IQR, 10-100) for homosexual men. The median level of the last HIV RNA determination within 2 years preceding death was 4.89 (IQR, 4.41-5.38), 5.16 (IQR, 5.02-5.12) and 5.08 (IQR, 4.54-5.50) log₁₀ copies/ml serum, respectively.

Table 2

Twelve haemophilic men, 17 IDU and 11 homosexual men died without being diagnosed with AIDS. The majority of the haemophilic and homosexual men died from natural causes, whereas among IDU a substantial proportion died from (un)intentional injuries (overdose (n = 4), suicide (n = 2), accident (n = 1) and homicide (n = 1)). Among the 12 haemophilic men who died without AIDS, five died (41.7%) in liver failure. The most common cause of death among IDU who died from natural causes was a bacterial infection. Among homosexual men, non-AIDS cancer was the most common cause of pre-AIDS death; three cases of cancer were identified (melanoma, cancer of pancreas and cancer of bowel). Four of the remaining five homosexual men who died from natural causes, died with severe immunosuppression (last CD4 cell count prior to death < 40 × 10⁶ cells/l) of exhaustion (n = 3) or fever (n = 1). The median CD4 cell count within 1 year preceding pre-AIDS death was 345 (IQR, 60-840) for haemophilic men, 380 (IQR, 260-1060) for IDU and 30 × 10⁶ cells/l (IQR, 20-560) for homosexual men. The median HIV RNA measurement was 4.54 (IQR, 4.06-4.95), 4.68 (IQR, 3.91-5.15) and 4.49 (IQR, 3.93-5.20) log₁₀ copies/ml, respectively. IDU without AIDS who died from natural causes had a much lower last CD4 cell count and a higher RNA level than those dying from (un)intentional injuries (median CD4 cell count 190 and 430 × 10⁶ cells/l; median RNA level 4.99 and 4.36 log₁₀ copies/ml, respectively).

Comparing age at death between those dying without and with AIDS, the age distribution appeared to be comparable between the two groups among IDU (median age 36.1 and 38.2 years, respectively, P = 0.384 by Mann-Whitney U-test) and homosexual men (median age 40.9 and 41.5 years, respectively, P = 0.707). Remarkably, haemophilic men dying without AIDS were much older (median, 49.1 years; IQR, 37.4-58.6) than those dying with AIDS (median, 34.9 years; IQR, 30.0-43.3) (P = 0.05).

Cumulative incidence of pre-AIDS mortality and AIDS

Considering pre-AIDS mortality and AIDS as competing risks, by 10 years after seroconversion 7.3% of the haemophilic men had died without AIDS and 38.2% (100-61.8) had developed AIDS (Fig 1). These figures were 20.2 and 30.5% (100-69.5) for IDU, and 8.0 and 55.0% (100-45.0) for homosexual men. Among IDU, pre-AIDS mortality occurred from seroconversion onwards, whereas among men with haemophilia and homosexual men the risk of pre-AIDS death appeared to be very small in the first 5-8 years after seroconversion, but increased thereafter. Median progression time to pre-AIDS mortality or AIDS was 11.64 years for haemophilic men, 9.93 years for IDU and 8.34 years for homosexual men.

Fig. 1

The risk of pre-AIDS death was stable during the different calendar periods among homosexual men (P = 0.964). Among IDU, the risk of pre-AIDS mortality tended to slow down over calendar time, although the trend was non-significant (P = 0.146). Amongst haemophilic men, an increase in pre-AIDS mortality was suggested in the course of calendar periods, although effects were not significant; the crude relative hazard (RH) was 10.49 for the period 1996 and later [95% confidence interval (CI), 0.18-609.39), 3.08 for 1994-1995 (95% CI, 0.15-61.49) and 1.10 (95% CI, 0.09-12.81) for 1990-1993 relative to the period before 1990.

Association between laboratory markers and pre-AIDS mortality

T cell counts were available for all haemophilic men, 112 of 118 IDU (94.9%) and 154 of 158 homosexual men (97.5%). The median number of CD4 cell count measurements available before pre-AIDS death or censoring was 17 (IQR, 8-37), 10 (IQR, 5-19), and 28 (IQR, 12-41) respectively. HIV RNA levels were measured in 107 haemophilic men (96.4%), 90 IDU (76.3%) and 156 homosexual men (98.7%). The median number of determinations was 7 (IQR, 4-10), 5 (IQR, 2-8) and 12 (IQR, 7-17), respectively.

In univariate analysis, elevated risks for progression to pre-AIDS mortality were observed with lower CD4+ cell counts for IDU (considering deaths from natural causes) and homosexual men [relative hazard (RH) 2.65 and 2.22 per 100 × 10⁶ cells/l lower, respectively] (Table 3). After adjustment for age at seroconversion, gender (IDU only) and calendar period of follow-up the effect of the most recent CD4 cell count remained significantly associated with pre-AIDS mortality in IDU and homosexual men; a 100 × 10⁶ cells/l lower CD4 cell count was associated with an 164 and 121% increase, respectively, in the hazard of pre-AIDS mortality. Among homosexual men and IDU, results using either the log or square-root transformed CD4 cell counts were comparable to those using the untransformed CD4 cell count; therefore these results have not been shown. Among haemophilic men, we found a significant relationship with the log-transformed CD4 count (crude RH 2.55 per log₁₀ × 10⁶ cells/l lower; 95% CI, 1.48-4.40), which remained after adjusting for age at seroconversion and calendar period of follow-up (adjusted RH, 2.75; 95% CI, 1.45-5.24). In univariate analysis, the relative hazards for the other transformations went in the same direction as that for the log transformation but were not significant. There was no relationship between the untransformed CD4 cell counts and pre-AIDS mortality (Table 3).

Table 3

The association between level of HIV RNA and pre-AIDS mortality in men with haemophilia was of borderline statistical significance (P = 0.06) in univariate analysis (Table 3). Adjusting for age at seroconversion and calendar period of follow-up reduced the RH from 2.09 in the crude analysis to 1.55 for each 1 × log₁₀ copies/ml increase in HIV RNA level. In addition, we did not find an association with HIV RNA when modelled categorically. Among IDU and homosexual men, we found that the risk increased with increasing HIV RNA levels in both univariate and multivariate analysis, although the effect did not reach statistical significance (Table 3). When modelled categorically, the RHs went in the same direction.

Restricting the analyses to prospectively identified seroconverters with a time interval of less than 2 years between the date of the last negative and first positive test resulted in similar associations in the three exposure groups.

Discussion

In the present study, as expected we demonstrated that both HIV-infected IDU and haemophilic men experience pre-AIDS mortality. Because we studied a well-defined population of subjects with documented intervals of HIV seroconversion, we were able to assess pre-AIDS death rates for the entire course from HIV seroconversion onwards. Surprisingly, the incidence rate observed in homosexual men was remarkably similar to that seen in men with haemophilia.

In contrast to IDU where pre-AIDS mortality is present from seroconversion onwards, haemophilic men and homosexual men become at risk of pre-AIDS death at a much later stage in their infection. Amongst IDU, pre-AIDS mortality from overdose and suicide has been shown to occur at a more or less stable rate independently of the duration of HIV, whereas pre-AIDS mortality from natural causes has been shown to be positively associated with time since seroconversion and decreasing CD4 cell count ^[2]. Hence, this might explain why pre-AIDS mortality is found much earlier in infection in IDU than in haemophilic and homosexual men since almost all pre-AIDS deaths are attributable to natural causes in the latter two groups.

Although the numbers are small, the major causes of pre-AIDS mortality appear to differ in the three exposure groups. As reported previously ^[18], HCV has led to substantial mortality in the cohort of men with haemophilia. Because progression of HCV is hastened by co-infection of HIV ^[5] and the recent introduction of HAART has increased the life expectancy of HIV-infected individuals ^[19], mortality from HCV is likely to further increase in haemophilic men as a result of the relatively long latency period of HCV. As IDU became infected with HCV later in time than haemophilic men, mortality from HCV infection is also likely to further increase among IDU. Moreover, chronic alcohol use and hepatitis B infection might additionally play a role in liver failure in IDU. At present, overdose and suicide contributed to most pre-AIDS deaths in the IDU, and most pre-AIDS death from natural causes were attributable to bacterial infections. These infections, which are associated with the life-style of IDU (e.g., endocarditis, sepsis, a single episode of pneumonia), have not been included in the AIDS case definition, although these infections are clearly related to immunosuppression ^[20-26].

The unexpected finding of a substantial proportion of pre-AIDS deaths in homosexual men is in contrast with previous reports, which were based on shorter follow-up ^[9,10]. In homosexual men pre-AIDS deaths were mainly due to cancer and severe immunosuppression. Cancer was the leading cause of death before 1990 (when AIDS became the leading cause) among men aged 25-54 years in Amsterdam ^[27] and therefore can be considered as a competing risk. In addition, several studies show an excess of malignancies that are not AIDS-defining in HIV-infected individuals ^[28-30], although an association with disease progression has only consistently been described for Hodgkin's disease. In our study the CD4 cell count prior to pre-AIDS death was much lower in the homosexual men than in the other groups, whereas HIV RNA levels preceding death were comparable. The fact that these seriously immunosuppressed homosexuals did not develop AIDS might, therefore, be due to their moderate HIV-RNA levels. Another possibility is that AIDS-defining conditions close to death were missed because these men died at home. In contrast to the USA, European countries decided not to include a CD4 cell count < 200 × 10⁶ cells/l in the 1993 expansion of the AIDS case definition ^[31], partly because this would label a considerable group of asymptomatic individuals as having AIDS. Nonetheless, based on the results of the present study, the inclusion of a CD4 cell count < 50 × 10⁶ cells/l as indicator of AIDS should be considered for discussion.

Consistent with a previous study ^[2], pre-AIDS mortality from natural causes was significantly associated with decreasing CD4 cell counts in IDU. In addition, our data suggested that higher HIV RNA levels are also predictive for pre-AIDS mortality, although the relationship was not statistically significant partly due to small numbers. Moreover, in the short term (i.e. when fitted as time-dependent covariates) and with more advanced HIV infection CD4 cell count has been shown to be a stronger predictor of disease progression than HIV RNA level ^[32,33]. Similar associations between these two markers and the risk of pre-AIDS death were found in homosexual men. In haemophilic men, however, these associations seem to be less consistent as results are dependent on the transformation applied. This finding may indicate that liver-associated deaths, although higher in HCV co-infected individuals ^[5], are not associated with HIV disease progression itself. Unfortunately, numbers of specific causes of death were too small to investigate whether in haemophilic men the relationship between laboratory markers and pre-AIDS mortality is the same for liver-associated deaths and deaths resulting from other causes. In conflict with this hypothesis, increasing rates of liver-associated deaths were observed with decreasing CD4 cell counts in IDU ^[2], but as mentioned previously, because of their life style and shorter HCV exposure, other hepatotropic viruses and alcohol-induced hepatitis are likely to contribute to liver-associated deaths in this group. Further longitudinal studies with sufficient number of endpoints are needed to determine whether laboratory markers of HIV disease progression are also predictive for HCV liver-disease deaths in co-infected individuals.

We are aware that different HIV RNA kits were used and that they may give different test results. In addition, the assay used in the Amsterdam cohorts has a higher detection limit than the assay used in London. However, it is very unlikely that the use of different assays has seriously biased our results because at both sites HIV RNA was determined in sera, in which lower viral load levels have been shown relative to plasma using PCR, and analyses were carried out separately for each exposure group. Furthermore, we compared the median HIV RNA levels at several time points from seroconversion (data not shown), and results appeared to be more or less comparable across exposure groups. And although we cannot rule out differences across assays, HIV RNA levels yielded similar prognostic information for progression to AIDS in all exposure groups (data not shown).

We postulated that the risk of premature death might have taken on greater significance due to the introduction of HAART. However, since the risk of pre-AIDS death remained stable over time in the homosexual men, in whom HAART was mostly prescribed, the follow-up time might yet be too short to observe its effect on pre-AIDS death rates. The tendency of a temporal decrease in IDU is likely to be due to the inclusion of pulmonary tuberculosis and recurrent pneumonia, diseases that might have led to pre-AIDS deaths in IDU in the past ^[1,2], in the 1993 expansion of the AIDS case definition. Furthermore, the use of HAART might have delayed bacterial infections in this group. Since the haemophilic men became infected earlier in time than the IDU and homosexual men in the present study, many of these men had died in an era when there was little one could do to treat HIV. Conversely, haemophilic men might have been more compliant with therapy than IDU once prophylaxis became available and therefore this may have prevented some pre-AIDS deaths. The suggestion of an increase over time in haemophilic men could reflect their longer exposure to HCV, or may simply reflect a gradual ageing of the cohort.

There are a number of methodological implications of our results. Firstly, Kaplan-Meier and Cox proportional hazards models are commonly used to study progression to AIDS. Often, in such analyses, follow-up on patients who die without AIDS is right-censored on the date of death. These methods make the assumption that these individuals, had they not died, would have had the same prognosis as those remaining in the study. If the risk of pre-AIDS death is associated with more advanced disease, as in our study, then the hazard of AIDS will be underestimated and any estimates of relative hazards for AIDS will be biased if pre-AIDS mortality differs across covariate levels (e.g., exposure group). Alternative methods ^[17,34-37] should be further developed and more often considered. Secondly, those applying backcalculation methods to estimate the HIV incidence from seroprevalence and AIDS data should consider pre-AIDS deaths not only in IDU and haemophilic patients but also in other risk groups for which it was assumed that their risk of premature death was low and not disease related.

In conclusion, pre-AIDS deaths occur and are at least partially related to HIV disease progression irrespective of how individuals became infected with HIV. Because of the longer life expectancy due to HAART and longer exposure to hepatotropic viruses, the number of pre-AIDS deaths is likely to further increase in the coming years. Our findings suggest that therapy directed against hepatotropic viruses as well as the prevention and treatment of other infections and cancers that are not AIDS-defining will become increasingly important in HIV-infected individuals. The interaction between these therapeutic interventions and HAART should be closely monitored.

Acknowledgements

We thank the laboratories collaborating with the original studies for determining lymphocyte subsets (Professor G. Janossy and the Department of Immunology, Royal Free Hospital, London, and Professor F. Miedema and the Department of Clinical Viro-Immunology, Central Laboratory of the Netherlands Red Cross Transfusion Service, Amsterdam) and HIV RNA (Professor C. Loveday and the Department of Retrovirology, Royal Free Hospital, London, and Professor J. Goudsmit and the Department of Human Retrovirology, Academic Medical Centre, Amsterdam), the clinicians and health workers who contributed by collecting data, and the participants for their ongoing participation. We especially acknowledge B.H.B van Benthem and R.B. Geskus for their contributions to this study.