Changes to AIDS dementia complex in the era of highly active antiretroviral therapy

Objectives: To determine the protective efficacy of highly active antiretroviral therapy (HAART) against AIDS dementia complex (ADC) relative to other initial AIDS-defining illnesses (ADIs), Australian AIDS notification data over recent years were examined.

Methods: All initial ADIs in Australia over the period 1992-1997 were included. Three initial ADI groups were established: ADC; other predominantly central nervous system (CNS) ADIs (toxoplasmosis and cryptococcosis); and non-CNS ADIs. For each ADI grouping, the proportion of total ADIs, and median CD4 cell count in the pre-HAART era (1992-1995) were compared with the HAART era (1996 and 1997).

Results: Initial ADIs peaked in Australia in 1994 (n=1049), with a gradual decline to 1996 (n=722), and a marked decline in 1997 (n=367). ADC constituted 4.4% of initial ADIs over the period 1992-1995, but increased after the introduction of HAART to 6.0% in 1996 and 6.5% in 1997 (P=0.02). In contrast, the proportion of other CNS ADIs (1992-1995, 8.1%; 1996, 6.0%; 1997, 8.2%; P=0.41) was stable over the period 1992-1997. The median CD4 cell count at ADC diagnosis increased from 70/mm³ in 1992-1995 to 120/mm³ in 1996 and 170/mm³ in 1997 (P=0.04). Although the median CD4 cell count also increased significantly over this period for both other CNS ADIs (40-60/mm³; P=0.02), and non-CNS ADIs (60-70/mm³; P=0.02), the increase was small.

Conclusion: A proportional increase in ADC compared with other ADIs and a marked increase in the median CD4 cell count at ADC diagnosis have occurred since the introduction of HAART in Australia. These changes suggest that HAART has a lesser impact on ADC than on other ADIs, with the poor CNS penetration of many antiretroviral agents a possible explanation.

Introduction

The era of combination antiretroviral therapy has produced considerable delays in HIV-1 disease progression in developed countries, on the basis of evidence from clinical trials [^1-4], clinical care [^5-8], and more population-based settings [^9,10]. Several studies [^5,7,11-13] have now demonstrated a recent decline in the incidence of major HIV-1-related opportunistic infections, in particular, since the introduction of protease inhibitor-containing triple combination therapy, alternatively described as highly active antiretroviral therapy (HAART). Declines in incidence of between 40 and 80% have been reported for several individual HIV-1-related opportunistic infections, including Pneumocystis carinii pneumonia [^5,11,13,14], oesophageal candidiasis [^5,15], cytomegalovirus disease [^5,11,16], Mycobacterium avium complex infection [^5,11,13,17], and cryptosporidiosis [^11,18].

Although the impact of HAART on the major HIV-1-related opportunistic infections appears to be relatively uniform, the effect on the incidence of AIDS dementia complex (ADC) and HIV-1-related malignancies is less clear. The poor central nervous system (CNS) penetration of many antiretroviral agents, in particular protease inhibitors [¹⁹], could lessen the impact of HAART on HIV-1-related CNS disease, including ADC.

To determine the protective efficacy of HAART against ADC compared with other AIDS-defining illnesses (ADIs) we have analysed routine AIDS surveillance data in Australia over the period 1992-1997.

Methods

National surveillance for AIDS is coordinated by the National Centre in HIV Epidemiology and Clinical Research from statutory notification by doctors of AIDS diagnoses to the Department of Health in each state and territory [²⁰]. Notifications are made by primary care and hospital-based HIV clinicians. Information on AIDS cases includes: HIV-1 exposure category, date of HIV-1 diagnosis, date of AIDS diagnosis, initial ADI(s), and CD4 cell count at AIDS diagnosis. Information on ADIs occurring after AIDS diagnosis are generally not available, and were not included in the analysis.

In 1992, AIDS was diagnosed on the basis of the occurrence of an illness or condition included in the 1987 AIDS case definition provided by the United States Centers for Disease Control and Prevention (CDC) [²¹]. Since 1993, the Australian AIDS case definition has been applied [²²], which includes the three ADIs (extra-pulmonary tuberculosis, recurrent bacterial pneumonia, invasive cervical carcinoma) added to the 1993 CDC revised AIDS classification [²³], the only difference being that it does not include a CD4 cell count of less than 200/mm³.

Initial ADIs diagnosed over the period 1992-1997, and reported to the National AIDS Registry by July 1998, were examined. The numbers of ADC cases and the proportion of total initial ADIs were determined by year. For comparison with ADC, other ADIs were grouped into those predominantly involving the CNS (toxoplasmosis and cryptococcosis; other CNS ADIs), and non-CNS ADIs.

To assess the relative impact of HAART on ADC, other CNS ADIs, and non-CNS ADIs, 1992-1995 was considered as the reference group, with odds ratios for proportions of each ADI grouping determined for 1996 and 1997. The groups were chosen to reflect changes in antiretroviral therapy uptake in Australia. In the period 1992-1995, the vast majority of people with HIV infection were receiving either monotherapy or no antiretroviral therapy. Although the use of two-drug combination antiretroviral therapy commenced in 1995, less than 25% were receiving this therapy by the end of 1995 [²⁴]. During 1996, antiretroviral therapy uptake changed dramatically, with the majority of people with HIV receiving triple combination therapy by the end of the year [²⁴]. Median CD4 cell counts at diagnosis of ADC, CNS ADIs, and non-CNS ADIs were also determined for the periods 1992-1995, 1996 and 1997. To examine changes in the CD4 cell count distribution for each ADI group, a non-parametric Wilcoxon-type test for trend in median CD4 cell count was used [²⁵].

Results

The incidence of AIDS peaked in Australia at 946 in 1994, with a gradual decline to 643 by 1996, and a marked decline to 317 in 1997. Because some AIDS cases have more than one ADI at diagnosis, total ADIs are higher than AIDS cases, but these also peaked in 1994 (n=1049), and declined by approximately 50% between 1996 (n=722) and 1997 (n=367).

Over the period 1992-1997, 4.8% (234/4870) of initial ADIs were ADC, and 7.8% (379/4870) were other predominantly CNS ADIs (toxoplasmosis and cryptococcosis). Median age at ADC diagnosis was 40 years, higher than for other CNS ADIs (36 years) and non-CNS ADIs (37 years; P<0.0005). The vast majority of ADC cases were men (93%), and male homosexual/ bisexual contact was the reported HIV-1 exposure category in 81% of cases. There was no significant difference between ADC, other CNS ADIs and non-CNS ADIs with respect to sex and HIV-1 exposure category proportions.

The proportion of ADIs presenting as ADC was stable over the period 1992-1995, with an overall mean proportion of 4.4%, but increased to 6.0% in 1996 and 6.5% in 1997 (P=0.02) (see Table 1). In contrast, the proportion of other CNS ADIs was relatively stable at 8.1% for 1992-1995, 6.0% in 1996 and 8.2% in 1997.

Table 1

CD4 cell counts at AIDS diagnosis over the period 1992-1997 were available for 85% (200/234) of ADC cases, 90% (342/379) of other CNS ADIs, and 78% (3315/4257) of non-CNS ADIs. The median CD4 cell count at ADC was stable over the period 1992-1995 (1992, 80/mm³; 1993, 60/mm³; 1994, 50/mm³; 1995, 80/mm³) with a median of 70/mm³, but increased to 120/mm³ in 1996, and 170/mm³ in 1997 (see Table 1; P=0.04 for trend). The median CD4 cell count was also stable for other CNS ADIs over the period 1992-1995 (1992, 37/mm³; 1993, 42/mm³; 1994, 30/mm³; 1995, 40/mm³) with a median of 40/mm³, but increased to 48/mm³ in 1996, and 60/mm³ in 1997 (P=0.02). Although significant, the change in median CD4 cell count for non-CNS ADIs was only from 60/mm³ in 1992-1995 to 70/mm³ in both 1996 and 1997 (P=0.02).

Discussion

Although ADC cases have declined in Australia in the era of HAART, this decline has been less marked than for other ADIs. An accompanying large increase in the median CD4 cell count at diagnosis of ADC has occurred, compared with moderate increases for other ADIs. These findings provide preliminary evidence that HAART produces less protection against ADC than other ADIs.

In the interpretation of our findings, several methodological limitations need to be addressed. First, our analysis was only performed on initial ADIs, which represent approximately 40% of total ADIs [²⁶]. A relative shift of ADC from subsequent to initial ADI diagnosis could contribute to an apparently increased proportion of ADC as the initial ADI. This limitation would be overcome by a study examining all ADI events. Second, individual antiretroviral therapy data was not available on the National AIDS Registry. Although other studies indicate that the uptake of antiretroviral therapy has changed dramatically in Australia since 1995, with the use of triple combination or HAART therapy increasing from approximately 10% in early 1996 to 40% in late 1996 and 70% in late 1997 [²⁴], the absence of antiretroviral therapy data in this study makes an assessment of the ADC protective efficacy of individual therapeutic agents and different therapeutic regimes impossible. Finally, the misclassification of ADC diagnosis is a possible limitation, but would have to alter over time to effect an analysis of temporal incidence relative to other ADIs. AIDS notifications in Australia are made by specialist and primary care doctors, with the demographic concentration of the HIV epidemic contributing to a relatively small, well-established group of HIV practitioners. It is unlikely that the ADC diagnostic capacity of these practitioners has changed over recent years, and the ADC case definition has also not altered over the study period.

Before the use of HAART, the incidence of ADC appeared to be stable among people with advanced HIV-1 infection [^26-29]. In the Australian AIDS Cohort study [²⁶], the cumulative risk of ADC one year after the diagnosis of AIDS was 11% for people diagnosed with AIDS in 1988-1990 and 1991-1994. Since the introduction of HAART, a declining incidence of ADC has been documented in some studies [^5,11], but neither of these studies compared the decline in ADC with other ADIs, nor related the temporal changes in the incidence of ADC to changes in CD4 cell count distribution at ADC diagnosis.

The lower rate of decline for ADC compared with other ADIs since 1995 may be because HAART provides relatively less protection against ADC, as a result of the poor CNS penetration of several antiretroviral agents, differential sensitivities of peripheral and central HIV-1 viral populations, or a delayed impact of HAART on the risk of ADC compared with other ADIs.

CNS penetration of antiretroviral nucleoside analogues is variable. Zidovudine, stavudine and lamivudine have a relatively high level of CNS penetration, and well demonstrated suppressive effects on cerebrospinal fluid (CSF) HIV-1-RNA concentrations [³⁰], whereas HIV-1-protease inhibitors are highly protein bound and have poor CSF penetration [¹⁹]. As CSF HIV-1-RNA has been shown to correlate with the severity of cognitive impairment [³¹], antiretroviral therapy regimes that produce maximal CSF HIV-1-RNA suppression would be expected to provide the greatest protection against ADC, although this is not yet known [³²].

Recent studies have pointed to virologically compartmentalized peripheral and CNS HIV-1 evolution, with independently evolving quasispecies [³³], and differential HIV-1 macrophage tropism [³⁴]. These findings indicate that for maximal CNS efficacy antiretroviral therapy may require characteristics different to those which determine efficacy in the lymphoid system.

The efficacy of HAART in protecting against ADC could be relatively delayed in comparison with other ADIs. Two log reductions in plasma HIV-1 viral load and increases in CD4 cell counts occur in a large proportion of people within 2 months of the commencement of HAART [⁴]. In contrast, the effect of the introduction of HAART on CNS HIV-1 levels may be more protracted, and plasma HIV-1 suppression may be required for several months before the risk of ADC development is reduced.

The greater protection of HAART against other CNS ADIs compared with ADC may indicate different immunological control mechanisms for HIV-1-related CNS diseases. Although both HIV-related cryptococcosis and toxoplasmosis present with predominantly CNS manifestations [^35,36], peripheral T-lymphocyte function, rather than CNS factors may be the key determinant of risk of infection. In contrast, CNS factors such as CNS HIV-1 level would appear to be crucial risk determinants for ADC, but once CNS HIV-1 infection is well established the restoration of immune function may provide relatively less protective benefit than for other ADIs.

The duration of HIV-1 infection, independent of the level of immunosuppression, may be a further important risk factor for ADC. The previously documented occurrence of ADC in advanced HIV-1 disease, and the continued development of ADC despite CD4 cell count-measured improved immune function are both consistent with an independent effect of duration of HIV-1 infection.

Important markers of ADC risk before the commencement of HAART may be a very low CD4 cell count and high plasma HIV-1-RNA, both of which may indicate considerable CNS HIV-1 infection, with subsequent HAART-induced improvements in peripheral immune function having limited ability to suppress CNS HIV-1 replication. Factors such as CD4 cell count nadir, maximal plasma HIV-1 viral load, and the duration of HIV-1 infection need to be included in future neuroepidemiological studies. At a patient management level, clinicians need to be aware that ADC may present at considerably higher CD4 cell count levels in the HAART era.

Further studies are required to clarify these preliminary findings and examine other possible predictors of ADC. In our institutions, both a cohort study of people with HIV and a case-control analysis of people with ADC are underway as components of an enhanced neuroepidemiological research agenda.

Acknowledgements

The authors thank doctors who reported AIDS cases under national surveillance procedures, and the National HIV Surveillance Committee for their collaboration.