Treatment outcome and cost-effectiveness of different highly active antiretroviral therapy regimens in the UK (1996-2002).

The aim of this study was to estimate the outcome and cost-effectiveness per life-year-gained (LYG) of first-, second- and third-line non-nucleoside reverse transcriptase inhibitors (NNRTI) versus protease inhibitor (PI) containing highly active antiretroviral therapy regimens. Hospital care costs (2002 US dollars discounted 3.5% per annum) were linked to treatment failure times. Results show that the median time-to-treatment failure for first-line (nucleoside reverse transcriptase inhibitors) 2NRTIs + NNRTI was substantially longer than that for 2NRTIs + PI(boosted), 2NRTIs + PI and 2NRTIs + 2PIs, whereas for second- and third-line they were similar. Comparing first-line 2NRTIs + NNRTI with 2NRTIs + PI(boosted) cost per LYG was US$ 12,375; US$ 12,139 per LYG when compared with 2NRTIs + PI and US$ 2948 per LYG when compared with 2NRTIs + 2PIs. For second-line cost per LYG comparing 2NRTIs + NNRTI with 2NRTIs + PI(boosted) was US$ 19,501; US$ 18,364 per LYG when compared with 2NRTIs + PI and cost-saving when compared with 2NRTIs + 2PIs. For third-line cost per LYG comparing 2NRTIs + NNRTI with 2NRTIs + PI(boosted) was US$ 2708; US$ 11,559 per LYG when compared with 2NRTIs + PI and cost-saving when compared with 2NRTIs + 2PIs. In conclusion, first-line 2NRTIs + NNRTI was cost-effective or cost-saving when compared with PI-containing regimens for all lines of therapy. Such information is required by clinicians and managers of HIV services to make appropriate treatment decisions based on clinical and financial grounds, and given the increasing number of people living with HIV, such information will become more important over time.

An immune control model for viral replication in the CNS during presymptomatic HIV infection.

The brain is targeted by human immunodeficiency virus type 1 (HIV-1) during the course of untreated infection, leading to cognitive impairment, neurological damage and HIV encephalitis (HIVE). To study early dynamics of HIV entry into the brain, we examined a unique autopsy series of samples obtained from 15 untreated individuals who died in the presymptomatic stages of infection from non-HIV causes. HIV was detected and quantified by limiting dilution PCR and genetically characterized in the V3 region of env. Limiting dilution was shown to be essential for correct estimation of genetic partitioning between brain- and lymphoid-associated HIV populations. While no actively expressing HIV-infected cells were detected by immunohistochemistry, variable and generally extremely low levels of proviral DNA were detected in presymptomatic brain samples. V3 region sequences were frequently genetically distinct from lymphoid-associated HIV variants, with association index (AI) values similar to those observed in cases of HIVE. Infiltration of CD8 lymphocytes in the brain was strongly associated with expression of activation markers (MHCII; R = 0.619; P < 0.05), the presence of HIV-infected cells (proviral load; R = 0.608; P < 0.05) and genetic segregation of brain variants from populations in lymphoid tissue (AI value, R = -0.528; P approximately 0.05). CD8 lymphocytes may thus limit replication of HIV seeded into the brain in early stages of infection. Neurological complications in AIDS occur when this control breaks down, due to systemic immunosuppression from HIV that destroys CD8 lymphocyte function and/or through the evolution of more aggressive neuropathogenic variants.

Delayed diagnosis of HIV infection in ophthalmic practice.

The introduction of potent antiretroviral drug combinations has transformed human immunodeficiency virus/acquired immunodeficiency disease syndrome (HIV/AIDS) from an intractable and rapidly fatal disease to a chronic manageable illness with prolonged life survival for many patients. This paper discusses the ophthalmic and medical histories of two young female patients who presented with 'atypical' optic neuropathy and toxoplasma gondii retinochoroiditis and were later found to be HIV infected. We discuss the need for increased vigilance in the ophthalmic community for suspecting HIV infection to allow optimal management of the ophthalmic and systemic manifestations of the disease spectrum associated with HIV/AIDS.

Changes in the cause of death among HIV positive subjects across Europe: results from the EuroSIDA study.

OBJECTIVES: The causes of death among HIV-positive patients may have changed since the introduction of highly active antiretroviral therapy (HAART). We investigated these changes, patients who died without an AIDS diagnosis and factors relating to pre-AIDS deaths. METHODS: Analyses of 1826 deaths among EuroSIDA patients, an observational study of 8556 patients. Incidence rates of pre-AIDS deaths were compared to overall rates. Factors relating to pre-AIDS deaths were identified using Cox regression. RESULTS: Death rates declined from 15.6 to 2.7 per 100 person-years of follow-up (PYFU) between 1994 and 2001. Pre-AIDS incidence declined from 2.4 to 1.1 per 100 PYFU. The ratio of overall to pre-AIDS deaths peaked in 1996 at 8.4 and dropped to < 3 after 1998. The adjusted odds of dying following one AIDS defining event (ADE) increased yearly (odds ratio, 1.53; P < 0.001), conversely the odds of dying following three or more ADE decreased yearly (odds ratio, 0.79; P < 0.001). The proportion of deaths that followed an HIV-related disease decreased by 23% annually; in contrast there was a 32% yearly increase in the proportion of deaths due to known causes other than HIV-related or suicides. Injecting drug users (IDU) were significantly more likely to die before an ADE than homosexuals (relative hazard, 2.97; P < 0.0001) and patients from northern/eastern Europe (relative hazard, 2.01; P < 0.0001) were more likely to die pre-AIDS than southern patients. CONCLUSIONS: The proportion of pre-AIDS deaths increased from 1994 to 2001; however, the incidence of pre-AIDS deaths and deaths overall declined. IDU and subjects from northern/eastern Europe had an increased risk of pre-AIDS death. HIV-positive patients live longer therefore it is essential to continue to monitor all causes of mortality to identify changes.

Identification of shared populations of human immunodeficiency virus type 1 infecting microglia and tissue macrophages outside the central nervous system.

Infection of microglia and other cells of the macrophage/monocyte lineage in the central nervous system (CNS) by human immunodeficiency virus type I (HIV-1) underlies the development of giant cell encephalitis (GCE). It is currently unknown whether GCE depends on the emergence of virus populations specifically adapted to replicate in cells of the monocyte/macrophage lineage and whether this also leads to the specific targeting of macrophages in other nonlymphoid tissues. Autopsy samples from lymph node, brain (frontal region), lung, and full-thickness colon sections were obtained from nine study subjects with GCE and from nine without. The two groups showed no significant differences in CD4 counts, disease progression, or treatment history before death. Genetic relatedness between variants recovered from lymph node and nonlymphoid tissues was assessed by sequence comparison of V3 and p17(gag) regions using a newly developed method that scores the sample composition at successive nodes in a neighbor-joining tree. The association index enabled objective, numerical comparisons on the degree of tissue compartmentalization to be made. High proviral loads and p24 antigen expression in the brain were confined to the nine individuals with GCE. GCE was also associated with significantly higher proviral loads in colon samples (median of the GCE(+) group: 1,010 copies/10(6) cells; median of GCE(-) group, 10/10(6) cells; P = 0.006). In contrast, there were no significant differences in proviral load between the GCE(+) and GCE(-) groups in lymph node or lung samples, where HIV infection was manifested predominantly by infiltrates of lymphoid cells. V3 sequences from brain samples of individuals with GCE showed the greatest compartmentalization from those of lymph node, although samples from other tissues, particularly the colon, frequently contained variants phylogenetically related to those found in brain. The existence of shared, distinct populations of HIV specifically distributed in cells of the monocyte/macrophage lineage was further indicated by immunocytochemical detection of CD68(+), multinucleated giant cells expressing p24 antigen in samples of lung and colon in two individuals with GCE. This study provides the basis for future investigation of possible phenotypic similarities that underline the shared distributions of HIV variants infecting microglia and tissue macrophages outside the CNS.

Mismatched human leukocyte antigen alleles protect against heterosexual HIV transmission.

Genetic variation at the human leukocyte antigen (HLA) loci has been shown to be an important risk factor for progression to HIV disease, but its significance in infection is less well understood. We have investigated its role in HIV transmission in a cohort of individuals at risk for heterosexual infection. Analysis of over 80 individuals revealed that that the degree of concordance at HLA A, B, and DR loci differs significantly between transmitting and nontransmitting couples at risk for heterosexual HIV transmission (p <.02), suggesting that allogeneic immune responses may confer a degree of protection against HIV infection. Analysis of the frequencies of specific alleles at the A, B, and DR loci revealed a significantly higher frequency of HLA DR5 among exposed uninfected individuals, relative to population controls.

Using phylogenetic analysis to trace HIV-1 migration among western European injecting drug users seroconverting from 1984 to 1997.

OBJECTIVE: To reconstruct the epidemiological relationships of the HIV epidemics among injecting drug users (IDU) in western Europe. METHODS: HIV env V3 sequences of and epidemiological data were obtained from 145 IDU who seroconverted in three sequential periods: 1984-1988, 1989-1992 and 1993-1997. The sequences were phylogenetically analysed and examined for signature patterns characteristic of northern European IDU, including the conserved GGC codon in the V3 loop. RESULTS: Subpopulations of genetically related HIV strains were observed in Italy, France, Scotland and Spain, in contrast to the Netherlands, Austria and Switzerland. This difference between the two groups of countries suggests that the HIV epidemics amongst IDU in the latter group was caused by multiple virus introductions. In Edinburgh and the surrounding area, most IDU were infected with the same GGC strain over the 12-year study period. The epidemic among IDU in north-western Europe started with GGC viruses, whereas in south-western Europe non-GGC viruses predominated. This geographical separation has faded during the course of the epidemic, most likely because of virus exchange among IDU populations.

Historical HIV prevalence in Edinburgh Prison: a database-linkage study.

BACKGROUND: The prevalence of HIV in prisons is often higher than in the surrounding community, because prisons contain a high proportion of injecting drug users (IDUs). Reliable estimation of HIV prevalence in UK prisons only began in the 1990s. Edinburgh, Scotland, experienced a major IDU-related HIV epidemic which began in 1983. We sought retrospectively to estimate HIV prevalence in Edinburgh Prison over the period 1983-94. METHODS: Prison records of all 477 male HIV-positive patients (332 IDUs) in the Edinburgh City Hospital Cohort (believed to include three-quarters of HIV-positive Edinburgh IDUs) were abstracted from Edinburgh Prison. Using this information and the seroconversion intervals of the patients, the number of person-years spent inside the prison by these individuals while HIV-positive was estimated for each calendar month. From this, HIV prevalence was inferred. RESULTS: HIV prevalence in the prison rose from January 1983, as prevalence among Edinburgh IDUs increased, reaching a peak of 8% in December 1984. Prevalence during 1985-86 was 5-6% and then gradually declined, as the surviving HIV-infected IDUs spent less time in the prison. DISCUSSION: These figures are probably underestimates, as some HIV-positive prisoners are not in the cohort. However, the degree of underestimation should not be great and trends over time are reliable. Our estimate for August 1991, 4.1%, compares favourably with the estimate 4.5%, from an anonymous unlinked survey conducted in the prison that month. Prevalence estimates from other UK prisons are reviewed and suggestions made for other uses of database linkage in HIV and IDU epidemiology.

Cerebral infarction in adult AIDS patients: observations from the Edinburgh HIV Autopsy Cohort.

BACKGROUND AND PURPOSE: Autopsy series of patients with AIDS have found a 4% to 29% prevalence of cerebral infarction. Little is known of the prevalence of cerebral infarction when not associated with non-HIV central nervous system (CNS) infection, lymphoma, or cardioembolic sources. Clinical correlation has seldom been available. We describe the pathological and clinical features of patients from the Edinburgh HIV Cohort Study found to have had cerebral infarcts without evidence of non-HIV CNS infection, CNS lymphoma, or cardioembolic sources at autopsy. METHODS: From 183 autopsy cases, 26 without evidence of opportunistic cerebral infection or lymphoma were selected. These 26 cases went through a second selection process in which the presence of cerebral infarction, in the absence of the conditions mentioned, was verified. Histology and clinical records for the remaining patients were reviewed. RESULTS: Ten (5.5%) cases fulfilled the inclusion criteria and demonstrated similar hypoxic-ischemic lesions. Small-vessel thickening was seen in all cases, and perivascular space dilatation, rarefaction, and pigment deposition, with vessel wall mineralization and perivascular inflammatory cell infiltrates, were seen in some cases. Vasculitis was not found. One patient had had a transient ischemic attack, and no patient had had a stroke. CONCLUSIONS: Cerebral infarcts in HIV-infected patients are not common in the absence of cerebral non-HIV infection, lymphoma, or embolic sources. We found an HIV-associated vasculopathy with similar features in all risk groups. In AIDS patients presenting with stroke or transient ischemic attack, potentially treatable causes, such as cerebral coinfection or tumor, should be sought.

Infections with Mycobacterium tuberculosis and Mycobacterium avium among HIV-infected patients after the introduction of highly active antiretroviral therapy. EuroSIDA Study Group JD.

The impact of highly active antiretroviral therapy (HAART) among human immunodeficiency virus (HIV)-infected patients on the incidences of mycobacterial infections has not been studied in detail. We assessed incidences of mycobacterial diseases among HIV- infected patients following the introduction of HAART, using data from the EuroSIDA study, a European, multicenter observational cohort of more than 7,000 patients. Overall incidences of Mycobacterium tuberculosis (TB) and Mycobacterium avium complex (MAC) were 0.8 and 1.4 cases/100 person-years of follow-up (PYF), decreasing from 1.8 (TB) and 3.5 cases/100 PYF (MAC) before September 1995 to 0.3 and 0.2 cases/100 PYF after March 1997. After adjustment for changes in CD4 cell count and use of antiretroviral treatment in Cox proportional hazards models, the risk of MAC decreased with increasing calendar time (hazard ratio per calendar year; HR = 0.58 [95% confidence intervals: 0.45-0.74], whereas this was not the case for TB; 0.95 [0.74-1.22]). In conclusion, we documented marked decreases in the incidence of TB and to an even larger extent of MAC among HIV-infected patients from 1994 to 1999. The decrease in TB was associated with the introduction of HAART and changes in CD4 cell count. These factors could also explain some of the decrease in MAC over time, though there remained a significantly lower risk of MAC than expected.

Do gender differences in CD4 cell counts matter?

OBJECTIVE: To examine the effect of gender on disease progression and whether gender differences in CD4 lymphocyte counts persisted for the entire course from HIV seroconversion until (death from) AIDS. METHODS: CD4 lymphocyte counts were modelled in 221 female and 443 male seroconverters following seroconversion, backwards from AIDS and backwards from death using regression analysis for repeated measurements. RESULTS: In the period before use of highly active antiretroviral therapy (HAART), progression to AIDS and to death were marginally slower in women than in men as assessed by proportional hazards analysis. Women seroconverted for HIV, developed AIDS and died at higher CD4 cell counts than men (women: 815, 146 and 44 x 10(6) cells/l, respectively; men: 727, 49 and 22 x 10(6) cells/l, respectively), although differences were only statistically significant at AIDS onset. Declines in CD4 lymphocyte counts were not significantly affected by gender and absolute differences between men and women were stable, with exception for the trajectory close to AIDS when the decline became steeper for men than women. CONCLUSION: These gender differences in CD4 lymphocyte counts suggest a delay of initiation of therapy in women compared with men (our model predicted that women reach the threshold of starting HAART at about 12 months later than men). If this delay unfavourably influences progression, treatment guidelines should be revised so that women can benefit equally from HAART.

Mosaic structure of the human immunodeficiency virus type 1 genome infecting lymphoid cells and the brain: evidence for frequent in vivo recombination events in the evolution of regional populations.

In addition to immunodeficiency, human immunodeficiency virus type 1 (HIV-1) can cause cognitive impairment and dementia through direct infection of the brain. To investigate the adaptive process and timing of HIV-1 entry into the central nervous system, we carried out an extensive genetic characterization of variants amplified from different regions of the brain and determined their relatedness to those in lymphoid tissue. HIV-1 genomes infecting different regions of the brain of one study subject with HIV encephalitis (HIVE) had a mosaic structure, being assembled from different combinations of evolutionarily distinct lineages in p17(gag), pol, individual hypervariable regions of gp120 (V1/V2, V3, V4, and V5), and gp41/nef. Similar discordant phylogenetic relationships were observed between p17(gag) and V3 sequences of brain and lymphoid tissue from three other individuals with HIVE. The observation that different parts of the genome of HIV infecting a particular tissue can have different evolutionary histories necessarily limits the conclusions that can be drawn from previous studies of the compartmentalization of distinct HIV populations in different tissues, as these have been generally restricted to sequence comparisons of single subgenomic regions. The complexity of viral populations in the brain produced by recombination could provide a powerful adaptive mechanism for the spread of virus with new phenotypes, such as antiviral resistance or escape from cytotoxic T-cell recognition into existing tissue-adapted virus populations.

Effect of chemokine receptor mutations on heterosexual human immunodeficiency virus transmission.

To assess the effect of mutations at the CCR-2 and CCR-5 loci on heterosexual human immunodeficiency virus (HIV) transmission, 144 persons heterosexually exposed to HIV (infected and uninfected [EU]) and 57 HIV-positive index partners were genotyped. A significantly higher frequency of 64I heterozygotes at CCR-2 was observed in HIV-positive than in EU women (P=.02, relative risk=1.6). The allele frequency of 64I in women was 8% in HIV-positive contacts and 1% in EUs (P<.02). At CCR-5, no difference in the frequency of Delta32 was seen between groups, and the CCR-5 genotypes did not differ in accumulated "at-risk" exposure in EUs. Combining the analysis of the Delta32 and 64I mutations in index partners suggested an additive effect on transmission (P=.10). Thus heterozygosity for 64I at CCR-2 acts as a risk factor for HIV infection of women after heterosexual contact but heterozygosity for Delta32 at CCR-5 has no detectable effect.

Geographical variation in disease progression in HIV-1 seroconverted injecting drug users in Europe?

BACKGROUND: Human immunodeficiency virus (HIV) disease progression might vary by geographical region due to differences in the spectrum of HIV-related illnesses and (access to) health care. Therefore, the effect of geographical region, next to the effect of other potential cofactors, on disease progression in 664 injecting drug users (IDU) with documented HIV seroconversion from eight cohorts in Europe was studied. METHODS: Kaplan-Meier methods and Cox proportional hazards analysis were performed to assess the effect of geographical region, other sociodemographics, drug use and repeated HIV exposure on progression from HIV seroconversion to immunosuppression, AIDS and death with AIDS. We considered the confounding effect of study-design related factors (e.g. setting of follow-up), and accounted for pre-AIDS death from natural causes by imputing when each endpoint would have occurred, had they not died without AIDS. RESULTS: Estimates of progression to AIDS and death with AIDS were substantially faster after taking pre-AIDS mortality into account. Median incubation time from seroconversion to the first CD4 count < 200 cells/microliter was 7.7 years (95% CI: 7.1-8.3) and to AIDS 10.4 years (95% CI: 9.8-infinity). The 10-year survival was 70.3% (95% CI: 62.8-76.6). The relative hazards (RH) of AIDS for IDU from central and southern Europe compared with IDU from northern Europe was 1.9 (95% CI: 1.2-3.0) and 1.2 (95% CI: 0.6-2.3), respectively, before, and 1.5 (95% CI: 0.7-3.2) and 1.1 (95% CI: 0.6-2.3) after taking differences in study-design related factors into account. Accounting for these factors, the RH of death with AIDS was 0.9 (95% CI: 0.3-2.5) for central and 1.2 (95% CI: 0.4-3.4) for southern Europe compared with northern Europe. For the first CD4 count < 200 cells/microliter these figures were 0.8 (95% CI: 0.5-1.4) and 0.8 (95% CI: 0.5-1.4). Age at seroconversion was the strongest predictor of disease progression. No statistically significant differences in disease progression were found by gender, foreign nationality, drug use and potential repeated HIV exposure. CONCLUSIONS: We found no evidence for regional variability in HIV disease progression among European IDU. Future studies evaluating geographical differences should consider the confounding effect of study-design related factors and differential non-AIDS mortality. As age is an important determinant of disease progression, it should be considered in recommending treatment.

Spinal cord pathology and viral burden in homosexuals and drug users with AIDS.

Unless treated with effective antiretroviral therapy many AIDS patients develop a characteristic vacuolar myelopathy of the spinal cord associated with moderate clinical disability. Opinion is divided as to whether vacuolar myelopathy is causally linked to HIV myelitis. To investigate this further, spinal cord pathology was assessed in 41 drug users, 33 homosexual men and 16 other patients, all with AIDS. Previous work has shown that HIV encephalitis is more common in Edinburgh drug users than in homosexual men. In the present study HIV myelitis (10% overall) was more common in drug users (17%) than in homosexual men (3%) (P = 0.05), whereas the incidence of opportunistic infections (7% v. 9%) and lymphomas (2% v. 6%) was comparable in the two groups, but with a slight trend in the reverse direction, reflecting similar findings in the brain. However, moderate or severe vacuolar myelopathy was equally represented in both groups (20% of drug users and 21% of homosexual men). The HIV proviral load, assessed by polymerase chain reaction in frozen samples of thoracic spinal cord in 37 cases, correlated closely with the presence of giant cells and/or with immunocytochemical evidence of productive HIV infection. In 13 cases, the proviral load was measured in cervical, thoracic and lumbar samples and proved to be uniformly high or low in individual cases. This study provides no evidence for direct involvement of HIV, cytomegalovirus, papovavirus or human foamy virus in the pathogenesis of vacuolar myelopathy.

The impact of chronic hepatitis C virus infection on HIV disease and progression in intravenous drug users.

OBJECTIVE: It is unclear whether co-infection with hepatitis C virus (HCV) can influence HIV related morbidity or mortality, either by accelerating HIV-related disease progression, or by contributing to end stage liver disease. The aim of this study was to examine the effect of HCV infection on the severity and progression of HIV disease in a cohort of Edinburgh intravenous drug users (IDUs). METHODS: In 240 (47%) out of 508 patients in the Edinburgh IDU cohort both HIV seroconversion dates and anti-HCV serology were available. Demographic variables and HIV-related progression between anti-HCV positive and anti-HCV negative groups were compared. Parameters assessed included clinical endpoints (time of development of significant symptoms attributable to HIV (CDC stage IV), time of development of AIDS, and time of death) and immunological endpoints (time of CD4+ counts dropping below 200/mm3, 100/mm3 and 50/mm3). RESULTS: Two hundred and two out of 240 patients (84%) had positive anti-HCV serology. There was no significant difference in the frequency of clinical and immunological endpoints between the anti-HCV positive and negative groups. Progression analysis from HIV seroconversion to HIV related clinical endpoints indicated that anti-HCV serology was not a significant factor influencing the rate of HIV progression (relative risks (RR) for anti-HCV positive group: seroconversion to CDC IV, 1.01; seroconversion to AIDS, 1.05; seroconversion to death, 0.90). Likewise, HCV serostatus did not significantly affect progression to immunological endpoints (RR for anti-HCV positive group: seroconversion to CD4+ < 200/mm3, 1.04; seroconversion to CD4+ < 100/mm3, 1.13; seroconversion to CD4+ < 50/mm3, 0.97. Overall mortality from end stage liver failure was 4% in HCV-seropositive patients without AIDS. This suggests that HCV has had a clinically (though not statistically) significant impact on overall survival in this cohort. CONCLUSIONS: This study demonstrates that HCV co-infection does not influence the rate of progression to either clinical or immunological endpoints in our population of HIV-infected drug users. Further data are required to assess the effect of HIV on thge progression of HCV-related liver disease.

HIV encephalitis, proviral load and dementia in drug users and homosexuals with AIDS. Effect of neocortical involvement.

In this consecutive autopsy study, the pathological evidence of HIV encephalitis, which included the presence of giant cells and/or HIV p24 immunopositivity, was found more frequently in drug users (25 of 45; 56%) than in homosexual men (6 of 35; 17%) with AIDS (P < 0.01). Productive infection, as shown by HIV p24 positivity, was found in frontal lobe white matter in 29 of the 31 HIV encephalitis cases, but was also present in grey matter in 50% of the HIV encephalitis cases. Immunopositivity was confined to microglia, monocytes and most but not all giant cells. HIV-1 proviral load was determined by quantitative PCR in 65 of the 80 cases (separately in grey and white matter in 49 of these), and correlated well with the presence of HIV encephalitis (P < 0.001). Twenty-five patients with AIDS (13 male homosexuals, 12 drug users) showed no HIV encephalitis, opportunistic infection or cerebral lymphoma, while 18 (2 male homosexuals, 16 drug users) showed pure HIV encephalitis. Cognitive function had been assessed prospectively in this cohort and graded as normal or mildly, moderately or severely impaired. Because opportunistic infections and lymphomas of the brain may also lead to dementia, patients found to have these conditions at autopsy were excluded from the final analysis of the cases with dementia, so that the precise correlation between cognitive impairment and pure HIV encephalitis could be determined in this cohort without possible confounding variables. Fourteen of 18 patients with pure HIV encephalitis had shown cognitive impairment. Severe dementia correlated better with pure HIV encephalitis in cases in which grey matter involvement was present (7 out of 9) than in those in which only white matter was involved (2 out of 9) (P < 0.05), although milder degrees of cognitive impairment had been present in a further 5 HIV encephalitis cases. No correlation was found between zidovudine therapy and the degree of cognitive impairment. Systemic and cerebral opportunistic infections and lymphoma showed a negative association with HIV encephalitis, being more common in homosexuals than in drug users, despite comparable CD4 counts in the two groups. These findings suggest that neocortical productive HIV infection is a significant factor in AIDS-related dementia, although this may reflect merely a higher overall viral burden in the brain.