Autopsy findings in a human immunodeficiency virus-infected population over 2 decades: influences of gender, ethnicity, risk factors, and time. (Editorial Consortium).

Two decades of the human immunodeficiency virus (HIV) epidemic in America have seen significant shifts in patient demographics, with increasing percentages of women, Hispanic individuals, and blacks affected. (1,2) Over the same time period, development of antiviral compounds has resulted in the eras of no antiretroviral therapy, limited antiretroviral therapy, and highly active antiretroviral therapy (HAART). (3,4) The intersecting evolutions of demographic and therapeutic factors have changed the spectrum of HIV-related diseases, as has been noted in many cohorts. (5-9) Unfortunately, despite the triumphs of the recent therapeutic era, HIV-related mortality continues to be a significant problem in the United States and other countries. (10,11)

Causes of mortality are best determined by the autopsy, and studies of many patient populations have demonstrated the utility of postmortem analysis (22-14) Before the HAART era, reviews of HIV-associated autopsy pathology were undertaken to elucidate the frequencies of various diseases in affected populations. (15-19) Most of these populations comprised relatively homogeneous groups, with only 4 large studies having sufficient heterogeneity for limited comparisons of disease in different gender, ethnicity, or risk groups. (20-23) There have been only 2 recent reviews of HIV-related autopsy pathology in patients from the HAART era, and both examined limited numbers of women and minorities. (24,25)

We undertook a review of pathologic findings in a heterogeneous cohort of 394 HIV-infected adults autopsied at a major metropolitan medical center and its affiliates between the years of 1979 and 2000. The diversity of gender, ethnicity, and risk factors in these patients, seen within a single health care system over time, permitted analysis of the differences in disease frequencies and age at death associated with different therapeutic eras and demographic factors. This analysis allowed insight into the differences at death in a heterogeneous HIV-infected population over time.

METHODS

We identified adult (aged 18 years or older) HIV-infected patients through review of the autopsy service records at the Mount Sinai Medical Center and its affiliates, the Queens General Hospital and Elmhurst Hospital. For years in which HIV serologies were not available, patients were identified for study if they fit standard criteria of the Centers for Disease Control and Prevention for the diagnosis of acquired immunodeficiency syndrome (AIDS). (26) We evaluated complete records of the autopsy, including the premortem clinical synopses. A total of 394 autopsies were reviewed. When slides of nervous system tissues were available, one of the authors (S.M.) examined the slides to confirm correct neuropathologic diagnoses. A total of 281 autopsies had neuropathologic findings available for review.

The general information recorded included patient age at death, year of patient death, sex, ethnicity, and risk group for HIV infection. Each autopsy was assessed for significant involvement of the major organ systems: lungs, liver, spleen, kidney, endocrine organs, gastrointestinal tract, genitourinary tract, heart, lymph node/bone marrow, and central nervous system (CNS). We recorded the presence of the following infectious diseases: cytomegalovirus (CMV) infection, herpes simplex virus (HSV) infection, varicella-zoster virus (VZV) infection, hepatitis B and C, human papillomavirus (HPV) infection, other viral infections, all bacterial infections, staphylococcal and streptococcal infections (a subset of all bacterial infections), Mycobacterium avium-intracellulare (MAI) infection, Mycobacterium tuberculosis (MTB) infection, invasive candidiasis, histoplasmosis, cryptococcosis, aspergillosis, other fungal infections, Pneumocystis carinii pneumonia (PCP), cryptosporidiosis, toxoplasmosis, and worm (nematode/trematode/cestode) infection. Cryptosporidiosis was positively identified in only 1 autopsy and was not considered in subsequent analyses. We recorded the following noninfectious diseases and conditions: systemic lymphoma, Kaposi sarcoma, other tumors, cirrhosis, end-stage renal disease, arteriosclerosis, cachexia, and traumatic wounds or injuries. Arteriosclerosis was assessed by either gross or microscopic visualization of the major coronary arteries and aorta and was deemed present when either lipidized plaques or significant intimal thickening were apparent. Cachexia was assessed by internal and external examination of the cadaver and was deemed present when normal fat stores appeared markedly depleted (abdominal wall panniculus [less than or equal to] 1 cm). Traumatic wounds or injuries were identified in concert with the clinical history available at autopsy; these wounds consisted of either residua of gun shot or stab wounds or cranial trauma consisting of contusions or traumatic subdural hematomas. We also recorded the presence of infectious and neoplastic CNS entities that were not encompassed in the preceding categories: HIV encephalitis, aseptic leptomeningitis, and progressive multifocal leukoencephalopathy, and CNS lymphoma. Other CNS abnormalities were recorded under organ system involvement and will be reported in detail elsewhere.

Information was initially collected in spreadsheets (Excel, version 8.0, Microsoft Corporation, Redmond, Wash) and then transferred to a statistics program (Statview, SAS Institute, Inc, Cary, NC) for analysis on an iMac computer (Apple Computer, Inc, Cupertino, Calif). Statistics included multiple logistic regressions and analysis of variance (ANOVA) with Student-Newman-Keuls post hoc tests of significance, unpaired comparisons (t tests), and contingency table analyses. Regressions and ANOVAs were performed after exclusion of patients of Asian and unknown/other ethnicities (these patients were a small subset of the overall cohort [n = 7]) and with coding of patients with both intravenous drug use (IVDU) and sexual risk factors primarily as patients with IVDU risk. The criterion for significance in all analyses was held at the P = .05 level or better. Ages are reported as mean [+ or -] SEM.

RESULTS

Patient Characteristics

A total of 394 adult patients ranging in age from 19 to 83 years comprised this HIV-infected autopsy population (Table 1 and Figure 1). The mean age at death was 41.6 [+ or -] 0.54 years. Eighty-eight (22%) of the patients were autopsied between 1979 and 1986 (group A), 207 (53%) between 1987 and 1995 (group B), and 99 (25%) between 1996 and 2000 (group C). The mean age at death for group C was significantly older than that for groups A and B (groups A, B, and C, 40.7 [+ or -] 1.2, 40.8 [+ or -] 0.75, and 43.9 [+ or -] 0.97 years, respectively; ANOVA, P = .0069). This significant difference in age persisted after adjustment for the demographic and risk composition of group C.

[FIGURE 1 OMITTED]

Overall, there were 98 women (25%) and 296 men (75%), with increasing percentages of women over time (17%, 26%, and 29% in groups A, B, and C, respectively). The mean age at death for women (38.9 [+ or -] 1.0 years) was significantly younger than that for men (42.5 [+ or -] 0.64 years) for the overall population (t test, P = .0039). This difference remained significant in post hoc tests (difference by Student-Newman-Keuls analysis, 3.6 years) when risk factors for HIV infection, ethnicity, and therapeutic era were factored into the analysis by ANOVA. In every therapeutic era, women died at younger ages than men.

Overall, there were 157 black patients (40%), 131 Hispanic patients (33%), 99 white patients (25%), and 7 patients (2%) of other or unknown ethnicity. The ethnic composition of the sample changed significantly over time, with consistently decreasing proportions of whites and increasing proportions of Hispanic individuals in time periods B and C compared with time period A (percentage of blacks in groups A, B, and C, 41%, 37%, and 44%, respectively; Hispanic individuals, 19%, 37%, and 37%, respectively; and whites 38%, 24%, and 17%, respectively; contingency table analysis, P = .0315). The mean age at death was significantly older for whites than for blacks and Hispanic individuals (t tests, P = .0009 for differences between whites and blacks, P = .0222 for differences between whites and Hispanic individuals). This difference remained significant after adjustments for variations in gender, risk factors, and therapeutic era (ANOVA, P = .0139; differences by Student-Newman-Keuls post hoc analysis, 4.9 years between whites and blacks and 3.3 years between whites and Hispanic individuals). However, there were significant changes in the survival of blacks over time, and in period C, their age at death was greater than that for other groups, although statistically no different from that for whites (period C t test, P = .6698 for differences between blacks and whites).

The population included 142 patients (36%) with IVDU risk, 58 (15%) with heterosexual risk, .106 (27%) with homosexual or bisexual risk, 13 (3%) with IVDU and homosexual risks, 17 (4%) with a history of transfusion, and 58 (15%) with unknown risk factors for HIV infection. The risk composition changed significantly over time, with steadily increasing percentages of patients with IVDU and heterosexual risks and decreasing percentages of patients with homosexual/bisexual risk (percentage with IVDU risk in groups A, B, and C, 31%, 33%, and 47%, respectively; heterosexual risk, 9%, 15%, and 18%, respectively; homosexual/bisexual risk, 42%, 27%, and 13%, respectively; contingency table analysis, P = .0001). For the overall group, patients with homosexual/bisexual risk were significantly older at death than were patients with heterosexual or IVDU risk (t tests, P = .0188 for difference between patients with homosexual/bisexual and IVDU risks, P = .0115 for difference between patients with homosexual/bisexual and heterosexual risks). However, when gender, ethnicity, and therapeutic era were factored into the analysis, these risk-related differences in age at death were not significant, and the only patients with a significantly older age at death were those with unknown risk factors (ANOVA, P = .0004; difference by Student-Newman-Keuls post hoc analysis, 7.6 years between unknown and IVDU risk groups, 8.9 years between unknown and heterosexual risk groups, but no difference between unknown and homosexual/bisexual risk groups).

Frequencies of Infectious Diseases

Over the entire time period, bacterial infections, CMV infection, and PCP were the most common infections, affecting 41%, 29%, and 22% of the overall population (Table 2). However, the relative frequencies of infectious diseases changed dramatically over time. These changes, when analyzed without regard to demographic and risk variables (contingency table), were statistically significant. Infectious diseases that were found less frequently at autopsy in period C when compared with periods A and B were CMV infection (P = .0099), HSV infection (P = .0137), VZV infection (P = .0100), MAI infection (P = .0215), PCP (P < .0001) (Figure 2), candidiasis (P = .0437), cryptococcosis (P = .0378), and histoplasmosis (P = .0118). Infectious diseases that were found more often at autopsy in period C when compared with periods A and B were hepatitis C (P < .0001) and staphylococcal and streptococcal infections (P = .0260). When the changing demographic and risk composition of the patient sample was taken into account in the analysis, the only significant time-related change in infectious disease at death was a decrease in PCP in period C (logistic regression, P = .0014 for period C vs period A difference; adjusted odds ratio [AOR] of 0.2712 for occurrence of PCP in period C when compared with period A; 95% confidence interval [CI], 0.1217-0.6042).

[FIGURE 2 OMITTED]

Only a few infectious diseases showed associations with demographic variables by logistic regression (Table 3). Hepatitis B occurred more frequently in men (9.5%) than in women (3.1%), CMV infection was found more frequently in whites (34.3%) and Hispanic individuals (37.4%) than in blacks (18.5%), and MAI infection was found more often in blacks (11.5%) than in whites (6.1%). Cytomegalovirus infection was more common in patients in the homosexual/bisexual (38.7%), heterosexual (41.4%), and unknown (32.8%) risk groups than in patients in the IVDU risk group (17.4%). The distribution of HSV infection was similar to that of CMV infection: HSV infection was found in 8.5% of the homosexual/bisexual risk group, 13.8% of the heterosexual risk group, and 2.6% of the IVDU risk group. Pneumocystis carinii pneumonia was more common in the sexual risk groups (32.1% of patients with homosexual/bisexual risk, 27.6% of patients with heterosexual risk) and in the transfusion risk group (41.2%) than in the IVDU risk group (14.2%). Toxoplasmosis was more common in the transfusion risk group (17.6%) than in the IVDU risk group (2.6%), and MAI was more common in the homosexual/bisexual (10.4%) and transfusion (23.5%) risk groups than in the IVDU risk group (5.2%). The only infectious diseases showing greater prevalence in the IVDU risk group than in the sexual or other risk groups were hepatitis C (16.1% in the IVDU risk group, 0.94% in the homosexual/bisexual risk group, 5.2% in the unknown risk group) and MTB infection (13.5% in the IVDU risk group, 1.9% in the homosexual/bisexual risk group).

Frequencies of Noninfectious Diseases

In the overall population, cachexia and arteriosclerosis were the most frequent noninfectious diseases at autopsy (46% and 35% of the entire population, respectively) (Table 4). There were statistically significant changes in the time-related frequencies of all disorders examined in this category when the data were analyzed without regard to demographic variables (contingency table analysis). Disorders that were found less frequently at autopsy in period C when compared with periods A and B were systemic lymphoma (P = .0385), Kaposi sarcoma (P = .0009) (Figure 2), and cachexia (P = .0128). Disorders that were found more often at autopsy in period C when compared with periods A and B were cirrhosis (P = .0014) (Figure 2), arteriosclerosis (P < .0001) (Figure 2), and traumatic lesions (P = .0062). However, after adjustments for the changing demographic and risk composition of the patient population over time, the only disorders with a significant change in prevalence were Kaposi sarcoma, which had a lower prevalence in period C than in period A (AOR, 0.253; 95% CI, 0.066-0.971; P = .0452); cirrhosis, which had a greater prevalence in period C than in period A (AOR, 2.632; 95% CI, 1.059-6.536; P = .0371); and arteriosclerosis, which had a greater prevalence in period C than in period A (AOR, 4.545; 95% CI, 2.283-9.091; P < .0001).

Cachexia and arteriosclerosis were significantly more common in men than in women, before and after adjustments for other demographic, time, and risk factors. At autopsy, 34.7% of women and 49.5% of men were cachectic, and 21.4% of women and 39.2% of men had significant arteriosclerosis. Systemic lymphoma was more common in Hispanic individuals (6.9%) and whites (8.1%) than in blacks (1.9%), and cachexia was more common in blacks (50.6%) than in Hispanic individuals (40.5%). Kaposi sarcoma was more common in the homosexual/bisexual (22.6%) and unknown (17.2%) risk groups than in the IVDU risk group (3.9%), and conversely, cirrhosis was more common in the IVDU risk group (25.2%) than in the homosexual/bisexual (7.6%) and heterosexual (6.9%) risk groups.

Frequencies of Specific Neuropathologic Entities

Aseptic leptomeningitis (20%) was the most common finding in the 281 neuropathology autopsies reviewed, followed by HIV encephalitis (15%) (Table 5). There were no statistically significant demographic- or risk group-related differences in the prevalence of neuropathologic entities when the data were analyzed by multiple logistic regression. HIV encephalitis was equally common in the IVDU (14.2%), homosexual/bisexual (12.9%), heterosexual (20.9%), transfusion (10.0%), and unknown (13.2%) risk groups.

Organ Systems Involvement

The organ system most frequently involved at autopsy was the lung (89%), followed in frequency by the CNS (64%), liver (45%), and gastrointestinal tract (44%) (Table 6). Changes over time in the frequency of organ involvement were statistically significant for all organ systems when the data were analyzed without regard to the demographic composition of the patient population. The most striking changes were a decrease in pulmonary pathology in period C compared with periods A and B and an increase in CNS abnormalities in the same period. Significant pathology was present in the lungs in 97%, 92%, and 76% of autopsies in groups A, B, and C, respectively (P < .0001), and in the CNS in 59%, 58%, and 80% of autopsies, respectively (P = .0020). However, after adjustments for the demographic and risk composition of the population, the only significant change in organ system involvement was a decrease in pulmonary pathology in period C compared with period A (AOR of pulmonary pathology in period C, 0.189; 95% CI, 0.049-0.722; P = .0148).

After adjustment for time period, gender, and risk group, there were several ethnicity-related differences in organ systems pathologies. Blacks (45.9%) more frequently had renal pathology than did Hispanic individuals (29.0%) (AOR, 2.150; 95% CI, 1.281-3.610; P = .0038) or whites (25.3%) (AOR, 2.585; 95% CI, 1.395-4.790;P = .0025). Blacks also more frequently had splenic and cardiac pathology at autopsy than did whites (splenic pathology in 30.6% and 19.2%, respectively; AOR, 2.197; 95% CI, 1.116-4.324; P = .0228; cardiac pathology in 39.5% and 27.3%, respectively; AOR, 1.883; 95% CI, 1.031-3.440; P = .0395). Liver pathology was more common in the IVDU risk group (53.5%) than in the homosexual/bisexual risk group (34.9%) (AOR, 2.710; 95% CI, 1.479-4.965; P = .0013).

COMMENT

The autopsy has played an important role in elucidating manifestations of HIV during the past 2 decades. As the epidemic has shifted in its demographic composition, autopsy populations have mirrored its transformation. Whereas studies in the 1980s were largely confined to men and individuals with homosexual/bisexual risk, those in the 1990s contained larger numbers of women, minorities, and other risk groups. (13,15-25,27) However, until the current study, no single reported American autopsy population has so closely reflected the overall shifts in HIV epidemiology over time. The population described herein, with increasing percentages of women, decreasing percentages of whites, and increasing percentages of patients with IVDU and heterosexual risks over time, parallels in its composition the overall trend of AIDS in the United States and reflects this disease's evolution in arguably the most diverse city in the nation, New York. (1,2) Another factor that is important to the current analysis is the use of a cohort obtained from a single health care system, in which the standards of medical care at any point in time were uniform for all groups. Thus, although lifelong access to health care and socioeconomic variables may have differed in subgroups of this population, during hospital visits, our patients encountered similar treatment philosophies and regimens.

Prior studies of the demographic- and risk-related differences in disease frequencies that contribute to death in patients with AIDS have been from the pre-HAART era. (6,20-23,28) The largest, from the Community Programs for Clinical Research on AIDS (CPCRA), examined a group of 1883 patients dying between 1990 and 1994 and ascertained through clinical records the frequency of AIDS-defining events in the 6 months before death. (6) Significant gender, risk, and ethnic differences were present in this population. Bacterial pneumonia was more common in women than in men; Kaposi sarcoma was more common in whites than in blacks and Hispanic individuals and more common in homosexual individuals than in individuals in other risk groups; CMV infection was more common in whites than in blacks and less common in individuals with IVDU risk than in individuals with other risks; and PCP was less common in individuals with IVDU risk than in individuals with other risks. (6) These findings are confirmed by the present autopsy study. In the present study, CMV infection showed similar associations with sexual risk factors and white and Hispanic ethnicities; Kaposi sarcoma, with homosexual/bisexual risk; and PCP, with sexual risk factors. However, in the present study, bacterial infections were not more common in women.

Several other features of the CPCRA study bear comment. Although one of its great strengths was the number of patients analyzed, the absence of confirmatory autopsy data could be viewed as a potential weakness, as AIDS autopsy analyses have shown significant discrepancies between premortem and postmortem diagnoses. (13,29) The CPCRA study also noted that disease frequencies varied more often with geographic location than with demographics or risk factors. (6) This important observation assists in interpretation of the present study and prior autopsy studies described subsequently.

Only 4 of the many pre-HAART era autopsy studies have had sufficient numbers of heterogeneously distributed patients to provide limited analyses of the influences of demographic and risk factors on HIV-related mortality. (20-23) One study of 395 patients from the south of France was composed entirely of whites but had good representation of women (32%) and an assortment of risk factors (65% IVDU, 20% homosexual, 10% heterosexual, 5% transfusion). (20) The conclusions for this population were that toxoplasmosis was more common with IVDU risk, that Kaposi sarcoma was associated with homosexual risk, and that there were no gender-related differences in disease frequencies. A Texas population of 279 patients with a preponderance of men (95%) had a heterogeneous composition of ethnicities (55% white, 37% black, 9% Hispanic) and risks (30% homosexual/bisexual, 22% IVDU, 5% transfusion, 4% heterosexual, 38% other or unknown). (22) The conclusions in this population were that MTB infection, histoplasmosis, and MAI infection were more common in the IVDU risk group than in the homosexual risk group, and invasive candidiasis was more common in the homosexual/bisexual risk group than in the IVDU risk group. A third study was done in a Los Angeles population of 565 patients of unspecified ethnicity; 94% of the patients were men, and more than half had homosexual/bisexual risk (54.2% homosexual/bisexual, 23.4% IVDU, 3.8% transfusion, 3.8% heterosexual, 13.5% other or unknown risk). (21) The only association found in this study was an association of Kaposi sarcoma with homosexuality. Finally, a study was done in a group of 252 autopsies from the Bronx (New York City); the patients in this population were of unstipulated race, 25.8% were women, and IVDU was the most common risk factor for HIV infection (48.8% IVDU, 18.7% heterosexual, 13.9% homosexual, 4.4% combined IVDU and homosexual, 2% transfusion, 12.3% unknown risk factors). (23) This study found the highest prevalence of CMV infection in patients with sexual risk factors and a higher prevalence of aspergillosis in women than in men.

Compared with the prior studies, the present analysis benefits from a more distributed apportionment of demographic and risk variables in defining significant differences in disease frequency. It also examines a broader spectrum of mortality-associated pathologies and is not restricted to those classically considered secondary to immunosuppression. Some results highlight the variability of infectious diseases in autopsy studies from differing geographic areas; for example, in our population, gender-based differences were not seen in the prevalence of aspergillosis, in contrast to the findings in the Bronx population, (23) and MAI infection was more common in homosexual patients than in patients with IVDU risk, in contrast to the findings in the Texas population. (22) However, other results identify commonalities; for example, we found associations between Kaposi sarcoma and homosexual risk in concurrence with findings in the Los Angeles population (21) and between CMV infection and sexual risk factors in concurrence with findings in the Bronx population. (23) Additionally, in the current study, examination of conditions that are not classically associated with immunosuppression enabled detection of diseases that are not directly attributable to HIV, yet that contribute significantly to mortality. For example, the greater frequency of arteriosclerosis in men than in women likely reflects the greater biologic propensity of this condition to occur in young men. In another example, the associations of hepatitis C and cirrhosis with IVDU are likely to reflect the repetitive and sustained hepatic insults inherent in substance use.

The present study spans 3 therapeutic eras, with appreciable patient populations in each era. Thus, the analysis of diseases associated with mortality can factor in not only demographic variables, but also the effects of the therapeutic era. With the advent of HAART, the time to development of AIDS and the time to death after seroconversion have been prolonged. (30) Correspondingly, in the present study, a significant increase in the mean age at death was seen for the period corresponding to the availability of these therapies. Between periods A and C, men and women had mean increases in the age at death of 3.7 and 4.3 years, respectively. A similar increase in mean age at death was found in 1 of the 2 prior autopsy studies in patients treated during the HAART era; in that study, linear regression demonstrated an increase in the age at death between the years 1984 and 1998. (25)

As reported in the 2 prior studies, the current analysis found a general decrease over time in many opportunistic diseases at death. (24,25) However, prior studies did not account for changing demographic compositions in their patient samples. In the present study, many diseases varied in frequency between different demographic and risk groups. Therefore, analysis of time-related changes would have to account for these influences, and indeed, when these influences were considered, only PCP and Kaposi sarcoma had declined significantly as causes of mortality in group C. Reductions in the prevalence of PCP at autopsy were noted before HAART era studies took place, and in those time periods, the reductions were thought to reflect the effect of antimicrobial therapies on causes of death. (18,21,25) The further decline in PCP in the HAART era may be an indication that potent antiretroviral therapies have shown an additive effect in the elimination of this organism. Similarly, the reduction in autopsy-associated Kaposi sarcoma shown in the present study echoes prior reports of a decline in incidence of the disease with enhanced therapeutics. (5,8) Thus, Kaposi sarcoma and PCP are distinct from other opportunistic conditions in the HAART era in that a decreased incidence during the course of HIV infection has translated into a decreased frequency at autopsy. In the present study, other opportunistic conditions appear still to significantly contribute to AIDS-related mortality in patients who do not experience the benefits of effective medical regimens.

Although virologic failure and opportunistic disease still have important roles in HIV-related morbidity and mortality in the HAART era, another concern has been the metabolic side effects of combination antiretroviral therapies. (31) One such adverse effect is the development of hyperlipidemia and hypertriglyceridemia. (31) With regard to this effect, the increased frequency of arteriosclerosis in period C shown in this study may suggest that treatment-related lipid abnormalities are a significant and realized problem. Concerns have also been raised about potential HAART-associated hepatotoxicity, as either a direct effect of cytochrome P450 inhibition or via immune-mediated injury to the liver in patients with hepatitis. (32) In this study, the increased frequency of cirrhosis in period C is in keeping with these concerns.

Finally, one of the striking features of the current study was the persistent disparity in age at death between men and women. Women in all time periods had a younger age at death, with the most statistically significant difference occurring in the years 1996-2000. This difference could not be accounted for by differences in disease frequency, as women had a similar spectrum of infectious disorders and actually lower prevalences of hepatitis B, arteriosclerosis, and cachexia in comparison with men. It could also not be accounted for by risk group, as the difference persisted within IVDU and heterosexual risk groups alone. Why a gender-related age difference at death would persist through 2 decades is unclear, but it is intriguing in light of publications suggesting that the course of HIV disease may be influenced by gender. (7,33,34) In one study, women with the same viral load as men had a 1.6-fold higher risk of AIDS, and when stratified by viral load, clinical disease progression was more rapid in women than in men. (33) In the present study, we were unable to examine viral loads at death, as these values were unavailable for the majority of years spanned by the study. We were similarly unable to determine if the mean age of seroconversion differed between the sexes. Furthermore, other important factors related to health care access and compliance could not be addressed. The CPCRA studies that have documented comparatively poorer survival for women have suggested that differential health care access or socioeconomic status may play a role in these findings. (34) The continuing HAART era, gender-related disparity in age at death seen in this study underscores the need to further investigate the differential impacts of HIV infection in men and women.

In summary, the current study suggests that significant differences in the timing and causes of AIDS-associated mortality exist between different demographic groups. It also underscores the shift in death-related pathologies from opportunistic conditions to those that are unrelated to the direct effects of HIV.

Table 1. General characteristics of the Study Population *

                                    Overall              Group A
                                   (N = 394)             (n = 88)

Gender, No. (%)
 Women                              98 (25)             15 (17)
 Men                                296 (75)            73 (83)

Age at death, y
 Entire population             41.6 [+ or -] 0.54  40.7 [+ or -] 1.2
 Women                         38.9 [+ or -] 1.0   36.3 [+ or -] 2.8
 Men                           42.5 [+ or -] 0.64  41.6 [+ or -] 1.4

Ethnicity, No. (%)
 Black                              157 (40)            36 (14)
 Hispanic                           131 (33)            17 (19)
 White                               99 (25)            33 (38)
 Other/unknown                        7 (2)              2 (2)

Age at death, y
 Black                         39.9 [+ or -] 0.80  35.1 [+ or -] 1.3
 Hispanic                      41.3 [+ or -] 0.87  42.9 [+ or -] 3.4
 White                         44.7 [+ or -] 1.2   45.8 [+ or -] 2.0

Risk factor for HIV infection
 IVDU                               142 (36)            27 (31)
 Heterosexual                        58 (15)             8 (9)
 Homosexual/bisexual                106 (27)            37 (42)
 IVDU and homosexual                 13 (3)              3 (3)
 Transfusion                         17 (4)              1 (1)
 Unknown                            58 (15)             12 (14)

Age at death, y
 IVDU                          40.0 [+ or -] 0.65  34.5 [+ or -] 0.83
 Heterosexual                  38.2 [+ or -] 1.5   32.2 [+ or -] 4.3
 Homosexual/bisexual           42.7 [+ or -] 1.0   43.4 [+ or -] 1.6
 IVDU and homosexual           38.2 [+ or -] 2.2   36.3 [+ or -] 6.6
 Transfusion                   43.2 [+ or -] 3.8   50
 Unknown                       47.0 [+ or -] 1.8   52.2 [+ or -] 4.9

                                    Group B             Group C
                                   (n = 207)            (n = 99)

Gender, No. (%)
 Women                              54 (26)             29 (29)
 Men                                153 (74)            70 (71)

Age at death, y
 Entire population             40.8 [+ or -] 0.75  43.9 [+ or -] 0.97
 Women                         38.6 [+ or -] 1.4   40.6 [+ or -] 1.6
 Men                           41.6 [+ or -] 0.88  45.3 [+ or -] 1.2

Ethnicity, No. (%)
 Black                              77 (37)             44 (44)
 Hispanic                           77 (37)             37 (37)
 White                              49 (24)             17 (17)
 Other/unknown                       4 (2)               1 (2)

Age at death, y
 Black                         38.7 [+ or -] 1.0   46.1 [+ or -] 1.6
 Hispanic                      41.0 [+ or -] 1.1   41.1 [+ or -] 1.3
 White                         43.9 [+ or -] 2.0   44.9 [+ or -] 2.2

Risk factor for HIV infection
 IVDU                               68 (33)             47 (47)
 Heterosexual                       32 (15)             18 (18)
 Homosexual/bisexual                56 (27)             13 (13)
 IVDU and homosexual                 7 (3)               3 (3)
 Transfusion                         16 (8)              0
 Unknown                            14 (18)             18 (18)

Age at death, y
 IVDU                          39.3 [+ or -] 0.95  44.1 [+ or -] 1.0
 Heterosexual                  40.0 [+ or -] 2.1   37.8 [+ or -] 2.0
 Homosexual/bisexual           41.7 [+ or -] 1.5   45.3 [+ or -] 2.8
 IVDU and homosexual           36.3 [+ or -] 2.6   44.7 [+ or -] 2.0
 Transfusion                   42.8 [+ or -] 4.0   NP
 Unknown                       43.6 [+ or -] 2.1   48.7 [+ or -] 3.2

* Percentages are within group; ages are presented as mean [+ or -]
SEM. HIV indicates human immunodeficiency virus; IVDU, intravenous
drug user; and NP, no patients.
Table 2. Frequencies of Infectious Diseases *

                          Overall    Group A    Group B   Group C
Disease                  (N = 394)   (n = 88)  (n = 207)  (n = 99)

Viral infection
 CMV                     116 (29)    33 (37)   61 (29)    22 (22)
 HCV                      32 (8.1)    0         8 (3.9)   24 (24)
 HBV                      31 (7.9)    8 (9.1)  13 (6.3)   10 (10)
 HSV                      25 (6.3)    5 (5.7)  17 (8.2)    3 (3.0)
 HPV                       7 (1.8)    2 (2.3)  3 (1.4)     2 (2.0)
 VZV                       4 (1.0)    0        4 (1.9)     0
 Other virus               8 (2.0)    1 (1.1)  2 (0.97)    5 (5.1)

Bacterial infection
 All bacteria            163 (41)    28 (32)   94 (45)    41 (41)
 Staphylococcus/
  Streptococcus           35 (8.9)    7 (8.0)  17 (8.2)   11 (11)

Mycobacterial infection
 MTB                      37 (9.4)    5 (5.7)  26 (13)    6 (6.1)
 MAI                      34 (8.6)   11 (12)   17 (8.2)   6 (6.1)

Fungal infection
 PCP                      88 (22)    35 (40)   42 (20)    11 (11)
 Candida                  55 (14)    13 (15)   29 (14)    13 (13)
 Cryptococcus             31 (7.9)    7 (8.0)  18 (8.7)    6 (6.1)
 Aspergillus              25 (6.3)    3 (3.4)  17 (8.2)    5 (5.1)
 Histoplasmosis            3 (0.76)   2 (2.3)   1 (0.48)   0
 Other fungus              9 (2.3)    4 (4.5)   2 (0.97)   3 (3.0)

Parasitic infection
 Toxoplasmosis            18 (4.6)    3 (3.4)  11 (5.3)    4 (4.0)
 Worms                     3 (0.76)   0         3 (1.4)    0

* Values are expressed as No. (%) of patients. CMV indicates
cytomegalovirus; HCV, hepatitis C virus; HBV, hepatitis B virus;
HSV, herpes simplex virus; HPV, human papillomavirus; VZV,
varicella-zoster Virus; MTB, Mycobacterium tuberculosis;
MAI, Mycobacterium avium-intracellulare; and PCP, Pneumocystis
carinii pneumonia.
Table 3. Demographic Associations With Disease Frequency at Autopsy *

                                                 Adjusted
Disease            Comparison                    Odds Ratio

CMV                Hispanic vs black               2.809
                   White vs black                  3.039
                   Heterosexual vs IVDU            3.448
                   Homosexual/bisexual vs IVDU     2.439
                   Unknown vs IVDU                 2.155

HSV                Heterosexual vs IVDU            4.950
                   Homosexual/bisexual vs IVDU     6.410

HBV                Men vs women                    3.610

HCV                Homosexual/bisexual vs IVDU     0.092
                   Unknown vs IVDU                 0.224

MAI                White vs black                  0.299
                   Homosexual/bisexual vs IVDU     3.690
                   Transfusion vs IVDU             6.623

MTB                Homosexual/bisexual vs IVDU     0.122

PCP                Heterosexual vs IVDU            2.801
                   Homosexual/bisexual vs IVDU     2.433
                   Transfusion vs IVDU             3.597

Toxoplasmosis      Transfusion vs IVDU            15.385

Lymphoma           Hispanic vs black               6.173
                   White vs black                  5.988

Kaposi sarcoma     Homosexual/bisexual vs IVDU     4.237
                   Unknown vs IVDU                 5.291

Cirrhosis          Heterosexual vs IVDU            0.220
                   Homosexual/bisexual vs IVDU     0.232

Arteriosclerosis   Men vs women                    3.378

Cachexia           Men vs women                    1.965
                   Hispanic vs black               0.581

                   95% Confidence
Disease               Interval       P Value

CMV                 1.592-4.950       0.0004
                    1.000-3.636       0.0498
                    1.681-7.042       0.0007
                    1.250-4.762       0.0089
                    1.050-4.425       0.0362

HSV                 1.383-17.857      0.0140
                    1.567-26.316      0.0090

HBV                 1.019-12.821      0.0467

HCV                 0.011-0.745       0.0254
                    0.060-0.841       0.0266

MAI                 0.102-0.874       0.0273
                    1.222-11.111      0.0205
                    1.374-32.258      0.0185

MTB                 0.265-0.564       0.0071

PCP                 1.279-6.135       0.0101
                    1.171-5.051       0.0171
                    1.103-11.628      0.0337

Toxoplasmosis       2.525-90.909      0.0030

Lymphoma            1.284-29.412      0.0231
                    1.142-31.250      0.0343

Kaposi sarcoma      1.433-12.500      0.0090
                    1.664-16.949      0.0048

Cirrhosis           0.073-0.664       0.0072
                    0.568-0.950       0.0014

Arteriosclerosis    1.802-6.329       0.0001

Cachexia            1.145-3.367       0.0141
                    0.351-0.961       0.0342

* Multiple logistic regression was used to obtain an adjusted
odds ratio, which refers to the risk of the first group when
compared with the second group. IVDU indicates intravenous drug
user; CMV, cytomegalovirus; HSV, herpes simplex virus; HBV,
hepatitis B virus; HCV, hepatitis C virus; MAI, Mycobacterium
avium-intracellulare; MTB, Mycobacterium tuberculosis; and PCP,
Pneumocystis carinii pneumonia.
Table 4. Frequencies of Noninfectious Diseases *

                           Overall   Group A    Group B   Group C
Disease                   (N = 394)  (n = 88)  (n = 207)  (n = 99)

Neoplastic disease
 Systemic lymphoma         20 (5.1)   7 (8.0)   9 (4.3)    4 (4.0)
 Kaposi sarcoma            44 (11)   15 (17)   26 (13)     3 (3.0)
 Other tumors ([dagger])   41 (10)   13 (15)   15 (7.2)   13 (13)

Other disease
 Cirrhosis                 61 (15)    8 (9.1)  31 (15)    22 (22)
 ESRD                      14 (3.6)   0        12 (5.8)    2 (2.0)
 Arteriosclerosis         137 (35)   26 (30)   58 (28)    53 (54)
 Trauma                     9 (2.3)   0         5 (2.4)    4 (4.0)
 Cachexia                 180 (46)   42 (48)   101 (49)   37 (37)

* Values are expressed as No. (%) of patients. ESRD indicates
end-stage renal disease.

([dagger]) Thirteen other tumors were central nervous system
lymphomas.
Table 5. Specific Neuropathologic Entities *

                Overall    Group A   Group B    Group C
                (N = 281)  (n = 71)  (n = 123)  (n = 87)

HIVE            41 (15)     7(10)     21 (17)    13 (15)
Aseptic lepto-
 meningitis     57 (20)    11 (15)    27 (22)    19 (22)
PML              5 (1.8)    3 (4)      0          2 (2)
CNS lymphoma    13 (4.6)    1 (1)      6 (5)      6 (7)

* Values are expressed as No. (%) of patients. HIVE indicates
human immunodeficiency virus encephalitis; PML, progressive
multifocal leukoencephalopathy; and CNS, central nervous system.
Table 6. Organ System Involvement *

                        Overall   Group A    Group B   Group C
Organ System           (N = 394)  (n = 88)  (n = 207)  (n = 99)

Lung                   350 (89)   85 (97)   190 (92)    75 (76)
Liver                  176 (45)   42 (48)    90 (44)    44 (44)
Spleen                 104 (26)   22 (25)    60 (29)    22 (22)
Gl tract               172 (44)   44 (50)    86 (42)    42 (42)
Kidney                 137 (35)   30 (34)    65 (31)    42 (42)
GU tract                62 (16)   25 (28)    18 (8.7)   19 (19)
Endocrine              130 (33)   31 (35)    71 (34)    28 (28)
Heart                  136 (35)   32 (36)    69 (33)    35 (35)
Lymphatic/bone marrow  130 (33)   33 (37)    68 (33)    29 (29)
CNS                    251 (64)   52 (59)   120 (58)    79 (80)

* Values are expressed as No. (%) of patients. GI indicates
gastrointestinal; GU, genitourinary; and CNS, central nervous system.

The authors thank Dr Jim Godbold for his guidance in statistical analysis and Dr Patrick Lento for his assistance in obtaining records. This study was supported by grant R24MH59724 from the National Institutes of Health (Dr Morgello).

References

(1.) Division of HIV/AIDS, Centers for Disease Control. The HIV/AIDS epidemic: the first 10 years. MMWR Morb Mortal Wkly Rep. 1991;40:357-369.

(2.) Wortley P, Fleming P. AIDS in women in the United States: recent trends. JAMA. 1997;78:911-916.

(3.) Flexner C. HIV-protease inhibitors. N Engl J Med. 1998;338:1281-1292.

(4.) Hirsch M, D'Aquila R. Therapy for human immunodeficiency virus infection. N Engl J Med. 1993;328:1686-1695.

(5.) Brodt H, Kamps B, Gute P, Knupp B, Staszewski S, Helm E. Changing incidence of AIDS-defining illnesses in the era of antiretroviral combination therapy. AIDS. 1997;11:1731-1738.

(6.) Chan I, Neaton J, Saravolatz L, Crane L, Osterberger J. Community Programs for Clinical Research on AIDS. Frequencies of opportunistic diseases prior to death among HIV-infected persons. AIDS. 1995;9:1145-1151.

(7.) Friedland G, Saltzman B, Vileno J, Freeman K, Schrager L, Klein R. Survival differences in patients with AIDS. J Acquir Immune Defic Syndr Hum Retrovirol. 1991;4:144-153.

(8.) Katz M, Hessol N, Buchbinder S, Hirozawa A, O'Malley P, Holmberg S. Temporal trends of opportunistic infections and malignancies in homosexual men with AIDS. J Infect Dis. 1994;170:198-202.

(9.) Palella F, Delaney K, Moorman A, et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. N Engl J Med. 1998;338:853-860.

(10.) Balter M. AIDS now world's fourth biggest killer. Science. 1999;284:1101.

(11.) Mitka M. Slowing decline in AIDS deaths prompts concern. JAMA. 1999; 282:1216-1217.

(12.) Burton E, Troxclair D, Newman WI. Autopsy diagnoses of malignant neoplasms: how often are clinical diagnoses incorrect? JAMA. 1998;280:1245-1248.

(13.) Wilkes M, Fortin A, Felix J, Godwin T, Thompson W. Value of necropsy in acquired immunodeficiency syndrome. Lancet. 1988;8602:85-88.

(14.) Zarbo R, Baker P, Howanitz P. The autopsy as a performance measurement tool--diagnostic discrepancies and unresolved clinical questions. A College of American Pathologists Q-probes study of 2479 autopsies from 248 institutions. Arch Pathol Lab Med. 1999;123:191-198.

(15.) Mobley K, Rotterdam H, Lerner C, Tapper M. Autopsy findings in the acquired immune deficiency syndrome. Pathol Annu. 1985;20:45-65.

(16.) Niedt G, Schinella R. Acquired immunodeficiency syndrome: clinicopathologic study of 56 autopsies. Arch Pathol Lab Med. 1985;109:727-734.

(17.) Reichert C, O'Leary T, Levens D, Simrell C, Macher A. Autopsy pathology in the acquired immune deficiency syndrome. Am J Pathol. 1983;112:357-382.

(18.) Sehonanda A, Choi Y, Blum S. Changing patterns of autopsy findings among persons with acquired immunodeficiency syndrome in an inner-city population: a 12-year retrospective study. Arch Pathol Lab Med. 1996;120:459-464.

(19.) Welch K, Finkbeiner W, Alpers C, et al. Autopsy findings in the acquired immune deficiency syndrome. JAMA. 1984;252:1152-1159.

(20.) Hofman P, Saint-Paul M, Battaglione V, Michiels J, Loubiere R. Autopsy findings in the acquired immunodeficiency syndrome (AIDS). A report of 395 cases from the south of France. Pathol Res Pract. 1999;195:209-217.

(21.) Klatt E, Nichols L, Noguchi T. Evolving trends revealed by autopsies of patients with the acquired immunodeficiency syndrome: 565 autopsies in adults with the acquired immunodeficiency syndrome, Los Angeles, Calif, 1982-1993. Arch Pathol Lab Med. 1994;118:884-890.

(22.) Lyon R, Haque A, Asmuth D, Woods G. Changing patterns of infections in patients with AIDS: a study of 279 autopsies of prison inmates and nonincarcerated patients at a university hospital in eastern Texas, 1984-1993. Clin Infect Dis. 1996;23:241-247.

(23.) Markowitz G, Concepcion L, Factor S, Borczuk A. Autopsy patterns of disease among subgroups of an inner-city Bronx AIDS population. J Acquir Immune Defic Syndr Hum Retrovirol. 1996;13:48-54.

(24.) Jellinger K, Setinek U, Drlicek M, Bohm G, Steurer A, Lintner F. Neuropathology and general autopsy findings in AIDS during the last 15 years. Acta Neuropathol. 2000;100:213-220.

(25.) Masliah E, DeTeresa R, Mallory M, Hansen L. Changes in pathological findings at autopsy in AIDS cases for the last 15 years. AIDS. 2000;14:69-74.

(26.) Centers for Disease Control. 1993 Revised Classification System for HIV Infection and Expanded Surveillance Case Definition for AIDS Among Adolescents and Adults. MMWR Morb Mortal Wkly Rep. 1992;41 (51):1-19.

(27.) Afessa B, Greaves W, Green W, et al. Autopsy findings in HIV-infected inner-city patients. J Acquir Immune Defic Syndr Hum Retrovirol. 1992;5:132-136.

(28.) Ambros R, Lee E, Sharer L, Khan M, Robboy S. The acquired immunodeficiency syndrome in intravenous drug abusers and patients with a sexual risk: clinical and postmortem comparisons. Hum Pathol. 1987;18:1109-1114.

(29.) Hui A, Koss M, Meyer P. Necropsy findings in acquired immunodeficiency syndrome: a comparison of premortem diagnoses with postmortem findings. Hum Pathol. 1984;15:670-676.

(30.) Detels R, Munoz A, McFarlane G, et al. Effectiveness of potent antiretroviral therapy on time to AIDS and death in men with known HIV infection duration. JAMA. 1998;280:1497-1503.

(31.) Volberding P, Deeks S. Antiretroviral therapy for HIV infection: promises and problems. JAMA. 1998;279:1343-1344.

(32.) Sulkowski M, Thomas D, Chaisson R, Moore R. Hepatotoxicity associated with antiretroviral therapy in adults infected with human immunodeficiency virus and the role of hepatitis C or B virus infection. JAMA. 2000;283:74-80.

(33.) Farzadegan H, Hoover D, Astemborski J, et al. Sex differences in HIV-1 viral load and progression to AIDS. Lancet. 1998;352:1510-1514.

(34.) Melnick S, Sherer R, Louis T, et al. Survival and disease progression according to gender of patients with HIV infection: the Terry Beirn community programs for clinical research on AIDS. JAMA. 1994;272:1915-1921.

Accepted for publication September 26, 2001.

From the Manhattan HIV Brain Bank, Department of Pathology, Division of Neuropathology, the Mount Sinai Medical Center, New York, NY.

Reprints: Susan Morgello, MD, Box 1134, Mount Sinai Medical Center, 1 Gustave L. Levy PI, New York, NY 10029 (e-mail: susan. [e-mail]).