SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

BACKGROUND

The introduction of highly active antiretroviral therapy (HAART) has radically changed the clinical course of human immunodeficiency virus (HIV) infection. The goals of the current study were to assess the change in the incidence of Kaposi sarcoma (KS) among European patients with HIV since the introduction of HAART and to identify the factors associated with the development of KS among patients receiving HAART.

METHODS

The incidence of KS and the factors associated with the development of this malignancy in patients receiving HAART were evaluated using Poisson regression. Patients examined in the current study were among the 9803 individuals with HIV who were enrolled in the EuroSIDA study, a pan-European multicenter investigation.

RESULTS

There was an estimated annual reduction of 39% (95% confidence interval [CI], 35–43%; P < 0.0001) in the incidence of KS between 1994 and 2003. The proportion of acquired immunodeficiency syndrome (AIDS) diagnoses made due to KS during prospective follow-up ranged from 4.1% to 7.5%, and there was no significant change over time in this figure (P = 0.97). Four thousand fourteen patients began receiving HAART during prospective follow-up; 41 of these 4014 were subsequently diagnosed with KS (1.0%). After adjustment in multivariate analyses, patients with higher current CD4 counts were found to have a decreased incidence of KS (incidence rate ratio [IRR], 0.60; 95% CI, 0.53–0.68; P < 0.0001), as were those for whom more time had elapsed since the initiation of HAART (IRR, 0.77; 95% CI, 0.60–0.98; P = 0.037). In contrast, homosexual men were found to have a significantly increased incidence of KS (IRR, 2.12; 95% CI, 1.00–4.54; P = 0.050)

CONCLUSIONS

The current incidence of KS among patients with HIV is less than 10% of the incidence reported in 1994; the proportion of AIDS diagnoses made on the basis of KS diagnoses remains near 6%. Most individuals who developed KS while receiving HAART began treatment with low CD4 cell counts and developed KS within 6 months of the initiation of HAART. There continues to be an increased incidence of KS among homosexual men and a greatly reduced incidence of KS among patients with higher CD4 counts. Cancer 2004. © 2004 American Cancer Society.

In 1981, Hymes et al.1 reported atypical occurrences of Kaposi sarcoma (KS) among homosexual men; these findings were among the first heralding the arrival of human immunodeficiency virus (HIV) and acquired immunodeficiency syndrome (AIDS). Since that time, KS has become one of the most common AIDS-defining illnesses; it has been diagnosed as the initial AIDS-defining illness for more than 20% of patients with AIDS across Europe.2 The incidence of KS, which has been diagnosed predominantly among homosexual men,3 was reported to be decreasing even before the introduction of highly active antiretroviral therapy (HAART).4, 5 Widely reported risk factors for KS include acquisition of HIV via homosexual contact6 and advanced immunosuppression,7 and KS risk also exhibits a certain degree of geographic variation.2, 8 Various studies have suggested that an infectious agent may be the cause of KS. Chang et al.9 found herpesviruslike DNA sequences in KS tissue obtained from a patient with AIDS. The virus characterized by these sequences was named human herpesvirus 8 (HHV-8), and the discovery of this virus led to suggestions that agents with activity against human herpesvirus (i.e., cytomegalovirus or recurrent herpes simplex infection) might be effective in the treatment of KS.10, 11

The rapid introduction and widespread use of HAART have led to a reduction in the incidence of all AIDS-defining illnesses,12, 13 with incidence rates decreasing to less than one-tenth of their levels before the introduction of this treatment. Since the introduction of HAART, several studies have reported decreased incidence of KS,14–17 although these studies have tended to be limited to patients within a single clinic or country or have primarily involved homosexual men. Because diagnosis of KS is now relatively rare, few studies, to our knowledge, have directly examined the risk factors associated with KS among patients for whom HAART has already been initiated. Therefore, the goals of the current study were to evaluate the change in the incidence of KS among European patients with HIV since the introduction of HAART and to identify the factors associated with the development of KS among patients receiving this treatment.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

The EuroSIDA study is a prospective, Europe-based study of patients with HIV-1 infection. The study includes 72 centers across Europe (including Israel) and has expanded to include an Argentinean center as well. For the purposes of the analysis, Europe was divided into the following regions: North (Denmark, United Kingdom, northern Germany, Ireland, The Netherlands, Norway, Scotland, and Sweden), Central West (Austria, Belgium, France, southern Germany, Luxembourg, and Switzerland), South (Argentina, Greece, Italy, Portugal, Spain, and Israel), Central East (Hungary, Czech Republic, Slovakia, and Poland), and East (Estonia, Latvia, Lithuania, Romania, and Ukraine).

Details of the study have been published elsewhere.18 In brief, each center provided data on consecutive patients who attended the center's outpatient clinic between May 2, 1994, and the date on which a predefined number of patients had enrolled in the study. The resulting population of 3115 patients was referred to as the EuroSIDA I cohort. Enrollment of a second cohort (EuroSIDA II), which included 1365 patients, began in December 1995. In April 1997, an additional 2839 patients were recruited, and this population was referred to as the EuroSIDA III cohort. Enrollment of an additional 1225 patients (EuroSIDA IV) began in April 1999, and the recruitment of 1256 patients into a fifth cohort (EuroSIDA V) began in September 2001. At each follow-up visit, viral load measurements and details on all CD4 lymphocyte counts evaluated since the last follow-up examination were obtained, and starting and ending dates for the administration of each antiretroviral agent and information on the use of prophylactic agents against opportunistic infection were recorded. Dates of diagnosis of all AIDS-defining conditions (according to the clinical definition of AIDS put forth in 1993 by the Centers for Disease Control), including diagnoses that followed the first diagnosis of such a condition, also were recorded. Members of the EuroSIDA Coordinating Office visited all centers to ensure that patient selection methods were appropriate; in addition, members of the Coordinating Office verified that data from each center were accurate by checking the data provided by the center against case notes for all reported clinical events as well as for a random sample of 10% of all other patients.

Statistical Methods

Patients were followed from the date of recruitment until the first diagnosis of KS or until the last follow-up examination, and the incidence of KS per 1000 person-years of follow-up (PYFU) was calculated according to calendar year. Patients with KS recurrences were excluded from the analysis, as were patients who were diagnosed with KS before recruitment. Poisson regression was used to test for trends over time. The analysis of trends over time was stratified according to homosexual exposure (yes vs. no) as well as other risk factors to determine whether observed trends were consistent among risk groups. In addition, changes in KS incidence were compared with changes in the incidence of non-Hodgkin lymphoma (NHL) and in the incidence of all other AIDS-defining illnesses combined. Calendar years were organized into three groups (as described elsewhere17), corresponding to the pre-HAART era (1994–1995), the early HAART era (1996–1997), and the late HAART era (1998–present).

Poisson regression also was used to analyze the incidence of KS among patients who had not yet begun receiving HAART and among patients who had begun receiving HAART during prospective follow-up. HAART was defined as treatment with a minimum of three antiretroviral agents, with at least one being a protease inhibitor, a nonnucleoside reverse transcriptase inhibitor, or abacavir. In the analysis of the incidence of KS among patients who had not yet begun to receive HAART, patient follow-up was censored at the date of initiation of HAART; variables analyzed included time-dependent covariates such as CD4 counts, diagnoses of other AIDS-defining illnesses, and initiation of treatment with antiherpesvirus agents (e.g., ganciclovir, foscarnet, acyclovir, adefovir, cidofovir), as well as demographic factors such as gender, exposure group, age, geographic region, and ethnic origin.

A similar analysis was conducted for patients who had begun to receive HAART; patients with KS at the date of initiation of HAART were excluded from the analysis. In addition to all of the variables analyzed for patients who had not begun receiving HAART, viral load was included in the analysis as a time-dependent covariate; also included were time since the initiation of HAART, date of initiation of HAART, and treatment variables such as previous antiretroviral therapy and number of new antiretroviral agents used.

All variables that were found to be significant on univariate analysis (P < 0.1) were included in a final multivariate model. Treatment variables were evaluated based on intention to treat; thus, all treatment changes have been ignored. Statistical analyses were performed using SAS software (Version 8.2; SAS Institute, Cary, NC), and all significance tests were two sided.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Patient characteristics are summarized in Table 1. Five hundred sixty patients had been diagnosed with KS at the time of recruitment (5.7%). Comparison of patients with and without KS at recruitment revealed that those with KS more commonly were male (97.0% vs. 77.2%), belonged to the homosexual exposure category (88.6% vs. 42.4%), and were from Northern Europe (46.3% vs. 31.7%) (P < 0.0001 for all comparisons). In successive recruitment cohorts, the percentage of patients with KS at recruitment decreased; for example, in Cohort I, 7.9% of patients had KS at recruitment, compared with 2.2% of patients in Cohort V (P < 0.0001). Patients with KS at recruitment had significantly lower CD4 counts (median, 103/μL vs. 245/μL; P < 0.0001), were significantly older (median, 39.0 years vs. 36.0 years; P < 0.0001), and were recruited to EuroSIDA more than a year earlier (median, December 1995 vs. February 1997; P < 0.0001) compared with patients who did not have KS at recruitment. Viral load at recruitment was similar (P = 0.93) for patients with KS (data available for 231 patients [41.3%]) and patients without KS (data available for 4939 patients [53.4%]).

Table 1. Patient Characteristics
CharacteristicNo. of patients (%)P
All patientsNo KS at recruitmentKS at recruitment
  1. KS: Kaposi sarcoma; IDU: intravenous drug user; ART: antiretroviral therapy; HAART: highly active antiretroviral therapy; AIDS: acquired immunodeficiency syndrome; IQR: interquartile range.

Total9803 (100)9243 (94.3)560 (5.7)
Gender    
 Male7678 (78.3)7135 (77.2)543 (97.0)< 0.0001
 Female2125 (21.7)2108 (22.8)17 (3.0) 
Exposure group    
 Homosexual4404 (44.9)3293 (42.4)481 (88.6)< 0.0001
 IDU2403 (24.5)2381 (25.8)24 (4.3) 
 Heterosexual2319 (23.7)2286 (24.7)33 (5.9) 
 Other675 (6.9)653 (7.1)22 (3.9) 
Race    
 White8400 (85.7)7897 (85.4)503 (89.8)0.0040
 Other1403 (14.3)1346 (14.6)57 (10.2) 
Location in Europe    
 South2990 (30.5)2878 (31.1)112 (20.0)< 0.0001
 Central West2609 (26.6)2429 (26.3)180 (32.1) 
 Central East725 (7.4)719 (7.8)6 (1.1) 
 East288 (2.9)285 (3.1)3 (0.5) 
 North3191 (32.6)2932 (31.7)259 (46.3) 
Recruitment cohort    
 I3115 (31.8)2870 (31.0)245 (43.8)< 0.0001
 II1365 (13.9)1227 (13.8)93 (16.6) 
 III2839 (29.0)2687 (29.1)152 (27.1) 
 IV1225 (12.5)1182 (12.8)43 (7.7) 
 V1256 (12.8)1229 (13.3)27 (4.8) 
Treatment before recruitment    
 None2183 (22.3)2127 (23.0)56 (10.0)< 0.0001
 Other ART4652 (47.4)4322 (46.8)330 (58.9) 
 HAART2968 (30.3)2794 (30.2)524 (31.1) 
Previous diagnosis of AIDS2993 (30.5)2433 (26.3)560 (100)< 0.0001
Median CD4 count (IQR)238 (108–380)245 (116–384)103 (21–228)< 0.0001
Median viral load in log10(copies/mL) (IQR)3.11 (2.30–4.37)3.12 (2.30–4.37)2.92 (2.30–4.39)0.93
Median age in yrs (IQR)36.2 (31.3–43.2)36.0 (31.2–43.0)39.0 (33.5–44.8)< 0.0001
Median recruitment date (IQR)Feb. 1997 (May 1994–Dec. 1998)Feb. 1997 (May 1994–Feb. 1999)Dec. 1995 (Apr. 1994–May 1997)< 0.0001

Changes in the incidence of KS by calendar year can be seen in Figure 1. In 1994, the incidence of KS was 24.7 per 1000 PYFU (95% confidence interval [CI], 17.2–32.2). By 1997, the incidence of KS had decreased by approximately 80%, to 4.7 per 1000 PYFU (95% CI, 2.7–6.7), and KS incidence subsequently decreased further, to 1.7 per 1000 PYFU (95% CI, 0.7–3.4), in 2002 and beyond. There was an estimated reduction of 39% in the incidence of KS with each successive calendar year (95% CI, 35–43%; P < 0.0001 [Poisson regression]). In a given calendar year, KS diagnoses were responsible for 4.1–7.5% of all AIDS diagnoses made during prospective follow-up; there was no significant variation over time in this figure (P = 0.97).

thumbnail image

Figure 1. Incidence of Kaposi sarcoma (1994–2003) in the EuroSIDA study cohort. CI: confidence interval; PYFU: person-years of follow-up; AIDS: acquired immunodeficiency syndrome; KS: Kaposi sarcoma.

Download figure to PowerPoint

Decreases in the incidence of NHL, KS (after stratification according to exposure group), and all other AIDS-defining illnesses combined are shown in Figure 2. Although the rate of reduction in KS incidence in both exposure groups was not significantly different from the rate of reduction in the incidence of all other AIDS-defining illnesses (P = 0.44 and P = 0.062 for the homosexual exposure and other groups, respectively), the rate of reduction in KS incidence was significantly higher than the rate of reduction in NHL incidence (P < 0.05 for both exposure groups). Among homosexual men, the incidence of KS decreased by 41% per calendar year (95% CI, 36–46%; P < 0.0001), whereas in all other exposure groups combined, KS incidence decreased by the significantly lower rate of 31% per year (95% CI, 22–39%; P < 0.0001) (P < 0.05 for comparison of the rates of change in incidence). After adjustment for current CD4 count, the decrease in incidence per calendar year was 29% (95% CI, 23–35%) for homosexual men and 22% (95% CI, 9–32%) for all other exposure groups combined, and the difference between these two groups was no longer statistically significant (P = 0.25).

thumbnail image

Figure 2. Incidence of Kaposi sarcoma (KS), non-Hodgkin lymphoma (NHL), and other acquired immunodeficiency syndrome (AIDS)-defining illnesses (1994–2003) in the EuroSIDA study cohort. CI: confidence interval; PYFU: person-years of follow-up.

Download figure to PowerPoint

Table 2 presents data on KS incidence in the three treatment eras. KS incidence decreased by approximately 60% between the pre-HAART era and the early HAART era and by another 80% between the early HAART era and the late HAART era. There was no significant change over time in CD4 count at diagnosis of KS (P = 0.091); the median CD4 count at diagnosis decreased slightly over time, from 40/μL in the pre-HAART era to 56/μL in the late HAART era. Data on viral load at diagnosis of KS were not available in the pre-HAART era, and there was no evidence of any change in viral load at diagnosis between the early and late HAART eras (median log10[copies/mL], 4.56 vs. 4.66; P = 0.85). Furthermore, there was no significant difference over time in the proportion of AIDS diagnoses that resulted from diagnoses of KS (P = 0.29). There also was no change over time in the percentage of definitive KS diagnoses, with approximately 75% of all cases of KS being diagnosed definitively in each treatment era (P = 0.52). As expected based on the results presented in Figure 2, there was a decrease over time, from 78% in the pre-HAART era to 63% in the late HAART era (P = 0.015), in the proportion of KS diagnoses that involved homosexual men.

Table 2. Incidence Rates and Other Data at Diagnosis of Kaposi Sarcoma According to Treatment Era
 Pre-HAART era (1994–1995)Early HAART era (1996–1997)Late HAART era (1998–present)P
  1. HAART: highly active antiretroviral therapy; KS: Kaposi sarcoma; PYFU: person-years of follow-up; CI: confidence interval; IQR: interquartile range; VL: viral load; AIDS: acquired immunodeficiency syndrome.

Follow-up (person-yrs)3948.77306.625179.6
No. of events997146
Incidence of KS per 1000 PYFU (95% CI)25.1 (20.2–30.0)9.7 (7.4–12.0)1.8 (1.3–2.3)< 0.0001
Median CD4 count at diagnosis (IQR)40 (12–90)32 (11–140)56 (17–225)0.091
Median VL at diagnosis in log10(copies/mL) (IQR)4.56 (2.86–5.21)4.66 (3.68–5.19)0.85
Initial AIDS diagnosis (%)55 (55.6)31 (43.7)22 (47.8)0.29
Presumptive diagnosis of KS (%)77 (77.8)49 (69.0)33 (71.7)0.52
Diagnosis of KS in a homosexual man (%)77 (77.8)61 (85.9)29 (63.0)0.015

Four thousand fourteen patients began receiving HAART during prospective follow-up. The median CD4 count at the beginning of HAART was 207/μL (interquartile range [IQR], 98–328/μL), the median CD4 nadir was 143/μL (IQR, 55–235/μL), and the median viral load (in log10[copies/mL]) was 4.20 (IQR, 3.24–4.94). The median date on which HAART was initiated was June 1997 (IQR, December 1996–July 1998), and the median age at the initiation of HAART was 37.2 years (IQR, 32.6–43.9 years). Eight hundred ninety-nine patients (22.4%) had never been treated with an antiretroviral agent before the initiation of HAART, and 2023 patients (50.4%) received at least 3 new antiretroviral agents (i.e., antiretroviral agents that they had not previously been exposed to) at the initiation of HAART. The incidence of KS decreased with increasing time from the beginning of HAART (Fig. 3). KS incidence was highest (6.7 per 1000 PYFU; 95% CI, 3.6–11.5 per 1000 PYFU) during the first 6 months after the initiation of HAART and decreased to 1.2 per 1000 PYFU (95% CI, 0.6–2.2 per 1000 PYFU) by 24 months, for a decrease of 41% per treatment period (95% CI, 24–51%; P < 0.0001). Kaplan–Meier survival analysis was used to estimate that 0.8% of patients receiving HAART had developed KS within 2 years after the initiation of treatment (95% CI, 0.5–1.1%) and that the corresponding rate at 48 months was 1.1% (95% CI, 0.8–1.4%).

thumbnail image

Figure 3. Incidence of Kaposi sarcoma according to time since the initiation of highly active antiretroviral therapy. CI: confidence interval; PYFU: person-years of follow-up; HAART: highly active antiretroviral therapy.

Download figure to PowerPoint

Table 3 summarizes the characteristics of patients who received HAART according to whether they subsequently developed KS. In general, at the initiation of HAART, patients who went on to develop KS had lower CD4 counts (median, 77/μL vs. 210/μL; P < 0.0001), higher viral loads (median log10[copies/mL], 5.22 vs. 4.32; P < 0.0001), and earlier HAART start dates (median, November 1996 vs. June 1997; P = 0.0013). Furthermore, patients who subsequently developed KS had lower CD4 count nadirs (median, 71/μL vs. 144/μL; P = 0.0003) and higher pretreatment viral loads (median log10[copies/mL], 5.38 vs. 4.68; P = 0.042). At 6 months after the initiation of HAART, 13 patients (31.7%) had already developed KS and therefore were excluded from further analysis; in addition, 230 patients (5.8%) who had not developed KS were lost to follow-up less than 6 months after the initiation of HAART. Nonetheless, among those who were free of KS and who continued to be followed at 6 months, there were certain differences in initial responses to HAART. For example, patients who subsequently developed KS had significantly lower CD4 counts at 6 months after the initiation of HAART (median, 116/μL vs. 267/μL; P < 0.0001), although the absolute increase (P = 0.32) and the percent increase (P = 0.37) in CD4 counts were similar in both groups. In addition, patients who subsequently developed KS had higher viral loads after 6 months (median log10[copies/mL], 4.18 vs. 2.70; P = 0.0005) and were less likely to have viral loads < 400 copies/mL (21.4% vs. 41.9%; P = 0.029).

Table 3. Comparison of Patients Who Did and Did Not Develop KS after the Initiation of HAART
VariableaDeveloped KSDid not develop KSP value
  • KS: Kaposi sarcoma; HAART: highly active antiretroviral therapy; VL: viral load; AIDS: acquired immunodeficiency syndrome; HIV: human immunodeficiency virus.

  • a

    For continuous variables, median values (with interquartile ranges in parentheses) are presented, with P values calculated using the Wilcoxon test. For categoric variables, numbers of patients/cases (with percentages in parentheses) are presented, with P values calculated using the chi-square test.

  • b

    Data were available in the 6 months before the initiation of highly active antiretroviral therapy for 38 patients who subsequently developed Kaposi sarcoma and 3632 patients who did not (P = 0.41).

  • c

    Data were available in the 6 months before the initiation of highly active antiretroviral therapy for 28 patients who subsequently developed Kaposi sarcoma and 2996 patients who did not (P = 0.29).

  • d

    Patients who developed Kaposi sarcoma or were lost to follow-up before this date were excluded from subsequent analysis.

At initiation of HAART   
 No. of patients (%)41 (1.0)3973 (99.0) 
 CD4 countb77 (21–138)210 (100–329)< 0.0001
  Nadir CD4 count before HAART71 (13–124)144 (55–236)0.0003
  Time since nadir CD4 count (mos)2 (0–6)7 (2–19)< 0.0001
 VLc (log10[copies/mL])5.22 (4.25–5.63)4.32 (3.54–4.98)< 0.0001
  No. with VL < 400 copies/mL (%)0 (0)239 (8.0)0.16
  Peak VL before HAART5.38 (4.37–5.64)4.68 (4.10–5.22)0.042
  Time since peak VL (mos)1 (0–6)3 (1–9)0.26
 Median age in yrs37.1 (31.3–44.8)37.2 (32.6–43.8)0.74
 Date at initiation of HAARTNov. 1996 (Sept. 1996–Oct. 1997)June 1997 (Dec. 1996–Aug. 1998)0.0013
 ART-naive at initiation of HAART11 (26.8)888 (22.4)0.49
 Previously diagnosed with AIDS15 (36.6)971 (24.4)0.072
 Time since HIV infection (mos)55 (33–96)71 (39–113)0.082
6 mos after initiation of HAARTd   
 No. of patients followed283744< 0.0001
 CD4 count116 (69–190)267 (148–412)< 0.0001
  Increase in CD4 count25 (1–56)44 (1–119)0.32
  % increase in CD4 count25.8 (0.1–165.6)24.3 (0.6–74.7)0.37
 VL (log10[copies/mL])4.18 (2.70–5.20)2.70 (1.70–3.76)0.0005
  No. with VL < 400 copies/mL6 (21.4)1567 (41.9)0.029
 No. withdrawn from all treatment1 (3.6)260 (6.9)0.48
 No. with any change in treatment10 (35.7)1255 (33.5)0.83

Table 4 summarizes the factors associated with the development of KS among patients receiving HAART and, for historical purposes, among patients who had not yet begun to receive HAART. Multivariate analysis with adjustment revealed that after the initiation of HAART, the development of KS was less likely among patients with higher current CD4 counts (incidence rate ratio [IRR], 0.60; 95% CI, 0.53–0.68; P < 0.0001) and among patients for whom a greater amount of time had passed since the beginning of HAART (IRR, 0.77; 95% CI, 0.60–0.98; P = 0.037). In addition, it was found that homosexual men were significantly more likely than others to develop KS (IRR, 2.12; 95% CI, 1.00–4.54; P = 0.050). Other factors did not have a statistically significant effect on KS risk after adjustment.

Table 4. Kaposi Sarcoma Incidence Rate Ratios before and after the Introduction of Highly Active Antiretroviral Therapy
VariableUnivariateMultivariate
IRR (95% CI)PIRR (95% CI)P
  • IRR: incidence rate ratio; CI: confidence interval; HAART: highly active antiretroviral therapy; AIDS: acquired immunodeficiency syndrome; CMV: cytomegalovirus.

  • a

    Forty-one cases of Kaposi sarcoma were observed among 4014 patients.

  • b

    Time-dependent covariate.

  • c

    Compared with female gender.

  • d

    Compared with all other risk groups combined.

  • e

    Compared with all other regions of Europe combined.

  • f

    One hundred fifty-two cases of Kaposi sarcoma were observed among 5503 patients.

  • g

    Compared with white race.

After initiation of HAARTa    
 CD4 countb (per 50% increase)0.58 (0.52–0.64)< 0.00010.60 (0.53–0.68)< 0.0001
 Viral loadb (per unit increase in log10[copies/mL])1.56 (1.14–2.12)0.00511.24 (0.92–1.67)0.16
 Time since initiation of HAART (per 12-month increase)0.61 (0.47–0.78)< 0.00010.77 (0.60–0.98)0.037
 Development of AIDSb2.09 (1.12–3.90)0.0201.22 (0.60–2.47)0.59
 Treatment with anti-CMV/antiherpesvirus agentsb2.49 (1.15–5.40)0.0211.56 (0.68–3.56)0.29
 Calendar date at initiation of HAART (per 6-month increase)1.05 (1.02–1.07)0.00131.01 (0.98–1.04)0.46
 Male genderc2.59 (0.92–7.29)0.0711.32 (0.42–4.17)0.64
 Homosexual exposured2.31 (1.21–4.43)0.0112.12 (1.00–4.54)0.050
 Residence in Northern Europee1.85 (0.99–3.44)0.0531.46 (0.75–2.84)0.26
Before initiation of HAARTf    
 CD4 count2 (per 50% increase)0.67 (0.63–0.71)< 0.00010.67 (0.63–0.72)< 0.0001
 Development of AIDSb2.92 (2.12–4.01)< 0.00011.00 (0.67–1.47)0.98
 Treatment with anti-CMV/antiherpesvirus agentsb1.65 (1.02–2.67)0.0420.76 (0.46–1.25)0.28
 Region of Europe    
  North1.00 (—)1.00 (—)
  South0.35 (0.22–0.54)< 0.00010.56 (0.35–0.89)0.014
  Central West0.95 (0.67–1.36)0.791.48 (1.01–2.16)0.042
  Central East and East0.19 (0.03–1.34)0.0950.62 (0.09–4.52)0.64
 Male genderc27.00 (6.67–108.7)< 0.00018.40 (2.14–37.50)0.0027
 Homosexual exposured5.53 (3.76–8.12)< 0.00013.13 (2.06–4.74)< 0.0001
 Age (per 10-year increase)1.22 (1.05–1.43)0.0111.02 (0.86–1.21)0.83
 Nonwhite race/ethnicityg0.51 (0.29–0.88)0.0160.56 (0.31–0.99)0.045

Certain similarities with respect to these findings became evident in the analysis of KS risk factors before the initiation of HAART; for example, even before the initiation of HAART, patients with lower CD4 counts had a higher risk of developing KS than did patients with higher CD4 counts. In addition, male patients had a significantly increased incidence of KS relative to female patients, and compared with white patients, patients of other races/ethnicities had a decreased risk of developing KS. Furthermore, before the initiation of HAART, KS risk also varied according to geographic location; relative to patients in Northern Europe, those in Southern Europe had a decreased incidence of KS, whereas an increased incidence of KS was observed among patients in Central Western Europe.

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

EuroSIDA, which is one of the largest observational cohort studies of patients with HIV-1, includes patients from across Europe and from various exposure categories. In the current investigation, we observed an overall annual decrease of 39% in the incidence of KS since 1994, with a significantly higher decrease among homosexual men compared with patients in other risk groups. Among patients receiving HAART, there remains an increased risk of KS for homosexual men and for patients with low CD4 counts.

The incidence of KS currently is less than 10% of the incidence reported in 1994. Although it is clear that KS incidence began to decrease before the introduction of HAART,4, 5 the incidence of KS continued to decrease following the rapid introduction of this treatment across Europe.19 It is possible that the exposure of individuals to HHV-8 and the risk of developing KS among those infected with HHV-8 have changed over time.20 The results of the current study are highly consistent with those of other studies, which have also noted decreases in KS incidence following the introduction of HAART.21, 22 Furthermore, in the current study, the rate at which KS incidence decreased was greater than the rate at which NHL incidence decreased, as has been demonstrated elsewhere,14–17, 21, 23, 24 although it should be noted that the observed rates of decrease differed for different types of NHL.25 It was reassuring to note that the decrease in KS incidence was consistent with the decrease in the incidence of all other AIDS-defining illnesses combined and that there was no change over time in the proportion of AIDS diagnoses attributable to KS. Since the beginning of 2000, we have observed fewer than 10 cases of KS per year (incidence, ∼1–2 per 1000 PYFU), but the proportion of diagnoses made definitively has remained constant at approximately 70–80%.

The decrease in KS incidence was significantly more pronounced among homosexual men compared with individuals in other risk groups, although after adjustment for current CD4 count, this difference ceased to be statistically significant. These results suggest that homosexual men have had higher CD4 counts compared with individuals in other exposure groups in recent calendar periods; there are several possible explanations for this finding. Approximately 25% of the EuroSIDA cohort was infected with HIV via intravenous drug use; previous studies have shown that patients infected in this manner are less likely to receive HAART or to access care,26, 27 may be more likely to encounter problems with adherence to treatment,28 and may have poorer responses to HAART compared with patients in other exposure groups.29, 30

In the current study, the incidence of KS among patients receiving HAART decreased with increasing time since the initiation of HAART, with approximately 31% of KS diagnoses occurring within 6 months of the start of therapy. Similar results have been reported by Ledergerber et al.31 The risk of developing KS appears to decrease significantly as the immune system recovers and CD4 counts increase. This finding clearly suggests that the primary factor associated with susceptibility to KS is the level of immunodeficiency, and it indicates that epidemiologically, KS behaves more like an opportunistic infection than a malignancy. Among patients who subsequently developed KS, the median CD4 count at the initiation of HAART was low. Current recommendations32 suggest that for best clinical response, HAART should be initiated before the CD4 count is < 200/μL, and the reasons for delaying therapy in some patients to lower CD4 levels are unclear. The finding of increased KS incidence among patients with low CD4 counts was not attributable to the occurrence of KS in patients with newly diagnosed HIV infections, as on average, patients had been infected with HIV for nearly 5 years before developing KS.

It is noteworthy that even after adjustment for current CD4 count, patients who had been receiving HAART for longer periods of time had a significantly reduced incidence of KS. This finding may be attributable to underadjustment for the effects of CD4 count and the frequency with which the CD4 count is monitored. In addition, it has been suggested that protease inhibitors can block several pathways involved in tumor growth, invasion, and metastasis33 and that these inhibitors may play a specific role in attenuating the activity and proliferation of endothelial cells.34 Recent work has indicated that protease inhibitor–based and nonnucleoside reverse transcriptase inhibitor–based regimens are equally effective in protecting against KS.35 In the current study, there was no difference in KS incidence between patients who received protease inhibitor–based HAART and patients who received nonnucleoside reverse transcriptase inhibitor–based HAART (data not shown), although the power to detect such a difference was limited.

To our knowledge, few studies have examined the various factors associated with the development of KS in the HAART era. The limited number of KS diagnoses, even in larger studies such as the current one, may shed light on why studies addressing this issue are relatively rare. Aside from time since the initiation of HAART, the factor most strongly related to the development of KS among patients receiving HAART was CD4 count, with higher CD4 counts being associated with reduced incidence of KS. Thus, current CD4 count remains one of the most significant prognostic factors with respect to KS; specifically, patients receiving HAART should experience a reduction in KS risk if their CD4 counts begin to increase. Current viral load was not related to KS incidence after adjustment for current CD4 count; this finding suggests that more prognostic information is captured by CD4 count than by viral load, as has been suggested elsewhere.36 KS risk was found to be significantly higher among homosexual men, as was also reported before the introduction of HAART2, 3, 7, 8; however, unlike the AIDS in Europe Study Group, which examined patients with AIDS before the advent of HAART,2 we did not find any difference in the incidence of KS according to gender or geographic location after adjustment for CD4 count. In addition, we could not confirm previous reports that patients receiving antiherpes agents had a reduced incidence of KS.10, 11 These discrepancies may be attributable in part to the limited power of the current study, in which only 41 diagnoses of KS were noted among all patients receiving HAART.

In the current study, the factors that were found to be related to an increased incidence of KS before the initiation of HAART were similar to those reported in other studies2; thus, men had an increased incidence of KS after adjustment for exposure group, as did patients from Central Western Europe, whereas decreased KS incidence was noted among patients from Southern Europe. It is possible that this finding resulted from differences in the prevalence of HHV-8 across Europe, as patients in different regions of Europe (including patients who belonged to the same exposure category) may have been differently exposed to HHV-8.37 The incidence of KS was not related to age in the current study cohort.

The current study has several limitations that should be noted. First, in considering the generalizability of our findings, the question of the extent to which the EuroSIDA population is representative of the patient population in European clinics in general arises. The current cohort differs slightly from a clinical population in that patients were recruited for the current study in five distinct enrollment periods, whereas new patients would continuously be enrolling at an open clinic. Nonetheless, it is likely that the selection of consecutive patients at various clinics resulted in a cohort that was broadly representative of patients who typically visit such clinics.

Another limitation that should be noted is that although many centers across Europe are represented in the EuroSIDA study, centers that are not represented may be different in terms of physician experience or the availability of certain HAART regimens. In addition, treatment decisions for individual patients generally are made on a center-to-center basis and may not be uniform across Europe. Data on the sites at which KS occurred, which would have allowed us to assess whether the observed decrease in incidence was consistent for all types of KS, were not available. Finally, the data obtained in the current study did not capture the extent to which KS lesions resolved following the initiation of HAART or whether the risk of reappearance of KS lesions was increased for patients who experienced such resolution if their CD4 counts subsequently decreased. Investigation of the latter question would be a significant step in addressing whether HAART clears HHV-8 infection or whether it suppresses the angiogenetic effects of the virus.38

In summary, we have demonstrated a sizeable decrease in the incidence of KS among European patients with HIV-1, despite the finding that KS continues to represent the AIDS-defining condition in approximately 6% of all AIDS diagnoses. Most patients who developed KS while receiving HAART had low CD4 counts at the start of treatment and developed KS within 6 months of the initiation of HAART. A greater incidence of KS among homosexual men continues to be observed, as does a significantly reduced incidence of KS among patients with relatively high current CD4 counts.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES