Purpose
To estimate the incidence of cytomegalovirus (CMV) retinitis in the era of highly active antiretroviral therapy (HAART) and to characterize the factors associated with increased risk of CMV retinitis.
Design
Prospective cohort study.
Methods
A total of 1600 participants with acquired immunodeficiency syndrome (AIDS) but without CMV retinitis at enrollment who completed at least 1 follow-up visit in the Longitudinal Study of the Ocular Complications of AIDS (LSOCA) were seen every 6 months to obtain disease and treatment history, ophthalmic examination, and laboratory testing. Incidence of CMV retinitis and risk factors for incident CMV retinitis were assessed.
Results
The incidence rate of CMV retinitis in individuals with AIDS was 0.36/100 person-years (PY) based upon 29 incident cases during 8134 PY of follow-up. The rate was higher for those with a CD4+ T cell count at the immediately prior visit below 50 cells/μL (3.89/100 PY, P < .01), whereas only 1 individual with a CD4+ T cell count of 50 to 99 cells/μL and 2 individuals with a CD4+ T cell count >100 cells/μL developed CMV retinitis. Having a CD4+ T cell count below 50 cells/μL at the clinical visit prior to CMV retinitis evaluation was the single most important risk factor (HR: 136, 95% CI: 30 to 605, P < .0001) for developing retinitis.
Conclusions
Patients with AIDS, especially those with severely compromised immune systems, remain at risk for developing CMV retinitis in the HAART era, although the incidence rate is reduced from that observed in the pre-HAART era.
Cytomegalovirus (CMV) retinitis is the most common ocular opportunistic infection in patients with acquired immunodeficiency syndrome (AIDS). Prior to the introduction of highly active antiretroviral therapy (HAART), approximately 30% of individuals infected with the human immunodeficiency virus (HIV) developed CMV retinitis at some point during their lifetime. The initiation of widespread use of HAART in the mid-1990s dramatically altered the course of AIDS. Treatment with HAART suppresses HIV replication, resulting in a drop in HIV load and in immune recovery, as evidenced by a rise in CD4+ T cell counts. As a result, the incidence rates of opportunistic infections, such as CMV retinitis, have declined, but they have not dropped to zero. Patients primarily remain at risk for developing CMV retinitis either because of delayed diagnosis of HIV infection or because they are noncompliant with, intolerant of, or unresponsive to HAART. Despite the substantial reduction in the incidence of CMV retinitis and the development of more effective therapies, CMV retinitis remains an important cause of visual loss in patients with AIDS during the HAART era. Therefore, quantifying the incidence rate and understanding the risk factors associated with the development of CMV retinitis remains important for both patients and health care providers.
The Longitudinal Studies of the Ocular Complications of AIDS (LSOCA) was initiated to study the occurrence and consequences of ocular complications among patients with AIDS during the HAART era. A previous analysis estimating the incidence of CMV retinitis was performed based on the prevalence data obtained at the time of enrollment. However, since the cohort was enriched deliberately with individuals having an opportunistic infection, such as CMV retinitis, that analysis provides an upper bound for the estimate of the incidence, which is still unknown. Herein we present an analysis to assess the incidence rate and risk factors for CMV retinitis based on data collected from patients free of CMV retinitis at enrollment in the LSOCA cohort.
Patients and Methods
LSOCA is a multicenter, prospective, observational study of patients with AIDS recruited from ophthalmology clinics specializing in infectious disease or retinal clinics that was initiated in 1998. Patients aged 13 or greater with a diagnosis of AIDS were eligible, regardless of immunologic or CMV retinitis status. AIDS was diagnosed according to the 1993 Centers for Disease Control and Prevention Revised Surveillance Case Definition. For this analysis, the data were frozen as of December 31, 2009.
In LSOCA, study visits were initially conducted every 6 months for individuals without an opportunistic ocular infection (OOI) and every 3 months for individuals with an OOI. As of December 1, 2008, all individuals, regardless of OOI status, were followed at 6-month intervals. To eliminate the potential for bias attributable to unequal follow-up, only the 6-month visits were included in this analysis. Individuals were considered to be lost to follow-up if 3 consecutive visits were missed (ie, no visits occurred within the year and a half prior to the data freeze). The data collected included medical and ophthalmic history, Karnofsky score, and a complete ophthalmologic examination. Standardized descriptions of classical clinical features were used to diagnose CMV retinitis by LSOCA-certified ophthalmologists. Laboratory measurements included CD4+ T cell counts, CD8+ T cell counts, and plasma HIV load. In addition to being considered as a continuous measurement, CD4+ T cell counts were categorized as ≥100, 50–99, and <50 cells/μL. The threshold of ≥100 cells/μL was chosen since that is the level at which discontinuation of anti-CMV therapy is recommended. Previous studies in the pre-HAART era have indicated that additional risk is present for individuals with a CD4+ T cell count below 50 cells/μL. CMV load was available at each visit through 2003. The results presented here are based on the 1600 participants who were free from CMV retinitis at enrollment and who completed at least 1 follow-up visit.
The incidence rates for CMV retinitis are expressed as rates per 100 person-years (PY) and were calculated for the cohort and stratified by CD4+ T cell status at the visit prior to CMV retinitis evaluation (≥100, 50–99, or <50 cells/μL). Staggered entry Kaplan-Meier curves were plotted to portray the cumulative probability of incident CMV retinitis over time and were anchored at the reported date of AIDS diagnosis. Univariate Cox proportional hazards models were used to compare the relative risks of acquiring CMV retinitis without adjusting for multiple comparisons. Risk factors with repeated measurements during the course of the study were analyzed as time-varying covariates. Of these, age at each visit, HAART, CD4+ T cell count, CD8+ T cell count, and HIV load were assessed at the visit prior to CMV retinitis evaluation (ie, lagged by 1 visit). Analyses based on the participants with complete data at baseline, missing values for time-varying covariates carried forward from the most recent clinic visit, and multiple imputation to account for missing data yielded similar results. Statistical analyses were performed with SAS (SAS/STAT User’s Guide, Version 9.1; SAS Institute, Cary, North Carolina, USA) and Stata (StataCorp, 2007, Stata Statistical Software: Release 10; StataCorp LP, College Station, Texas, USA) statistical packages.
Results
As of December 31, 2009, 2271 participants with AIDS were enrolled in LSOCA, of whom 492 had CMV retinitis in at least 1 eye at enrollment. Of the remaining 1779 individuals free of CMV retinitis at enrollment, 1600 completed at least 1 follow-up visit. These individuals constitute the basis for our report. Of those 1600 subjects, a total of 1430 (89%), 1098 (69%), 797 (50%), and 80 (5%) completed 1, 3, 5, and 10 years of follow-up, respectively, with a total of 8134 person-years of follow-up. The missed visit rate was 13%, 337 individuals died, and an additional 329 were lost to follow-up (ie, had not been seen for a minimum of 18 months).
Characteristics of the Study Population at Enrollment and During Follow-up
The population characteristics at study entry and over the course of follow-up based on the observed visits for individuals free from CMV retinitis at enrollment are reported in Table 1 . The demographic characteristics observed in this cohort are comparable to those observed in other AIDS cohorts used to track CMV retinitis. In general, patients had long-standing disease with a median time from diagnosis of AIDS to enrollment of 4.4 years (25th–75th percentile: 1.7–7.3). Most individuals were male (N = 1288, 80%), in particular men having sex with men (N = 933, 58%), and the majority were nonwhite (N = 866, 54%). The median CD8+ T cell count and nadir CD4+ T cell count as of enrollment were 776 cells/μL (25th-75th percentile: 503–1135) and 40 cells/μL (25th-75th percentile: 12–104), respectively. The distributions of the time-varying CD8+ T cell counts and the time-updated nadir CD4+ T cell counts were similar to those observed at enrollment. Eighty-five percent of individuals were on HAART (N = 1356) and 72% (N = 411) had a CD4+ T cell count of 100 cells/μL or greater at enrollment with a higher fraction of each during the follow-up interval (93% [N = 13 105] and 86% [N = 12 065], respectively). The median duration of HAART usage time since HAART initiation at enrollment was 2.9 years. However, the date of initiation is missing for 36% (N = 570) of the participants since this question was added during the course of the study. Despite widespread HAART usage, 34% (N = 519) of participants had an HIV load greater than 10 000 copies/mL (>4 on the log 10 scale) at enrollment; however the percentage of the follow-up visits with HIV load greater than 10 000 copies/mL was lower (19%, N = 2517). CMV load was available for 67% (N = 1101) of individuals at enrollment and 33% (3351) of the follow-up visits. The percentage with detectable samples (≥400 copies/mL) was 2% (N = 20) and 1% (31) for enrollment and follow-up visits, respectively.
| Characteristic | Participants | Visits |
|---|---|---|
| (N = 1600) | (N = 14 311) | |
| Median age, years a | 43 (38 to 49) | |
| Median years since AIDS diagnosis a | 4.4 (1.7 to 7.3) | |
| Missing, n (%) b | 17 (1%) | |
| Median years since HAART initiation at enrollment a | 2.9 (1.0 to 5.1) | |
| Missing, n (%) b | 570 (36%) | |
| Sex, n (%) b | ||
| Female | 312 (20%) | |
| Male | 1288 (80%) | |
| Race, n (%) b | ||
| White, non-Hispanic | 734 (46%) | |
| Nonwhite | 866 (54%) | |
| HIV exposure category, n (%) b | ||
| Other | 667 (42%) | |
| Men having sex with men | 933 (58%) | |
| AIDS-defining illness, n (%) b | ||
| T-cell lymphopenia | 1031 (65%) | |
| Opportunistic infection/neoplasm | 558 (35%) | |
| Missing | 12 (1%) | |
| Karnofsky score, n (%) b | ||
| 90–100 | 819 (51%) | |
| ≤80 | 781 (49%) | |
| HIV retinopathy, n (% yes) b | 46 (3%) | 131 (1%) |
| Missing | 0 (0%) | 248 (2%) |
| HAART, n (% yes) b | 1356 (85%) | 13 105 (93%) |
| Missing | 1 (0%) | 235 (2%) |
| Nadir CD4+ T cell count | ||
| Median (cells/μL) a | 40 (12 to 104) | 37 (11 to 97) |
| Cells/μL, n (%) b | ||
| ≥100 cells/μL | 411 (26%) | 3405 (24%) |
| 50–100 cells/μL | 292 (19%) | 2629 (19%) |
| <50 cells/μL | 875 (55%) | 8007 (57%) |
| Missing | 22 (1%) | 270 (2%) |
| CD4+ T cell count | ||
| Median (cells/μL) a | 197 (81 to 360) | 324 (172 to 491) |
| Cells/μL, n (%) b | ||
| ≥100 cells/μL | 1142 (72%) | 12 065 (86%) |
| 50–100 cells/μL | 178 (11%) | 824 (6%) |
| <50 cells/μL | 268 (17%) | 1093 (8%) |
| Missing | 12 (1%) | 329 (2%) |
| CD8+ T cell count | ||
| Median (cells/μL) a | 776 (503 to 1135) | 827 (568 to 1152) |
| Cells/μL, n (%) b | ||
| ≥400 cells/μL | 1327 (84%) | 12 044 (88%) |
| <400 cells/μL | 253 (16%) | 1637 (12%) |
| Missing | 20 (1%) | 630 (4%) |
| HIV load | ||
| Median (log 10 [copies/mL]) a | 2.7 (1.9 to 4.6) | 2.3 (1.7 to 3.3) |
| Log 10 [copies/mL], n (%) b | ||
| <2.6 (undetectable) | 716 (47%) | 8921 (67%) |
| 2.6–4.0 | 298 (19%) | 1847 (14%) |
| >4.0 | 519 (34%) | 2517 (19%) |
| Missing | 67 (4%) | 1026 (7%) |
| CMV load, n (%) b | ||
| Undetectable (<400 copies/mL) | 1081 (98%) | 3320 (99%) |
| Detectable (≥400 copies/mL) | 20 (2%) | 31 (1%) |
| Missing | 500 (33%) | 10 962 (77%) |
a The 25th-75th percentiles are included in parentheses with the median.
b Percentages of missing values are calculated out of the total sample size. Percentages for categories are calculated out of the total sample of non-missing values.
Epidemiologic Risk of Cytomegalovirus Retinitis
During the 8134 person-years of follow-up for individuals free from CMV retinitis at enrollment, 29 incident cases of CMV retinitis were observed, yielding a rate of 0.36 per 100 PY with a cumulative incidence of 1.2% (95% CI: 0.5% to 3.1%) at 4 years and 4.2% (95% CI: 2.7% to 6.3%) at 10 years ( Table 2 ). The rate was constant over time since AIDS diagnosis ( Figure 1 ) . The Kaplan-Meier estimates of cumulative incidence of CMV retinitis stratified by time-varying CD4+ T cell count are presented in Supplemental Figure 1 . The incidence rate for those individuals with a CD4+ T cell count below 50 cells/μL at the preceding visit was 3.89/100 PY, significantly higher than the overall rate ( P < 0.0001). Twenty-six of the individuals who developed incident CMV retinitis (90%) had a CD4+ T cell count below 50 cells/μL at the visit prior to detection as compared to 1 (3%) with 50 to 99 cells/μL and 2 (7%) with ≥100 cells/μL. For the 3 individuals who had CD4+ T cell counts ≥50 cells/μL at the visit prior to the diagnosis of CMV retinitis, the onset of disease was well after the initial diagnosis with AIDS (4.6 and 4.9 years for the 2 individuals with CD4+ T cell counts ≥100 cells/μL and 12.9 years for the individual with CD4+ T cell count between 50 and 99 cells/μL). All 29 individuals who developed incident CMV retinitis had a CD4+ T cell count of less than 50 cells/μL at some time prior to developing CMV retinitis. However, not all individuals who experienced that level of immunodeficiency went on to develop the disease. In fact, when restricting the analysis to all follow-up visits that occurred after a CD4+ T cell count less than 50 cells/μL was observed, the incidence rate was 0.65/100 PY, slightly less than double the rate observed for population as a whole.
| Number of CMV Retinitis Incidence Cases | 4-Year Cumulative Incidence d (95% CI) | 10-Year Cumulative Incidence d (95% CI) | |
|---|---|---|---|
| Overall | 29 | 1.2% (0.5% to 3.1%) | 4.2% (2.7% to 6.3%) |
| CD4+ T cell count a | |||
| ≥100 cells/μL b | 2 | 0 | 0.3% (0.1% to 1.7%) |
| 50–99 cells/μL c | 1 | 0 | 0 |
| <50 cells/μL | 26 | 12.3% (5.1% to 27.9%) | 33.6% (23.2% to 47.0%) |
a CD4+ T cell count is evaluated at the visit prior to CMV retinitis evaluation (ie, lagged by 1 visit).
b The first incident case of CMV retinitis for an individual with a CD4+ T cell count of 100 cells/μL or more occurred after 4.6 years.
c The first incident case of CMV retinitis for an individual with a CD4+ T cell count between 50 and 99 cells/μL occurred after 12.9 years.
d Cumulative incidences with 95% confidence intervals are calculated using staggered entry Kaplan-Meier estimates anchored at the reported date of AIDS diagnosis.
Factors Influencing Risk of Cytomegalovirus Retinitis
The analysis of the risk factors associated with incident CMV retinitis is summarized in Table 3 . In the univariate analysis, a Karnofsky score <80 (hazard ratio [HR]: 2.2, 95% CI: 1.01 to 4.6, P = .045), the presence of HIV retinopathy (HR: 11.6, 95% CI: 3.3 to 40.2, P = .0001), CD8+ T cell count <400 cells/μL at the previous visit (HR: 15.2, 95% CI: 6.8 to 33.6, P < .0001), HIV load >10 000 copies/mL (>4 on the log 10 scale) at the previous visit (HR: 37, 95% CI: 8 to 162, P < .0001), CMV load at the previous visit (HR: 26, 95% CI: 3 to 186, P = .0014), and CD4+ T cell count less than 50 cells/μL at the previous visit (HR: 136, 95% CI: 30 to 605, P < .0001) were associated with a significant increase in the risk of developing incident CMV retinitis. CD4+ T cell counts below 10 were not significantly associated with increased incidence when compared to counts of 10 to 49 (HR: 2.0, 95% CI: 0.9 to 4.4, P = .07). CD4+ T cell counts of 50 to 99 cells/μL at the previous visit were not significantly associated with an increase in incidence (HR: 6.4, 95% CI: 0.5 to 73.2, P = .14), nor was treatment with HAART at the previous visit (HR: 0.6, 95% CI: 0.1 to 1.6, P = .27) or race (HR for nonwhite: 0.6, 95% CI: 0.3 to 1.4, P = .22). The HIV exposure men having sex with men was associated with an increased risk that was of borderline statistical significance (HR: 2.2, 95% CI: 0.9 to 5.2, P = .064). Older age was associated with a significant decrease in risk (HR per 10 years: 0.6, 95% CI: 0.4 to 0.8, P < .0001). Large multivariate models were not considered appropriate for analysis because of the potential for multiple risk profiles with zero events based on the small number of events and the co-linearity of the risk factors. Pairwise models, including CD4+ T cell counts plus 1 other variable, were fit for each of the significant risk factors. CD4+ T cell counts remained significant but the magnitude of the risk for the other variables was attenuated and became nonsignificant for all other variables (results not shown). Analyses restricting the risk set to those who had a nadir CD4+ T cell count below 50 cells/μL produced similar findings (results not shown).
| Risk Factor | Events per 100 PY | (# Events/PY) | Crude Hazard Ratio c (95% CI) | P Value |
|---|---|---|---|---|
| Age (per 10 years) a b | n/a | n/a | 0.6 (0.4 to 0.8) | <.0001 |
| Race | ||||
| White, non-Hispanic | 0.44 | (17/3908) | 1.0 | |
| Nonwhite | 0.29 | (12/4145) | 0.6 (0.3 to 1.4) | .22 |
| HIV exposure category | ||||
| Other | 0.22 | (7/3167) | 1.0 | .064 |
| Men having sex with men | 0.45 | (22/4885) | 2.2 (0.9 to 5.2) | |
| Karnofsky score | ||||
| 90–100 | 0.24 | (11/4607) | 1.0 | .045 |
| ≤80 | 0.52 | (18/3445) | 2.2 (1.01 to 4.6) | |
| HIV retinopathy a | ||||
| No | 0.33 | (26/7973) | 1.0 | |
| Yes | 3.80 | (3/79) | 11.6 (3.3 to 40.2) | .0001 |
| HAART a | ||||
| No | 0.61 | (4/654) | 1.0 | |
| Yes | 0.31 | (25/7398) | 0.6 (0.1 to 1.6) | .27 |
| CD4+ T cell count a | ||||
| ≥100 cells/μL | 0.03 | (2/6826) | 1.0 | |
| 50–99 cells/μL | 0.18 | (1/551) | 6.4 (0.5 to 73.2) | .14 |
| <50 cells/μL | 3.89 | (26/669) | 136 (30 to 605) | <.0001 |
| CD8+ T cell count a | ||||
| ≥400 cells/μL | 0.13 | (9/7076) | 1.0 | |
| <400 cells/μL | 1.99 | (19/955) | 15.2 (6.8 to 33.6) | <.0001 |
| HIV load [log 10 (copies/mL)] a | ||||
| <2.6 copies/mL | 0.04 | (2/5017) | 1.0 | |
| 2.6–4.0 copies/mL | 0.08 | (1/1250) | 2.1 (0.1 to 22.8) | .55 |
| >4.0 copies/mL | 1.47 | (25/1696) | 37 (8 to 162) | <.0001 |
| CMV load, n (%) a | ||||
| Undetectable (<400 copies/mL) | 0.52 | (9/1659) | 1.0 | |
| Detectable (≥400 copies/mL) | 6.67 | (1/15) | 26 (3 to 186) | .0014 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
