The Epidemiology and Incidence of Cytomegalovirus Retinitis in the HIV Population in Singapore over 6 Years

Abstract

Purpose.: Wereport the epidemiologic trend of cytomegalovirus (CMV) retinitis among human immunodeficiency virus (HIV) patients in Singapore.

Methods.: Our study included a retrospective case series of HIV patients with newly diagnosed CMV retinitis (CMVR) at the Singapore CDC between 2005 and 2010. Demographics, symptoms, signs, and laboratory results, including CD4 counts, were collected. Prevalence and disease trends over time were measured.

Results.: Of 224 new patients, 92.9% were male and 96.0% were on antiretroviral therapy (ART). Median age was 43.0 years, with a median CD4 count of 38.0 cells/μL at HIV diagnosis. There was a decline in CD4 counts at diagnosis, and median duration of ART initiation from HIV diagnosis generally was earlier. Overall incidence rate was 10.4 cases per 1000 person-years (PY) and the mortality rate was 19.4 per 1000 PY. Mean survival time was 160.2 months (95% confidence interval [CI] 150.3–170.2). The 25 patients who died during the study period were older at the time of CMVR diagnosis (P = 0.003) and had lower CD4 counts (P = 0.030). Worse prognoses were associated with older age (hazard ratio [HR] 1.06, 95% CI 1.02–1.10) and shorter time lag from HIV to CMVR diagnosis (HR 0.97, 95% CI 0.95–0.99). Patients who did not receive highly active antiretroviral treatment (HAART) had a higher mortality risk (HR 4.70, 95% CI 1.54–14.33).

Conclusions.: The incidence rate of CMVR was decreasing with earlier initiation of ART and lower CD4 counts at HIV diagnosis. Poor prognostic factors included older age, shorter time lag from HIV to CMVR diagnosis, and the absence of HAART. Constant ophthalmologic surveillance until immune reconstitution is recommended.

Introduction

The incidence of human immunodeficiency virus (HIV) infection has been on the rise in Singapore since its index case diagnosis in 1985, even though Singapore is classified as a low-level epidemic. ¹ By December 2010, 4845 individuals have been infected, with 23.5% having acquired immunodeficiency syndrome (AIDS)-related illnesses. ² Opportunistic infections (OIs) are common in HIV patients with advanced disease and form part of the definition for clinical AIDS. Several OIs may occur concomitantly, many of them affecting the eye. Cytomegalovirus (CMV) retinitis is the most common ocular AIDS-related opportunistic infection and accounts for up to 75 to 85% of CMV disease in these patients. ³ It has been reported to affect up to 25% to 40% of HIV patients and is the most common cause of visual loss. ⁴ Although common, it is a relatively late manifestation of HIV disease, usually associated with T-lymphocyte CD4 counts of <50 cells/μL. Before the advent of highly-active antiretroviral therapy (HAART), the estimated lifetime probability of a patient with AIDS suffering CMV retinitis (CMVR) was 30%. ⁵ With the advent of HAART, this has led to improved survival and ocular outcomes. However, in countries where HAART and anti-CMV treatment are not readily available, progression of CMVR is universal, often resulting in bilateral extensive retinal necrosis and eventual blindness. 

In Singapore, most patients with AIDS and CMVR have been treated with intravitreal ganciclovir since 1997. In addition, over the last 5 years HAART has become more readily available and since has replaced the 2-drug antiretroviral therapy (ART) as standard of care. Screening for CMVR in HIV patients with low CD4 counts and pre-HAART commencement also is routine practice at our institution. The presentation of CMVR also has been changing with the introduction of HAART. The change in trends is important and useful in the planning of medical services for HIV patients. However, epidemiologic data on patients with HIV and CMVR are lacking in our local context. We describe the changing epidemiologic trend, and clinical features of patients with AIDS and CMVR between 2005 and 2010 in Singapore. 

Materials and Methods

This is a retrospective case series on all consecutive patients diagnosed with clinical CMVR due to AIDS between 2005 and 2010 based on the criteria defined by the Centers for Disease Control (CDC). ⁶ All patients were seen and treated at the Communicable Disease Center, the main dedicated center to treat patients with CMVR and AIDS patients in Singapore. Clinical data obtained from case records included patient demographics, clinical presentation and outcomes, concomitant opportunistic infections, morbidity and mortality data, as well as laboratory data, and were collected on a standardized form. Approval was obtained from the local ethics review board in accordance with the Declaration of Helsinki. 

CMVR was diagnosed clinically only based on indirect ophthalmoscopy findings of the typical appearance of the disease with concurrent documentation on digital fundal photography (Visupac; Carl Zeiss Meditec AG, Jena, Germany). No aqueous or vitreous samples were obtained for viral PCR to confirm or exclude the diagnosis. The lesion typically consisted of an area of retinal opacification (necrosis or edema) surrounded by granular infiltrates and a silvery-white border marking the edge of the active borders, with variable amounts of retinal hemorrhage and inflammatory vascular sheathing. CMVR lesions were described based on lesion size and location. Lesion size was classified categorically as <25%, 25% to 49%, and >50% based on accurate clinical drawings. The location of CMVR lesion in each eye was categorized into 3 zones. Zone 1 was defined as the area within 1500 μm of the optic nerve or within 3000 μm of the center of the macula. Zone 2 extended from zone 1 to the vortex veins and zone 3 lay anterior to the vortex veins. For patients with CMVR involving multiple zones, the zone nearest to the macula was reported. All patients were treated upon detection and diagnosis of clinical active CMVR regardless of the location of the lesion. Bilateral treatment was instituted when both eyes were involved. No prophylactic treatments were done. All patients were treated with intravitreal ganciclovir (Cymevene; Roche, West Sussex, UK) in an open-label manner as described previously. ⁷ Informed consent was obtained from all patients for each eye treated. 

Frequencies and medians of demographics and clinical characteristics were calculated. Comparisons were done for survivors and deceased patients, as well as for patients with and without CMVR at HIV diagnosis. Continuous variables, such as age, CD4 counts, and durations, were compared using the Mann-Whitney test. Categoric variables, such as sex, lesion location, and size, were compared using the Pearson χ² test, or Fisher's exact test if expected counts within a category were less than 5 for more than 20% of cells. Post hoc analysis of adjusted standardized residuals was done if the Pearson χ² test indicated statistically significant results. The mortality rate was analyzed by enumerating the deaths due to HIV during the study period per unit follow-up time of all patients from the diagnosis of HIV. Cox regression analysis was performed univariately to elucidate the association of predictor variables, such as age, sex, time lag from HIV diagnosis to CMVR diagnosis, complications, and CD4 counts at CMVR diagnosis, with shorter time to death. A final multivariate Cox regression model then was built using all predictors found significant in the univariate analysis at the 5% level of significance. We used the likelihood ratio test to evaluate whether the inclusion of a new covariate improved the fit of the model. A P value less than 0.05 was considered to indicate statistical significance. Data analysis was done using IBM SPSS Statistics (version 19; IBM Corp., New York, NY). 

Results

Demographics

Of 224 new patients with CMVR, 208 (92.9%) were male and 215 (96.0%) were on HAART. The median age at presentation was 43.0 years (interquartile range [IQR] 34.3–52.0 years), with similar distribution between men (median 43.0 years, IQR 34.3–52.0 years) and women (median 42.5 years, IQR 34.8–47.8 years). There was no trend in the sex differences over the years. Bilateral disease was seen in 33.5%. The overall median CD4 count at HIV diagnosis was 38.0 cells/μL (IQR 20.0–110.8 cells/μL) and the median CD4 count at CMVR diagnosis was 33.0 cells/μL (IQR 13.0–63.8 cells/μL). The CD4 counts remained constant with no significant increase or decrease between 2005 and 2010. The overall median duration of HAART initiation from HIV diagnosis was 7.0 months (IQR 1.0–33.0 months) and median time lag from HIV diagnosis to CMVR was 26.0 months (IQR 5.0–60.0 months, Table 1). All patients were CMV IgG positive. Of 224 cases with CMVR, 14 were diagnosed at initial HIV diagnosis. Of the remaining 210 eyes, 31 were diagnosed before HAART commencement (median 2.0 months, IQR 1.0–3.0 months) and 166 had CMVR during HAART treatment (median 13.0 months, IQR 4.0–31.3 months). 

Table 1. 

 

Demographics of HIV Patients with Cytomegalovirus (CMV) Retinitis and Clinical Characteristics of CMV Retinitis Seen in Singapore from 2005–2010

	2005	2006	2007	2008	2009	2010	Overall
CMV Retinitis cases, n (%)*
Total	31 (13.84)	39 (17.41)	38 (16.96)	41 (18.30)	36 (16.07)	39 (17.41)	224 (100.00)
Male	29 (12.95)	37 (16.52)	34 (15.18)	37 (16.52)	34 (15.18)	37 (16.52)	208 (92.86)
Female	2 (0.89)	2 (0.89)	4 (1.79)	4 (1.79)	2 (0.89)	2 (0.89)	16 (7.14)
Cumulative HIV cases, n	2703	3062	3485	3941	4404	4845
Incidence, per 1000 PY	11.6	13.0	11.2	10.8	8.5	8.4	10.6
Age, y; median [IQR]
At CMVR diagnosis	38.0 [28.0–51.0]	38.0 [27.0–55.0]	44.5 [37.8–52.3]	42.0 [37.5–53.5]	46.5 [39.0–51.0]	42.0 [30.0–52.0]	43.0 [34.3–52.0]
At HIV diagnosis	35.0 [26.0–45.0]	35.0 [24.0–52.0]	43.0 [35.8–51.0]	41.0 [33.0–51.0]	41.5 [35.0–48.8]	40.0 [29.0–48.0]	40.0 [30.0–49.0]
At HAART initiation	37.0 [26.0–49.0]	36.0 [23.3–50.8]	44.0 [36.0–51.0]	43.0 [38.0–53.0]	44.0 [36.5–49.8]	40.0 [29.0–50.3]	41.0 [32.0–50.0]
CD4 count, cells/μL; median [IQR]
At CMVR diagnosis	36.0 [13.0–66.0]	37.0 [16.0–60.0]	39.0 [12.8–82.8]	20.0 [8.0–50.0]	29.5 [10.3–78.5]	37.0 [20.0–71.0]	33.0 [13.0–63.8]
At HIV diagnosis	110.0 [26.0–183.0]	39.0 [19.0–87.0]	33.5 [16.5–83.0]	21.0 [11.0–57.0]	37.5 [20.0–62.5]	41.0 [28.0–126.0]	38.0 [20.0–110.8]
At HAART initiation	46.0 [23.0–84.0]	50.0 [28.3–75.8]	32.0 [13.0–49.0]	22.0 [7.0–53.0]	31.5 [14.3–62.5]	37.0 [15.5–61.0]	36.0 [14.0–66.0]
Patients with CMVR and ADI at CMVR diagnosis, n (%)
Total, n (%)	24 (77.42)	32 (82.05)	34 (89.47)	37 (90.24)	33 (91.67)	36 (92.31)	196 (87.50)
Duration, mo; median [IQR]
HIV diagnosis to HAART initiation	26.0 [7.0–50.0]	13.5 [2.0–33.8]	5.0 [1.0–16.0]	2.0 [1.0–12.0]	2.0 [1.0–23.0]	10.0 [2.0–39.3]	7.0 [1.0–33.0]
HAART initiation from CMVR diagnosis	12.0 [4.0–20.0]	21.5 [9.3–36.3]	4.0 [1.0–22.0]	1.0 [0.0–13.0]	2.0 [0.0–37.8]	7.5 [−0.3–21.8]	7.0 [1.0–24.0]
HIV diagnosis to CMVR diagnosis	32.0 [18.0–69.0]	37.0 [24.0–74.0]	13.0 [4.0–53.0]	9.0 [2.0–48.5]	15.0 [2.0–62.8]	27.0 [10.0–57.0]	26.0 [5.0–60.0]
Location of CMVR, n (%)
Zone 1	9 (22.5)	16 (29.6)	27 (45.8)	29 (55.8)	13 (28.9)	15 (30.6)	109 (36.5)
Zone 2	26 (65.0)	27 (50.0)	25 (42.4)	21 (40.4)	26 (57.8)	25 (51.0)	150 (50.2)
Zone 3	5 (12.5)	11 (20.4)	7 (11.9)	2 (3.8)	6 (13.3)	9 (18.4)	40 (13.4)
Lesion size of CMVR, n (%)
<25%	11 (27.5)	18 (33.3)	34 (58.6)	33 (62.3)	27 (60.0)	14 (28.6)	137 (45.8)
25–50%	21 (52.5)	23 (42.6)	20 (34.5)	17 (32.1)	11 (24.4)	25 (51.0)	117 (39.1)
>50%	8 (20.0)	13 (24.1)	4 (6.9)	3 (5.7)	7 (15.6)	10 (20.4)	45 (15.1)

*   Expressed as percentage of new CMV retinitis cases.

Characteristics of CMVR

Incidence rates of HIV patients with CMVR ranged from 10.8 to 13.0 cases per 1000 person-years (PY) between 2005 and 2008, and declined to 8.5 cases per 1000 PY with an overall incidence rate of 10.6 cases per 1000 PY. This correlated with an increase in prevalence of HIV cases locally from 91.4 to 125.2 cases per million from 2005 to 2008, and a decrease to 124.0 and 116.9 cases per million from 2009 to 2010. The total number of new HIV cases also increased by 10% annually from 2005 to 2008, with a slight decline from 2009 to 2010. Lesions were found most commonly in zone 2 (50.2%) and 45.8% were <25% of retinal area in size (Table 1). There was no trend in the location and size of CMVR lesions over the period of study. The most common symptoms at CMVR diagnosis were blurring of vision (47.8%) and floaters (18.3%), but 32.1% were asymptomatic. 

Comparing the patients with and without CMVR at HIV diagnosis, there was a similar sex distribution in both groups. The median CD4 counts at HIV diagnosis were 26.5 and 38.5 cells/μL, respectively (P = 0.110) and at CMVR diagnosis they were 26.5 and 33.0 cells/μL, respectively (P = 0.517, Table 2). The median age at HAART initiation for patients with CMVR at HIV diagnosis was 49.0 years compared to 41.0 years for those without CMVR (P = 0.023). There was a significant difference in the median duration of HAART initiation after HIV diagnosis between both groups. Patients without CMVR commenced HAART later than those with CMVR at HIV diagnosis (8.0 vs. 1.0 months, P = 0.001). Eleven patients (78.6%) with CMVR at HIV diagnosis compared to 185 (88.1%) without CMVR had presence of concomitant AIDS-defining illnesses (ADI) at diagnosis of HIV (P = 0.393). There was no statistical difference in the location of CMVR lesions between both groups. 

Table 2. 

 

Comparison between Patients with and without CMVR at HIV Diagnosis

	With CMVR	Without CMVR	P Value
Number of patients, n (%)	14 (6.25)	210 (93.75)
Sex, n (%)			>0.999*
Male	13 (92.86)	195 (92.86)
Female	1 (7.14)	15 (7.14)
Age, y; median [IQR]
At CMVR diagnosis	49.0 [40.3–56.8]	42.0 [34.0–51.0]	0.071†
At HIV diagnosis	49.0 [40.3–56.8]	39.0 [30.0–48.0]	0.011†
At HAART Initiation	49.0 [40.3–56.8]	41.0 [32.0–50.0]	0.023†
CD4 count, cells/μL; median [IQR]
At CMVR diagnosis	26.5 [11.5–57.3]	33.0 [13.0–65.3]	0.517†
At HIV diagnosis	26.5 [11.5–57.3]	38.5 [20.0–115.0]	0.110†
At HAART initiation	30.0 [15.8–57.3]	36.0 [14.0–67.0]	0.650†
Duration, mo; median [IQR]
HIV diagnosis to HAART initiation	1.0 [1.0–2.0]	8.0 [1.0–34.0]	0.001†
HAART initiation to CMVR diagnosis	−1.0 [−2.0 to −1.0]	9.0 [1.0–27.0]	<0.001†
HIV diagnosis to CMVR diagnosis	0.0 –	29.0 [8.0–62.3]	<0.001†
Commencement of HAART, mo			<0.001*
HAART before CMVR diagnosis, n (%)	14 (100.00)	35 (17.41)
Duration, mo; median [IQR]	1.0 [1.0–2.0]	1.0 [0.0–3.0]
HAART after CMVR diagnosis, n (%)	0 (0.00)	166 (82.59)
Duration, mo; median [IQR]	– –	13.0 [4.0–31.3]
Presence of concomitant AIDS defining illness at diagnosis of HIV, n (%)			0.393†
Yes	11 (78.57)	185 (88.10)
No	3 (21.43)	25 (11.90)
Location for both eyes, n (%)			0.549‡
Zone 1	8 (42.11)	101 (36.07)
Zone 2	10 (52.63)	140 (50.00)
Zone 3	1 (5.26)	39 (13.93)
Lesion size in both eyes, n (%)			0.032‡
<25%	11 (61.11)	126 (44.84)
25–50%	2 (11.11)	115 (40.93)
>50%	5 (27.78)	40 (14.23)

*   Fisher's Exact test.

†   Mann-Whitney test.

‡   Pearson χ² test.

Mortality

The overall mortality rate was 19.4 deaths per 1000 PY and overall mean survival time was 160.2 months (95% confidence interval [CI] 150.3–170.2 months). Among the 25 patients who died of HIV-related complications during the study period, which made up 11.2% of all our CMVR patients, the median duration from HIV diagnosis to death was 13.0 months (IQR 6.0–67.0 months), and that from CMVR diagnosis to death was 6.0 months (IQR 4.0–14.0 months). 

Comparing the patients who died and those who survived, there was no difference in the sexes. The median age of patients who died were older at time of HIV diagnosis (45.0 vs. 39.0 years, P = 0.003) and at diagnosis of CMVR (48.0 vs. 42.0 years, P = 0.005), had a lower median CD4 count at diagnosis of CMVR (12.0 vs. 34.0, P = 0.030), and tended to have more zone 1 disease (adjusted standardized residual = 2.8). However, there was no statistical difference in the median CD4 count at the time of HIV diagnosis (37.0 vs. 38.0, P = 0.705). There also was no statistical difference in the proportion of patients with ADIs and CMVR lesion size at the time of HIV diagnosis (Table 3). 

Table 3. 

 

Comparison between Survivors and Deceased Patients in our Study Population

	Survivors	Deceased	P Value
Number of patients, n (%)	199 (88.84)	25 (11.16)
Sex, n (%)			>0.999*
Male	184 (92.46)	24 (96.00)
Female	15 (7.54)	1 (4.00)
Age, y; median [IQR]
At CMVR diagnosis	42.0 [33.0–51.0]	48.0 [40.5–59.5]	0.005†
At HIV diagnosis	39.0 [29.0–48.0]	45.0 [35.5–59.5]	0.003†
At HAART initiation	41.0 [32.0–50.0]	45.0 [39.5–57.0]	0.028†
CD4 count, cells/μL; median [IQR]
At CMVR diagnosis	34.0 [14.0–66.0]	12.0 [8.0–50.5]	0.030†
At HIV diagnosis	38.0 [20.0–111.0]	37.0 [20.0–129.0]	0.705†
At HAART initiation	37.0 [15.0–67.0]	18.0 [3.5–50.5]	0.037†
Duration, mo; median [IQR]
HIV diagnosis to HAART initiation	7.0 [1.0–33.3]	4.0 [1.0–13.0]	0.399†
HAART initiation from CMVR diagnosis	9.0 [1.0–25.5]	0.0 [−2.0–4.5]	0.002†
HIV diagnosis to CMVR diagnosis	28.0 [6.0–60.0]	6.0 [1.0–52.5]	0.032†
Presence of concomitant ADI at diagnosis of HIV, n (%)			0.331*
No	27 (13.57)	1 (4.00)
Yes	172 (86.43)	24 (96.00)
Number of eyes, n (%)	260 (86.96)	39 (13.04)
Location for both eyes, n (%)			0.015‡,§
Zone 1	87 (33.46)	22 (56.41)
Zone 2	135 (51.92)	15 (38.46)
Zone 3	38 (14.62)	2 (5.13)
Lesion size in both eyes, n (%)			0.093‡
<25%	114 (43.85)	23 (58.97)
25%–50%	103 (39.62)	14 (35.90)
>50%	43 (16.54)	2 (5.13)

*  Fisher's Exact test.

†  Mann-Whitney test.

‡  Pearson χ² test.

§  Post hoc analysis of adjusted standardized residuals showed that there are more CMVR lesions in Zone 1 than expected (adjusted standardized residual = 2.8) among patients with CMVR who died in our study.

A multivariate Cox regression model consisting of age at CMVR diagnosis, duration from HIV diagnosis to CMVR treatment, and HAART treatment was built based on predictors found significant in the univariate analysis. It was found that a worse prognosis was associated with older age at CMVR diagnosis (hazard ratio [HR] 1.06, 95% CI 1.02–1.10, P = 0.003), and shorter time lag from HIV to CMVR diagnosis (HR 0.97, 95% CI 0.95–0.99, P = 0.001). Patients with CMVR who did not or were not able to receive concomitant HAART were more likely to experience a shorter time to death (HR = 4.70, 95% CI 1.54–14.33, P = 0.006) compared to those who received treatment (Table 4). The cumulative survival of patients with CMVR is presented in the Figure. Median survival time was not defined, as more than 50% of the subjects were alive at the end of the study. 

Figure. 

 

Survival curve of patients with CMV retinitis in our study.

Table 4. 

 

Hazard Ratios for Time-to-Death after HIV Diagnosis in Patients with CMV Retinitis in Our Study Population

	n	Unadjusted HR (95% CI)*	P Value	Adjusted HR (95% CI)*	P Value
Age at CMVR diagnosis	224	1.06 (1.03, 1.10)	0.001	1.06 (1.02, 1.10)	0.003
Duration from HIV diagnosis to CMVR diagnosis	224	0.97 (0.95, 0.99)	0.001	0.97 (0.95, 0.99)	0.001
HAART treatment		5.10 (1.74, 14.93)	0.003	4.70 (1.54, 14.33)	0.006
No	9
Yes (reference)	215

*  HR (95% CI) = Hazard Ratio (95% Confidence Interval).

Discussion

Since its first recognition in 1981, the AIDS pandemic has killed an estimated 2.1 million people despite increased access to antiretroviral therapy. Most of them succumb to opportunistic infections during advanced stages of the disease, of which CMV disease remains a major cause of AIDS-related morbidity and mortality. ⁸ Opportunistic CMVR accounts for 75 to 85% of CMV disease and occurs in approximately 30% of AIDS patients, especially when CD4 <50/μL. ⁹ It accounts for over 90% of cases of HIV-related blindness and in developing countries where ART is inaccessible, the reported risk of mortality is 30% per year. ¹⁰ As the CD4 T-lymphocyte cells are the primary targets of HIV, it is the most common surrogate laboratory indicator of the degree of immunocompromise in the patient. Its count is used to stage the disease and prognosis, determine the risk and likelihood of various types of opportunistic infections, and guide decisions about the timing of ART commencement. 

During our 6-year study period, the incidence of CMVR among AIDS patients decreased from 13.0 to 8.4 cases per 1000 PY, although the rate of CMVR present at the time of AIDS diagnosis was only 6.3%. This is similar to rates in other reports that ranged between 3% and 40%. ^3,11–16 Various studies from Asia reported a prevalence of 7% to 20% of CMVR in HIV patients in the first decade of the 21st century. ^17–19 In Taiwan, another low-endemic country for HIV like Singapore, the incidence of CMVR declined from 26.6% in 1995 to 2% to 5% in the 2000s, presumably with the advent of HAART. In the United States, Hoover et al. reported a 4-year incidence of CMVR of 25%, ⁵ but this had dropped to 7% in the LSOCA studies reported by Sugar et al. ²⁰ This worldwide downward trend of CMVR patients relative to new AIDS patients is attributable to the increase in accessibility and decreased cost of HAART worldwide and in Singapore over the years, albeit some earlier than others, as well as the changing trend of infectious disease physicians to treat AIDS patients at higher CD4 counts. ²¹ All of our patients were tested to be CMV IgG positive, which was not surprising given the high seroprevalence of CMV of up to 87% in Singapore. ²² There was a gradual shortening of the time of initiation of HAART after HIV diagnosis from 26 to less than 10 months, reflective of the changing criteria for HAART initiation as well as the late presentations of our population. However, there still was a significant number of patients who presented late or had CMVR after diagnosis of HIV and commencement of HAART. This also was confirmed from the median CD4 counts at the time of CMVR diagnosis (33.0 cells/μL) and at HAART initiation (36.0 cells/μL). However, compared to the patients with CMVR at time of HIV diagnosis, there was no significant difference in the severity of their diseases whether in terms of CD4 counts or presence of concomitant ADIs, suggesting that there was no difference in the populations of patients who suffer CMVR early or late. In the Asian population, social stigmata and financial reasons are the most frequent reasons for not seeking earlier diagnosis and treatment. We noted that HAART was commenced at a median of 7 months after CMVR diagnosis. However, some patients also already were on treatment before development of CMVR. This was likely attributable to treatment failures from various reasons, including resistance and noncompliance as well as financial reasons for noncompliance. Our patients suffered CMVR at a median of 26.0 months after HIV diagnosis (IQR 5.0–60.0 months), a rate not dissimilar to that of other studies reporting a median of 18 months. ²³ As such, one should be aware of the importance of a regular monitoring program for HIV patients with low CD4 counts even if on HAART therapy, as many may suffer retinitis 1 to 2 years after HIV has been diagnosed. 

Analyzing the trend of disease over the 5 years, a general decline in CD4 count was observed at HIV diagnosis and at CMVR diagnosis, which was likely due to increasing proportion of late presentations, even though HAART was increasingly accessible in Singapore. HAART was being initiated at higher CD4 counts over time and current guidelines state that HAART should be initiated in all patients with an ADI (e.g., CMVR) or CD4 threshold counts <350 cells/μL in asymptomatic, treatment-naïve individuals. ²¹ Our median duration of HAART initiation from HIV diagnosis had, indeed, become shorter with time. Many physicians are leaning towards starting treatment earlier in the attempt to halt progression and replication of HIV, reducing its attending comorbidities. 

Our mortality rate of patients with CMVR was 19.4 per 1000 PY and mean survival was 160 months. This is substantially longer than the mean 35.2 months survival reported in a series by Kempen et al. ²⁴ This is particularly interesting as our patients with CMVR were treated with intravitreal ganciclovir injections until CD4 counts were >100 cells/μL. ⁷ Only patients with systemic manifestations were given the standard systemic anti-CMV therapy regimen of 3 to 4 weeks. Previous studies have suggested that the development of CMV disease is a predictor of increased risk of death. ^3,25,26 In our study, a shorter time lag from HIV to CMVR diagnosis corresponded with shorter time to death (P = 0.032). It has been reported that there was a 30% mortality risk per year in the HAART-naïve patient; however, this was reduced by 80% with the use of HAART and anti-CMV therapy. ²⁴ Our study showed that patients with lower CD4 counts at diagnosis of CMVR, but not at HIV diagnosis, were more likely to die (P = 0.030 and P = 0.705). This was independent of the presence of an ADI at diagnosis of HIV (P = 0.331). High CMV DNA load was associated with higher mortality and treatment with anti-CMV reduces this risk. ²⁴ However, we noted that even with HAART and anti-CMV treatment, there still is a higher risk of mortality when patients contract CMVR. This trend was consistent even in patients with CMV disease in non-HIV disease. ²⁷ Patients with CMVR alone still can have detectable cytomegaloviremia that predisposes to increased mortality. Systemic anti-CMV therapy also has been shown decrease mortality independent of HAART. ^14,24 This highlights the advantage of systemic anti-CMV treatment for CMVR as this has impact on the eventual long-term mortality rate of these patients. Although patients with concomitant systemic CMV disease were treated with systemic anti-CMV medications, our patients with CMVR alone were treated with intravitreal ganciclovir therapy alone due to financial considerations. In addition, among our patients who were older and where duration of HIV to CMVR was shorter, there also was an increased risk of mortality. This most likely was a reflection of the poorer immune status and, thus, susceptibility to systemic complications with associated morbidity and mortality in these individuals. Thus, there may be value in further evaluating the role of CMV in the disease progression of HIV and mortality. ²⁸  

Our study described a gradually later presentation of HIV infections in Singapore. In this era, our people still are not seeking early medical attention due to financial considerations and the social stigmata associated with the diagnosis of HIV in a traditionally and culturally conservative society. In addition the high cost of treatment adds to the burden of being HIV-positive. This dissuades high-risk individuals from seeking medical attention, blood testing and treatment, deterring progress on controlling the spread of virus. Our study showed that patients who were diagnosed with HIV but otherwise were relatively asymptomatic, or had no ADI or CMVR still were reluctant to start treatment compared to those who already had CMVR. The difference was statistically significant. The relative lack of education and awareness of the disease, and its complications and implications also are factors that contribute to our changing epidemiologic trends. 

In summary, despite aggressive medical therapy, the incidence rate of CMVR in patients with HIV has remained constant over the last 5 years. The median duration of HAART initiation after HIV diagnosis, however, has decreased from 26.0 to 10.0 months and CMVR lesions generally were getting smaller at diagnosis, although patients still were presenting late with a general decline in CD4 counts at HIV diagnosis. A shorter time lag from HIV to CMVR diagnosis had correlated with shorter time to death, supporting earlier reports of CMV disease as a predictor of increased risk of mortality. There still is a need for regular monitoring of surveillance of HIV patients even at a time when HAART is increasingly accessible and available. The changing patterns and presentations of HIV patients with CMVR over the last 5 years, including the early detection of the disease with early institution of anti-CMV therapy to prevent permanent visual loss, complications of immune reconstitution after HAART institution, and concomitant OIs in this group of patients, indicate that a comprehensive ophthalmic surveillance program should be enforced and developed for the local population. 

Footnotes

 The authors alone are responsible for the content and writing of this paper.

Footnotes

 Disclosure: S.C. Teoh, None; P.X. Wang, None; E.P.Y. Wong, None