Retinal manifestations of HIV-1 and HIV-2 infections among hospital patients in The Gambia, West Africa

Introduction

In developed countries, most AIDS patients develop retinal complications. Non-infectious HIV retinopathy characterized by the presence of cotton-wool spots and retinal haemorrhages is the most common, and is found in 30–50% of patients examined in cross-sectional studies ( Jabs 1995; Spaide et al. 1995 ) and in 75–100% of autopsies ( Pepose et al. 1985 ; Harkins 1995). Cytomegalovirus (CMV), which can cause severe necrotizing retinitis and blindness, can be detected in 10–15% of patients at the time of AIDS diagnoses ( Ballinger 1995), and it is the AIDS-defining infection in 2–3% of patients ( Sison et al. 1991 ). A cross-sectional study conducted in New York found that CMV retinitis was present in 34% (58/171) of patients with CD4 counts < 50/μl and in 7% (9/131) of patients with CD4 counts of 51–200/μl. Other studies have reported CMV retinitis in 20% to 40% of AIDS patients ( Holland et al. 1983 ; Jabs et al. 1989 ; Jabs 1995; Sarraf & Ernest 1996; Greenwood & Graham 1997).

There are few reports of the prevalence of retinal complications in sub-Saharan Africa. In a prospective study conducted in Rwanda among 20 AIDS patients between February and July 1984, cotton wool spots were detected in 6 patients (30%), and Roth spots and haemorrhages in 2 patients (10%) ( Kestelyn et al. 1985 ). No cases of CMV retinitis were found. Diagnoses were based on colour photographs of the fundi of patients in whom abnormal fundal findings had been identified by examination. CD4 + count was not tested. A second series of 9 patients with AIDS-related complex were also examined in this study but the number of retinal complications among these patients was very low. In a second study, conducted in Malawi among 99 AIDS patients in 1993, the prevalence of noninfectious retinopathy was 13%. One patient, who also had TB, had necrotizing retinitis in one eye. Diagnoses were by indirect ophthalmoscopy and CD4 + count was not tested ( Lewallen et al. 1994 ).

Both HIV-1 and HIV-2 exist in The Gambia and their prevalence among pregnant women is 0.6% and 1.1%, respectively ( O'Donovan et al. 1996 ); the number of cases of HIV-1 is rising. In order to estimate the prevalence and nature of retinal complications in HIV-infected and AIDS patients in the Gambia, we took photographs of the fundi of a large number of patients who attended hospital, had these reviewed by an independent ophthalmologist, and related the findings to the percentage CD4 + cells. This paper reports the findings.

Patients and methods

Patients were recruited from The Medical Research Council Laboratories, Banjul, The Gambia, as part of their ongoing long-term studies on the natural history and on the pathogenesis of HIV-1 and HIV-2 infections. The Laboratories are the country's main referral centre for HIV. Patients are identified either because they present with symptoms and/or signs suggesting chronic HIV infection or by routine screening in a sexually transmitted disease clinic. The latter are more often asymptomatic.

Testing for HIV-1 or HIV-2 infection was by two type-specific competitive enzyme linked immunosorbent assays (ELISA; Wellcome Laboratories, Dartford, Kent, UK) and by type-specific Western blot (New LAV Blot I and New LAV Blot II, Diagnostics Pasteur, Marnes-la-Coquette, France) or by type-specific peptide strips (Peptilav, Diagnostics Pasteur, Marnes-la-Coquette, France). All patients received appropriate treatment for diagnosed infections such as tuberculosis, but no routine prophylaxis was given for opportunistic infections. Percentage CD4 + cells were estimated using Facscan after staining blood with standard flourescein labelled mono-specific antibodies (Becton-Dickinson, Cockeysville, USA). Because an automated cell counter was not available, absolute numbers of CD4 + cells were calculated from the white blood cell count, estimated in a haemocytometer, and the differential white cell count. This estimate proved more variable than the percentage CD4 + cells, so we chose to analyse our data according to percentage CD4 + cells. The percentage CD4 + cells category for diagnosing AIDS, according to CDC criteria, is < 14% ( Castro et al. 1992 ). The CD4 cell count can be estimated from the percentage CD4 + cells using the formula CD4 cell count per μl = 7.1212 × (CD4%)^1.2914 ( Kidd et al. 1993 ).

During July 1997, all HIV-positive patients over the age of 14 years whose percentage CD4 + cells were < 14% at their last visit to clinic and every second patient from the remaining attendees were invited to join the study. Patients had a basic eye examination and then a detailed fundal examination. Visual acuity was tested by an E-chart, which is used for illiterate people. Five photographs of each retina were taken with a fundus camera (KOWA 3-dx) after the pupils had been dilated with mydiatric drops (tropicamide BP 1% w/v, phenylephrine 10%). All of the photographs were read in November 1997 by a consultant ophthalmologist experienced in interpreting fundal appearances of HIV-positive patients (PF). The ophthalmologist was kept unaware of the patients' clinical status or percentage CD4 + cells.

The study was approved by the Medical Research Council/Gambia Government Ethical committee.

Results

One hundred and twenty-five patients were asked to join the study, of whom one refused and 4 were too ill to be recruited. The patient who refused was well and had 44% CD4 + cells (CD4%). Three of the 4 paients who were too ill to be recruited had had CD4% tests done, resulting in counts of 1, 4 and 14. One patient had HIV-1 while the other three were HIV-2 infected.

Table 1 shows the patient characteristics of the 56 HIV-1, 52 HIV-2 and 12 dually infected patients who were recruited into the study. Thirty-six (30.0%) patients were male. The median time between the date of CD4 test and recruitment into the study was 3 days (range 0–488) days. The CD4% had been determined for 117 patients: 17 (15%) had CD4% < 7; 40 (34%) patients had < 14; and 28 (24%) had > 28. HIV-2 patients were significantly older and had significantly higher CD4% than HIV-1 patients. Twenty-eight of 54 (52%) HIV-1 compared with 6 of 51 (12%) HIV-2 and 6 of 12 (50%) dually infected patients had CD4% below 14. Visual acuity was tested in 118 patients; the other 2 were too ill to be tested. One hundred and fifteen patients had vision better than 6/18 in either eye while the vision of 2 patients was between 6/60 and 6/18 in both eyes.

Table 1. Patient characteristics

Table 2 lists the important fundal findings. Five fundus photographs per eye were taken for 111 (93%) patients. Five (4.2%) patients had 5 or fewer photographs taken from both eyes, of whom 2 were blind in one eye from cataract or injury (their CD4% were 25 and 14), one had an iris scar in one eye (patient's CD4% was 10), and the other 2 were unable to fix their eyes for photography (their CD4% were 46 and 6). The remaining 4 patients had a total of 10 photos missing between them, of whom only 2 had CD4% tests, which were 54 and 21. In addition, 29 photographs were not usable because of poor quality. Thus it was possible to read 1132 (94%) of 1200 photos taken.

Table 2. Abnormal fundal findings among 56 HIV-1, 52 HIV-2 and 12 dually infected hospital patients

No cases of CMV retinitis were found. Four patients had HIV-associated abnormal retinal findings, giving an overall prevalence of 3.3% (95% CI, 0–6.6%). Each of these patients had cotton-wool spots, which varied from 1 to 14 in any one patient. One of them, patient 031, had associated microaneurysm and blot haemorrhages typical of more advanced HIV microvasculopathy ( Figure 1). Three of the patients were HIV-1 infected while the fourth had dual infection. One patient had a current diagnosis of tuberculosis while 2 had had tuberculosis in the past. The prevalence of abnormal retinal findings was 8% (95% CI, 0–16%) among patients who had CD4% < 14 and 12% (95% CI, 0–27%) among patients who had CD4% under 7.

Incidental findings included drusen and inactivated chorioretinal scars in 13 patients and patches of retinal pigment epithelium irregularity in 6 patients.

Discussion

The aim of this study was to determine the prevalence and nature of retinal complications in a range of HIV-1 and HIV-2 infected patients who presented to hospital, particularly those whose CD4% was < 14, as all of these would have had AIDS ( Castro et al. 1992 ). As there were no data relating CD4 to the prevalence of retinal complications in Africa, we decided to also include a proportion of the hospital attendees whose CD4% was 14 or higher. We chose to stage patients according to CD4% rather than according to the WHO or CDC staging criteria, as the application of either is very difficult in African settings. We took fundal photographs of all patients, irrespective of the clinical findings.

We found no cases of CMV retinitis. This finding is consistent with that from a study in Rwanda in which no cases were reported from among 20 AIDS patients studied prospectively for 6 months (Kestlyn et al. 1985). In that study, diagnoses were supported by fundal photographs of any abnormal findings. A cross-sectional study conducted in Malawi, in which diagnoses were by indirect ophthalmoscopy, reported necrotizing retinitis in one eye of one patient among 99 AIDS patients ( Lewallen et al. 1994 ). We have not found any other reports of retinal findings in HIV-infected people from sub-Saharan Africa. In contrast to this very low prevalence in Africa, CMV retinitis has been reported in about 20–40% of AIDS patients in developed countries in a number of studies ( Kupperman et al. 1993 ; Jabs 1995; Spaide et al. 1995 ; Sarraf & Ernest 1996; Montforte et al. 1997 ). The highest prevalence occurred in patients with CD4 counts < 50/μl. In much of sub-Saharan Africa few AIDS patients survive to these low CD4 counts. In the Gambia, for example, the CD4 cell count at the time of death is on average equivalent to approximately 140/μl ( Jaffar et al. 1997 ) and survival is considerably shorter than in developed countries ( Whittle et al. 1994 ). It is also our experience that those who are most severely ill and thus presumably more likely to suffer from retinal complications do not attend hospital and prefer to stay at home. Seventeen of our patients did have CD4% < 6 and 5 had CD4% < 3; these CD4% are equivalent to cell counts of 72 and 29 CD4/μl ( Kidd et al. 1993 ). Thus it is probably reasonable to assume that at least 10 of our patients were severely immunosuppressed; if the risk of CMV retinitis was 30%, then the probability of detecting zero cases of CMV retinitis from 10 patients is just 0.03. Even this is likely to be an underestimate, since African patients have the same level of immunosuppression as Europeans at about a 25% higher CD4 cell count ( Anglaret et al. 1997 ).

There are no data comparing rates of CMV retinitis between African and non-African people living in developed countries. However, among 313 African and 314 non-African AIDS patients living in London, the number of diagnoses of CMV disease, of which retinitis is by far the most common manifestation ( Danner 1995), was 45 (14%) among Africans compared with 75 (24%) among non-Africans ( Amo et al. 1996 ). Taken together, these data suggest that the incidence of CMV retinitis is lower in Africans than in non-Africans, and that among Africans, it is lower among those living in Africa than among those living in developed countries. These differences are likely to be associated with a number of factors including

As well as the low frequency of CMV retinitis, we found that the prevalence of noninfectious retinopathy in Gambian HIV-infected patients was just 3% overall, 8% among patients whose CD4% was < 14, and 12% among patients whose CD4% was < 7%. It is possible that these are underestimates, as 4 very ill patients were not recruited, and fewer than the required photos were taken of two patients who could not focus their eyes. If we assume that all 6 of these patients had fundal abnormalities, then the prevalence of abnormal fundal findings would have been 8% (10/124) among all patients, 14% (6/42) among those whose CD4% was < 14 and 21% among those whose CD4% was < 7. These figures contrast with those from developed countries where a prevalence of 30% to 50% among AIDS patients has been reported ( Jabs 1995; Spaide et al. 1995 ; Sarraf & Ernest 1996). Studies conducted among AIDS patients in Africa have shown varying rates of noninfectious HIV retinopathy: 30% (6/20) in Rwanda and 13% (13/99) in Malawi (Kestlyn et al. 1985; Lewallen et al. 1994 ). Our data suggest that the prevalence of noninfectious HIV retinopathy is probably lower in African patients than in patients living in developed countries.

In The Gambia both types of HIV exist ( Wilkins et al. 1991 ) and our research suggests that survival is shorter and immune depletion faster in HIV-1 patients than in HIV-2-infected individuals ( Whittle et al. 1994 ; Jaffar et al. 1997 ). All cases of noninfectious retinopathy were infected with HIV-1, one of whom was dually infected. This is likely to be due to differences in immunosuppression, as HIV-2 patients were much less immunosuppressed than HIV-1 patients (6/51 of the HIV-2 compared with 28/52 of the HIV-1 patients had CD4% < 14%), or it may reflect differences in the pathogenesis of HIV-1 and HIV-2. A larger study will be needed to answer this question.

In summary, we believe that the prevalence of noninfectious HIV-retinopathy and of CMV retinitis is probably considerably lower in The Gambia than in developed countries generally. Research is now required on the ocular manifestations of HIV so that appropriate treatment and prevention strategies may be devised.

Acknowledgements

We would like to thank Ken Joof, Ramu Njie and Ansumana Fatty for help with the fieldwork, Dr Maarten Schim van der Loof and Professor Richard Hayes for helpful discussions, and Dr Margaret Parker for support. Special thanks to the patients who took part in this study. The study was funded by the London School of Hygiene and Tropical Medicine Trust Fund and by the United Kingdom Medical Research Council AIDS strategic programme.