Article Text


Postmortem histological survey of the ocular lesions in a British population of AIDS patients
  1. I Pecorellaa,b,
  2. A Ciardib,
  3. A Garnera,
  4. A C E McCartneya,
  5. S Lucasc
  1. aInstitute of Ophthalmology, Department of Pathology, Bath Street, London EC1V 9EL, UK, bDepartment of Experimental Medicine and Pathology, University “La Sapienza”, Viale Regina Elena, 324-00161 Rome, Italy, cDepartment of Histopathology, St Thomas's Hospital, Lambeth Palace Road, London SE1 7EH, UK
  1. Dr Irene Pecorella, Dipartimento di Medicina Sperimentale e Patologia, Università degli Studi “La Sapienza”, V le Regina Elena, 324 00161 Rome, Italyirenepecorella{at}


AIMS To study ocular pathology and systemic correlations in a series of 73 postmortem eyes from British patients who died from AIDS before the introduction of a HAART regimen.

METHODS The eyes were studied with conventional histology, special histochemical stainings, and immunohistochemistry.

RESULTS 72.6% of the cases showed chronic uveal inflammation, caused by opportunistic agents in 37.7% of them (cytomegalovirus (CMV) in 30.1%,C neoformans in 5.6%, and Gram positive bacteria in 1.8%). Cytoid bodies were noted in 10/73 eyes, three linked to CMV retinitis. Six retinal haemorrhages, four of which were secondary to CMV, were found. 14 specimens (19.1%) showed foci of calcification, and a further 11 (15%) calcium oxalate deposits. In no cases were the calcific deposits suspected clinically. Six eyes (8.2%) did not show any abnormality.

CONCLUSIONS CMV retinitis is the most common (28.7%, 21/73) ocular infection in this series and may occur either during or in the absence of systemic dissemination. Conversely, ocular cryptococcosis appears to be an epiphenomenon of systemic and CNS disease. No other opportunistic ocular infections were present in this series. Interesting findings were the presence of intraocular precipitates of calcium oxalate and calcium phosphate or carbonate in a significant number of cases (15% and 19%, respectively), and the high prevalence of idiopathic uveal inflammation (43.8%).

  • eye
  • post mortem
  • histology
  • AIDS

Statistics from

Ocular involvement in the acquired immunodeficiency syndrome (AIDS) is widely recognised. The introduction of a highly active antiretroviral regimen (HAART) has not only improved survival of AIDS patients, but has induced a dramatic downturn in the number of cases of cytomegalovirus (CMV) retinitis1 and other ocular infections. Before the HAART era, most commonly, ocular lesions developed late in the course of the infection, with the effect that the increased survival rates resulting from improved therapeutic measures were accompanied by an increased incidence of ocular disease.2 The prevalence of different populations at risk for AIDS varies between countries, as does the endemicity of the opportunist pathogens responsible for many of its ocular complications, which means that incidence rates from different centres are not necessarily comparable. Moreover, variations in clinical diagnostic criteria can add a further degree of uncertainty.

To date there are only a few histopathological studies of the postmortem ocular findings in AIDS patients,3-6 and only two of them3 5 correlate these findings with the systemic involvement of infections or tumours. We now present findings at postmortem examination in a British population composed exclusively of homosexual or bisexual males. We believe that a documented clinicopathological study of the ocular lesions observed in a patient population before the HAART era might help in the comparison of the two groups of patients and provide a baseline for measuring the efficacy of treatments. Furthermore, because our series is composed exclusively of homosexual individuals, the data we provide are different from other histological studies from other countries with a different prevalence of risk factors.

Material and methods

Seventy three subjects with clinically and immunologically confirmed AIDS underwent a full postmortem examination between 1989 and 1994 at University College Hospital, London with histopathological study of one eye (in two instances both eyes were processed). The eyes were selected according to either a clinical history of ocular disease or randomly, when such evidence could not be obtained. The eyes and a short segment of the intraorbital portion of the optic nerve were fixed in formalin, before being opened horizontally at the superior edge of the cornea and examined macroscopically with a dissecting microscope. Unless the calottes of the eyes appeared to be macroscopically abnormal, only the central portions of the eyeballs were processed to paraffin embedding, and sectioned at three or more levels before staining with haematoxylin and eosin, periodic acid Schiff, Gram stain, mucicarmine, Ziehl–Neelsen, Warthin–Starry, and Gomori's methenamine stain. In cases with microscopic features consistent with a possible calcium salt deposition, von Kossa's and Yasue's stains for calcium phosphate and oxalate, respectively, were also performed. Immunohistochemistry, using a standard biotin-streptavidin-peroxidase technique, was used in all of the cases to demonstrate the presence of HIV p24 antigen (Dupont), whereas searches forToxoplasma gondii (Biogenex), herpes simplex virus types 1 and 2 (Dako) and paramyxovirus (antibody kindly donated by Dr A Warnes), and in situ hybridisation with a probe for cytomegalovirus (CMV) (Enzo Diagnostics) and for varicella zoster virus (Keratech Biotechnology BV) were undertaken in cases with consistent histological findings. In five instances transmission electron microscopy was performed.

All the 73 patients were male and homosexual (two bisexual) with overt AIDS, while one was also an injecting drug user. Five of the subjects originated from Barbados, Jamaica, and Trinidad (one each), and Spain (two). Apart from those coming from the West Indies, all the cases were white. Ages ranged from 26 to 59 years (mean 37.8 years). The notes were available for 47 patients and were reviewed retrospectively with regard to the ophthalmological findings in the last year of life.



In the two instances in which both eyes were examined no difference was detected on the two sides and the results are presented with respect to the number of cases rather than the number of eyes examined.

The most frequent finding was uveal inflammation in 53 of the 73 cases (72.6%). In seven it was confined to the anterior uvea, in 12 to the choroid, while in 34 it involved the entire uvea. Generally, the inflammation was of mild or moderate severity, showing focal aggregates of mononuclear cells in the uveal interstitium, but in four there was a severe panuveitis. The inflammation appeared to be secondary to CMV infection in 16 instances, to Cryptococcus neoformans infection in three, and Gram positive bacillary infection in one. Although one patient with severe panuveitis and associated necrosis of the entire retina and acute keratitis had had a history of shingles affecting the ophthalmic branch of the trigeminal nerve and causing ophthalmic arteritis, no aetiological agent could be demonstrated in the remaining 32 cases (Table 1). Five cases of retinal necrosis (retinitis) attributable to CMV infection were identified in the absence of uveitis. Cryptococcal infection was invariably mild with occasional organisms located next to the choriocapillaris (Fig 1) and rarely in the retina: a mild to moderate granulomatous uveal response in the site of cryptococcal invasion was observed in three cases, but retinal inflammation was not seen. Special stains in these three cases detected no other associated intraocular infections. Solitary cryptococcal involvement of the optic nerve meninges in the absence of inflammation was noted in one instance. A few yeasts suggestive ofHistoplasma capsulatum were seen in the ciliary body and blood vessels of the anterior choroid in one instance. Three eyes (4.1%) showed mild stromal keratitis of obscure cause.

Table 1

Non-infectious and non-neoplastic histological ocular features in our series of AIDS patients

Figure 1

Note multiple cryptococci (arrows) in the lumina of the vessels in the choriocapillaris. The organisms are stained dark red with mucicarmine and this allows us to distinguish C neoformans from Histoplasma (mucicarmine, ×400).

Cytoid bodies were a feature of the retinas in 10 patients (13.6%) (Table 1), three of whom also had CMV retinitis (Fig 2A, B). In one of the cases with associated CMV retinitis, the cytoid bodies were adjacent to the foci of vasculitis but in the other two there was no such clear relation. There were six cases (8.2%) of single or multiple retinal haemorrhages, four of them associated with CMV infection in other parts of the retina. While the CMV related retinal bleeding affected both the superficial and deep parts, other haemorrhages were entirely superficial. A solitary example of subconjunctival haemorrhage was also noted. Organised thrombi were observed in an occasional blood vessel of the retina and ciliary body of two cases (2.7%) without CMV infection and another two showed focal coagulative necrosis of the retina without demonstrable circulatory deficiency (Table 1).

Figure 2

Histologically, cytoid bodies (A) (arrows) can resemble CMV infected cytomegalic cells (B) (arrowheads). However, the latter have a clear halo around the nuclear viral inclusion and the cytoplasmic borders are often recognisable (haematoxylin and eosin, ×250).

Ocular calcification was a feature of 14 eyes (19.1%), affecting the stroma of the ciliary processes in eight instances, the sclera in five, the peripheral choroidal blood vessels in two, the cornea in three (Fig3A, B), the bulbar and palpebral conjunctiva in one, Bruch's membrane in one, the optic nerve body and an extraocular muscle in another one. A giant drüsen of the optic nerve head and calcified drüsens of the peripheral retinal pigment epithelium were also observed. The deposits were amorphous, basophilic, and stained with von Kossa's method. Also observed were 11 eyes (15%) with focal or more widespread greyish crystals located in the walls of the retinal blood vessels and deposited on the inner limiting lamina of the retina. Similar deposits were also seen in the retinal pigment epithelium of one case, along the surface of the ciliary processes and the ciliary body in two cases, and in another they involved the sclera and anterior orbital soft tissues and palpebral blood vessels. On the basis of polarising microscopy and Yasue's staining method the crystals were identified as calcium oxalate (Fig 4A, B). Of the 24 eyes with calcific deposits, 18 showed other microscopic abnormalities (see Table2).

Figure 3

Two examples of corneal calcification: histology disclosed a rather localised bilateral mid stromal deposit in this 34 year old bisexual AIDS patient. The corneal epithelium (asterisks) is artefactually lost (A). Calcification was more superficial and diffuse in another 44 year old homosexual individual, and included focal groups of basal epithelial cells (arrows). Other deposits were observed in the sclera, optic nerve body, and iridociliary pigmented epithelium of this patient (von Kossa's stain, ×250 and ×25, respectively).

Figure 4

Oxalate crystals were deposited mainly on basal membranes. In (A), the oxalates outline the contour of the ciliary processes (×100, crossed polarised light). In (B), they involve the wall of the retinal blood vessels. Deposits are also seen in the vitreal cavity (asterisk), next to the nerve fibre layer (Yasue's stain, ×25).

Table 2

Sites of calcium (Ca) deposition and associated ocular microscopical abnormalities

Other miscellaneous findings included pingueculae in eight cases, elastotic degeneration of the forniceal and palpebral conjunctiva in one, and benign choroidal melanocytic naevi in three. These lesions are possibly purely coincidental, although the finding of stromal elastosis in an unexposed region of the conjunctiva is unusual and pinguecula is relatively rare in Britain.7 Six eyes (8.2%) failed to show any abnormality (Table 1).


Thirty nine of the 47 patients whose notes were reviewed had had an ophthalmological examination in the year before their death.

The clinical notes were available for 18 of the 32 patients diagnosed histologically as having idiopathic uveal inflammation. However, there was no objective abnormality in 15 patients who were examined ophthalmoscopically: the remaining three had no clinical complaint and were not examined. It should be noted that only eight of the 18 patients had been examined in the last month of life. Of the two patients with severe idiopathic panuveitis, one was blind and had associated evidence of extensive retinal necrosis; the other was reported as having a normal fundus when examined 1 month before death, but subsequent histological study revealed a calcific giant drüsen of the optic disc, a calcified posterior scleral plaque, and focal calcification of the optic nerve and an extraocular muscle, as well as elastotic degeneration of the palpebral conjunctiva. The causes of death for patients with idiopathic uveal inflammation are listed in Table 3.

Table 3

Postmortem findings in patients with uveitis of unknown origin

The histological findings confirmed the presence of CMV retinitis in 14 patients diagnosed antemortem and also showed that there were small foci of activity in two cases considered to be clinically quiescent. Of the remaining seven patients with microscopic CMV retinitis, one had not been examined, one had been seen before death and the fundi not visualised, and another one with only choroidal CMV infection showed no abnormalities at funduscopy. Four patients had no clinical notes available.

No abnormality was recorded in the three patients with histological evidence of corneal calcification when they were examined 1 month before death.

The notes of seven of the 11 cases with posterior segment calcium oxalate deposition were available: five had been examined ophthalmoscopically, and three of them on the last admission, but in none was oxalosis suspected.


The overall postmortem histology of the 73 patients was reviewed and data relating to the ocular histology in those with identified systemic disease or with non-ocular neoplasms are detailed in Table4.

Table 4

Histologically proved systemic infections and neoplasms (%) and concomitant ocular involvement in our series compared to two other studies

Systemic CMV infection was recognised in 42 (57.5%) cases and of these 15 had associated ocular involvement. Thirty two patients had terminal bacterial infections, seven having septicaemia, but ocular pyogenic bacterial disease (necrotising iritis) was observed in only one instance. Four of five patients with disseminated cryptococcosis had ocular or optic nerve sheath involvement. Two patients had herpetic ulcers of the upper respiratory tract and skin, respectively, but in neither case were the eyes involved. Other systemic infections were candidiasis (23 cases), aspergillosis (two cases), histoplasmosis (one case), mycobacteriosis (19 cases), pulmonary and extrapulmonary pneumocystosis (18 cases), and cerebral toxoplasmosis (one case). In none of these were the eyes implicated.

Twenty five patients died with Kaposi's sarcoma: 20 had cutaneous lesions, including two with eyelid involvement. No conjunctival tumours were seen although one patient had a subconjunctival mass which histology showed to be a large haemorrhage. Ten patients developed lymphomas but without ocular or orbital involvement.


This is the largest European histological study performed on eyes removed at necropsy from AIDS patients and includes 73 British cases with complete postmortem and clinical data. The benefit of histopathological examination is that it adds to the reliability of purely clinical diagnosis, but at the time of writing only few other such series have been documented. Two from the USA5 8described the ocular findings at post mortem in 35 and 206 patients, respectively, and another three were reports of 25 French,3 43 Danish,4 and 48 Swedish2 cases. However, three of these studies2 4 8 do not present full data on the systemic findings in their patients.

In respect of CMV retinitis, cryptococcosis, and toxoplasmosis, the incidence of ocular infection by the various pathogens encountered in the three complete series, including the present one, reflects the incidence of systemic involvement, albeit at lower levels (Table 4). Thus, a 57.5% incidence of systemic CMV infection in the present study was associated with a 20.5% retinal involvement, which compares with the corresponding figures of 71.4% and 34.2%, respectively, in the American series5 and of 36% and 20%, respectively, in the French study.3 In all three series CMV retinitis was much the most common ocular infection in AIDS sufferers and, as in three (12%) of the French patients, in six of our cases (8.2%) it was apparently confined to the eye. The prevalence of CMV retinitis in homosexual men does not appear to differ from other series of patients with different risk factors, as shown by other studies where a prevalence of 28%9 and 31%10 were recorded.

By contrast, ocular cryptococcosis in AIDS patients was never seen in either the present series or that from the USA in the absence of systemic or CNS involvement, which suggests that the eye is unlikely to be a primary infection site. However, relative to the levels of ocular infection accompanying systemic candidiasis, aspergillosis, and other mycoses there may be an enhanced risk that AIDS patients with systemic cryptococcal infection will succumb to ocular complications (four out of five cases in the present study and two out of five cases in the American series: an overall incidence between the three series of 50%). The low level of ocular cryptococcosis in AIDS patients as a whole (six of 133 cases (4.5%)) appears to be a reflection of the prevalence of systemic infection. Although C neoformans has been shown to destroy the anterior visual pathways but only produce minimal inflammation in the process,11 in our three cases with intraocular cryptococcal infection and no other micro-organism detectable at the special stains, a mild to moderate choroiditis was observed.

Rare instances of ocular candidiasis in association with AIDS have been recorded12 and infection appears to be more likely in injecting drug user with candidaemia. The low number of injecting drug users in our series probably explains the absence of ocular infection by Candida orAspergillus spp, despite the moderately high incidence of systemic infection of the former in both the British and American studies. There are a few well documented reports of ocular histoplasmosis13 and, contrary to our experience, of pneumocystosis14, toxoplasmosis,12 15tuberculosis,16 and Mycobacterium avium intracellulare choroiditis17 in AIDS patients in the literature. They are, nevertheless, as all three postmortem studies suggest, uncommon infections. The case of histoplasmosis reported by Specht13 indicated a long delay between systemic and ocular infections and, given the rapid death of some AIDS patients, this may possibly contribute to a lowered incidence of ocular involvement. It may be relevant that one of our patients who eventually died from disseminated histoplasmosis had only a very few yeasts in the choroid with minimal evidence of an inflammatory response. Infection in the solitary case of toxoplasmosis in our series was confined to the brain, thus resembling the American experience,5 and although the French study3 revealed a much higher incidence of cerebral involvement, there was only one case in which the eyes were affected. This is interesting since a retinochoroiditis incidence of 3% is cited in respect of AIDS sufferers on clinical grounds.14 Possibly the reliability of such diagnoses should be questioned.15 Moreover, a large neuropathological study of 269 AIDS cases revealed CNS toxoplasmosis in just 2.6%,18 which taken with the observation that only 10–20% of those who have intracranial disease can be expected to incur ocular infection,10 suggests that the true incidence is considerably lower. The absence of ocular pneumocystis infection in any of the three histopathological series despite a moderately high level of pulmonary involvement may reflect the infrequency of extrapulmonary dissemination overall (demonstrated in only three of the 18 patients with lung infection in the present series). It should be noted that several of our patients had been treated with prophylactic aerosolised pentamidine therapy. Nevertheless, even in patients with no prophylactic therapy, the incidence of Pneumocystis carinii choroiditis appears to be low, as shown by Morinelli's study,8 where only four out of 20 (1.9%) AIDS patients examined post mortem had suffered from ocular pneumocystosis (data from systemic involvement not reported).

Uveitis in the absence of an identifiable pathogen was a feature of 32 eyes. The possibility that the affected patients had a directly HIV related inflammation19 20 must be allowed. On the other hand, immunohistochemical staining for the p24 antigen was consistently negative. A response to circulating immune complexes is another possibility. The current study includes patients dying before the availability of protease inhibitor combination regimens, which are known to cause an ocular inflammatory syndrome characterised by uveitis and cystoid macular oedema, as a result of restored specific immune reactivity. Recently, HIV, but no other micro-organism, was isolated in cultures of aqueous or vitreous from eyes of patients with idiopathic HIV associated uveitis. In these patients, symptoms resolved promptly with systemic zidovudine, strongly suggesting that HIV could have been the causative agent for their ocular inflammation.21Differently from Levinson et al,22 we observed no associated retinal infiltrates in the eyes of patients with chronic uveitis of unknown origin.

American experience gives an incidence for Kaposi's sarcoma in 10–24% of AIDS cases,6 but in the 133 patients studied histologically in the combined British, French, and American series there were only three cases of ocular (conjunctival) involvement. Diagnosis can be difficult in the absence of histological confirmation due, as in one of our cases, to the possibility of confusion with a localised haemorrhage.

We found no instances of intraocular lymphoma in our series, although five (6.8%) had a non-Hodgkin's lymphoma in the brain. De Girolami and colleagues described a single instance in association with a CNS tumour.3 Overall, however, the incidence of intraocular lymphoma in AIDS patients appears to be very low.

In our study we noted focal collections of cytoid bodies, the histological counterpart of cotton wool spots, in 10 eyes, three linked to CMV retinitis. As such they represented the outcome of focal retinal ischaemia and can reasonably be attributed to occlusion of a terminal retinal arteriole. Where they were associated with CMV infection vasculitis could be implicated. Those occurring as part of a retinal microvasculopathy in the absence of CMV or other demonstrable infection are less easily explained but proffered mechanisms include vascular occlusion by circulating immune complexes, increased blood viscosity secondary to raised plasma fibrinogen levels, or even a direct effect of HIV on the vascular endothelium. The retinal haemorrhages observed in four cases with CMV retinitis and two without could also be the outcome of vasculitis. Other clinically described features of retinal vasculopathy such as capillary closure, vascular attenuation, and microaneurysms are not readily recognised in conventional histological sections and their absence in our findings is not significant.

An unexpected and previously unreported finding was the calcification affecting 14 eyes. Calcification of the sclera and ciliary processes is common in elderly subjects, but it is difficult to implicate senescence in generally young AIDS population. Nevertheless, the pattern and character of the deposits in most instances closely resembled those associated with ageing and an accelerated degenerative process seems likely, although the stimulus is obscure. The three patients with substantial corneal deposits, however, are not explained on this basis since the cornea does not normally share the sclera's predisposition to age related calcification. Furthermore, should dystrophic calcification affect the cornea for whatever reason in non-AIDS individuals it primarily involves Bowman's zone and the superficial stroma, whereas in the AIDS patients it was mainly located in the deeper lamellae with sparing of Bowman's zone. The patients' routine investigations indicated normal plasma calcium and phosphate levels. Even stranger was the finding that a further 11 eyes contained calcium oxalate crystals, although it might be noted that oxalates can occasionally present in the lens and in relation to long standing retinal detachments in non-AIDS patients as a presumed degenerative phenomenon. Although two of the 24 patients with ocular calcification or oxalosis also had CMV retinitis, none had received foscarnet, which has a propensity to disturb calcium balance.


The authors wish to thank Mr Ian Rhodes for his technical support. Dr Pecorella was given a grant by the Istituto Superiore di Sanità (Rome, Italy).


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.