Exfoliation syndrome (XFS) is an age-related disease common to many populations, but there is a significant geographic and racial variation in its prevalence.1 It is characterised by the progressive accumulation of white, fibrillogranular material within the anterior segment of the eye.2 XFS is a major risk factor for the development of glaucomatous optic neuropathy (GON), especially when it is associated with elevated intraocular pressures.3 Glaucoma associated with XFS, as the Early Manifest Glaucoma Trial demonstrated, is also associated with more aggressive disease progression than primary open-angle glaucoma (POAG).3

Recent differential proteomic and genetic studies have significantly advanced our understanding of the molecular pathophysiology of XFS glaucoma, not least of which is a genome-wide association study of Icelandic and Swedish patients with glaucoma and XFS.4 Thorleifsson et al identified two non-synonymous single nucleotide polymorphisms in the lysyl oxidase-like 1 gene (LOXL1) with a strong association with XFS. However, the high-risk haplotype, as subsequent studies demonstrated, had a similar prevalence in communities with, proportionally, a much lower incidence of XFS.5 The implication of this work is that other factors, whether genetic or environmental, affect the penetrance of the LOXL1 sequence variants.

The World Health Organization’s (WHO) initiative to eliminate preventable blindness—Vision 2020—identified the importance of collecting accurate ophthalmic epidemiological data from Asia in order to optimise the use of limited resources.6 Recent studies from neighbouring communities in Southern India have reported relatively high rates of XFS, and shown an association with glaucoma, cataract and visual morbidity.7^–9 Data regarding the Sri Lankan community are not available. The Kandy Eye Study (KES) was created with the intention of providing a robust epidemiological survey with respect to the Sri Lankan community. As a part of the KES, this study aims, first, to determine the prevalence of XFS in rural Sri Lanka. Second, it aims to describe the correlates of XFS in this community.

MATERIALS AND METHODS

The KES was a population-based, cross-sectional ophthalmic survey of the inhabitants of rural villages in central Sri Lanka. The study was conducted within the rural districts of the Central Province, an area encompassing 5674 km2. Subjects were randomly selected using a multistage cluster sampling process. A sampling frame consisting of the list of all districts and villages in the Kandy area with their populations was obtained from the 2001 Sri Lankan Census. The city of Kandy was excluded, and three districts were randomly selected from a total of 20 districts; within each district a random sample of villages was taken. Households were then randomly selected from each village and all inhabitants 40 years of age and over were invited to participate. The study sample size of the Kandy Eye Study was calculated to estimate with precision the causes of blindness in the community, not specifically the prevalence of XFS. Data were collected between June 2006 and February 2007. A medical and ophthalmic history was obtained from each participant in their own language by qualified healthcare workers who used a standardised questionnaire. Each participant received a comprehensive vision and eye examination, which included: presenting and best-corrected visual acuity (VA) using a front-illuminated illiterate E logMAR acuity chart; Goldmann applanation tonometry (Haag-Streit AT 900, Koeniz, Swizerland); pupil reflex examination; slit-lamp examination of the eyelids, ocular surface and anterior segment; static and dynamic gonioscopy; ocular biometry (Quantel Medical Axis II PR); pachymetry (Quantel Medical Pocket II); dilatation followed by slit-lamp lens assessment using a Lens Opacities Classification System (LOCS) III grading, and stereoscopic fundus examination.10 If >90° of posterior trabecular meshwork was visible, the pupil was dilated with tropicamide 1% and phenyephrine 2.5%. If >90° of posterior trabecular meshwork was visible, the pupil was dilated with tropicamide 1% and phenyephrine 2.5%. An eye with ⩽90° of posterior trabecular meshwork visible was deemed a primary angle closure suspect (PACS), and thus dilated with tropicamide 0.5% only and kept under observation for 4 h. If this was not possible, or if either eye had evidence of previous acute angle-closure glaucoma, then they were examined with an undilated pupil.

XFS was defined as the presence of dandruff like material on the pupillary margin (undilated examination), on the anterior lens capsule (dilated examination), and/or on the trabecular meshwork (on gonioscopy); the presence of a pigmented angle or Sampaolesi line was considered insufficient to diagnose XFS. Visual acuity (VA) was tested monocularly, and two attempts were made per line. Blindness was defined (using WHO criteria) as corrected VA of <3/60 in the better eye. Visual impairment was defined as corrected VA <6/18 but ⩾3/60 in the better eye. Nuclear cataracts were defined as a nuclear opacity LOCS III score of ⩾4. Cortical cataracts were defined by a LOCS III score of ⩾2.

Ethics

The KES had ethics approval from the Royal Adelaide Hospital Ethics Committee. Consent was obtained from the head of each village, and informed consent was obtained from all participants.

Statistical analysis

The prevalence of XFS in one or more eyes was calculated as a ratio estimate incorporating weights for each of the sampled villages (weights were calculated as the inverse of the probability of selection). Univariate analyses were performed to determine whether age, gender, ethnicity, blindness, impaired vision, intraocular pressure (IOP), nuclear and/or cortical opacity of the lens, lens thickness, anterior chamber depth (ACD), axial length, a history of smoking, a history of hypertension or of diabetes (type I or II) were significantly related to XFS. Logistic GEE regression models incorporating weights from the sampled villages were fitted to the data. The GEE portion of the model was used to adjust variance estimates for the dependence in outcomes between eyes of the same patient. Odds ratios (ORs) and 95% CIs for each of the predictor variables were calculated in univariate analyses. All prevalence estimates were performed using SAS Version 9.1 (SAS Institute, Cary, NC).

RESULTS

There were 1721 eligible subjects, and 1375 participated (79.9% participation rate). The mean age of the participants was 57.0 (SD 10.6) years. The mean age of the male participants was 57.9 (10.6) years. The mean age of the female participants was 56.4 (10.5) years. A total of 2734 eyes (785 females, 582 males) had sufficient data to be included in the final analysis. One-thousand and forty-four (87%) participants identified themselves as Singhalese, 112 (8%) as Tamil and 111 (8%) as Muslim.

Of the total eligible population, the prevalence of XFS was estimated to be 1.1% (95% CI 0.5 to 1.5%; 22 eyes). XFS was bilateral in eight subjects and unilateral in six subjects. Table 1 shows the age-specific prevalence.

A global test comparing all four age groups simultaneously demonstrated a significant association between increasing age and the prevalence of XFS (p<0.001). When age groups were also tested separately, a significant increase in the prevalence of XFS, when compared with those between 40 and 49, was only noted in the 70 and above age group (see table 2).

In this study, the mean IOP of those with XFS was 16.1 (3.73), and the mean IOP in those without XFS was 12.9 (2.92). IOP was analysed as a continuous predictor for XFS and found to be statistically significant (p<0.001) in a univariate model (see table 3). The odds of having of XFS increased at a multiple of 1.174 for every 1 mm Hg rise in IOP. So, the OR of having XFS in an eye with an IOP of 22, versus an eye with an IOP of 12, would be 4.98 (ie, 1.17410). Axial length, anterior chamber depth and lens thickness also measured as a continuous predictor were not statistically significant.

Corrected visual impairment, nuclear cataract and hypertension were significant predictors for the presence of XFS using univariate analysis (table 4). Among other factors (see table 4), ethnicity, gender and cortical cataract were not associated with XFS.

DISCUSSION

Although XFS has a worldwide distribution, there is a well-recognised tendency for the syndrome to cluster both geographically and within racial and ethnic subgroups.2 The highest reported prevalence is in the Scandinavian population.11 Conversely, XFS is uncommon in East Asian populations—it has not been reported in the Inuit population, and a rate of less than 0.5% (age 40 and over) was observed in two Mainland Chinese populations.12 13 No prior study of the prevalence of XFS has been made in Sri Lanka. Epidemiological data from neighbouring regions in Southern India report a prevalence (in a population 40 years or older) of 3.0–6.0%.7^–9 Care must be taken with this comparison, as the homogeneity of the Sinhalese population (the largest ethnic cohort in the KES) and populations studied in Southern India is not ensured. This study demonstrates that the prevalence of XFS in rural Sri Lanka (1.1%; 95% CI 0.5 to 1.5) was small in regional terms.

The correlation between XFS and increasing age and increasing IOP is well accepted in the literature, and this study confirms that, in this respect, XFS in Sri Lanka is not atypical.2 Several researchers have suggested an association between XFS and increasing risk of lenticular opacification.14 15 This study has demonstrated a correlation between XFS and nuclear cataracts, but not cortical cataracts. Although this finding is controversial, it is consistent with several other studies reporting a higher frequency of nuclear cataracts in association with XFS.16 17 Recent studies from India, not limiting their analysis to cataract, have highlighted the association between XFS and visual morbidity in general.7 8 Similarly, a significant association was seen with visual impairment.

Deposits of fibrillar extracellular material, similar to that found in the eye, have been identified in specimens of heart, lung, liver, kidney, gall bladder and skin of patients with XFS, prompting research into the potential systemic effects of the disease.18 Several small-scale studies have suggested a relationship between XFS and cardiovascular and cerebrovascular disease.19^–21 These results, however, have not been consistently reproduced; nor has an increase in the mortality of XFS individuals been demonstrated.20 22 In this paper, a correlation was seen between arterial hypertension and XFS. A similar observation has been reported in an Australian population-based study of XFS individuals.19 The importance of our finding is unclear, and drawing a conclusion regarding the systemic complications of XFS is beyond the scope of the study design.

The small number of participants with XFS limited the analysis of associated factors and prevented multivariate analysis, and hence confounding factors such as age, cannot be excluded. Hypertension, IOP and nuclear cataracts are known to increase with age and may therefore not prove to be significant in a multivariate model.23^–25 Further, large-scale multicentre epidemiological research is required to validate potential systemic associations of XFS. Future studies may also involve genetic testing of an XFS cohort for the LOXL1 haplotypes. Of particular interest will be first the correlation between the XFS phenotype and high-risk LOXL1 homozygous haplotype and second the prevalence of the high-risk haplotypes in the non-XFS population. Pairing a focused genetic survey within a broader cross-sectional population-based study may help clarify the environmental contribution to LOXL1 penetrance.

In conclusion, this study has described a prevalence of XFS in rural Sri Lanka of 1.1% (age over 40). To the best of our knowledge this is the first such study in Sri Lanka. Age and IOP have been demonstrated to be strong correlates of XFS in Sri Lanka; accordingly, sufficient consideration should be given to the diagnosis of XFS when these risk factors are present. Correct diagnosis is crucial because of the known aggressive course of XFS glaucoma, the implications for cataract surgery and, as this study has demonstrated, the association of XFS with visual morbidity in general.