Fasting blood glucose levels and the long-term incidence and progression of cataract - the Blue Mountains Eye Study

Purpose: To assess the associations between fasting blood glucose and the long-term incidence and progression of cataract. Methods: A total of 3654 persons aged 49+ years were examined at baseline with fasting blood glucose measured, and 2454 re-examined after 5 and/or 10 years. Lens photographs from each visit were used to assess cataract incidence and progression. Associations between baseline fasting blood glucose and incidence and progression of cortical, nuclear and posterior subcapsular (PSC) cataract were assessed using discrete logistic regression and change-point models. Results: After adjusting for potential confounders, baseline fasting blood glucose was associated with the 10-year incidence of cortical cataract with a threshold at blood glucose level of 6.0 mm (odds ratio [OR] 1.79, 95% confidence interval [CI] 1.25–2.57 for fasting glucose ≥6.0 mm compared to fasting glucose <6.0 mm). Each 1.0 mm increase in fasting glucose was also associated with higher 5-year progression of PSC (OR 1.25, 95% CI: 1.15–1.35), 10-year progression of cortical (OR 1.14, 95% CI: 1.01–1.27) and nuclear (OR 1.20, 95% CI: 1.01–1.43) cataract, with no thresholds detected. Conclusions: In this primarily nondiabetic older population, baseline fasting blood glucose was associated with the long-term incidence of cortical cataract and long-term progression of all three cataract subtypes. Diabetes is a well-established risk factor for the development of cataract, as documented by several clinic-based (Harding et al. 1989, 1993; Szmyd & Schwartz 1989; Leske et al. 1991; Miglior et al. 1994; Donnelly et al. 1995; Tavani et al. 1995) and population-based studies (Klein et al. 1998; Leske et al. 1999; Delcourt et al. 2000; Rowe et al. 2000; Saxena et al. 2004; Mukesh et al. 2006). However, it remains unclear whether the association between fasting blood glucose levels and the risk of cataract is linear (i.e. also evident in persons without diabetes) or has a threshold at certain fasting glucose levels (i.e. evident in persons only with diabetes). Previous studies have shown that intensive blood glucose control in persons with diabetes is associated with a reduced risk of microvascular complications such as diabetic nephropathy, diabetic neuropathy and diabetic retinopathy (Klein et al. 1988, 1994, 2008, Diabetes Control and Complications Trial Research Group 1993; UK Prospective Diabetes Study Group 1998). Given that cataract and cataract surgery represent a significant eye health care burden in both developed and developing countries, any delay in the development or progression of age-related cataract could lead to considerable cost savings. In this report, we aimed to examine the patterns of association between baseline fasting blood glucose and the 10-year incidence and progression of cortical, nuclear and posterior subcapsular (PSC) cataract, in a population-based cohort of generally healthy, older Australians, mostly without diabetes. Details of the Blue Mountains Eye Study (BMES) population and its methods are reported elsewhere (Attebo et al. 1996; Kanthan et al. 2008). In brief, the BMES is a population-based cohort study of vision and common eye diseases in an urban older population comprising two postcode areas in the Blue Mountains region, west of Sydney, Australia. This geographically well-defined area has a stable population, representative of Australia in socioeconomic status and other measures. At baseline examinations (1992–1994), 4433 residents were identified as eligible to participate, of whom 3654 (82.4%) were interviewed and examined. Differences between participants and nonparticipants at baseline are reported (Mitchell et al. 1997). All surviving participants were invited for follow-up eye examinations after 5 (1997–9) and 10 years (2002–2004). Participants who did not return to the 5-year follow-up were also invited to attend the 10-year examinations; 2335 participants (63.9% of the original cohort or 75.1% of survivors) were examined after 5 years, and 1952 participants (53.4% of the original cohort or 75.6% of survivors) were examined after 10 years. Thus, a total of 2564 participants were examined at least once after the baseline examination (70.2%). The study was approved by the Western Sydney Area Health Service Human Research Ethics Committee, and written informed consent was obtained from all participants. At baseline, a detailed demographic and medical history was taken using an interviewer-administered questionnaire. All participants underwent detailed eye examinations. Slit-lamp lens photographs were taken from each eye using Ektachrome 200 colour film (Kodak, Rochester, New York, NY, USA) on a Topcon SL-7E photograph slit-lamp camera (Topcon, Tokyo, Japan) to assess presence of nuclear cataract. Retro-illumination lens photographs were taken using similar film on a Neitz CT-R cataract camera (Neitz Instruments, Tokyo, Japan) to assess presence of cortical and PSC cataract. Blood samples were taken from each participant at baseline to record fasting venous blood glucose level. At the 5- and 10-year follow-up visits, similar questionnaires were used to collect updated demographic and medical history for the past 5 years, and lens photographs were taken. Participants were examined in approximately the same order as at baseline, using the same procedures and equipment. The Wisconsin Cataract Grading System, first developed in 1990 for the Beaver Dam Eye study, was used to perform a masked grading of the lens photographs. Details of this method were previously reported by the BMES and found to have good reproducibility (Klein et al. 1990; Panchapakesan et al. 1997). A five-point scale with one decimal place was used to assess the presence and severity of nuclear cataract. This was performed by comparing participant photographs with four standards of increasing opacity. Nuclear cataract was considered present if the nuclear opacity was worse than Standard 3. The presence and severity of cortical cataract and PSC were graded from Neitz photographs using a circular grid divided into eight equal wedges and a central circle. Graders estimated the percentage area involved by cataract in each of the nine-grid segments. The percentage of opacities in each segment was then summated to give a score for the whole lens area. Cortical cataract was considered present when at least 25% of the lens area was involved. PSC was defined if any was present. Incident cataract was defined as the appearance of nuclear, cortical or PSC cataract subtypes in bilaterally phakic participants without the corresponding cataract subtype in either eye at baseline. Progression of nuclear cataract was defined as an increase by 1.0 or more in the grading score of nuclear opacity. Progression of cortical cataract was defined as an increase by at least 10% of the lens area affected by cortical cataract. Progression of PSC cataract was defined as an increase by at least 2% of the lens area affected by PSC cataract. sas software (SAS Institute, Cary, NC, USA) was used for data analyses. Discrete logistic models estimated odds ratios (OR) and 95% confidence intervals (CI) for associations between baseline fasting glucose and incidence and progression of cataract subtypes. Change-point models were used to assess whether there was a threshold when any associations were found. Table 1 shows a comparison of participants and nonparticipants at the 10-year follow-up examinations. Nonparticipants were significantly younger (p < 0.0001), more likely to be current smokers (p < 0.0001) and to have been diagnosed with diabetes (p = 0.049). They were also less likely to have tertiary level education qualifications (p = 0.041), and to live in their own home (p = 0.0006), than those who participated in follow-up examinations. After adjusting for age, gender, socioeconomic status, smoking, myopia and steroid use, no significant associations were found between fasting blood glucose at baseline and the 5-year incidence of nuclear (p = 0.93), cortical (p = 0.60) or PSC (p = 0.54) cataract. Similarly, no significant associations were found between baseline fasting blood glucose and the 10-year incidence of nuclear (p = 0.29) or PSC (p = 0.48) cataract subtypes. However, after adjusting for these confounders, an overall significant association was found between baseline fasting blood glucose and the 10-year incidence of cortical cataract (p = 0.03). This association had a threshold at a fasting glucose of approximately 6.0 mm. Participants with baseline fasting blood glucose ≥6.0 mm had an 80% higher risk of cortical cataract (OR 1.79, 95% CI 1.25–2.57), compared to those with fasting glucose <6.0 mm. When this analysis was repeated with alternative cut-offs (baseline fasting blood glucose levels 4, 5, 7, 8 and 9 mm), no significant difference was observed between the two groups using these alternative cut-offs (Table 2). Table 3 shows associations between fasting glucose and the 5-year progression of each of the three cataract subtypes. Of 1296 participants, 371 (28.6%) showed progression of nuclear cataract, 154 (8.3%) of 1843 participants showed progression of cortical cataract and 40 (2.2%) of 1837 participants showed progression of PSC cataract. After adjusting for age, gender, smoking, myopia, socioeconomic status and steroid use, fasting glucose at baseline was significantly associated with the 5-year progression of PSC cataract. No threshold effect was observed in this association. For each mmol/L higher baseline fasting glucose, the 5-year risk of PSC cataract progression increased by 25% (OR 1.25, 95% CI 1.15–1.35). The associations between baseline fasting glucose and the 5-year progression of cortical (per mmol/L increase, OR 1.09, 95% CI 0.96–1.25) or nuclear cataract (OR 1.09, 95% CI 0.98–1.22) were not significant. Table 4 shows associations between fasting glucose and the 10-year progression of the three cataract subtypes. Of 671 participants, 365 (54.4%) showed progression of nuclear cataract, 262 (20.5%) of 1278 participants showed progression of cortical cataract and 28 (2.4%) of 1163 participants showed progression of PSC cataract. After adjusting for the above confounders, baseline fasting blood glucose was not significantly associated with 10-year progression of PSC cataract (per mm increase, OR 0.94, 95% CI 0.60–1.45). Nevertheless, baseline fasting blood glucose significantly predicted the 10-year progression of both cortical and nuclear cataract subtypes. For each mm increase in baseline fasting blood glucose, the 10-year risk of cortical cataract progression was 14% higher (OR 1.14, 95% CI 1.01–1.27) and the 10-year risk of nuclear cataract progression was 20% higher (OR 1.20, 95% CI: 1.01–1.43). No threshold effect was evident in these associations. Of the 1319 participants, 494 (37.4%) demonstrated progression of at least one type of cataract after 5 years and 557 (70.6%) of 789 participants demonstrated progression of at least one type of cataract after 10 years. After adjusting for age, gender, socioeconomic status, smoking, myopia and steroid use, progression of any cataract was significantly associated with baseline fasting blood glucose level at both the 5-year (per mm increase in baseline fasting blood glucose, OR 1.13, 95% CI 1.01–1.26,) and 10-year (OR 1.23, 95% CI 1.01–1.48) follow-up visits. (Tables 3 and 4). In this older Australian cohort of a population, primarily without diabetes, we showed that baseline fasting blood glucose was significantly associated with the long-term incidence of cortical cataract, with a threshold at blood glucose level of 6 mm. We also found that fasting blood glucose was associated with the long-term progression of all three cataract subtypes, without a threshold, indicating an effect of glucose levels even within the normal range. An association between diabetes and cataract was initially suggested by some clinic-based case–control studies (Harding et al. 1989, 1993; Szmyd & Schwartz 1989; Leske et al. 1991; Miglior et al. 1994; Donnelly et al. 1995; Tavani et al. 1995) and later confirmed by many population-based studies (Klein et al. 1998; Leske et al. 1999; Delcourt et al. 2000; Rowe et al. 2000; Saxena et al. 2004; Mukesh et al. 2006), including the BMES. A population-based cross-sectional study from southern France reported that known diabetes of 10 or more years’ duration was associated with cortical and PSC cataract (Delcourt et al. 2000). Cross-sectional data from the BMES found that diabetes was associated with an increased prevalence of PSC cataract and cataract surgery, after adjusting for potential confounders (Rowe et al. 2000). In the Barbados Eye Study population, participants with a history of diabetes had a significantly higher prevalence of cortical cataract (Leske et al. 1999). Beaver Dam Eye Study reported that diabetes was associated with an increased 5-year incidence and progression of cortical and PSC cataract (Klein et al. 1998). The Melbourne Visual Impairment Project reported a positive association between diabetes and the 5-year incidence of PSC cataract (Mukesh et al. 2006). Although the association between diabetes and the risk of cataract has been well documented, the exact pattern of the relationship between blood glucose and the development of cataract in nondiabetic populations was unclear. In other words, we are unsure whether the effect of blood glucose on cataract development and progression extends beyond the range of diabetes and is also applicable to persons with blood glucose levels within the normal range. Few studies have examined the relationship between blood glucose and the risk of different cataract subtypes in nondiabetic population samples. The Barbados Eye Study reported that glycated haemoglobin level (a measure of blood glucose levels over the past 8 weeks) was positively associated with the prevalence of cortical cataract (Leske et al. 1999). The Beaver Dam Eye Study reported that baseline glycated haemoglobin predicted an increased 5-year incidence of cortical and nuclear cataract (Klein et al. 1998). We previously reported that impaired fasting glucose (6.0–7.0 mm) was a risk factor for the 5-year incidence of cortical cataract (Saxena et al. 2004). Our present study provides consistent evidence supporting a positive association of fasting glucose and the progression of age-related cataract. We also found that the association between fasting blood glucose and 10-year incidence of cortical cataract had a threshold at blood glucose levels around 6 mm. This threshold level corresponds to the cut-off value for a clinical diagnosis of impaired fasting glucose (IFG). In this study, the 5-year progression of PSC was associated with baseline blood glucose, but this association was no longer present after 10 years. One possible explanation for this observation is the influence of cataract surgery. Nevertheless, when all incident cases of cataract surgery were assumed to be caused by the progression of PSC, which is the most visually disabling type of cataract and most likely to lead to cataract surgery, we still could not document a significant association between baseline fasting glucose level and 10-year progression of PSC (adjusted OR 0.97, 95% CI: 0.84–1.12). On the other hand, the association of blood glucose with the progression of cortical and nuclear cataract was significant only in the longer term, at the 10-year follow-up visit. This suggests that the effect of blood glucose on the progression of cortical and nuclear opacities is likely to be relatively subtle and may take several years for any significant impact to be detected. The findings from our present study have important clinical implications. Previous studies have shown that intensive glycaemic control in people with diabetes is associated with a reduced risk of microvascular complications such as diabetic nephropathy and neuropathy (Diabetes Control and Complications Trial Research Group, 1993, UK Prospective Diabetes Study (UKPDS) Group, 1998) and a reduced incidence and progression of diabetic retinopathy (Klein et al. 1988, 1994, 2008, Diabetes Control and Complications Trial Research Group, 1993). The linear relationship found in our study between fasting glucose and progression of all three cataract subtypes suggests that tight glycaemic control in patients with diabetes may help to slow cataract progression, as well as reducing the risk of other diabetic complications. Further, this linear relationship suggests that fasting glucose may also affect cataract progression in persons without diabetes. Even in this larger group, relatively lower blood glucose levels may help to reduce the burden of cataract surgery in older persons. While further studies are needed to confirm this hypothesis, support may be found from an allied BMES study on the glycaemic index of food consumed and incident cataract (Tan et al. 2007). The exact mechanism by which hyperglycaemia contributes to cataract is not clear but two possible mechanisms have been proposed. Lens fibres are not dependent on insulin for glucose transport. As a result, hyperglycaemia may lead to high concentrations of glucose within the lens fibres leading to an accumulation of sorbitol in the lens. This may cause an imbalance in osmotic equilibrium and subsequent cataract formation (Jedziniak et al. 1981). Previous studies have shown significantly high level of sorbitol in human lens that correlates with the degree of hyperglycaemia (Varma et al. 1979). Alternatively, increased levels of glucose in the lens fibres result in accelerated nonenzymatic glycosylation and subsequent alteration of lens proteins, leading to cataract formation (Kasai et al. 1983). Either or both mechanisms could explain the association between hyperglycaemia and the development and progression of age-related cataract. Strengths of our study include its long-term follow-up of a large population-based sample with reasonable surveillance rates and the use of consistent, masked methods at each study visit to assess cataract. This study has several limitations. First of all, a single measurement of fasting blood glucose at baseline may not adequately reflect long-term glycaemic control. HbA1c gives a better indication of long-term blood glucose levels (over 8–10 weeks) and is mostly used for patients with diabetes. In our study, we did not assess HbA1c as our study sample is a primarily nondiabetic population. Secondly, participants lost to follow-up were more likely to have diabetes than those who returned to follow-up examinations (Table 1). This selective loss of participants with diabetes could have reduced the magnitude of association reported from our study. Thirdly, we did not collect any information about cataract surgery during the study period. In particular, we do not have any data on the type and extent of cataract that led to cataract surgery. We therefore excluded participants who had surgery when we analysed progression of cataract subtypes. Nevertheless, we also performed further analysis by combining incident cataract surgery with 10-year progression of each cataract subtype with the assumption that incident cataract surgery was caused by progression of the corresponding cataract subtype. No significant associations were found for cortical (adjusted OR 1.05, 95% CI 0.95–1.14), nuclear (adjusted OR 1.02, 95% CI 0.93–1.11) and PSC (adjusted OR 0.97, 95% CI: 0.84–1.12) cataract subtypes in these supplementary analyses. Our findings should be interpreted with caution. Although the associations of fasting glucose levels with the incidence and progression of cortical cataract are internally consistent, the associations with nuclear or PSC cataract types are not. The incidence of nuclear and PSC cataract subtypes was not found to be associated with baseline fasting blood glucose level, whereas the progression of these two cataract types showed a significant association with baseline fasting blood glucose. The lack of association between fasting glucose and the incidence of nuclear and PSC cataract could have been because of insufficient study power. Nevertheless, the possibility of a chance finding for some of the observed associations cannot be ruled out, and only confirmation from other studies can help to clarify. In conclusion, in this population-based longitudinal study of a primarily nondiabetic population, we showed that fasting blood glucose was significantly associated with the long-term incidence of cortical cataract, with a threshold evident. In addition, fasting blood glucose significantly predicted the long-term progression of all three cataract subtypes. The study was supported by the Australian National Health & Medical Research Council, Canberra, Australia (Grant Nos. 974159 and 211069).