Introduction

Diabetic maculopathy continues to be the leading cause of new onset vision loss among working age populations.1 The Early Treatment of Diabetic Retinopathy Study (ETDRS) demonstrated that focal or grid laser photocoagulation reduced the risk of moderate visual loss in patients with clinically significant macular oedema (CSMO) by 50% (from 24 to 12%) at 3 years, although visual acuity (VA) improvement was observed in <3% of cases, based on 15-letter gain at 3 years.2 Despite the unsatisfactory outcomes, this treatment remains the gold standard of the treatment for CSMO. Indeed, recent clinical trials conducted by the Diabetic Retinopathy Clinical Research Network (DRCRN.net) indicate that the outcomes associated with macular laser treatment have improved significantly.3,4 Advances in laser technology and optimisation of glycaemia and blood pressure (BP) control have been attributed to these beneficial outcomes.5 Similarly, contemporary studies also suggest that the prevalence of diabetic retinopathy is decreasing when compared with the Wisconsin Epidemiologic Study of Diabetic Retinopathy published in 1984.6 This decline in diabetic retinopathy prevalence is also thought to be because of the enhanced control of systemic factors.7,8,9

For more than a decade, the lessons from the UK Prospective Diabetes Study (UKPDS)10 and the Diabetes Control and Complications Trial11 studies have governed our clinical practise with regard to the management of diabetic retinopathy and macular oedema (DMO). Strict glycaemic and BP control remain the most effective interventions to date. Given that contemporary clinical trials and prevalence data suggest an improvement in visual outcomes and better control of risk factors, we conducted a retrospective study to assess the 5-year visual outcome associated with macular laser photocoagulation (2003–2009) in a clinic-based setting catering to a multiethnic inner city population. We also determined the effect of systemic factors on visual outcomes to evaluate whether similar outcomes are obtained in real-life settings.

Methods

The protocol for this study was approved by the Chair of the institutional review board. The project was also registered in the Clinical Effectiveness Department of the institution. The study adhered to the tenets of the Declaration of Helsinki.

Study population

This study was carried out at the King's College Hospital, London, UK, where an established diabetic retinopathy screening programme caters to a multi-racial community with high levels of social and material deprivation. One-third of the total study population was drawn from Black and Ethnic minority groups. Individuals were graded as having diabetic maculopathy based on post-mydriatic two-field colour fundus photographs. These screen-positive patients were referred to the retinal clinics where a clinical examination and additional investigations (eg, fundus fluorescein angiography) were performed before laser photocoagulation. Optical Coherent tomography (OCT) was not available at baseline examinations.

Study design

Consecutive patients with type 2 diabetes and DMO who required their first macular laser photocoagulation in 2003–2004 were identified from the laser register. In bilateral cases, the first eye treated in each patient was included in the study. In cases in which both the eyes were treated during the same session, the eye with the poorer baseline VA was included. Patients who did not complete the 5-year follow-up were excluded from the study and the reasons for being lost to follow-up were recorded.

Laser photocoagulation

The focal/grid photocoagulation protocols used in the department mirror the DRCRN.net protocols (modified from the original ETDRS protocol).1 In brief, the treatment was performed with a 514 nm green laser light Iridex Oculite GLx (Iridex Corp., Mountain View, CA, USA) with a spot size of 75–125 μm and an exposure time of 100 ms to obtain a light grey–white (just visible) burn and applied in a focal or grid pattern to cover the area of oedema.

The patients were reviewed every 4–6 months, unless they failed to attend an appointment. Laser treatment was repeated if clinical, angiographic and more recently, OCT evidence indicated a persistence of macular thickening. No distinction was made between focal or grid lasers in this study, because in clinical practise, many patients tend to have both on long-term follow-ups.

Visual acuity

VA was recorded using the Snellen VA charts in the early years, followed by the ETDRS charts at 2 m. As a result, all VA recordings were converted to ETDRS scores for this study. The VA examiners were not certified and the VA measurements were recorded in busy clinic settings. Under these circumstances, it is possible that the examiner did not spend enough time to encourage the patient to read as far as possible. As a result, the best-corrected VA may have been underestimated at times. The mean annual visual outcome was defined as the average of all VA measurements recorded per year.

Ocular and medical co-morbidity

All annual clinical data regarding ocular and medical history, including laboratory values, were obtained retrospectively from the electronic patient record, clinical files and laboratory records. Data collected that was related to systemic factors included age at first laser treatment, gender, ethnicity, length of duration of diabetes at baseline, date of initiating insulin therapy, average annual HbA1c levels, mean annual systolic and diastolic BP, number of anti-hypertensive medications at baseline and annually, average annual BMI, history of being on statins, history of cardiovascular co-morbidity, peripheral neuropathy, and foot ulcers.

Data collected that was related to ocular features included mean annual visual outcome, grade of diabetic retinopathy, date of cataract surgery (if carried out), number of macular laser treatments in 5 years, date of initiating pan-retinal photocoagulation (if required), history of any other surgical procedures including date, other ocular co-morbidity, number of retinal clinic appointments in 5 years, and the number of appointments the subject failed to attend in 5 years.

Statistical analyses

The primary outcome measures in our patients (KCH cohort) included the mean annual change in visual outcomes up to 5 years; the 3 year outcomes were compared with the outcomes of the laser arm in the DRCRN randomised controlled study that compared intravitreal triamcinolone acetonide (IVTA) with laser photocoagulation for DMO.12 The last observation carried forward method was used to assign 45 missing values over the 5-year study period. Data were expressed as percentages, mean values (with standard deviations) or median values. In the univariate analyses, we compared each of the variables using t-tests, Mann–Whitney U-test and Fisher's exact test where appropriate. After the univariate analysis, a multivariate logistic regression model of patient characteristics and outcomes was performed to identify the clinical variables associated with gain of vision (ie, losing ≤5 ETDRS letters). To correct for multiple comparisons, results were only included in the multivariate analyses when the corresponding P<0.01.1

Results

The baseline characteristics of the study cohort are summarised and compared with the DRCRN study population in Table 1. The mean age of the patients at study baseline was 68.8 years (range 38–91 years), with 47 (31%) female and 53 (69%) male patients. A total of 201 clinical notes were screened to identify patients who met the criteria for enrolment; causes for exclusion included lack of adequate follow-up (n=54), lost to follow-up (n=32), and mortality (n=15).

Table 1 Baseline characteristics of the KCH cohort relative to the DRCRN laser group12

Visual outcomes

The mean change in VA at 3 years was −4.15 ETDRS letters in the KCH cohort relative to a gain of 5 ETDRS letters in the DRCRN study. In the first year, the percentages of gainers (ie, patients who experienced a loss of ≤5 ETDRS letters) were similar in both groups (73% in the KCH cohort vs 74% in the DRCRN laser group). However, by the third year, only 50% of the KCH group patients were gainers compared with 83% in the DRCRN laser group. The proportion of gainers in the KCH cohort was relatively similar from the third to fifth years after the first laser treatment (47–50%; Table 2). Only 1 out of 10 KCH cohort members gained ≥15 ETDRS letters at year 1, and this result was maintained to year 5. However, in the DRCRN laser group, the number of patients that gained ≥15 ETDRS letters nearly doubled from 14% in the first year to 26% in the third year. The results with the KCH cohort are superior to those of the ETDRS study,2 in which only 3% gained ≥15 ETDRS letters. When we considered the proportion of patients with moderate visual loss at 3 years (loss of ≥15 ETDRS letters), the outcomes with the KCH cohort are inferior (27%) to those of the DRCRN laser group (8%). Taken together, the results of these comparisons show that the visual outcomes of the KCH cohort are inferior to the visual outcomes of the laser group in the contemporary DRCRN study.

Table 2 Annual mean visual outcomes of the KCH cohort compared with the DRCRN laser group outcomes12

Mean number of laser treatments

The mean number of laser treatments over the 5-year study period for the KCH cohort was 2.74±1.6. Table 3 shows the number of laser treatments for the KCH cohort compared with the DRCRN laser group. The mean number of laser treatments performed was less for the KCH cohort, and more patients in the KCH cohort had only one laser session compared with the DRCRN laser group, despite the fact that 60% of the DRCRN group had previous laser treatment and 13% of the DRCRN group had additional treatments other than laser (eg, IVTA and bevacizumab). All of the patients in the KCH cohort were treatment naive and none of the KCH cohort patients received any additional intravitreal treatments. Notably, the proportion of patients having four or more laser sessions in the KCH group was less than that in the DRCRN group.

Table 3 Number of laser treatments in the KCH cohort compared with the DRCRN laser group

Influence of systemic factors on visual outcomes at 5 years

Table 4 shows the mean annual changes in HbA1C and systolic and diastolic BP in the KCH cohort over 5 years in the current era of improved glycaemia and BP control relative to the DRCRN cohort. Although the mean HbA1C and BP values in the KCH cohort improved slowly over the 5-year study period, the overall control of risk factors for the KCH cohort was inferior to the baseline data for the DRCRN laser group.

Table 4 Changes in HbA1C and blood pressure in KCH cohort over the 5-year study period

Univariate analyses of the known risk factors are shown in Table 5. Gainers were defined as those who lost ≤5 ETDRS letters; the rest were termed losers. Insulin users, BMI ≥25, better baseline VA (≥55 ETDRS letters), number of laser treatments (a surrogate marker of severity of DMO), and more number of failed appointments were associated with poorer visual outcome. However, the multivariate model showed that better baseline VA and insulin users were the only poor prognostic indicators (Table 6).

Table 5 Univariate analysis of the prognostic systemic and ocular factors for gain in vision after macular laser treatment for DMO
Table 6 Multivariate model for visual outcome at 5 years

Discussion

Laser photocoagulation remains the standard treatment for patients with CSMO. The main objective of laser treatment is to prevent visual loss, rather than improve vision. Nevertheless, contemporary studies on laser photocoagulation for CSMO indicate that the visual outcomes with macular laser treatment are much better than those obtained with the ETDRS study with 1 in 4 gaining ≥15 ETDRS letters by 3 years.1,13,14 The suggested reasons for this improvement include better glycaemic and BP control and perhaps early detection and prompt treatment of cases compared with a decade ago. However, our study in a clinical setting catering to a multiracial inner city population shows that the long-term results (3–5 years) are inferior to those obtained in clinical trials with 12% showing improved vision and 26% suffering moderate vision loss at 5 years.

We assessed a number of factors that may determine the poorer outcome. These factors included demographics, ocular and systemic factors and issues associated with healthcare provisions. Compared with the DRCRN study (baseline data comparing IVTA to laser for DMO), the KCH cohort was younger and contained more ethnic minority groups. But the treatment outcomes for Caucasians and other ethnic groups were not dissimilar ruling out inequalities to access to health care (data not shown).

The mean HbA1C of our group was 8.5% compared with 7.5% in the DRCRN group. HbA1c levels of ≥8 are associated with an increased risk of macular oedema, irrespective of the ethnic group.15 In a recent report in our population, we found that the risk of diabetic maculopathy independent of the ethnic group is significantly higher in subjects registered with family practices with the lowest quartile of HbA1c achievement.16,17

However, the present study results reflect those of the DRCRN group, indicating that the levels of HbA1C do not influence the outcomes of macular laser treatment. Thus, decreasing HbA1C levels is more important with regard to the prevention of maculopathy than with maculopathy treatment. This finding suggests that over time, other factors such as increased vascular endothelial growth factor levels may dominate the course of the disease.18 Patients in the KCH cohort also had higher systolic and diastolic BP compared with the DRCRN group. Again, this difference may be explained in part by the differential susceptibility of the African–Caribbean group to high BP. However, unlike the ETDRS study, the DRCRN study reported that baseline systolic BP and mean arterial BP did not influence VA outcomes. Despite the higher BP in the KCH cohort, univariate analyses did not reveal BP as a predictive factor. As discussed above with regard to HbA1C, epidemiological studies and clinical trials strongly support hypertension as an important modifiable risk factor for diabetic retinopathy.19 In the UKPDS study, tight BP control reduced the risk of retinopathy progression by about one-third, visual loss by one-half and the need for laser treatment by one-third in patients with type 2 diabetes. Similarly, the EUCLID study,20 DIRECT study,21 and RASS study22 show positive outcomes for anti-hypertensives on retinopathy risk. However, these are risk reduction strategies for the development and progression of DR. Although both HbA1C and BP must be optimally controlled to decrease the rate of incident DR and maculopathy, they do not appear to influence laser treatment outcomes as shown in the current study and based on the analysis of the DRCRN group.1

Owing to the large number of variables evaluated, we only considered associations with a P<0.01 to be significant. Poor prognostic indicators included insulin users, less number of laser treatment sessions, more number of missed clinic appointments, better baseline VA, and BMI ≥25. Nevertheless, only a few of the variables met a P<0.05 value threshold in multivariate analyses; they were the univariants, being on insulin medication and baseline VA. Similar to the analyses of the DRCRN group,1 we found that visual improvement was better in eyes with poorer baseline VA (<55 ETDRS letters). These types of ceiling and floor effects have been reported for treatment outcomes associated with both diabetic maculopathy and age-related macular degeneration.1,23 The duration of oedema may be an important determinant of final visual outcomes, but this factor was not analysed directly in the current study. Nevertheless, the poorer results in year 3–5 may serve as a surrogate marker of chronicity of disease.

Despite the fact that all our patients were treatment naive at baseline, the mean number of laser applications was only 2.7 at 5 years compared with 2.9 at 3 years in the DRCRN group. Although it did not reach a significant level in the multivariate model, the number of laser applications is an important factor that may have influenced our outcomes. The high threshold among retinal specialists to perform more lasers when 2–3 attempts have not shown a positive response should change based on recent data reported by the DRCRN indicating that the probability of improvement of eyes treated previously with laser ≥3 times had a similar chance of VA improvement as eyes that had not had previous laser treatment.4 Taken together, these findings suggest that it is useful to proceed with further laser treatment if there is sufficient space to apply more burns. It is also important to note that the response to laser treatment is slow, and that persistent oedema after one to two laser treatments should not deter physicians from re-treating.

Another significant problem in the real-life setting of urban populations is the lack of awareness of diabetic retinopathy and its associated complications. In all, 22% of the patients in our recent study of urban populations failed to attend screening appointments16 with the highest non-attendance reported among 18–34 year olds. In this study, 50% of the subjects were not followed-up regularly, and 16% were lost to follow-up. Therefore, our results may be worse than reported if the outcomes for the lost patients were known. Our current screening and treatment guidelines ensure that patients with sight-threatening disease are promptly referred and treated. However, the major challenge of providing timely monitoring and treatment appointments for these patients remains unaddressed.

The strength of this study is that it included the largest number of patients with DMO who had macular laser treatments in real-life settings with long-term follow-up, thereby allowing the results to be compared with outcomes from contemporary clinical trial results. However, a limitation of this study is its retrospective nature. Despite the fact that we recruited consecutive patients with 5-year follow-up, 50% of the patients did not complete the 5-year follow-up or did not have at least one annual follow-up visit during this time period. Therefore, we can only postulate that the results may be inferior to our present data if all patients would have been followed. Finally, we did not differentiate between focal and diffuse macular oedema in this study, as angiograms were not available in all cases.

In summary, this study shows that retinal specialists should contemplate further laser treatments in patients with persistent oedema despite potential initial non-responsiveness to laser treatment. Rigorous measures should be initiated to ensure timely follow-up to avoid non-attendance and resultant loss of vision of these high risk individuals.