Background
People with diabetes are at risk of experiencing permanent sight loss because of complications of diabetic retinopathy, including diabetic macular oedema (DMO), proliferative diabetic retinopathy (PDR), and macular ischaemia.1
In DMO (Figure 1) fluid, and at times blood and lipid (fat), leaks from the retinal blood vessels, damaged by the chronically high glucose environment and its consequences, and builds up in the centre of the retina, the macula, the area responsible for central vision. As a result, the macular function is compromised, and the person’s sight is reduced.
FIGURE 1 Infrared and optical coherence tomography (OCT) images of patients with and without diabetic macular oedema (DMO). (a) Infrared image and (b) optical coherence tomography (OCT) scan of the macula (left eye) of a patient with DMO. Fluid is observed as areas or reduced reflectance (arrows) on the OCT scan. For comparison, (c) infrared image and (d) OCT scan of the macula (right eye) of another patient in whom DMO cleared following treatment. The normal depression at the centre of the macula (fovea) is present (arrowhead). (Figure not reproduced here; see the original article.)
Currently, depending on the amount of fluid present [this can be measured using a diagnostic technology called optical coherence tomography (OCT) (Figures 1b and 1d)], different treatment options are recommended for patients with DMO. If the amount of fluid is considerable [i.e. people with a central retinal subfield thickness (CRT), as measured with OCT, of ≥ 400 µm], the National Institute for Health and Care Excellence (NICE) recommends treatment with intravitreal eye injections of drugs; this is known as anti-vascular endothelial growth factor (anti-VEGF) therapy.2,3 Anti-VEGF therapy is required monthly during the first year of treatment for most patients and at less frequent intervals thereafter, until the fluid clears. If the amount of fluid in DMO is less severe (a CRT of < 400 µm on OCT) macular laser treatment is recommended by NICE as the treatment of choice as, for this group, macular laser is as clinically effective as anti-VEGF therapy but less costly.1,2,4 Macular laser is applied in a single session with topical anaesthetic drops; injections of local anaesthetic are not required as it causes no discomfort/pain to patients. The topical anaesthesia is used as to apply the laser, a viewing contact lens needs to be placed on the surface of the eye, so an appropriately magnified and sharp view of the macula is obtained to perform the treatment. Laser sessions may need to be repeated every 3–4 months until the fluid clears.
The first randomised clinical trial (RCT) that showed a benefit of laser treatment for people with diabetes and macular oedema was the Early Treatment Diabetic Retinopathy Study (ETDRS).5 The ETDRS demonstrated that macular laser reduced the risk of visual loss (≥ 3 ETDRS line loss) by 50% at 3 years in patients with what was defined as clinically significant diabetic macular oedema (CSMO).5 The definition of CSMO was based on the presence of the following characteristics on clinical examination (slit-lamp biomicroscopy): (1) thickening of the retina at or within 500 µm from the centre of the macula, (2) hard exudation at or within 500 µm from the centre of the macula provided that it was associated with adjacent retinal thickening or (3) thickening of the retina of the size of the optic nerve head or larger within one disc diameter from the centre of the macula. In the ETDRS only 3% of patients had improvement in vision of ≥ 15 letters, but 85% of eyes included in the study had excellent vision at baseline (Snellen equivalent ≥ 20/40) which may have accounted for the limited visual improvement observed. More recent RCTs on laser treatment for centre-involving DMO showed laser can indeed improve vision, with improvements of ≥ 10 ETDRS letters observed in 32% of eyes at 2 years and in 44% of eyes at 3 years.6,7
It is important to note that centre-involving DMO may not necessarily mean CSMO. In regards to this, trials conducted comparing anti-VEGF therapy with macular laser included people with centre-involving DMO. Centre-involving DMO is defined based on OCT findings. The natural history of centre-involving DMO diagnosed by OCT is less clear than that of CSMO.
Macular laser is also used in patients with DMO who do not fully respond to anti-VEGF therapy. In a randomised trial comparing different anti-VEGF drugs to treat centre-involving DMO, 41–64% of eyes receiving anti-VEGF therapy required macular laser by 2 years following treatment initiation to control the disease.8
In the ETDRS study,5 laser treatment was undertaken using a continuous wave laser. This laser [referred to as ‘threshold’ laser, and here, throughout this report, as standard threshold macular laser (SL)] produces a visible burn in the retina and is considered the ‘standard’ manner in which to perform macular laser. The laser energy when using this SL is predominantly absorbed by one of the layers of the retina, the retinal pigment epithelium (RPE), and converted into heat. Although the mechanism of action of macular laser is not completely understood, it is believed that it has its effect, at least partly, by acting on still viable RPE cells around the area of the burn. Given that heat spreads by conduction, there could be damage to the retinal layers overlying the RPE, including photoreceptor cells (i.e. the cones and rods which are the visual cells of the retina). SL can ‘burn’ the retina and, if applied to the centre of the fovea (the centre of the macula, which is the area of the retina that provides maximal central vision) could cause marked central sight loss. Therefore, this form of laser requires considerable expertise by the clinician administering it as the fovea may not be easily identifiable when thickened by DMO. Ideally, as advised by the ETDRS,5 a fundus fluorescein angiogram (FFA) would be performed to identify areas of leakage that should be aimed at by the laser treatment. However, clinicians may opt to use SL to treat areas of macular thickening as observed on OCT.6,7 Side effects of SL, besides the potential burning of the fovea as explained above, could include paracentral scotomas (areas of loss of sight around the centre) which can potentially affect driving ability, colour vision deficits and formation of scarring over or under the retina (i.e. epiretinal membrane/subretinal fibrosis, respectively). If strong laser is applied close to the centre of the macula, the size of the ‘burn’ (i.e. the area where retinal cells are lost) could expand over time and, even if initially the centre is not affected, central loss of vision could occur over time.
Macular laser treatment can also be delivered using what is called ‘subthreshold’ micropulse laser (SML). In SML, a series of repetitive, very short pulses of laser are delivered, with each pulse of active ‘on’ laser separated by a long ‘off’ time. This ‘off’ time allows for cooling of the retina, preventing the development of a ‘burn’ and, thus, leaving the RPE and overlying neurosensory retina, including photoreceptors, intact. It is believed SML acts by directly stimulating the RPE. As there is no destruction of the retina, this treatment could be applied to larger areas of the retina (not only those with ‘leaking’ blood vessels or thickened by DMO) in a standardised fashion and repeated as many times as needed.
Although the lack of a ‘burn’ in the retina would appear to be of clear advantage, the absence of clinically objective changes in the retina following SML has led to some clinicians/researchers being sceptical of SML having equally beneficial effects as SL. Earlier studies, however, including small RCTs, have shown comparable or superior results using SML. Lavinsky et al.9 showed superiority regarding improvement in vision and reduction in CRT at 12 months following high-density SML in a three-arm trial that included 123 participants with CSMO (n = 42 and n = 39 randomised to high-density and low-density SML, respectively; and n = 42 to SL). Similarly, in a smaller RCT which included 62 eyes from 50 patients with centre-involving CSMO (32 eyes received SML and 30 eyes received SL) Vujosevic et al.10 found no statistically significant differences in vision or CRT between laser groups, but a statistically significant increased retinal sensitivity (i.e. better retinal function) on microperimetry testing following SML. Other trials by Figueira et al. (n = 53 patients with CSMO according to the ETDRS definition),11 Venkatesh et al.12 (n = 33 patients with CSMO), Kumar et al.13 (n = 20 patients with CSMO according to ETDRS criteria) and Laursen et al.14 (n = 16 patients with CSMO), with follow-up of 18 weeks and 8, 12 and 5 months, respectively, found no statistically significant differences in vision and CRT between both types of laser. A more recent RCT by Xie et al.15 which included 84 patients (99 eyes) receiving standard argon laser (n = 49 eyes) or SML (n = 50 eyes) also found comparable results between laser groups.
In a systematic review and meta-analysis by Chen et al.,16 which included all but one (Kumar et al.13) of the RCTs mentioned above, and based on data from 398 eyes (203 eyes in the SML group and 195 in the SL group), SML was found to be statistically significantly superior to SL in terms of mean change of best-corrected visual acuity (BCVA) at 3, 9 and 12 months following treatment, with no statistically significant difference in change in CRT. The largest difference in BCVA between laser arms was observed at 12 months (0.1 log-MAR) favouring SML, but this seemed to be due to the effect of one trial (Lavinsky et al.9), and to the high-density arm of this trial; the difference would not be considered, in any case, clinically important (0.1 log-MAR, equivalent to 5 ETDRS letters).
Qaio et al.17 subsequently published another systematic review comparing SML laser with SL, which included seven RCTs, all six included in Chen et al.’s review16 and, in addition, the study by Kumar et al.,13 with a total of 467 eyes from 379 participants. The primary outcomes were change in BCVA and CRT. Secondary outcomes included contrast sensitivity and retinal damage; neither quality of life nor cost-effectiveness of the treatments were evaluated. Meta-analysis found no statistically significant differences in outcomes at any time point (longest follow-up was 12 months). There was comparable preservation of contrast sensitivity with SML with less retinal damage. The review authors concluded that SML was as good as SL but caused less retinal damage.
The results of the review by Qiao et al.17 differ somewhat from those of Chen et al.16 owing to the choice of arm and outcome from the RCT conducted by Lavinsky et al.9 The review by Qiao et al.17 reported mean BCVA at 12 months with three trials, Figueira et al.,11 Lavinsky et al.,9 and Vujosevic et al.,10 whereas the review by Chen et al.16 reported change from baseline with only two trials at 12 months, Lavinsky et al.9 and Vujosevic et al.10 The Lavinsky et al.9 RCT had two SML arms, one with normal and another with high-density SML treatment, and reported better results with the latter. The Chen et al.16 meta-analysis used only the high-density arm whereas the Qiao et al.17 meta-analysis used only the normal density arm. Their different conclusions were due to the Lavinsky et al. trial,9 deemed to give a significant result in Chen et al.16 (based on change from baseline) but not in Qiao et al.17 (based on mean BCVA at 12 months).
A Bayesian network meta-analysis by Wu et al.18 found no significant difference in mean change in BCVA and CRT between SL and SML. A 2018 Cochrane systematic review19 and meta-analysis by Jorge et al. concluded SML may be as effective as SL, but the GRADE (Grading of Recommendations Assessment, Development and Evaluation) assessment was of low certainty.
All conducted meta-analyses were limited by the inherent limitations of the available RCTs, including the short follow-up (longest follow-up of 12 months). Furthermore, it is unclear what the proportion of participants with a CRT of < 400 µm was in the RCTs included, which based on NICE guidelines would be the participants most likely to respond to macular laser.1,2,4
Recently (and thus not included in any of the systematic reviews mentioned above) a small (68 participants, 34 in each group), very short-term (4 months) RCT by Fazel et al.20 compared SML with SL in people with CSMO and with a CRT of less than 450 µm. Changes in CRT were statistically significantly higher in the SML group than the SL group. Changes in macular volume and visual acuity were only significant in the SML group but not in the SL group.
In summary, published data suggest that SML is comparable, and potentially even superior, to SL, but stronger evidence is required. None of the trials conducted comparing SML and SL included patient-reported outcome measures (PROMs) or the cost-effectiveness of the treatments. Furthermore, outcomes were measured, at the longest, at 12 months. Given the lack of destruction of retinal tissue, SML can be given to larger areas of the macula and repeated as needed, potentially indefinitely. SML may allow for a more standardised delivery of treatment (e.g. using set grids that cover the entire macula), which could, in turn, minimise variability in the quality of treatment delivered, with successful delivery being less dependent on the surgeon’s skills. The lack of deleterious effects when used over the fovea21 makes SML very safe (i.e. sight loss as a result of an inadvertent foveal burn is obviated), potentially facilitating training on its application to junior ophthalmologists and ophthalmic-allied non-medical staff. This would be highly advantageous considering the major problems with capacity currently faced in ophthalmic clinics throughout the world.
On this basis, the DIAMONDS (DIAbetic Macular Oedema aNd Diode Subthreshold micropulse laser) trial was designed as a non-inferiority trial of the efficacy of SML when compared with SL, although the trial was also powered to test equivalence and superiority (if this were to exist). We chose central vision as the primary outcome, as this is very important to people with diabetes and DMO, and set the non-inferiority margin at 5 ETDRS letters (equivalence margin as ± 5 ETDRS letters), as visual changes of this size should not be clinically relevant to patients and could even be due to test/re-test variability.