Clinical effectiveness

Participating sites and characteristics of DIAMONDS trial participants

Table 2 lists the 16 participating sites, the date when they were opened to recruitment and the total number of participants screened and recruited at each site.

Participants

Participant flow

The CONSORT flow diagram in Figure 2 details the flow of patients through the DIAMONDS trial.

FIGURE 2 Participant flow in the DIAMONDS trial. a, Total number of reasons can be greater than number of patients excluded as patients can have multiple exclusion reasons; b, n = 1 patient withdrew permission for their data to be used and will not be included in any analysis going forward. Reproduced with permission from Lois et al.47 This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The figure above includes minor additions and formatting changes to the original figure. (Figure not reproduced here; see the original article.)

Patient recruitment took place between 18 January 2017 and 20 November 2018. A total of 336 participants were assessed for eligibility and 266 (79%) of those assessed as eligible agreed to join the trial and were randomised (intervention, n = 133; control, n = 133). One participant withdrew consent for their data to be used and was excluded from all analyses. The first month 24 follow-up visit was conducted on 25 January 2019 and the final month 24 follow-up visit was conducted on 22 December 2020. Recruitment was initially due to be completed by April 2018 but, at that time, the recruitment of participants had not been completed. An extension was subsequently approved to extend this timeline until December 2018; recruitment was then completed. The cumulative patient recruitment against the anticipated pre-trial sample size is shown in Figure 3.

FIGURE 3 Patient recruitment. (Figure not reproduced here; see the original article.)

Patient baseline characteristics

Patient baseline characteristics were broadly similar across intervention groups (Table 3).

With respect to the trial inclusion criteria, all participants had DMO present in the study eye with an overall mean duration of diagnosis of 2.5 years (SD 4.5 years). In addition, 24.2% (n = 64) of participants had received previous macular laser treatment prior to joining the trial, with a mean number of laser sessions of 1 [median 1, interquartile range (IQR) 1–2] and a mean length of time since the last laser session of 4.2 years (SD 4.8 years). The mean CRT was 329.2 µm (SD 37.3 µm) and the mean BCVA was 80.2 ETDRS letters (SD 8.4 ETDRS letters).

The randomisation minimisation factors were as follows:

• site – sites participating in the DIAMONDS trial are shown in Table 2
• distance BCVA at baseline – 91.7% (n = 243) of participants had ≥ 69 ETDRS letters and 8.3% (n = 22) of participants had 24–68 ETDRS letters at baseline
• previous use of anti-VEGF therapy in the study eye – 11.7% (n = 31) of participants reported previous use of anti-VEGF therapies in the study eye, with a mean number of injections of 8 (median 8, IQR 4–12), and a mean length of time since last injection of 0.9 years (SD 0.9 years)
• previous use of macular laser in the study eye – 24.2% (n = 64) of participants reported previous use of macular laser in the study eye.

The majority of participants were male (70.2%, n = 186) and sex distribution was similar between treatment groups. The overall mean age was 62.2 years (SD 10.3 years). A total of 85% (n = 226) of participants had type 2 diabetes with a mean duration of diabetes since diagnosis of 15.7 years (SD 7.6 years); type 1 diabetes was noted in 14.3% (n = 38) of participants, with a mean duration of 26.4 years (SD 11.4 years) [in one (0.4%) participant ‘other’ type of diabetes was noted in the CRF with a duration of 14.3 years). Most participants were overweight, obese, or morbidly obese (88%), with a mean HbA_1c value of 69.5 mmol/mol (SD 18.4 mmol/mol) or 8.5% (SD 3.8%).

Table 3 presents the mean and SD or median and IQR for continuous data, and the number and percentage for categorical data for the baseline information in the treatment groups.

Treatment after trial entry

Participants were followed up at 4-month intervals (baseline, and at months 4, 8, 12, 16, 20 and 24) for a total of seven visits. Table 4 summarises protocol deviations reported by treatment group for both, number of events and number of patients. There were a total of 83 protocol deviations (4%) reported relating to missed visits out of the total possible number of visits (1855) which affected 58 participants, balanced across both groups [SML, n = 31 (23%) and SL, n = 41 (21%)].

Overall, there were 35 participants not available for the main primary outcome analyses: 12 participants were lost to follow up, 11 participants withdrew consent to continue in the trial but gave permission for their data to be used, 1 was lost to follow-up because of an adverse event, 1 was lost to follow-up because of a serious adverse event, 7 patients died and 2 were lost for ‘other’ reasons. One participant withdrew completely from the trial, also withdrawing permission for any of their data to be used.

Of the 265 participants randomised and included in the analysis, 50% (n = 133/265) of participants were randomised to receive SML and 50% (n = 132/265) were randomised to receive SL. The majority received treatment as allocated, with 0.8% (n = 2) of participants not receiving allocated treatment at baseline: one participant was found to be ineligible and did not receive any laser and another received SL because the local team considered the SML not to be working appropriately.

Treatment after trial entry for the study eye is shown in Table 5 (and for the fellow eye in Appendix 1, Table 24).

A total of 512 protocol deviations were recorded, 255 (49.8%) in the SML group and 257 (50.2%) in the SL group (Table 4); 22 of these were classified as major by the chief investigator and NICTU, 11 (50.0%) in the SML group and 11 (50.0%) in the SL group. The two major deviations classified as ‘study intervention not administered as per protocol’ related to one participant who was randomised to SML but received SL in error at the month 4 visit, and one participant who was randomised to SML but received SL at the baseline visit. The three major deviations classified as “eligibility” related to one participant who met the inclusion criteria and was randomised but when rechecked during the baseline visit was found to have OCT > 400 µm and so laser was not performed; one participant who was randomised but later identified as being treated with pioglitazone; and one participant who was mistakenly randomised with a baseline OCT of 401 µm. The remaining 17 major protocol deviations related to rescue treatment, that is the patient did not meet eligibility criteria for rescue treatment.

The 50 minor deviations classified as ‘randomisation error due to incorrect data entry by site’ related to an issue that was identified with data entry into the Sealed Envelope electronic randomisation system following a reconciliation with the clinical trial database on MACRO version 4.9.1 (Ennov UK, Eaton Socon, UK). It was found that Sealed Envelope contained some data entry errors relating to the minimisation criteria, and it was not possible to amend the data on the Sealed Envelope system. Protocol deviations were recorded on the clinical trial database for all patients affected. These deviations were classed as minor as they did not affect the randomisation process itself. The four minor deviations classified as ‘study intervention not administered as per protocol’ related to one patient who received SML administered with lower power than what was required at one of the laser sessions and three patients who received treatment only on certain areas of the macula, rather than the full grid as advised in the pertinent SOP.

Primary outcome: mean change in best-corrected visual acuity in the study eye at 24 months after treatment

The primary analysis was PP and included all participants who satisfied the PP criteria and had BCVA data at baseline and month 24. Of the 265 participants randomised (excluding the single patient that fully withdrew consent for their data to be used), primary outcome data were available for 87% (n = 231; SML, n = 116 and SL, n = 115).

The primary outcome was the mean change in BCVA in the study eye from baseline to month 24. The difference between lasers in change in BCVA (with a 95% CI) from baseline to month 24 (primary endpoint) was compared with the permitted maximum difference of 5 ETDRS letters (0.1 log-MAR) in the non-inferiority analysis. The mean change in BCVA in the study eye from baseline to month 24 was –2.43 ETDRS letters (SD 8.2 ETDRS letters) in the SML group and –0.45 ETDRS letters (SD 6.72 ETDRS letters) in the SL group. The difference in the mean change in BCVA in the study eye from baseline to month 24 was –1.98 ETDRS letters (95% CI –3.9 to –0.0 ETDRS letters; p = 0.046), which although statistically significant was not clinically relevant. Therefore, SML was deemed to be non-inferior to SL because the lower limit of the 95% confidence interval of the treatment difference (–3.9 ETDRS letters) was above the non-inferiority margin (–5.0 ETDRS letters). Furthermore, SML was also deemed equivalent to SL as the 95% CI (–3.9 to –0.0 ETDRS letters) was wholly within both the upper and lower margins of the permitted maximum difference (–5.0 to 5.0 ETDRS letters). An ITT analysis was also undertaken, which supported the findings from the PP analysis. Table 6 displays the analysis results for the primary outcome for both the PP and ITT analyses.

Figure 4 shows graphically the non-inferiority margin of –5 ETDRS letters; the shaded area represents the equivalence zone. It illustrates that the 95% CIs from both the PP and ITT analyses lay wholly within the equivalence zone and were also above the non-inferiority margin of –5 ETDRS letters.

FIGURE 4 Primary outcome (observed values). Dashed line represents the non-inferiority margin and shaded area represents the equivalence zone. PP difference is –1.98 ETDRS letters (95% CI –3.93 to –0.035 ETDRS letters); ITT difference is –1.96 ETDRS letters (95% CI –3.90 to –0.022 ETDRS letters). (Figure not reproduced here; see the original article.)

In accordance with the statistical analysis plan, the primary outcome was also adjusted for baseline BCVA score, baseline CRT, and previous cataract surgery in the study eye prior to enrolment in the trial as well as the minimisation factors on both the PP and the ITT populations. These results support the findings from the unadjusted analyses in Table 6. The adjusted mean change in BCVA in the study eye from baseline to month 24 was –2.36 ETDRS letters (SE 0.67 ETDRS letters) in the SML group and –0.53 ETDRS letters (SE 0.67 ETDRS letters) in the SL group. The difference of –1.84 ETDRS letters (95% CI –3.72 to 0.047 ETDRS letters; p = 0.056) was found to be neither statistically significant nor clinically important.

Table 7 displays the analysis results for the primary outcome, for both PP and ITT analyses, after adjusting for baseline BCVA score, baseline CRT, cataract surgery in the study eye prior to enrolment in the trial, and the following minimisation factors/covariates: centre, distance BCVA at baseline, previous use of anti-VEGF therapies and previous use of macular laser in the study eye.

A secondary analysis was performed on the subset of participants with both eyes included in the trial (Table 8), including study eye as a random effect within the mixed model. The mean change in BCVA in both eyes from baseline to month 24 was –2.34 ETDRS letters (SE 8.10 ETDRS letters) in the SML group and –0.52 ETDRS letters (SE 6.81 ETDRS letters) in the SL group. The difference of –1.82 ETDRS letters (95% CI –2.42to –1.23 ETDRS letters; p = < 0.001) was statistically significant but is not considered clinically important.

Secondary outcomes

Secondary outcomes reported in this section are mean change from baseline to month 24 in binocular BCVA, CRT, MD of the Humphrey 10–2 visual field in the study eye, percentage of people meeting driving standards; adverse effects, number of laser treatments carried out and additional treatments. EQ-5D-5L, NEI-VFQ-25, and VisQoL scores and cost per QALY gained are reported in Chapter 4.

Results for the secondary outcomes are presented in Table 9. There was no statistically significant difference in the following secondary outcomes: mean change in binocular BCVA (mean difference 0.32 ETDRS letters, 95% CI –0.99 to 1.64 ETDRS letters; p = 0.63), CRT (mean difference –0.64 µm, 95% CI –14.25 µm to 12.98 µm; p = 0.93), MD of the 10–2 Humphrey visual field (0.39 dB, 95% CI –0.23 dB to 1.02  dB; p = 0.21), percentage meeting driving standards (percentage point difference 1.6%, 95% CI –25.3% to 28.5%; p = 0.91), side effects [risk ratio (RR) 0.28, 95% CI 0.06, 1.34; p = 0.11] and additional treatments (percentage point difference –2.8%, 95% CI –13.1% to 7.5%; p = 0.59). The number of laser treatments performed was higher in the SML group (mean difference 0.48 treatments, 95% CI 0.18 to 0.79 treatments; p = 0.002). On closer inspection of the data, this difference appeared to be driven by a small number of participants requiring a larger number of laser treatments in the SML group. Specifically, 13 participants required six or seven sessions in the SML group compared with two participants in the SL group who needed this number of sessions.

Adverse events

A total of 70 SAEs were reported, affecting 46 (17%) participants. Participants randomised to the SML group had 0.8 times the risk of a SAE of those randomised to the SL group (RR 0.8, 95% CI 0.5 to 1.4; p = 0.50).

There were no SAEs reported that were deemed to be related to study treatment (i.e. SARs).

A total of 418 AEs were reported, affecting 157 (59%) participants. Participants randomised to the SML group had 0.9 times the risk of an AE of those randomised to the SL group (RR 0.9, 95% CI 0.8 to 1.1; p = 0.48).

Appendix 1, Table 25 displays safety by treatment group, overall and by the reported System Organ Class. There were no statistically significant differences identified between treatment groups.

There were nine AEs reported that were deemed to be related to study treatments [i.e. adverse reactions (ARs)], affecting six (2%) of participants (Table 10). Participants randomised to SML had 0.5 times the risk of an AR of those who randomised to the SL group (RR 0.5, 95% CI 0.1 to 2.7; p = 0.45)

A total of 161 unanticipated eye-related AEs occurring in each of the laser groups were reported, affecting 38% of participants (n = 102) as detailed in Table 11.

The primary outcome was analysed within pre-specified subgroups for which there was a clinical rationale (centre, distance BCVA at baseline, previous use of anti-VEGF therapies and macular laser in the study eye) by including the corresponding interaction term in the regression model and using 99% CIs. Analyses were carried out to identify whether or not any groups of participants [e.g. those with higher HbA_1c or those having had previous cataract surgery (i.e. those with pseudophakic eyes)] were at high risk of poorer outcomes (Table 12). There was a statistically significant interaction for the centre subgroup analysis (p = 0.013), but this result was unreliable because of wide variability in the number of participants recruited at each centre. No other statistically significant interactions were identified.

Sensitivity analyses

Sensitivity analyses were undertaken to assess the impact of missing data by imputing extreme values (lowest and highest) and last observation carried forward; the impact of including patients who were not treatment naive (i.e. excluding those who had had previous laser treatment for DMO in the study eye or previous anti-VEGF therapy for DMO or PDR in the study eye); the impact of including patients who had received cataract surgery in the study eye prior to entering into the trial (i.e. those with pseudophakic eyes); and the impact of using month-24 data collected outside ± 14 days of the due date (Table 13). These results support the findings of the unadjusted analyses in Table 6, and the lower limit of the 95% CIs for each of the treatment differences in these sensitivity analyses lie above the non-inferiority margin (–5.0 ETDRS letters) and within the equivalence margins (± 5.0 ETDRS letters).