Objectives
- To explore whether intravitreal aflibercept compared with PRP causes measurable regression of retinal NV at 12 and 52 weeks in terms of decimal disc area units in 4-field colour photographs and FFA: difference in means.
- To explore differences in the mean change in retinal vessel calibre and oxygen saturation in eyes treated with intravitreal aflibercept compared with PRP at 12 and 52 weeks: difference in means.
- To explore whether intravitreal aflibercept reduces angiographically quantifiable areas of retinal non-perfusion compared with panretinal photocoagulation through 52 weeks: means and proportions.
Participants for the mechanistic study
The mechanistic substudy was carried out on 40 patients recruited at Moorfields Eye Hospital. These patients also underwent retinal oximetry in addition to the other study assessments carried out at baseline, 12 and 52 weeks. A total of 20 patients were in the aflibercept arm and 20 were in the PRP arm. A total of 20 patients were treatment-naive and 20 patients had been previously treated with PRP but had persistent new vessels at baseline.
Retinal oximetry
Retinal oximetry was performed using the retinal oximeter (Oxymap T1 device connected to Topcon TRC50-DX fundus camera; Oxymap ehf., Reykjavik, Iceland).78
The retinal oximeter consists of a fundus camera with an attached image splitter as well as a digital camera that captures images at two wavelengths, 605 nm and 586 nm. Optical density is sensitive to oxygen saturation at 605 nm but not at the reference wavelength of 586 nm. The oxyhaemoglobin saturation (SO2) of a vessel can therefore be calculated because the optical density ratio at these wavelengths has been shown to have an approximately inverse linear relationship with SO2. Although the saturation data are not absolute, the oximeter has been shown to give reproducible results and to be sensitive to changes in oxygen saturation. The comparison of data in the same eye before and after treatment is therefore valid.79
Optic disc-centred images were captured through dilated pupils. The images obtained were 1200 × 1600 pixels and covered a 50° field of central retina. The images were captured for both eyes but only the study eye was used in this analysis. Images were analysed by two independent observers (Luke Nicholson and Roxanne Crosby-Nwaobi) using the Oxymap Analyser software (Oxymap ehf., Reykjavik, Iceland) version 2.5, which automatically detects vessels > 8 pixels in diameter. The area analysed was selected by centring quadrant lines on the optic disc. An initial central circle was used to delineate the optic disc. Two additional measurement circles, an inner and outer circle, two and four times the diameter of the central circle, respectively, were then demarcated. The area between the circles centred on the optic disc was analysed. The width of this ring was two disc diameters (DDs). Vessels beyond the area of analysis, certain areas where vessel detection would prove inaccurate (branching, overlapping or intersecting) and segments of vessels < 19 pixels in length were excluded. Vessels to be measured were selected manually. Measurements were first taken to yield arteriolar SO2 (SaO2) by quadrant. This was repeated for venules to yield venular SO2 (SvO2) by quadrant. Overall saturation was computed by averaging saturation values of each quadrant. Because Oxymap software measurements are calibrated to non-diabetic young individuals, results are relative to that calibration, occasionally resulting in SO2 measurements > 100%. These values were not truncated to 100%, per established oximetry protocol. The retinal oximetry data streams were extracted from the source data file. The data streams underwent multistep preprocessing to eliminate missing or invalid data. The data were inspected for (1) interrupted regions of the recording (as noted in the research record) and (2) regions where it was not able to properly measure saturations. The entire data epoch was rejected if the data stream failed one or more of these checks or if continuous measurements were not available. Data collected from the Oxymap software included the mean oxygen saturation in the identified segments of first and second branch retinal arterioles (SaO2) and venules (SvO2) within 0.5–1.0 optic DDs from the disc margin, and their difference, the Sa – vO2. The mean arterial and venous diameters were also recorded.
Ultrawide field colour fundus photography
All 40 patients underwent CFP and fluorescein angiography using the Optos 200TX (Optos Plc, Dumfermline, UK) ultrawide field system.
The 200° images were captured with steering, that is aligning images with three axes, eyes looking at centre, superiorly and inferiorly. The pattern of vessel regression was assessed by one investigator (JR) and confirmed by a second investigator (SS or LN) (Table 32).
| Ring | Original area per segment, DA | Enlargement factor | Modified area per segment, DA |
|---|---|---|---|
| M | 2.00 | 1.08 | 1.85 |
| 1 | 6.25 | 1.20 | 5.21 |
| 2 | 10.42 | 1.34 | 7.78 |
| 3 | 14.58 | 1.54 | 9.47 |
| 4 | 18.75 | 1.81 | 10.36 |
| 5 | 22.92 | 1.97 | 11.63 |
Ultrawide field fundus fluorescein angiography
The FFA images were acquired after intravenous bolus infusion of 5 ml of 20% fluorescein sodium. The images were acquired at transit phase (up to 45 seconds), arteriovenous phases (3–4 minutes) and late frames at 5–7 minutes. A single investigator (LN) identified the best macular-centred fluorescein angiography (FA) image in the arteriovenous phase from the FA series of each eligible eye. A correction factor was applied for the flattening of the three-dimensional image to a two-dimensional image using the Optos V2 Vantage Pro software (Dunfermline, Scotland). Capillary non-perfusion was defined as an area of the capillary network that failed to fill with fluorescein by the late arteriovenous phase, with minimum linear dimension of at least 63 µm (width of a retinal venule at the disc margin).
In order to assess spatial location and quantity of non-perfusion and area of new vessels, a concentric rings template was applied to each image in accordance with previously described methods.80 In brief, this validated method incorporates a macular ring with a radius of 2.5 DDs and five additional concentric rings (rings 1–5), each with a 2.5-DD increment in radius. Each of the six rings (ring M and rings 1–5) was divided into 12 segments. Each segment is graded as ungradeable, not perfused or perfused (if ≥ 50% of the segment is involved). In addition to quantifying non-perfusion, the concentric rings method allows documentation of location of non-perfusion. The area of each cell in each concentric ring was modified based on the enlargement factor identified using 3D printed model eyes.81 The enlargement factors for rings M, 1, 2, 3, 4 and 5 were 1.08, 1.20, 1.34, 1.54, 1.81 and 1.97, respectively. The modified area of each segment in each ring is represented in Table 32.
Statistical analysis
The oximetry measurements, new vessel area and area of retinal non-perfusion were described using means and SD at cross-sectional time points within each arm, and the change from baseline to follow-up time points in each arm was estimated by the mean change and its standard error (SE). The outcomes were compared between aflibercept and PRP arms at both 12 and 52 weeks for oximetry and area of new vessels and at 52 weeks for area of retinal non-perfusion using ANCOVA. If the data contained highly skewed outcomes, the Mann–Whitney U-test was used instead and data were also described using medians and interquartile ranges (IQRs). The statistical significance was set at 0.05.
Mechanistic results
The baseline characteristics of the 40 patients were evenly matched. Retinal oximetry measurements reveal fairly similar SaO2 and SvO2 saturations in both arms at every visit. The mean Sa – vO2 at baseline was 36.7% in the PRP group and 33.4% in the aflibercept group. At week 12, the Sa – vO2 was 36.1% and 35.5% for the PRP and aflibercept group, respectively. At week 52, the Sa – vO2 increased in the PRP arm to 39.7% and minimally decreased in the aflibercept arm to 32.5%. However, the –4.0% (95% CI –8.8% to 0.8%) adjusted difference in the change from baseline between the two groups was not found to be statistically significant (p = 0.10). Mean arterial and venous diameter decreased in both groups at week 52, again with no significant difference between groups. Table 33 shows the detailed retinal oximetry values in the whole cohort and Tables 34 and 35 show the oximetry values in the treatment-naive and non-naive groups, respectively.
| Sa – vO2 | Treatment arm, mean (SD); n | Change from baseline, mean (SE) | Adjusted difference between arms (95% CI) | p-value | ||
|---|---|---|---|---|---|---|
| PRP | Aflibercept | PRP | Aflibercept | |||
| Mean arterial SaO2 (%) | ||||||
| Baseline | 98.0 (9.7); 19 | 98.0 (9.8); 19 | ||||
| 12 weeks | 98.6 (10.0); 18 | 100.4 (10.2); 18 | 0.7 (2) | 2.3 (1.3) | 1.6 (–2.9 to 6.2) | 0.47 |
| 52 weeks | 102.5 (12.8); 18 | 98.9 (7.1); 18 | 3.7 (2) | 0.9 (1.3) | –3.0 (–7.8 to 1.9) | 0.22 |
| Mean venous SaO2 (%) | ||||||
| Baseline | 63.2 (7.0); 19 | 67.2 (8.4); 19 | ||||
| 12 weeks | 63.8 (7.8); 18 | 66.7 (6.6); 18 | 0.4 (1.8) | –0.3 (1.4) | 1.0 (–3.2 to 5.1) | 0.64 |
| 52 weeks | 63.4 (7.9); 18 | 67.7 (4.4); 18 | 0.3 (1.4) | 0.4 (1.4) | 2.0 (–1.4 to 5.5) | 0.23 |
| Mean arterio venous SaO2 (%) | ||||||
| Baseline | 36.7 (10.2); 19 | 33.4 (8.2); 19 | ||||
| 12 weeks | 36.1 (14.7); 18 | 35.5 (6.7); 18 | –0.3 (2.5) | 1.9 (1.4) | 1.8 (–4.1 to 7.6) | 0.54 |
| 52 weeks | 39.7 (11.9); 18 | 32.5 (6.9); 18 | 2.1 (2.1) | –0.9 (1.2) | –4.0 (–8.8 to 0.8) | 0.10 |
| Mean arterial diameter (µm) | ||||||
| Baseline | 113.5 (10.2); 19 | 111.8 (16.3); 19 | ||||
| 12 weeks | 111.0 (8.5); 18 | 108.8 (11.0); 18 | –3.3 (2.1) | –2.7 (3.6) | –1.3 (–7.3 to 4.8) | 0.67 |
| 52 weeks | 107.4 (9.3); 18 | 109.3 (14.4); 18 | –5.5 (2.1) | –3.4 (2.4) | 2.0 (–3.8 to 7.8) | 0.48 |
| Mean venous diameter | ||||||
| Baseline | 143.4 (13.1); 19 | 146.7 (16.6); 19 | ||||
| 12 weeks | 144.9 (14.8); 18 | 144.1 (12.3); 18 | 0.2 (2.6) | –3.3 (4.6) | –1.7 (–10.4 to 7.0) | 0.69 |
| 52 weeks | 140.5 (12.7); 18 | 143.9 (15.4); 18 | –3.8 (2.0) | –1.4 (3.8) | 2.8 (–5.0 to 10.7) | 0.47 |
| Sa – vO2 | Treatment arm, mean (SD); n | Change from baseline, mean (SE) | Adjusted difference between arms (95% CI) | ||
|---|---|---|---|---|---|
| PRP | Aflibercept | PRP | Aflibercept | ||
| Mean arterial SaO2 (%) | |||||
| Baseline | 96.4 (10.7); 11 | 99.3 (9.0); 13 | |||
| 12 weeks | 98.2 (8.0); 11 | 102.2 (10.9); 12 | 1.8 (2.1) | 2.8 (1.2) | 0.2 (–5.4 to 5.8) |
| 52 weeks | 101.5 (13.8); 10 | 100.7 (6.5); 13 | 4.0 (1.8) | 1.4 (1.4) | 2.7 (–2.0 to 7.3) |
| Mean venous SaO2 (%) | |||||
| Baseline | 61.9 (8.4); 11 | 69.1 (9.0); 13 | |||
| 12 weeks | 65.0 (8.4); 11 | 69.0 (6.5); 12 | 3.2 (2.2) | 0.0 (2.1) | 0.2 (–5.4 to 5.8) |
| 52 weeks | 62.0 (9.4); 10 | 69.1 (4.4); 13 | 0.5 (1.8) | –0.1 (1.7) | 2.7 (–2.0 to 7.3) |
| Mean arteriovenous SaO2 (%) | |||||
| Baseline | 35.3 (10.7); 11 | 33.7 (8.6); 13 | |||
| 12 week | 34.1 (10.4); 11 | 35.7 (7.9); 12 | –1.2 (2.9) | 1.5 (1.8) | 2.3 (–4.0 to 8.6) |
| 52 weeks | 39.3 (10.7); 10 | 33.3 (6.8); 13 | 2.7 (2.4) | –0.5 (1.5) | –4.0 (–9.2 to 1.2) |
| Mean arterial diameter (µm) | |||||
| Baseline | 114.3 (12.2); 11 | 111.3 (17.6); 13 | |||
| 12 weeks | 112.3 (9.2); 11 | 108.8 (11.5); 12 | –2.0 (3.0) | –2.0 (4.6) | –2.3 (–10.2 to 5.7) |
| 52 weeks | 108.4 (9.1); 10 | 109.7 (15.7); 13 | –5.0 (2.9) | –1.6 (2.1) | 2.8 (–3.6 to 9.3) |
| Mean venous diameter (µm) | |||||
| Baseline | 146.4 (11.8); 11 | 142.6 (12.9); 13 | |||
| 12 weeks | 145.4 (12.4); 11 | 145.9 (13.6); 12 | –1.0 (3.4) | 2.4 (4.3) | 1.9 (–8.5 to 12.4) |
| 52 weeks | 144.5 (8.4); 10 | 145.0 (15.7); 13 | –3.8 (1.6) | 2.3 (2.7) | 5.4 (–1.9 to 12.8) |
| Sa – vO2 | Treatment arm, mean (SD); n | Change from baseline, mean (SE) | Adjusted difference between arms (95% CI) | ||
|---|---|---|---|---|---|
| PRP | Aflibercept | PRP | Aflibercept | ||
| Mean arterial SaO2 (%) | |||||
| Baseline | 100.3 (8.3); 8 | 95.2 (11.6); 6 | |||
| 12 weeks | 99.3 (13.3); 7 | 96.7 (8.2); 6 | –0.9 (3.9) | 1.5 (3.2) | 1.9 (–5.2 to 9.0) |
| 52 weeks | 103.7 (12.3); 8 | 94.4 (7.0); 5 | 3.5 (11.8) | –0.2 (3.4) | –1.0 (–7.3 to 5.3) |
| Mean venous SaO2 (%) | |||||
| Baseline | 65.0 (4.0); 8 | 62.9 (4.9); 6 | |||
| 12 weeks | 61.7 (6.9); 7 | 61.9 (4.0); 6 | –4.0 (2.7) | –1.0 (0.9) | 1.9 (–5.2 to 9.0) |
| 52 weeks | 65.1 (5.5); 8 | 64.2 (1.9); 5 | 0.1 (6.6) | 1.6 (2.8) | –1.0 (–7.3 to 5.3) |
| Mean arteriovenous SaO2 (%) | |||||
| Baseline | 38.7 (9.9); 8 | 32.6 (8.1); 6 | |||
| 12 weeks | 39.1 (20.4); 7 | 35.3 (4.1); 6 | 0.9 (4.7) | 2.7 (2.5) | 3.0 (–10.5 to 16.5) |
| 52 weeks | 40.1 (14.0); 8 | 30.6 (7.5); 5 | 1.4 (10.6) | –1.9 (1.9) | –4.2 (–16.5 to 8.1) |
| Mean arterial diameter (µm) | |||||
| Baseline | 112.4 (7.1); 8 | 112.9 (14.5); 6 | |||
| 12 week | 109.0 (7.4); 7 | 108.9 (11.1); 6 | –5.2 (3.0) | –4.0 (6.3) | 0.2 (–11.4 to 11.8) |
| 52 weeks | 106.2 (10.0); 8 | 108.1 (11.7); 5 | –6.2 (9.6) | –8.2 (6.8) | 0.6 (–13.2 to 14.3) |
| Mean venous diameter (µm) | |||||
| Baseline | 139.2 (14.6); 8 | 155.3 (21.6); 6 | |||
| 12 weeks | 144.2 (19.0); 7 | 140.6 (8.9); 6 | 2.0 (4.0) | –14.8 (9.6) | –8.4 (–27.9 to 11.1) |
| 52 weeks | 135.4 (15.8); 8 | 141.0 (16.2); 5 | –3.8 (11.7) | –11.1 (11.5) | 0.7 (–19.9 to 21.2) |
The mean total area of retinal non-perfusion increased in both PRP and aflibercept groups with no statistically significant difference between groups. At baseline, the total area of retinal non-perfusion was 125.1 disc areas in the PRP group and 131.2 disc areas in the aflibercept group. At week 52, this increased to 156.1 disc areas and 158.4 disc areas in the PRP and aflibercept groups, respectively. The proportion of eyes that experienced an increase in retinal non-perfusion of more than 20 disc areas was 51.6%, while 22.6% had an increase in retinal non-perfusion of 5–20 disc areas and 25.8% did not experience a significant increase (< 5 disc area change). This is further detailed in Table 36.
| Sa – vO2 | Treatment arm, mean (SD); n | Change from baseline, mean (SE) | Adjusted difference between arms (95% CI) | p-value | ||
|---|---|---|---|---|---|---|
| PRP | Aflibercept | PRP | Aflibercept | |||
| M + 1 | ||||||
| Baseline | 15.4 (12.2); 18 | 16.9 (15.6); 18 | 0.24 | |||
| 52 weeks | 19.6 (14.7); 18 | 14.8 (11.7); 17 | 4.0 (3.6) | –1.2 (2.6) | –5.0 (–13.4 to 3.5) | |
| Treatment naive | ||||||
| Baseline | 16.0 (11.0); 11 | 12.3 (8.0); 12 | ||||
| 52 weeks | 18.2 (17.5); 9 | 16.5 (8.9); 11 | 4.3 (5.9) | 3.0 (2.0) | –2.2 (–14.6 to 10.2) | |
| Non-naive | ||||||
| Baseline | 14.5 (14.7); 7 | 26.1 (23.1); 6 | ||||
| 52 weeks | 21.1 (12.2); 9 | 11.7 (16.1); 6 | 3.7 (3.9) | –9.4 (5.0) | –11.3 (–24.3 to 1.6) | |
| 2 + 3 + 4 | ||||||
| Baseline | 109.6 (48.7); 18 | 114.3 (65.7); 18 | 0.23 | |||
| 52 weeks | 136.4 (62.1); 18 | 143.6 (53.7); 17 | 23.5 (11.5) | 39.2 (9.7) | 17.4 (–11.8 to 46.5) | |
| Treatment naive | ||||||
| Baseline | 116.5 (55.1); 11 | 117.3 (69.1); 12 | ||||
| 52 weeks | 146.6 (69.6); 9 | 149.1 (62.6); 11 | 37.7 (16.6) | 37.0 (12.4) | 2.9 (–38.6 to 44.4) | |
| Non-naive | ||||||
| Baseline | 98.9 (37.7); 7 | 108.3 (64.1); 6 | ||||
| 52 weeks | 126.3 (55.9); 9 | 133.5 (34.5); 6 | 5.2 (13.5) | 43.5 (17.1) | 30.8 (–11.6 to 73.3) | |
| Total | ||||||
| Baseline | 125.1 (55.1); 18 | 131.2 (75.4); 18 | 0.46 | |||
| 52 weeks | 156.1 (74.7); 18 | 158.4 (54.5); 17 | 27.5 (13.3) | 38.0 (9.6) | 12 (–20.7 to 44.7) | |
| Treatment naive | ||||||
| Baseline | 132.5 (60.7); 11 | 129.7 (74.3); 12 | ||||
| 52 weeks | 164.8 (85.4); 9 | 165.6 (63.6); 11 | 41.9 (19.7) | 40.0 (12.3) | 0.3 (–47.6 to 48.2) | |
| Non-naive | ||||||
| Baseline | 113.3 (46.9); 7 | 134.4 (84.8); 6 | ||||
| 52 weeks | 147.4 (66.2); 9 | 145.1 (33.1); 6 | 8.9 (15.3) | 34.0 (16.7) | 21.3 (–21.6 to 64.3) | |
The total area of new vessels was reduced from baseline with treatment in both arms with more rapid and significant regression in the aflibercept arm at 12 weeks (p = 0.019). By 52 weeks, the PRP arm also showed similar regression of new vessel area (p = 0.45). This is further detailed in Table 37.
| Sa – vO2 | Treatment arm, mean (SD); n | Change from baseline, mean (SE) | p-valuea | ||
|---|---|---|---|---|---|
| PRP | Aflibercept | PRP | Aflibercept | ||
| Total new vessel area (mm2) | |||||
| Baseline | 1.54 (1.75); 19 | 1.68 (1.99); 18 | |||
| 12 weeks | 1.13 (1.36); 19 | 0.00 (0.00); 18 | –0.42 (0.16) | –1.68 (0.47) | 0.019 |
| 52 weeks | 0.35 (0.42); 16 | 0.11 (0.23); 18 | –1.03 (0.41) | –1.57 (0.45) | 0.45 |
| Total new vessel area: naive subgroup (mm2) | |||||
| Baseline | 1.22 (1.28); 11 | 1.09 (1.34); 12 | |||
| 12 weeks | 0.98 (1.09); 11 | 0.00 (0.00); 12 | –0.25 (0.22) | –1.09 (0.39) | |
| 52 weeks | 0.45 (0.48); 9 | 0.04 (0.06); 12 | –0.93 (0.52) | –1.05 (0.39) | |
| Total new vessel are: non-naive subgroup (mm2) | |||||
| Baseline | 1.98 (2.28); 8 | 2.87 (2.63); 6 | |||
| 12 weeks | 1.33 (1.73); 8 | 0.00 (0.00); 6 | –0.66 (0.25) | –2.87 (1.07) | |
| 52 weeks | 0.23 (0.33); 7 | 0.25 (0.38); 6 | –1.17 (1.8) | –2.62 (1.03) | |