The CATT trial randomised patients with neovascular AMD to bevacizumab or ranibizumab, monthly or as-needed. At two years bevacizumab and ranibizumab produced equivalent effects on visual acuity when given on the same schedule, and as-needed dosing produced less gain than monthly dosing for both drugs.

Protocol T compared aflibercept, bevacizumab and ranibizumab for centre-involved diabetic macular oedema. At one year all three improved vision; aflibercept was more effective than bevacizumab where baseline acuity was 20/50 or worse, and the three were comparable where baseline acuity was better than that.

Comparative case series of 40 children aged 3 to 16 who underwent both routes within one week. All 40 completed the oral study against 24 of 40 for the intravenous study (100% versus 60%, P<0.0001), and among preschoolers 19 of 19 against 7 of 19 (P<0.0001); the 16 intravenous failures were needle fear in 14 and a positive fluorescein skin test in 2. Graded image quality was similar for foveal avascular zone visualization (P=0.8972), for clinically important findings (P>0.9999), and overall (P=0.2500). Median time to dye entry was 5.2 minutes and to the optimal angiogram 10.5 minutes. No adverse event occurred in either arm, but the authors state their sample was too small to evaluate the safety of the oral route.

A systematic review with network meta-analysis of needling revision compared mitomycin C, 5-fluorouracil and no antimetabolite at all. It found no significant difference between them in intraocular pressure at 12 months or latest follow-up, complication rates, qualified or complete success, or the number of re-needling cases. Six studies met inclusion and only one was randomised, and the authors attribute an overall low quality of evidence to that design mix and call for further research. So this is no difference SHOWN, not a difference ruled out.

A network meta-analysis of 17 studies (12 retrospective, 5 randomized) in 1492 patients with stage II to IV idiopathic macular hole found that peeling with 0.25% of this dye performed worse for hole closure than peeling with no dye at all, while 0.05% brilliant blue G, 0.15% trypan blue, 0.5% and 0.05% of this dye all outperformed 0.25%; the authors suggest 0.05% as a good choice for surgeons who prefer this dye, and call for large high-quality randomized trials to confirm the ranking. A separate meta-analysis of 9 studies and 846 eyes found better postoperative acuity with brilliant blue G than with this dye (mean difference -0.10 logMAR, 95% CI -0.16 to -0.03, P=0.004, negative favoring brilliant blue G) but no significant difference in closure rate (OR 1.98, 95% CI 0.71 to 5.48, P=0.19).

A prospective open-label phase 2 trial enrolled 23 patients with refractory autoimmune uveitis. 18 of 23 (78%) met a composite success endpoint at week 10, and all patients who completed 50 weeks maintained it. The safety signal is the reason to read this row carefully rather than the efficacy: 5 of the 14 enrolled for a year stopped early for significant adverse events, serious events potentially related to infliximab included pulmonary embolus, congestive heart failure, lupus-like reaction in 2 patients and vitreous haemorrhage in 2, and antinuclear antibodies developed in 15 of 20 patients receiving 3 or more infusions. The authors call the rate of serious toxic effects unexpectedly high.

Open-label extension of OPTIC, no control arm and very small numbers: of 8 initial responders who flared, 5 (62.5%) responded again on retreatment with a mean proptosis reduction of 1.9 mm (SD 1.2) from the extension baseline, and of 5 initial nonresponders retreated, 2 responded, 1 fell short at 1.5 mm, and 2 discontinued early. The 37 previously placebo-treated patients, whose median disease duration was 12.9 months versus 6.3 months in OPTIC, responded at 89.2%, which is what supports treating longer-duration disease. Mild hearing impairment recurred in 2 patients on retreatment.

STOP-Uveitis randomised 37 patients to 4 mg/kg or 8 mg/kg intravenous tocilizumab every 4 weeks to month 6. Its primary endpoint was safety, and repeated infusions were well tolerated; the sight measures below are all secondary. Among those with room for it, 43.5% achieved a 2-step reduction in vitreous haze; mean visual acuity improved 8.22 letters and mean central macular thickness fell 83.88 micrometres. Read the design before the numbers: the trial compared two doses of tocilizumab against each other, with no placebo or active control arm, so it shows the drug is tolerated and that treated eyes improved, not that it beats an alternative.