PURPOSE: To assess the safety of the subretinal delivery of a recombinant adeno-associated virus serotype 2 (AAV2) vector carrying a human choroideremia (CHM)-encoding cDNA in CHM. DESIGN: Prospective, open-label, nonrandomized, dose-escalation, phase I/II clinical trial. PARTICIPANTS: Fifteen CHM patients (ages 20-57 years at dosing). METHODS: Patients received uniocular subfoveal injections of low-dose (up to 5 × 1010 vector genome [vg] per eye, n = 5) or high-dose (up to 1 × 1011 vg per eye, n = 10) of a recombinant adeno-associated virus serotype 2 (AAV2) vector carrying a human CHM-encoding cDNA (AAV2-hCHM). Patients were evaluated preoperatively and postoperatively for 2 years with ophthalmic examinations, multimodal retinal imaging, and psychophysical testing. MAIN OUTCOME MEASURES: Visual acuity, perimetry (10-2 protocol), spectral-domain OCT (SD-OCT), and short-wavelength fundus autofluorescence (SW-FAF). RESULTS: We detected no vector-related or systemic toxicities. Visual acuity returned to within 15 letters of baseline in all but 2 patients (1 developed acute foveal thinning, and 1 developed a macular hole); the rest showed no gross changes in foveal structure at 2 years. There were no significant differences between intervention and control eyes in mean light-adapted sensitivity by perimetry or in the lateral extent of retinal pigment epithelium relative preservation by SD-OCT and SW-FAF. Microperimetry showed nonsignificant (< 3 standard deviations of the intervisit variability) gains in sensitivity in some locations and participants in the intervention eye. There were no obvious dose-dependent relationships. CONCLUSIONS: Visual acuity was within 15 letters of baseline after the subfoveal AAV2-hCHM injections in 13 of 15 patients. Acute foveal thinning with unchanged perifoveal function in 1 patient and macular hole in 1 patient suggest foveal vulnerability to the subretinal injections. Longer observation intervals will help establish the significance of the minor differences in sensitivities and rate of disease progression observed between intervention and control eyes.

Refraction predictions from intraocular lens (IOL) calculation formulae are inaccurate in children. We sought to quantify the relationship between age and prediction error using a model derived from the biometry measurements of children enrolled in the Infant Aphakia Treatment Study (IATS) when they were ≤7 months of age. We calculated theoretical predicted refractions in diopters (D) using axial length, average keratometry, and IOL powers at each measurement time point using the Holladay 1 formula. We compared the predicted refraction to the actual refraction and calculated the absolute prediction error (APE). We found that the median APE was 1.60 D (IQR, 0.73-3.11 D) at a mean age (corrected for estimated gestational age) of 0.20 ± 0.14 years and decreased to 1.11 D (IQR, 0.42-2.20 D) at 10.60 ± 0.27 years. We analyzed the association of age with APE using linear mixed-effects models adjusting for axial length, average keratometry, and IOL power and found that as age doubled, APE decreased by 0.25 D (95% CI, 0.09-0.40 D). The accuracy of IOL calculations increases with age, independent of biometry measurements and IOL power.