We analyzed visual acuity data from three clinical trials conducted by us among patients with typical retinitis pigmentosa from 1984–1991 (clinical trial I), from 1996–2001 (clinical trial II), and from 2003–2008 (clinical trial III). Each had been approved by the Institutional Review Boards of the Massachusetts Eye and Ear Infirmary and Harvard Medical School and each had been conducted in accord with the guidelines of the Declaration of Helsinki. In each trial informed consent had been obtained from the patients after explanation of the nature and possible consequences of the trial. We included nearly all available patients (ages 18–60) in these analyses. We identified 7 patients (3 from trial I, 3 from trial II, and 1 from trial III) as outliers for decline in ETDRS acuity based on the Generalized Extreme Studentized Deviate Test (GESD)23
and excluded their data from analyses. In addition, for 9 patients who participated in two of the trials and 1 patient who participated in all three trials, data used in the present analyses were restricted to the first trial in which they participated. From trial I we included data from 143 of the 146 patients ages 18–49 on vitamin A as retinyl palmitate 15,000 IU/day for 4–6 years,16
from trial II we included data from 101 of the 108 patients ages 18 to 56 on this dose of vitamin A for 4–5 years,18
and from trial III we included data from 113 of the 121 patients ages 18–60 on this dose of vitamin A for 4–5 years for a total sample of 357 patients. 19
In each trial patients were screened according to comparable preset eligibility criteria. These eligibility criteria included a best corrected Snellen distance visual acuity of 20/100 or better in at least one eye. In each trial patients agreed not to know their group assignment or the course of their condition until the end of the study. Eligible patients had been reexamined 6–8 weeks after the screening examination before treatment and then annually thereafter during treatment. In each trial all staff in contact with the patients had been masked to treatment group assignment and each examination had been conducted without prior information on results of previous examinations. In each trial the data had been monitored by an independent Data and Safety Monitoring Committee selected by the National Eye Institute.
In all three trials patients were evaluated under the same test conditions. They completed the Willett food frequency questionnaire24
at screening and at annual follow-up visits with a clinical coordinator. They then had an ocular examination including a measure of Early Treatment Diabetic Retinopathy Study (ETDRS) distance acuity at 3.2 meters25
and, after dilation, Snellen retinal acuity with a Guyton-Minkowski retinal potential acuity meter (PAM) which projected a numerical acuity chart on the retina in a narrow beam that can bypass a central cataract. 26,27
The ETDRS chart, the standard method for measuring visual acuity in clinical trials, contained 5 letters of comparable difficulty on each line; adjacent lines differed in letter size by 0.23 loge
unit. A different chart was used for each eye to reduce the likelihood of memorization. The ETDRS visual acuity was scored as the total number of letters identified. A numerical chart was used for measuring retinal acuity also to minimize memorization; retinal acuities were then transformed to the natural log scale. We performed a slit-lamp examination to quantify area of posterior subcapsular cataracts, if present, with a slit-lamp beam, and then a fundus examination with an ophthalmoscope.
From the responses to the food frequency questionnaire, we calculated intake of long chain omega-3 fatty acids (primarily docosahexaenoic acid [DHA]), total energy intake, and other nutrients as described elsewhere.28
The validity of long chain omega-3 fatty acid intake calculated from this questionnaire has been documented by comparison with levels in adipose tissue.29
In trial II, the validity of this questionnaire for estimating omega-3 intake was confirmed by the moderate correlation between dietary omega-3 intake and red blood cell phosphatidylethanolamine docosahexaenoic acid (RBC PE DHA) levels (r=0.53, p<0.01). A level of RBC PE DHA ≥5% of total RBC PE fatty acids corresponded to an estimated dietary omega-3 intake of ≥ 0.20 g/day.18
Using the food frequency questionnaire, we estimated omega-3 intake for each patient by averaging the data from all examination visits to minimize measurement error. Using these values the entire study population (n=357) was divided into those with high (≥0.20 g/d, n=215) or low (<0.20 g/d, n=142) omega-3 intake. The value of 0.20 g/d represents the median intake observed during the course of trial II. Furthermore, in trial II, patients with ≥ 0.20 g/d of omega-3 intake were found to have a slower loss of visual field than those with omega-3 intake < 0.20 g/d. 18
Eyes with a Snellen distance acuity <20/100 at baseline were excluded from analysis of ETDRS or PAM data to reduce the likelihood of a floor effect. If patients became pseudophakic (n = 6), or if ETDRS acuity declined to 0 letters (n = 8) (comparable to ≤ 20/300), data from these eyes at subsequent visits were censored to eliminate any effect of cataract surgery on acuity or any possible floor effect. Visual acuity testing at 1 meter was not performed. Longitudinal regression analyses30,31
using PROC MIXED of SAS 9.1.332
were performed to compare rates of decline in distance acuity and retinal acuity by high (≥0.20 g/d) versus low (<0.20 g/d) omega-3 intake. Since there were significant differences in baseline visual acuity among the 3 clinical trials, we included indicator variables for trial in the models used for analyses.33
In addition, since patients in the high omega group were slightly older than those in the low omega group (see ), we also included a term for age at baseline. These analyses also took into account variable lengths of follow-up and intraclass correlations of visual acuity between fellow eyes of individual patients at a single visit and between the same eye over time.
Baseline Patient Characteristics of Study Population with Retinitis Pigmentosa
Since Goldmann fields were performed in trial I and Humphrey fields in trials II and III, we could not combine data from the three trials for field analyses. With respect to full-field cone ERGs, among the 375 patients taking vitamin A, 10 participated in more than one trial and data from the first trial only were included. In addition, 99 patients had initial values <0.68 μV and could not be followed due to a floor effect which would be expected to occur when they reached 0.34 μV. Thus the sample for ERG analysis included 266 patients with pretreatment amplitudes ≥0.68 μV and analysis of the combined data was performed; annual rate of decline of 163 patients with high omega-3 intake was compared with rate of decline of 103 patients with low omega-3 intake.