The greatest rates of exponential decline were obtained with V30, CS, and VTotal metrics, followed by BCVA, OMS, and V5 metrics. Pooling all functional data together to allow for exponential analyses allowed us to obtain an arguably truer and precise estimate of progression with the various metrics studied, as subjects spanned a range of ages from 9 to 56 years.
Our exponential BCVA decline rate of 5.02% is greater than the annual decline in visual acuity reported by two studies of
RPGR-RP subjects (4.0%
17 and 3.5%
19). Our perimetry-derived exponential decline rates, in particular OMS rate of 4.7% per annum, is similar to that obtained by Sandberg et al.
17 who quantified perimetric progression (albeit with kinetic Goldmann V4e perimetry) in a cohort of molecularly confirmed subjects with
RPGR-associated RP. Sandberg et al.
17 have found an exponential decline rate of 4.7% per annum, which is greater than the progression previously reported in their
RHO-RP cohort of 2.9% per annum.
53 Huang et al.,
18 in another study of
RPGR-associated RP subjects, report an annual perimetry (Goldmann V4e) decline rate of 9%; this is the average value from individual rates of only 13 subjects. Bellingrath et al.
20 have recently published a retrospective cross-sectional study of
RPGR subjects in which they are unable to adequately assess perimetric progression in patients with increasing disease severity beyond the second decade of life. This may be due to their use of the Goldmann III4e isopter as compared to the more commonly used V4e for kinetic perimetry in RP studies.
12–18,54
The RF as an arbitrary construct is nevertheless a useful and succinct tool to summarize test reliability with the specific purpose of identifying reliable tests.
55 The RF threshold for test inclusion (RF ≤ 25) has been used in previous work on RP subjects to good effect.
27 Additional scrutiny was placed on tests with RF scores between 21 and 25 with the exclusion of tests bearing a false-positive rate exceeding 10% on the basis of the subject being “trigger-happy” (i.e., responsive in the absence of stimuli). The inclusion of unreliable tests can be corrosive to results, as a 10% false-positive answer rate can give rise to an erroneous 1.5-dB increase in mean sensitivity.
56 The number of tests lost owing to unreliability was however small, as shown in
Table 7.
In our study, we additionally relied on the use of
R2 ≥ 0.4 as a method to exclude subjects with great fluctuation in performance, in order to calculate individual progression rates. We included all individual slopes with an
R2 ≥ 0.4 regardless of slope steepness or how quickly or slowly we found their rate of decline to be. By using this approach, we sought to minimize the bias of selecting for subjects with great progression rates—different from the approach taken by others whereby a significant proportion of subjects are excluded on the basis of slow or insufficient progression.
13–15 In addition, all reliable perimetry data were included into our calculations for overall exponential decline rates. Our approach in determining progression rates of visual function can be applied to other forms of RP, particularly in instances where variability and fluctuation in performance can pose a challenge to the interpretation of progression without bias. However, if we were to reanalyze individual progression rates obtained from the linear trend lines of our subjects with ≥1-year follow-up, disregard our
R2 (goodness-of-fit) criterion, and exclude those with positive rates of change (i.e., improvement in function), we would obtain the following mean percentage annual rates of decline for V
Total, V
30, OMS, BCVA, and CS,: 12.3%, 9.4%, 8.1%, 7.6%, and 6.9%, respectively.