This study describes TAC and HOTV testing in a patient population known to have comorbid medical conditions (i.e., attention deficit–hyperactivity disorder [ADHD]) and developmental delay that could affect both the success rate of their testing as well as the relationship between grating and recognition acuity. A relatively small portion of the population (8.7%) could not complete monocular TAC testing for both eyes. Nearly one-third of our subjects could not complete HOTV testing, most of whom were younger than 5 years old. The relatively high rate of HOTV failure highlights both the association of age on testing success and the difficulty in transitioning from grating acuity to recognition acuity in young children. Even though children with NF1 are known to have a much higher rate of developmental delay and ADHD, we were surprised that the success rate of completing either test was not associated with the diagnosis of NF1 when accounting for subject age.
The success rate of the ATS computer-based HOTV testing has previously been evaluated, but primarily as a screening tool for vision loss in population-based studies of otherwise well children.
5,7,26,27 In these population-based studies,
7,26,27 approximately 37 to 47% of children younger than 3 years of age could complete the HOTV testing, compared to none of the subjects in our study. In the same studies, the success rate for children who were 3 and 4 years old markedly improved to nearly 85%,
7,26,27 whereas only half of our subjects could complete HOTV testing. Even when half of the study subjects have eye conditions resulting in decreased VA,
5 the success of computer-based HOTV testing is similar to larger studies comprised of primarily well children.
7,26,27 Our subjects typically have two to four eye exams per year for their clinical care, so their success rate may potentially have been lower if they were completing the testing for the first time. Unfortunately, we did not record the number of previously completed ophthalmologic exams or the testing method used. For children with NF1, we suspect our low success rate (i.e., 30.1%) is due to the known cognitive and behavioral comorbidities of this condition.
22,28 The treatment of OPGs, especially radiation, is known to cause cognitive side effects in young children and would likely be associated with VA testing success.
For children who could complete both TAC and HOTV testing, there was a strong correlation and coefficient of determination between logMAR scores. More specifically, the strength of this relationship is supported by the high adjusted coefficient of determination, suggesting that other factors included in the regression model did not significantly alter this result. Our correlation coefficients were nearly identical to those from a study of premature infants who completed both TAC and HOTV testing at 4 years of age, although their analysis included mostly children with normal VA.
29 Birch and Spencer find an even higher correlation between TAC and Snellen acuity in a small group of children with a history of cicatricial retinopathy of prematurity (ROP).
30 Despite the differences in testing protocols as well as patient populations, our study and others
29,30 suggest that grating acuity and recognition acuity are closely related. On the other hand, data from a large multicenter study of children followed for ROP suggest that the ability of TAC to predict future recognition acuity is much more limited, primarily due to the test–retest variability.
11,31 When categorizing grating and recognition acuity into a binary category (i.e., normal versus abnormal), most investigators agree that an abnormal TAC is predictive of future abnormal recognition acuity.
11,17 Contrary to Kushner,
17 Dobson and colleagues contend that a normal TAC has a high likelihood of normal Snellen recognition acuity in the future since over 85% of their subjects with normal TAC eventually demonstrated normal Snellen recognition acuity.
11
Our study had a number of limitations that should be considered when interpreting the results. Having four different examiners from two separate study sites perform the VA testing could certainly have affected the VA results. For example, some examiners may have had a higher success rate of completing the VA testing. Although a single designated examiner may have reduced testing variability, the correlation between TAC and HOTV remained strong and the success of testing did not differ between centers. Second, we did not routinely document the incidence of developmental delay and ADHD since it has been well established that at least half of all children with NF1 have some degree of intellectual and cognitive impairment.
22,23 Most clinicians defer making a formal diagnosis of ADHD until the child has been enrolled in elementary school. Since approximately half of our subjects had not yet entered elementary school, limiting this diagnosis to half of our study population would have been problematic. While the incidence of developmental delay and ADHD is likely associated with both the success and comparison between VA testing methods, our prospective study design and large sample size should have resulted in a representative population of children with NF1 and/or OPGs. Since both of our hospitals have dedicated NF1 and OPG centers, it is conceivable that some amount of referral bias resulted in our subjects having a higher incidence of these comorbid conditions.
The longitudinal assessment of VA in young children with OPGs is complicated by a number of factors. First, children with OPGs may enter a clinical trial at various ages and developmental stages, resulting in subjects enrolled with different success rates in performing TAC or recognition acuity. The strong relationship between TAC and HOTV logMAR in our current study provides further support to our previous recommendation that all pediatric OPG clinical trials test patients using TAC and, when the subject is developmentally able, also include HOTV.
24 Having subjects transition from grating acuity testing to recognition acuity testing, or even transition within recognition acuity methods (i.e., HOTV to Snellen), is problematic since their results are not equivalent, thereby complicating the analysis and interpretation of both short-term and long-term results.
8,11,16,17 Second, the mechanism and natural history of vision loss in children with OPGs is dynamic compared to vision loss from ROP, which is static after the perinatal period. Vision loss from OPGs can occur at any age, can be separated by many months or years, and can vary in severity. Therefore, categorizing VA loss as normal or abnormal, as in ROP, is not acceptable because a binary rather than continuous VA outcome would eliminate any meaningful longitudinal assessment of VA change. Some investigators have calculated the interocular difference in TAC to detect and measure the magnitude of vision loss.
8 Unfortunately, a majority of OPGs are located in the optic chiasm and tracts, putting both eyes at risk for vision loss. Last, an overwhelming majority of children who are being monitored for their OPG are younger than 8 years of age and are still experiencing normal visual maturation. It is unclear how the study design and statistical analysis plan should account for normal visual development.
In conclusion, the aforementioned factors highlight the difficulties in the longitudinal assessment of VA in children with OPGs and the need for a different approach to assessment than is used with other pediatric ophthalmologic diseases. For these reasons, and based on the data presented in the current study, it is still important to limit the VA testing methods to TAC and, when the children are old enough, to also test using HOTV, until a larger multicenter longitudinal clinical trial of children with OPGs can confirm that the VA measures are comparable and that the transition between methods will not significantly alter the results.