Johnson & Johnson Vision myopia research showcased at AAO 2023

Authors: Noel Brennan¹, Xu Cheng¹, Mark Bullimore²

1. Johnson and Johnson Vision, Jacksonville, Florida, United States
2. University of Houston, Houston, Texas, United States

This study explored whether one-year data could predict three-year outcomes in myopia control treatment through a meta-analysis. Studies with both one-year and three-year axial elongation or refractive progression data, covering different interventions, were reviewed. Using variables like age, axial length, refractive error, and demographic factors, meta-regression was used to predict the primary outcome at year three based on year-one results. For optical interventions, reliable estimates for three-year axial elongation reduction can be obtained by doubling the one-year group mean value, assuming robust data at the one-year mark. This was true independent of age, demographic and other variables. Three-year efficacy is not triple the one-year value due to an initial boost in myopia control effect, and age-related reduction in myopia progression.

Purpose: Demonstrating efficacy of a myopia control treatment over a 3-year period is difficult because of enrolment challenges, ethical issues, non-compliance, loss to follow-up and cost. We examined the feasibility of using 1-year data to predict 3-year outcomes among myopes in a meta-analysis.

Methods: Our initial review of the literature did not provide sufficient data to model efficacy robustly. Instead, we systematically reviewed the literature to identify papers that contained both 1-year and 3-year axial elongation (AE) or refractive progression (RP) data (including control and treatment groups with any type of intervention considered). Covariates included baseline age, axial length and refractive error, study type, sex, race and measurement method. Meta-regression was used to model the primary outcome at year 3 with that measure at year 1 as the primary predictor variable. Studies were weighted by inverse variance, based on standard deviations at 1-year. Hierarchical models controlled for study as a random effect and created additional meta-regression models to control for study-level characteristics, as available, to determine if such additions improved overall predictive power. Validation employed a holdout sample approach (bootstrapping), where a dataset with 10 randomly selected studies was withheld and the remainder assigned as a training dataset. This approach was replicated 1000 times to create competing meta-regression models and estimate total squared predictive error (TSPE) for each when applied to the validation sample, allowing uncertainty intervals to be created for competing models.

Results: AE at both years 1 and 3 was reported in 36 data sets (interventions were limited to optical). The table presents results for different models. Knowledge of 1-year AE was sufficient to predict 3-year AE in the absence of other covariates with impressive accuracy and the mean multiplier was 2.00. Perhaps surprisingly, different equations were not required for treated and untreated groups. Thus, the value of ‘2’ can be directly applied to estimate 3-year myopia control efficacy from 1-year data with considerable certainty. Models for RP showed notably more variability in terms of results and performance.

Conclusion: Mean cumulative absolute reduction in axial elongation at 3 years can be estimated for optical interventions with a high degree of confidence by simply doubling the group mean value at 1 year, provided robust intervention and control data sets at 1 year are available. The reason that 3-year efficacy is not triple the 1-year value relates to the initial boost in slowing progression that we have previously reported, and the general, known reduction in progression with age.

Authors: Mark Bullimore¹, Xu Cheng², Noel Brennan²

1. University of Houston, Houston, Texas, United States
2. Johnson and Johnson Vision, Jacksonville, Florida, United States

In this study, the researchers aimed to investigate the relative changes in the prevalence of high myopia compared to myopia by analyzing data from 43 sets of data in 23 papers, which were originally used to create a universal calculator for myopia prevalence. The analysis focused on adult data and used a consistent approach to examine how the prevalence of high myopia (defined as -6.00D) changes concerning myopia (defined as -0.50D) prevalence. The results revealed that at approximately 80% myopia prevalence, the prevalence of high myopia increases at a greater rate than myopia, with a slope exceeding 1.0. This observation suggests that the reported disproportionate increase in high myopia prevalence compared to myopia can be explained by the inherent characteristics of refractive error probability distributions, obviating the need for further investigation into its origin.

Purpose: In their 2021 IMI report,1 Sankaridurg et al. noted that “the rate of change in prevalence of high myopia appears to be disproportionately greater compared to the rate of change in the prevalence of myopia”, citing illustrative examples. They suggested that this change over time may be attributable to a number of possible factors: 

• Children in Asian countries increasingly developing myopia at earlier ages than before, 
• A faster rate of progression of myopia in children of Asian ethnicity compared to Caucasian counterparts, or 
• A faster rate of myopia progression in younger compared to older children.

Here we examine the relative change in the prevalence of high myopia compared to myopia from a sample of studies containing prevalence data to draw inferences on the origin of the putative disproportionate increase in the prevalence of high myopia.

Method: We used 43 sets of data from 23 papers previously used to develop a preliminary universal calculator for myopia prevalence.2 These data sets consisted of prevalence estimates for at least two refractive error thresholds (eg, -0.50D and -6.00D). The sample was untargeted other than incorporating a wide spread of overall myopia rates and, for the purpose of the current analysis, using only adult data. Our previous work, using logarithm of the odds of having myopia at different refractive thresholds, has shown that the slope of ln(odds) versus refractive error threshold is highly consistent across populations within specific age groups, even in the face of vastly different underlying myopia prevalence. We applied a weighted slope of ln(odds) at a given prevalence versus prevalence for the adult population to estimate rates of high myopia (-6.00 D) across the myopia (-0.50 D) prevalence range and the corresponding ratio of change in high myopia to myopia prevalence.

Results: The left hand figure plots ln(odds of high myopia) versus ln(odds of myopia) for the populations considered. The right hand figure plots the predicted change in the prevalence of high myopia for every one percent increase in the prevalence of myopia. It is noticeable that, at around 80% myopia prevalence, high myopia prevalence will increase by a greater increment than for myopia; that is, the slope is above a value of 1.0.

Conclusion: The claimed disproportionate increase in the prevalence of high myopia compared to that of myopia as the latter rises is merely a function of, and entirely explainable by, the underlying nature of refractive error probability distributions and requires no further investigation as to its origin.

1. Sankaridurg et al. IOVS 2021; 62:2.
2. Brennan et al. Optom Vis Sci 2020: e-abstract 205355

Authors: Alex Nixon¹, Mark Bullimore², Noel Brennan¹, Xu Cheng¹

1. Johnson and Johnson Vision, Jacksonville, Florida, United States
2. University of Houston, Houston, Texas, United States

This study examines the distinction between "responders" and "non-responders" in the context of myopia control interventions by analyzing data from the first year of the Low-concentration Atropine for Myopia Progression (LAMP) study. The LAMP study was a well-designed, placebo-controlled trial with a substantial sample size. The researchers calculated the efficacy of various low-concentration atropine treatments in terms of reducing myopia progression and axial elongation within different age groups and compared it to overall progression metrics. Surprisingly, the data showed that so-called "non-responders" (faster progressors) achieved similar reductions in myopia progression and axial elongation as "responders" (slower progressors), challenging the validity of these labels. The study suggests that the use of such terms is inaccurate and that focusing on treatment adjustments, combination therapies, and improved compliance may be more beneficial strategies in myopia progression interventions.

Purpose: It is common to hear talk of ‘responders’ and ‘non-responders’ with respect to myopia control interventions. We consider the reality of distinguishing these sub-groups using data from the first year of the Low-concentration Atropine for Myopia Progression (LAMP) study.1, 2 

Methods: The first year of the LAMP study was a robustly-designed, placebo-controlled trial comparing different low concentrations of atropine with a large sample size (N > 100 randomized to each group).1 Li et al. subsequently published adjusted mean axial elongation and myopia progression rates generated in generalized estimating equations by age group.2 We used these data to calculate efficacy in terms of both absolute reduction in myopic progression and absolute reduction in axial elongation for each of the different atropine concentrations at each age group. We then compared these efficacy data to the overall progression for each of the two progression metrics.

Results: The figure plots efficacy as a function of overall myopia progression and axial elongation rates for each of three different age groups. The data demonstrate the invariant nature of efficacy, from the perspective of clinical importance of absolute reduction in progression, despite a substantial range of overall progression. That is, faster progressors, or so-called non-responders, achieved similar reduction in axial elongation and myopia progression as the slower progressors (so called responders) within the various atropine treatment groups.

Conclusion: The use of the terms responders and non-responders during myopia progression interventions is erroneous. Those designated as such are simply slower or faster progressors but the real response to treatment is unknown. While changes in treatment regimens for fast progressors will seem to achieve better results through regression to the mean, it is unclear that arbitrary switching of treatments provides any real benefit and is likely due to seasonal or annual variations in untreated progression. Switching to treatments with proven superiority in terms of efficacy, addition of therapies (i.e., combination therapy) and improving compliance are more likely to be beneficial remedies.

1. Yam et al. Ophthalmology. 2019;126:113
2. Li et al. Ophthalmology. 2021;128:1180