Introduction

Age-related macular degeneration (AMD) is a common global cause of blindness and is associated with vision impairment, falls, loss of independence and depression1,2. Although treatments are currently available only for late AMD – i.e. the disease stage when such outcomes occur3,4 – reductions in visual function and vision-related quality of life already appear in earlier AMD stages5,6. Visual domains affected in early and intermediate AMD include night vision, contrast vision, and dark adaptation, which are rarely assessed in routine patient care and large-scale clinical trials. Also, multiple visual function assessments are time- and cost-intensive and difficult to implement in trials. Patient-reported outcome measures (PROMs), on the other hand, are easy to obtain on site or remotely, standardized, and measure concepts relevant to patients, which is increasingly requested by regulatory authorities7.

The Vision Impairment in Low Luminance (VILL) questionnaire has recently been introduced and validated to assess vision-related quality of life in AMD6,8,9. It overcomes disadvantages of other previously used instruments. The VILL was developed in accordance with the requirements of the United States Food and Drug Administration (FDA) guideline for PROM development10 and has been shown to be internally consistent, test–retest reliable, inter-mode reliable, content valid, and able to differentiate between different AMD stages6,8,9.

Several visual function assessments including best-corrected and low-luminance visual acuity, low luminance deficit, contrast sensitivity, Moorfields acuity, and rod intercept time from adaptometry11,12,13,14,15 are associated with vision-related quality of life and patient-reported symptoms of AMD. However, most of these data were obtained from visual function assessments of only one eye, which limits the ecological (i. e., ‘real-life’) external validity of these analyses.

Despite increasing recognition of the importance of patient-reported outcomes in AMD, evidence supporting the construct validity of newer, specific instruments like the VILL questionnaire remains limited. In particular, associations between bilateral functional vision measures and patient-reported difficulties under low luminance have not been thoroughly investigated, leaving a critical gap in validating such tools for use in clinical trials. We thus analyzed the association between vision-related quality of life, as assessed by the VILL questionnaire, and visual function in both eyes in the MACUSTAR study, a multi-centre European cohort study on intermediate AMD and neighbouring disease stages.

Methods

Participants

We included participants of the cross-sectional part of the MACUSTAR study. The MACUSTAR study is a multi-centre cohort study investigating functional, structural and patient-reported outcomes in intermediate AMD. It is conducted at 20 study sites across seven European countries, from which 18 sites recruited participants for the cross-sectional part referred to in this report16,17.

In summary, 586 participants with intermediate AMD and 133 control participants with bilateral early AMD (n = 34), late AMD (n = 43) or no AMD (n = 56) were recruited for the MACUSTAR study from early 2018. The disease categorization of AMD followed the Beckman classification18. Comprehensive six- to twelve-monthly assessments including visual function testing, multimodal imaging and administration of PROMs were performed in all participants with intermediate AMD or early AMD within the longitudinal part of MACUSTAR.

Within the cross-sectional part of the MACUSTAR study, 168 participants with bilateral intermediate AMD and all control participants attended an additional retest visit scheduled 2 weeks (± 1 week) after the baseline visit. Acquisition of functional, structural and patient-reported data followed procedure manuals specifically developed for the study and all functional data underwent a formal, pre-specified review procedure on a semi-annual basis to ensure that the multi-centre setting of MACUSTAR does not compromise data quality. Imaging data were graded by a central reading centre (GRADE Reading Center, University Hospital Bonn, Germany)16,17.

We included only participants with AMD in our analysis and thus excluded healthy controls (Supplementary Figure). We also excluded participants who only participated in the longitudinal part, where disease stage was allowed to be asymmetric.

The study protocol and additional information on the conduct of the MACUSTAR study have been published previously16,17 and can be found on the study website www.MACUSTAR.eu. Ethics committees at all participating study sites approved the conduct of the MACUSTAR study and it followed the tenets of the Declaration of Helsinki. All participants provided written informed consent. The clinicaltrials.gov registration identifier is NCT03349801.

Vision Impairment in Low Luminance (VILL) questionnaire

The VILL questionnaire is a 33-item PROM assessing vision-related quality of life in individuals with AMD, specifically designed for future use in intermediate AMD clinical trials. Its item pool was developed with extensive patient input (in-depth interviews, focus group discussions, cognitive debriefs in people with AMD) and it focuses on the visual difficulties in low-luminance and low-contrast situations which is frequently reported by people with AMD8. An initial 37 item-version of the VILL was reduced to a final set of 33 items based on the MACUSTAR data6,8. The VILL includes items from the three independent subscales “reading and accessing information” (17 items, “reading”), “mobility and safety (12 items, “mobility”) and “emotional well-being” (4 items, “emotional”). The response scale of the VILL includes four response options per item and a separate response option ”not applicable”. The VILL is scored through Rasch models which have been established and psychometrically validated in the MACUSTAR cohort previously6,8.

The VILL has been shown to be an internally consistent, test–retest reliable, content and construct valid PROM which yields similar scores across modes of administration (interview administration, self-administration via paper and electronic self-administration)6,9. Within MACUSTAR, the VILL was administered in a standardized manner, using a PROM administration manual by the clinical site personnel in the respective local language6,17.

Visual function assessments

Visual function assessments implemented in the MACUSTAR study include best-corrected (BCVA) and low-luminance visual acuity (LLVA), Moorfields acuity (MA), Pelli-Robson chart contrast sensitivity (PR-CS), mesopic and scotopic fundus-controlled perimetry (microperimetry; S-MAIA, CenterVue, Padova, Italy) as well as dark adaptometry (AdaptDx, Maculogix, Pennsylvania, USA). The assessments available from both eyes were selected as main read-outs for in this study, i.e. BCVA [logMAR], LLVA [logMAR], MA [logMAR] and PR-CS [logCS], which are all chart-based. In a supplemental analysis, average threshold from mesopic and scotpic microperimetry (mAT and sAT [dB]) as well as rod intercept time from dark adaptometry (RIT [min]) were also considered. However, mAT, sATm, and RIT were only available for one eye, which was not necessarily the better eye (i.e., the eye with higher BCVA). Reading performance was also assessed in the MACUSTAR study but not considered in our analysis due to the performance-based nature of reading speed assessments, contrasting the other visual function metrics assessed.

All functional testing procedures were conducted by specifically trained and tested staff that followed specific testing protocols as reported previously5,17,19. Visual function testing was conducted separately on both eyes in the MACUSTAR study. Following the study protocol, both eyes had the same AMD stage. Previous research supports that visual function of both eyes impacts vision-related quality of life20,21,22,23, which was the rationale for including only the visual function assessments that were available from both eyes in our main analysis. In a supplemental analysis, data from only the study eyes were considered, which were selected based on better visual acuity at screening if both eyes were considered eligible.

Statistical analysis

To investigate the relationship between the VILL subscale person measures and the above-mentioned visual function parameters of both eyes in all participants with AMD, we fitted linear regression models with VILL person measures as dependent variables and the respective functional parameters as independent variables, controlling for age, sex, and AMD stage. Next, we fitted linear regression models using the same approach in the intermediate AMD subgroup, which is the main target population of the MACUSTAR study. All analyses were conducted using R (version 4.3.0, R Core Team, Vienna, Austria). P-values < 0.05 were considered statistically significant.

Results

We included 245 participants with AMD in our analysis (early AMD, n = 34, 79% female, mean age 72 ± 6 years; intermediate AMD, n = 168, 65% female, mean age 71 ± 8 years; late AMD, n = 43, 64% female, mean age 72 ± 7 years). Mean logMAR BCVA was 0.15 ± 0.32 for study eyes and 0.20 ± 0.36 for fellow eyes (Table 1).

Table 1 PROM and visual function data from all included participants (n = 245).
Full size table

Vision-related quality of life and visual function across AMD stages

Absolute Pearson correlation coefficients between VILL subscale scores and chart-based visual function parameters from both eyes in all 245 participants with AMD ranged between 0.50 and 0.67 for the VILL reading subscale, between 0.48 and 0.58 for the mobility subscale and between 0.28 and 0.36 for the emotional subscale (Fig. 1). When introducing these variables into a regression model adjusted for age, sex and AMD stage, all visual function parameters were significantly associated with the VILL reading, mobility, or emotional subscale scores (Table 2). Specifically, BCVA, LLVA, MA and PR-CS were associated with VILL-Reading and VILL-Mobility in both the better and the worse eye. VILL-Emotional was associated with MA and PR-CS in the worse-seeing eye.

Fig. 1
Fig. 1
Full size image

Correlation heatmap of VILL subscale scores (displayed in rows) and visual function testing results from both eyes (columns) across AMD stages (n = 245). BCVA, best-corrected visual acuity; BE, better eye; LLVA, low-luminance visual acuity; MAT, Moorfields acuity test; PR-CS, Pelli-Robson contrast sensitivity; WE, worse eye.

Table 2 Linear regression analysis of functional assessments (independent variables) in all participants with AMD (n = 245) against VILL subscales (dependent variables).
Full size table

Vision-related quality of life and visual function in intermediate AMD

In the subgroup of 168 participants with intermediate AMD, BCVA, LLVA, MAT and PR-CS were associated with VILL mobility subscale scores in the worse-seeing eye when controlling for age and sex (Table 3). PR-CS was also significantly associated with patient-reports in the better eye (i.e., the VILL mobility subscale) and with the VILL reading subscale, while MAT was associated with the VILL emotional well-being subscale (Table 3). Notably, effect sizes were generally larger in models that included visual function data from the worse eye compared to the better eye.

Table 3 Linear regression analysis of summed functional assessments (independent variables) in all participants intermediate AMD (n = 168) against VILL subscales (independent variables), controlled for age and sex.
Full size table

In the MACUSTAR study, microperimetry and dark adaptometry were performed only in the study eye. Therefore, we additionally evaluated associations between VILL scores and unilateral visual function in the study eye (Supplementary Table). This analysis revealed significant associations between the VILL reading and emotional well-being subscales and average thresholds on microperimetry.

Discussion

Our findings show that patient-reported vision impairment in low luminance, as measured by the VILL questionnaire, is generally associated with bilateral visual function deficits across all AMD stages – and specifically in intermediate AMD. This further supports the construct validity of the VILL as a tool for assessing visual functioning and vision-related quality of life in AMD. Additionally, our results underscore contrast sensitivity as a key factor contributing to everyday visual difficulties, particularly in individuals with intermediate AMD.

This is the first study to investigate the association between visual function measures and patient-reported low-luminance vision impairment in individuals with intermediate AMD using the VILL questionnaire. These findings are clinically and regulatorily relevant, as they support the use of the VILL as a potential endpoint for assessing functional vision and vision-related quality of life in this patient population. Other patient-reported instruments assessing general vision-related quality of life, as well as those specifically targeting mesopic and scotopic visual functioning, have previously been compared to visual function measurements11,24,25,26. The Low Luminance Questionnaire (LLQ) was associated with dark adaptation, as reported by Owsley and colleagues27. This was consistent with a later study by another group, in which rod intercept time from dark adaptometry was highly associated with LLQ scores, alongside BCVA and contrast sensitivity, which were linked to some of the subdomains28. In a more recent paper, LLVA and low luminance deficit – defined as the difference between BCVA and LLVA – were significantly associated with LLQ composite scores, whereas BCVA, rod intercept from dark adaptation, and microperimetry thresholds were not13. Wu and coleagues used the 10-item Night Vision Questionnaire (NVQ)29 in 100 individuals with bilateral intermediate AMD and found that the PROM scores were not significantly associated with BCVA, LLVA, or retinal sensitivity measured by microperimetry12. However, the low luminance deficit was significantly greater in individuals reporting night vision difficulties in their study.

Our analysis adds to these results by showing that chart-based visual function tests were indeed associated with vision-related quality of life across AMD stages, and specifically with visual functioning of the worse eye in intermediate AMD. In contrast to previous studies, we included MAT and PR-CS in addition to BCVA and LLVA. Notably, PR-CS was significantly associated with visual functioning in intermediate AMD (i.e., reading and accessing information, mobility and safety). The association between the VILL’s mobility subscale and PR-CS was significant for both the better and worse eyes. The particular importance of contrast vision is also supported by previous qualitative data8,30. In addition to chart-based visual function tests, unilateral mean sensitivity reduction on microperimetry was associated with the reading and emotional well-being subscale scores of the VILL questionnaire in our data, contrasting the findings of Thompson et al. in a smaller cohort with varying disease stages13. Dark adaptometry in the study eye, however, did not have significant associations with any of the VILL scores, which could be related to the different testing spot in MACUSTAR compared to other protocols27. Interpreting our data in the context of similar PROMs further supports the VILL’s construct validity and suggests that the VILL is a suitable tool for future clinical studies and drug trials in the context of AMD.

The results of our study may serve as a baseline interpretation of the minimally important difference (MID) for the VILL questionnaire, using visual function as an anchor. Regulators commonly accept a three-line change in BCVA as clinically meaningful7, which could be translated into patient-relevant differences of 0.41 logits in the VILL-Reading subscale, 0.55 logits in the VILL-Mobility subscale, and 0.72 logits in the VILL-Emotional subscale. Previous data support that visual function in the better-seeing eye – used for this estimation – is a stronger predictor of vision-related quality of life than visual function in the worse-seeing eye23. However, this interpretation of our results as MIDs is limited by the absence of longitudinal data and a patient-reported anchor variable.

Our main results are based on the analysis of both eyes, which is known to better reflect vision-related quality of life than monocular vision in the context of visual impairment20,21,22,23. The particular importance of contrast vision of both eyes for vision-related quality of life in intermediate AMD is strongly supported by our findings, where PR-CS in both better and worse eyes was associated with VILL mobility subscale scores. Similar studies to ours have stratified the ‘better’ and ‘worse’ eyes based on BCVA11,12,27,28,31, while the relative value of other AMD-affected function domains (e.g., contrast vision, dark adaptation) to patients remains unknown.

Strengths of our study include the large, multicentric and thoroughly phenotyped sample with the same disease stage in both eyes; a broad battery of visual function tests performed under highly standardized settings by experienced study teams; and the validated VILL questionnaire, which was specifically developed to be used in the context of AMD, using both qualitative and quantitative methods and is scored with Rasch models. Limitations include the lack of binocular visual function testing, which we addressed by separately analysing data from the eye with better and worse BCVA. Some assessments, including microperimetry and dark adaptometry were performed in only one eye, and were thus used for generating supplementary findings only. Lastly, we were unable to directly compare associations between the VILL and visual function assessments with other PROMs, as the VILL is the only ophthalmic PROM included in the MACUSTAR study.

In conclusion, our data show that vision-related quality of life in AMD, as measured by the VILL questionnaire, is associated with a variety of visual function assessments. In intermediate AMD, contrast vision – when considering both eyes – is particularly linked to visual functioning. Our findings support the validity of the VILL questionnaire as a patient-relevant measure for future studies.