**2. Materials and Methods**

#### *2.1. Population Study*

The Montrachet study (Maculopathy Optic Nerve and nutrition neurovAsCular and HEarT) was an ancillary study of the population-based Three Cities (3C) study, which had been previously reported [20]. In the 3C cohort study, 9294 persons were randomly selected from the electoral rolls of three French urban cities (Bordeaux, Dijon, and Montpellier) and aged 65 years. Ten years later, the subgroup of participants from Dijon was invited to participate in the Montrachet study.

The methodology of the Montrachet study and baseline characteristics of volunteers have already been described [21]. From October 2009 to March 2013, 1153 volunteers underwent complete eye examination in the Department of Ophthalmology of the Dijon University Hospital, France. All participants were asked to complete a questionnaire about their lifestyle (alcohol consumption and smoking status) and environment (sun exposure). This examination included the collection of self-reported eye disease and treatment history, visual acuity measurement, intraocular pressure measurement (Tonoref II, Nidek, Gamagori, Japan), central corneal thickness measurement with an ultrasonic contact pachymeter (DGH 500, DGH Technology, Exton, PA, USA), visual field examination with a screening program (Frequency-Doubling Technology, Carl Zeiss Meditec, Dublin, CA, USA), and spectral-domain optical coherence tomography (SD-OCT; software version 5.4.7.0; Spectralis, Heidelberg Engineering Co., Heidelberg, Germany) for both the macula and the optic nerve head after pupil dilation. The high-speed resolution mode and the eyetracking system were activated to acquire the images. For the optic disc, the retinal nerve fiber layer (RNFL) thickness acquisition was obtained with a circle diameter of 3.5 mm. Moreover, each participant benefited from two retinal photographs: one centered on the macula and one on the optic nerve (TRC NW6S, Topcon, Tokyo, Japan). In addition, fasting blood samples were drawn to measure plasma carotenoids and fatty acids. We also used a semiquantitative food frequency questionnaire to quantify the dietary intake of lutein and zeaxanthin among participants. The study was approved by the regional ethics committee and was registered as 2009-A00448-49. All participants gave their informed consent. This study adhered to the tenets of the Declaration of Helsinki and we followed the STROBE policy according to the EQUATOR guidelines [22].

#### *2.2. Glaucoma Diagnosis*

Glaucoma diagnosis in the Montrachet study has already been reported [23,24]. Briefly, optic disc photographs were reviewed by two trained ophthalmologists (LA and PHG) masked for clinical and RNFL thickness information. In case of discrepancy, a senior glaucoma specialist (AMB) made an adjudication. Suspected-glaucoma eyes benefited from a new examination with gonioscopy and a Humphrey Swedish Interactive Thresholding Algorithm 24-2 visual field (Carl Zeiss Meditec, Dublin, CA, USA). Glaucoma was defined by the glaucoma classification of the International Society of Geographical and Epidemiological Ophthalmology (ISGEO) [25]. In case of bilateral disease, the eye with the most severe presentation was kept for analysis. Only cases of primary open-angle glaucoma (POAG) were considered for the present analysis. Secondary glaucoma, angle-closure glaucoma, and non-glaucomatous optic neuropathy were excluded from the analysis.

#### *2.3. MPOD Measurements*

The MPOD was measured with the two-wavelength autofluorescence method using a modified Heidelberg Retina Angiograph (HRA; Heidelberg Engineering, Heidelberg, Germany). After pupil dilation with tropicamide 0.5% (Thea, Clermont-Ferrand, France), two acquisitions were performed at a 30 s interval and captured at 488 and 514 nm excitation wavelengths by a trained technician for both eyes (Figure S1). Glaucomatous status was masked to the operator. MPOD maps were generated by digital subtraction of the log autofluorescence images. We recorded MPOD at 0◦, 0.5◦, 1◦, 2◦, and 6◦ eccentricities using the software provided by the manufacturer of the device. MPOD was expressed in optical density units (DU). In the control participants group without optic neuropathy, the eye with the best image quality was retained for analysis. The right eye was chosen when image quality was similar in both eyes. We excluded participants with poor image quality in both eyes and those who suffered from late age-related macular degeneration (AMD). Classification of late AMD was based on both color fundus and OCT images [26]. From the graphs generated by the software of the modified HRA, we divided the different MP spatial distribution profiles into three groups (ring-like, no ring, and intermediate). A second investigator (PHG) analyzed fifty eyes of our population randomly chosen independently from the first investigator (LA).

### *2.4. Blood Sampling*

Blood samples were collected from fasted volunteers for plasma lipids and fatty acids analysis [27]. Lipids extracted from plasma were stored under inert gas and then transmethylated with boron trifloride in methanol [28]. Finally, fatty acid methyl esters were isolated with hexane and analyzed by gas chromatography using the Hewlett Packard Model 5890 (Palo Alto, CA, USA) with a CPSIL-88 column (100 m × 0.25 mm i.d., fim thickness 0.20 μm; Varian, Les Ulis, France) equipped with a flame ionization detector. The carrier gas used was hydrogen.

#### *2.5. Statistical Analysis*

Categorical variables were expressed as a number (*n*, percentage) and continuous variables as mean ± standard deviation (SD) or median (interquartile range (Q1–Q3)) according to their distribution. We performed our analysis with one eye per individual as the unit of analysis. We displayed two groups: control eyes and eye with POAG. Bivariate analysis was performed with Pearson's chi-square or Fisher exact tests for percentage comparisons as appropriate. Student test or analyses of variance, or the Kruskal–Wallis test was used for comparison of mean or median when appropriate. Cohen's kappa coefficient was used to measure the concordance between the two eyes for MPOD and for MP spatial distribution profiles between the two investigators. For multivariable analysis between MPOD at 0.5◦ and the presence of POAG as a dependent variable, all variables associated with glaucoma in the bivariate analysis with a *p*-value < 0.20 were included in the model. The smoking status variable was forced in the model. Then, a final model adjusted for age, sex, smoking status, lens status, plasma alpha-linoleic acid (ALA), and eicosapentaenoic acid (EPA) n-3 polyunsaturated fatty acids (PUFAs) was obtained by manual backward selection. For all analyses, the tests were two-tailed and the results were considered significant when *p*-values were less than 0.05. Analyses were performed using SAS software (version 9.4; SAS Institute, Inc., Cary, NC, USA).
