Seventy patients with DM1 with mild signs of NPDR (37 male patients and 33 female patients; mean age ± standard deviation, 48.36 ± 6.34 years) were screened for enrollment in the study.
The diagnosis of mild NPDR was determined by two retinal specialists (MP and FS) through indirect ophthalmoscopy, slit lamp stereo biomicroscopy, and stereoscopic fundus photography. Mild NPDR was defined as the presence of at least one microaneurysm and/or hemorrhage in the central retina in the absence of peripheral lesions (stage 2 of the International Clinical Diabetic Retinopathy Disease Severity Scale [ICDRSS]) .
Inclusion criteria were patients with DM1 with mild NPDR and impairment of retinal sensitivity based on FDT findings of mean deviation (MD) P < 5%, or two locations with P < 5% and one location with P < 1% in the total, or pattern deviation plots, and best corrected visual acuity (BCVA) > 0.1 logarithm of the minimum angle of resolution (logMAR).
Exclusion criteria were hyperopia higher than + 3 diopter (D), myopia higher than − 6 D, or astigmatism higher than 2 D and BCVA < 0.1 logMAR, media opacity, previous ocular surgery, previous diagnosis of glaucoma, uveitis, other retinal disease, and ocular or systemic disease other than diabetes or mild DR; in addition, those with signs of retinopathy more advanced than mild (e.g., macular edema or any sign of proliferative DR) and poor metabolic status (hemoglobin A1c [HbA1c] > 9%) at baseline or during follow-up (see below) were excluded.
When both eyes fulfilled the inclusion criteria, the eye with worse BCVA was selected; when both eyes showed the same BCVA, the right eye was selected for analysis. As a result, 20 eyes with mild NPDR of 20 patients with DM1 (11 male patients and nine female patients; mean age ± standard deviation, 47.35 ± 8.25 years) were enrolled in the study.
This is a prospective, interventional, randomized, double-masked, monocentric pilot study. It was conducted according to the tenets of the Declaration of Helsinki and approved by the local ethics committee (Comitato Etico Centrale IRCCS Lazio, Sezione IFO/Fondazione Bietti, Rome, Italy). The study was registered at ClinicalTrials.gov (NCT04009980). In all patients, signed informed consent forms were obtained at the time of recruitment.
Patients with DM1 were randomly divided into two age-matched groups of 10 patients each: (a) The DC group comprised those who received treatment with citicoline and vitamin B12 eye drops (10 eyes of five male and five female patients with mean age ± standard deviation of 46.86 ± 8.78 years; mean duration of disease of 25 ± 12.82 years; mean percentage of HbA1c of 7.36 ± 0.96; and mean BCVA (Early Treatment Diabetic Retinopathy Study [ETDRS] letter score) of 87.57 ± 3.51). (b) The DP group comprised those who received placebo treatment (10 eyes of six male patients and four female patients with mean age ± standard deviation of 47.89 ± 7.74 years; mean duration of disease of 21.78 ± 9.42 years; mean percentage of HbA1c of 7.81 ± 0.79; and mean BCVA (ETDRS letter score) of 86.33 ± 5.85.
The patients with DM1 with screening by MP and PG were randomly separated into two groups by an electronically generated randomization system according to age, duration of disease, percentage of HbA1c, and similar values of FDT (mean sensitivity [MS] of 10-2, described below).
During 36 months of trial, the patients in the DC group were treated with citicoline and vitamin B12 eye drops (OMK2® containing citicoline 2%, hyaluronic acid 0.2%, and cyanocobalamin 0.05%; Omikron Italia srl, Italy), 1 drop thrice daily, while those in the DP group were treated with placebo (eye drops containing hypromellose 0.3%), 1 drop thrice daily.
During the trial period, three eyes each of the DC and DP groups were lost to follow-up as a result of patients’ unavailability at the time of examination. Finally, seven eyes in the DC group and seven eyes in the DP group that completed the study were analyzed.
Compliance with eye drops treatment was assessed through a questionnaire administered by the study personnel at each visit. As expected in clinical study, self-reported adherence to treatment was high, and in all patients, rating of compliance as good to very good (regular use of eye drops in at least 80% of the trial period) was obtained.
The randomization key was opened to all investigators at the end of the follow-up.
In each patient of the DC and DP group, examinations were performed at baseline and 12, 24, and 36 months as follows.
Psychophysical Evaluation (FDT)
In both the DC and DP groups, two FDT strategies using thresholds of 24-2 and 10-2 were used, according to the method in our previous study . At baseline and during follow-up, to assess test–retest variability and rule out relevant learning curve, FDT examinations were performed at two separate recording sessions within 2 weeks. Reliability criteria were defined according to fixation errors less than 33%, false positives, and false negatives. Values obtained at the second examination were submitted to statistical analysis. If the second test was deemed unreliable, a third examination was performed, which was included in the statistical analysis. In both FDT 24-2 and 10-2 strategies, the values of the following parameters were considered: mean deviation (MD), pattern standard deviation (PSD), and MS measured as the average sensitivity at all test locations divided by the total number of locations (except the fovea). MD P value of less than 5% was considered as the threshold of abnormality.
Spectral Domain Optical Coherence Tomography (SD-OCT) and OCT Angiography (OCTA)
In all patients of the DC and DP groups, SD-OCT and OCTA were conducted three times according to the method described in our previous studies [9, 13], and the average values for each parameter (see below) were considered for subsequent analysis (Spectralis; Heidelberg Engineering GmbH, Heidelberg, Germany; OCTA AngioVue XR Avanti; Optovue Inc., Fremont, California, USA).
In SD-OCT, the macular measurements (subfoveal, parafoveal, and total values) were obtained using in-built Spectralis mapping software, and segmentation software  was used for individual retinal layer thicknesses: RNFL, ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL), retinal pigment epithelium (RPE), and photoreceptor layer (PR), and central retinal (CRT).
In OCTA, en face OCT angiograms were segmented to identify the superficial capillary plexus (SCP) and DCP. The SCP slab was acquired from the internal limiting membrane (ILM) to IPL/INL (9 μm above), while the DCP slab was acquired from the IPL/INL junction (9 μm above) to the OPL/ONL junction (9 μm below), of which, the latter additionally included the intermediate capillary plexus. Moreover, parafoveal vessel density (VD), defined as the percentage of total area occupied by the vessels and microvasculature, was quantified in SCP and DCP within an annulus centered on the fovea with inner and outer ring diameters of 1 and 2.5 mm. The area of the foveal avascular zone (FAZ) was measured using the no flow function of the software. Accuracy of image segmentation in SD-OCT and OCTA was verified independently by two readers (MP and FS).
Adaptive Optics (AO)
In all patients of the DC and DP groups, AO was conducted using Rtx1 (Imagine Eyes, Orsay, France) according to the method in our studies [18, 19]. On the basis of our previous findings by AO , the following parameters were considered: cone density, linear dispersion index, and heterogeneity packing index.
Sample size was estimated on the basis of a pilot study including four eyes of four patients with DM1 without signs of DR who were not included in the present study (unpublished data). Inter-individual variability is expressed as standard deviations (SD) for FDT 10-2 MS; in particular, the size of patient groups was calculated using mean value of 29.77 decibel (dB) and SD of 1.58 dB. Assuming a change of FDT 10-2 MS of about 10% with SD of 1.5 dB during the follow-up period, six patients in each group were required at α = 0.05 and β = 0.20 (power 1 − β = 0.80). On the basis of a predicted dropout rate of 40% of patients during follow-up, 10 patients per group were enrolled at baseline.
The results of the patient’s age, BCVA, and instrument measurements are expressed as mean ± SD. Normal data distribution was assessed using one-sample Kolmogorov–Smirnov test. The differences in values at each timepoint between the DC and DP groups were compared using independent-samples Mann–Whitney test. Differences in values over time were assessed using MAnova (general linear model [GLM] within factor time and between factor group) repeated measures or Friedman test, as appropriate. Intention-to-treat analysis was performed to avoid bias due to missing information. In the comparison of treatment groups, patients’ allocation after randomization was considered in the last value carried forward method.
In all analyses, P < 0.05 was considered as statistical significance. Statistical evaluation was performed using SPSS software (version 17.0; SPSS Inc, Chicago, IL).