Investigating the Neuroprotective Effect of Oral Omega-3 Fatty Acid Supplementation in Type 1 Diabetes (nPROOFS1): A Randomized Placebo-Controlled Trial

Diabetic sensorimotor polyneuropathy (DSP) affects ∼50% of individuals with diabetes (1). Progressive nerve damage, which can lead to debilitating symptoms and disability, has profound quality-of-life and socioeconomic impacts (2). Early-stage DSP is associated with loss of small sensory nerve fibers in the cornea (3). Quantification of corneal subbasal nerve plexus parameters, from in vivo confocal microscopy (IVCM) imaging, provides a noninvasive, sensitive, and reliable marker for monitoring DSP progression (4). IVCM has also been used to monitor corneal nerve regeneration following pancreas and kidney transplant in individuals with type 1 diabetes (5).

Intensive glycemic control is the only established method for slowing DSP progression in type 1 diabetes (6). There is a need for disease-modifying therapies to reduce progressive peripheral nerve damage. A promising intervention is omega-3 (n-3) polyunsaturated fatty acids (PUFAs), derived from the diet or via supplementation. Long-chain n-3 PUFAs, eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), impart a range of biological effects, including modulation of cellular signaling and production of anti-inflammatory and neuroprotective mediators (7,8). In rodent models, n-3 PUFA supplementation improves sensory function recovery after peripheral nerve injury (9) and reduces corneal nerve loss in experimental diabetes (10).

In a pilot randomized controlled trial (RCT), long-chain n-3 PUFA supplementation (1,500 mg/day for 3 months) significantly increased corneal nerve density, relative to placebo, in dry eye disease (11). An open-label, single-arm trial found that long-chain n-3 PUFA supplementation for 12 months (2,330 mg/day) increased corneal nerve fiber length (CNFL) by 29% relative to pretreatment levels in type 1 diabetes (12). However, our recent systematic review identified a paucity of RCTs for evaluating the effects of oral n-3 PUFA supplementation on peripheral nerve structure and/or function in individuals with diabetes (13).

To address this evidence gap, we conducted an RCT with double masking to investigate the effects of 6 months of long-chain n-3 PUFA supplementation on peripheral nerve parameters in individuals with type 1 diabetes.

This was a single-center, double-masked, randomized, two-arm, parallel-group, superiority, placebo-controlled trial. Study procedures were undertaken in the Department of Optometry and Vision Sciences, University of Melbourne, and Department of Neurology/Neurophysiology, St. Vincent’s Hospital Melbourne. This study was approved by St. Vincent’s Hospital Human Research Ethics Committee (HREC/17/SVHM/236), registered with University of Melbourne Human Research Ethics Committee (identifier 1851526), and undertaken in accordance with the Declaration of Helsinki. The trial was registered prospectively on the Australian New Zealand Clinical Trials Registry (ACTRN12618000705280) and conducted according to an a priori protocol (available as Supplemental Material). Written informed consent was obtained from all participants prior to performing any study-related procedures. Trial reporting conforms with the guidelines of the 2010 Consolidated Standards of Reporting Trials (CONSORT) statement (14).

Participants were enrolled at the University of Melbourne, where they underwent eligibility assessment and ocular examinations. A full list of eligibility criteria is provided in Table 1.

Participants were randomized (1:1) to receive either oral n-3 PUFA or placebo (olive oil) supplements, according to a computer-generated randomization list in block sizes of four. To ensure allocation concealment, an independent data manager developed the randomization code and held the code until unmasking. The independent data manager provided the randomization code to a pharmacist (Dartnell’s Pharmacy, Surrey Hills, Australia), who packaged the supplements into identical opaque containers that were consecutively numbered to administer to participants. Participants and personnel, including outcome assessors and data analysts, were masked to treatment allocations.

Participants were advised to consume two supplement capsules per day for 180 ± 24 days. The n-3 PUFA (active) group received long-chain triglyceride n-3 PUFA supplements (Triple Strength Fish Oil concentrate; Caruso’s Natural Health, Eastern Creek, New South Wales, Australia), totaling 1,080 mg/day EPA and 720 mg/day DHA. The placebo group received olive oil supplements (BJP Laboratories, Yatala, Queensland, Australia) (15), equivalent to 600 mg/day. Olive oil is an appropriate, inert control for n-3 PUFA investigations; its major component (oleic acid, an n-9 monounsaturated fatty acid) does not affect systemic PUFA levels (16).

Participants attended St. Vincent’s Hospital Melbourne, within 35 days, for baseline neurophysiology examinations. They were instructed to begin taking their assigned supplements on the day of the neurophysiological examination (day 0). Participants attended follow-up visits at days 30 ± 14, 90 ± 14, and 180 ± 24 (study end point) from initiating treatment.

The primary outcome measure was change in central CNFL (mm/mm²), measured from IVCM images, at day 180. To minimize potential bias from manual quantification, the image analysis was performed using automated software (ACCMetrics) (17). For each participant, central CNFL was recorded as the average measure from 12 randomly selected images (see Neuroprotective Effect of Oral Omega-3 Fatty Acid Supplementation in Type 1 Diabetes [nPROOFS1] protocol, available as Supplemental Material).

Key secondary efficacy outcomes were change from baseline at day 180 for central corneal nerve branch density (CNBD) (branches/mm²) and corneal nerve fiber density (CNFD) (nerves/mm²), quantified from IVCM images, and central corneal sensitivity thresholds (mbar) to room temperature and cooled stimuli, measured with a noncontact corneal aesthesiometer (SDZ Electronics, Auckland, New Zealand).

Secondary outcomes were change from baseline at day 180 for patient-reported quality of life, with use of the five-level EuroQol five-dimensional descriptive system (EQ-5D-5L) questionnaire (18); cutaneous silent period (CSP) onset latency and duration in the upper and lower limbs; sweat volume and latency in the foot, with use of the quantitative sudomotor axonal reflex test (QSART); Michigan Neuropathy Screening Instrument (MNSI) score and the Michigan Diabetic Neuropathy Score (MDNS) (19); nerve conduction study (NCS) parameters for sural sensory amplitude, peroneal motor conduction velocity, and tibial minimum F-wave latency; and median nerve axonal excitability, with use of the TROND protocol (20), for the following parameters: depolarizing threshold electrotonus (TEd) peak, TEd 10–20 ms, TEd 90–100 ms, hyperpolarizing threshold electrotonus (TEh) 90–100 ms, superexcitability, subexcitability, resting current-voltage slope, and strength duration time constant. Safety outcomes were the incidence of adverse events and change from baseline at day 180 in habitual distance visual acuity, intraocular pressure, diabetic retinopathy grading, and blood pathology parameters.

Compliance was assessed by counting of unused supplement capsules, returned at the final visit, by an independent technician. Investigators involved in the collection and analysis of data were not involved in reconciling returned investigational product. Participants with compliance <75% or >125%, or who did not return remaining supplements, were considered noncompliant. Change in erythrocyte fatty acid concentrations were quantified with dried blood spot analysis (PUFAcoat technology; Xerion Diagnostics Pty Ltd, Brighton, Victoria, Australia), analyzed by an independent laboratory (Waite Lipid Analytical Service, Adelaide, South Australia, Australia) using validated protocols (21). Erythrocyte PUFA parameters were analyzed for Omega-3 Index (%), a metric of erythrocyte EPA and DHA concentrations; EPA (%); DHA (%); total n-6 (%); total n-9 (%); and the n-6:n-3 ratio.

Exploratory outcomes included change from baseline at day 180 for sweat volume and latency at the forearm, proximal leg, and distal leg, with use of QSART; motor NCS of the median, peroneal, and tibial nerves; sensory NCS of the median, sural, and ulnar nerves; axonal nerve excitability of the median nerve; and tear film and ocular surface parameters.

A total of 21 participants per group was estimated to give 80% power at a two-sided 5% level of significance, assuming a difference between intervention arms in CNFL at day 180 of 2.9 mm/mm²; an SD of 3.2 mm/mm², equal in both groups; and no correlation between baseline and postbaseline measurements (11).

Statistical analyses were performed according to the intention-to-treat (ITT) principle, as defined in a statistical analysis plan (available as Supplemental Material), finalized prior to unmasking. The primary outcome is presented as the absolute difference in change from baseline, obtained using a constrained longitudinal data analysis model. Other continuous outcomes were analyzed using the same model, with outcomes log base e transformed prior to model fitting, where applicable, presenting the geometric mean ratio. For binary and categorical variables, Fisher exact test was used to compare the proportion of participants from each treatment group regarding the change from baseline during the study. Analyses of continuous outcomes provide valid inference when the missing data are at most missing at random. Binary and categorical variables were analyzed using complete cases. Multiple imputation was not applied, given the low proportion of missing data and the sample size.

Two adjusted models were prespecified for primary, key secondary, and secondary outcomes, with adjustment for age, diabetes duration, and HbA_1c (model 1) and, additionally, potential imbalances in baseline CNFL and Omega-3 Index (model 2). A post hoc sensitivity analysis was performed for the primary and key secondary outcomes with adjustment for baseline Omega-3 Index as a potential prognostic factor. Primary and key secondary outcomes were also analyzed with exclusion of participants who were noncompliant. For outcomes with baseline imbalance, we performed a sensitivity analysis by removing the constraint of balance at baseline. Preplanned subgroup analyses were performed for the primary outcome for subgroups with CNFL >12.5 mm/mm² and ≤12.5 mm/mm² (22) and presence/absence of small fiber neuropathy (SFN) at baseline (from QSART and CSP) (23). All 95% CIs and P values were reported two sided. Multiple testing was accounted for using the Benjamini-Hochberg stepwise method with a false discovery rate of 5% for key secondary outcomes.

Masking efficacy was assessed at the final visit with a forced-choice guess (n-3 PUFA or placebo) by the participant and outcome assessor. We used Cohen κ to assess the success of masking by calculating the level of agreement between actual treatment assignments relative to guesses made by both the participant and outcome assessor. Statistical analyses were performed in Stata 16 (StataCorp, College Station, TX).

Deidentified data that underlie the primary, key secondary, and safety outcomes are available from the corresponding author upon reasonable request, subject to approval of an ethics amendment by the relevant human research ethics committee. No applicable resources were generated or analyzed during the current study.

Forty-five participants were recruited between 6 July 2017 and 17 September 2019. Forty-three participants were deemed eligible and randomly assigned to receive either n-3 PUFA (n = 21) or placebo (n = 22) supplements. All participants in the n-3 PUFA group and 20 participants in the placebo group completed the study (Supplementary Fig. 1); two participants were lost to follow-up.

Participants in each study group were similar for all key baseline demographics parameters (Table 2), although those in the n-3 PUFA group were, on average, 8 years older than those in the placebo group. Age was adjusted for in the preplanned sensitivity analysis. Concomitant medications are summarized in Supplementary Table 1.

For the primary outcome, the estimated between-group change from baseline in central CNFL was 1.80 mm/mm² (95% CI 0.83, 2.78) at day 90 and 2.70 mm/mm² (95% CI 1.64, 3.75; P < 0.001) (Table 3 and Fig. 1) at day 180, favoring the n-3 PUFA group.

The relative change in CNFD in the n-3 PUFA group at day 180 compared with placebo was 4.98 nerves/mm² (95% CI 2.51, 7.44; adjusted P = 0.004) and in CNBD 11.23 branches/mm² (95% CI 7.01, 15.45; adjusted P = 0.002).

At day 180, there was no between-group difference in the relative change in geometric mean ratio for corneal sensitivity thresholds to a room temperature (0.84 [95% CI 0.60, 1.16]; adjusted P = 0.39) or cooled (0.81 [95% CI 0.54, 1.20]; adjusted P = 0.29) stimulus.

At day 180, the estimated change in quality-of-life score in the n-3 PUFA group relative to placebo was 0 units (95% CI −0.08, 0.08) for the EQ-5D-5L index score and −4.9 units (−10.9, 1.2) for the EQ-5D-5L visual analog scale score. For small nerve fiber function at day 180, the estimated change from baseline in the n-3 PUFA group did not significantly differ from that for the placebo group (Table 4).

At day 180, there was no between-group difference in the relative change in geometric mean MNSI score for the n-3 PUFA group compared with placebo (0.78 [95% CI 0.55, 1.10). At baseline, most (86%) participants had no neuropathy as defined with MDNS. At day 180, there was no significant between-group difference for change in MDNS (Table 5). For NCS, there was no between-group difference in change from baseline for sural sensory conduction amplitude (−0.20 µV [95% CI −2.26, 1.86]), peroneal motor conduction velocity (0.01 ms⁻¹ [95% CI −2.43, 2.46]), or tibial minimal F-wave latency (0.04 ms [95% CI −1.19, 1.27]).

For nerve excitability, there was a significant difference in the estimated relative change in TEh at 90–100 ms for the n-3 PUFA group compared with placebo (−9.33 units [95% CI −18.41, −0.25]) (Supplementary Fig. 2). No between-group differences were found in change from baseline at day 180 for the other nerve excitability variables (Table 5).

The estimated between-group change in Omega-3 Index at day 180 was 3.3%, favoring the n-3 PUFA group (95% CI 2.4, 4.2) (Supplementary Table 2). Mean (SD) Omega-3 Index for the treatment group increased from 4.9% (0.8%) at baseline to 8.2% (1.7%) at day 180. In the placebo group, the Omega-3 Index was similar at both time points (baseline, 4.6% [1.0%], vs. day 180, 4.8% [1.3%]). With the n-3 PUFA group used as a reference, the estimated between-group difference in change from baseline at day 180 for n-6:n-3 ratio was −3.5% (95% CI −4.3, −2.6).

There were 23 adverse events (Supplementary Table 3), comprising 12 in the n-3 PUFA group and 11 in the placebo group. Seventeen participants reported at least one adverse event (AE), including one serious AE in the placebo group that was deemed unrelated to the study treatment. At day 180, the change from baseline in the n-3 PUFA group did not differ from that in the placebo group for habitual visual acuity, intraocular pressure, blood chemistry parameters, or diabetic retinopathy grading (Table 6).

A prespecified subgroup analysis showed an interaction effect for subgroups with a baseline CNFL ≤12.5 mm/mm² at day 90, with the difference in estimated effect being 3.11 mm/mm² (95% CI 1.37, 4.86; P < 0.001) (Fig. 2) relative to the subgroup with baseline CNFL >12.5 mm/mm². At day 180, the interaction effect was 1.68 mm/mm² (95% CI −0.45, 3.82; P = 0.12) between subgroups. For participants with clinically defined SFN at baseline, the interaction effect at day 90 was −0.17 mm/mm² (95% CI −2.29, 1.96; P = 0.88) and at day 180 was 1.67 mm/mm² (95% CI −0.46, 3.82; P = 0.12) in comparison with those without SFN at baseline.

Prespecified analysis adjusted for baseline variables (age, diabetes duration, HbA_1c) did not alter study findings. A second prespecified model, with additional adjustment for baseline imbalances in CNFL and Omega-3 Index, was not used as these variables were not imbalanced. However, post hoc sensitivity analysis with adjustment for baseline Omega-3 Index did not alter the primary or key secondary results. Sensitivity analysis excluding individuals who were noncompliant (n = 3) did not alter the outcomes. Sensitivity analyses accounting for baseline imbalances did not alter the findings, with the exception of the n-6:n-3 ratio (Supplementary Table 2).

Exploratory outcomes are presented in Supplementary Tables 4–10. There were no between-group differences in estimated mean change from baseline for sweat latency and volume (measured using QSART) at the distal leg, proximal leg, or forearm. For median motor latency, there was a between-group difference of −0.16 ms (95% CI −0.32, −0.01); however, no other NCS parameters showed a between-group difference in change from baseline. For ocular surface parameters, the relative change in geometric mean ratio for tear break-up time was 1.31 (95% CI 1.01, 1.71) seconds, favoring the n-3 PUFA group.

Confirming the integrity of the masking procedures, there was poor agreement for the treatment assignment guess by the participant (κ = −0.03 [95% CI −0.33, 0.28]) and outcome assessor (κ = −0.18 [95% CI −0.48, 0.12]).

In this RCT with double masking, we have identified that, relative to placebo, triglyceride long-chain n-3 PUFA supplements, dosed at 1,800 mg/day for 180 days, promote corneal nerve regeneration (as indicated by an increase in the CNFL primary outcome measure and other corneal nerve parameters, CNFD and CNBD) in individuals with type 1 diabetes. Corneal sensitivity thresholds, to both room temperature and cooled stimuli, did not differ between treatment groups at day 180. Compared with placebo, n-3 PUFA supplementation did not alter small (measured with CSP and QSART) or large (measured with routine NCS) nerve fiber function.

The observed corneal neurotrophic effects with n-3 PUFA supplementation likely relate to the effect(s) of lipid metabolites, resolvins and protectins, which are anti-inflammatory and neuroprotective. Treatment with topical DHA and the DHA-derived lipid mediator neuroprotectin D1 improved corneal nerve density and sensitivity in rabbits following corneal nerve injury (24). PUFAs are key components of membrane phospholipids, affecting membrane fluidity and regulating protein and receptor function. n-3 PUFAs affect gene expression by activating the transcription of peroxisome proliferator–activated receptor and retinoid X receptor genes, which modulate neuronal apoptosis and neuroinflammation (25). The present findings are consistent with results from rodent studies that oral n-3 PUFA supplementation (menhaden oil) can reverse CNFL loss and corneal sensitivity impairment in experimental diabetes (10).

The mean natural history CNFL change in type 1 diabetes is estimated to be −0.8% per year (90% CI −14.0, 9.9) (4). Rapidly progressive CNFL loss, defined as ≥6%/year, occurs in 17% of individuals and is associated with a higher risk of DSP development and progression. In this subpopulation, the mean (SD) annual reduction in CNFL is estimated as −14.7% (11.5%) (4). Consistent with the established heterogeneity of CNFL changes in type 1 diabetes, the mean effect evident in the placebo group in the current study fell within this range, with a CNFL change of −5.8% (14.7%) over 6 months.

The magnitude of improvement in CNFL in the n-3 PUFA group (mean [SD] 2.06 [1.73] mm/mm²) in the present study is similar to that described in an open-label study of seal oil supplements (750 mg/day EPA, 1,020 mg/day DHA, and 560 mg/day docosapentaenoic acid) in a population with type 1 diabetes (12). An increase in CNFL from 8.3 (2.9) mm/mm² to 10.1 (3.7) mm/mm² was reported after 12 months of n-3 PUFA supplementation (12). In this study, compared with the reported mean treatment effect (29% increase in CNFL from baseline), a greater increase in CNFL (56%) was reported in individuals with baseline CNFL <15 mm/mm² (12). In the current study, CNFL showed time-dependent changes that were influenced by baseline CNFL. The estimated change in CNFL at day 90 relative to placebo was predominantly driven by the effect in the subgroup with baseline CNFL ≤12.5 mm/mm² (Fig. 2). This finding suggests that baseline CNFL may be a treatment predictor that identifies individuals who will derive more rapid benefits from n-3 PUFA supplementation. At day 180, the heterogeneity in effect between subgroups was not significant. Thus, with a longer supplementation period, the relative CNFL increase is comparable between severity groups, potentially suggesting a ceiling effect that is achieved with temporal characteristics dependent on baseline corneal nerve integrity. It has recently been proposed that baseline n-3 fatty acid levels may also influence the extent of corneal nerve regeneration; a secondary analysis of a single-arm, open-label trial found a positive association between the change in CNFL and baseline systemic n-3 fatty acid levels (26). However, in the current study, the treatment effect did not change in a post hoc sensitivity analysis with adjustment for baseline Omega-3 Index.

Nerve excitability abnormalities associated with diabetes are likely due to axonal membrane sodium/potassium pump dysfunction and reduced nodal sodium channel conductance. Increased threshold change in TEh, evident in the n-3 PUFA group, could be associated with reversal of these mechanisms (27). However, the clinical significance of this finding is unclear without an associated increase in threshold change in TEd or improvement in other nerve excitability parameters. Although nerve excitability abnormalities can occur in type 1 diabetes without clinically evident neuropathy (28), the mean baseline curves in our cohort appeared normal and, together with normal NCS, suggest unaffected large fiber axonal function. Thus, there may have been little capacity for improvement with intervention.

There were no between-group differences for change in small nerve fiber function, measured using CSP and QSART, or in most NCS parameters at day 180. The absence of detectable improvement in peripheral nerve function may relate to sample size, as the study was not powered to detect differences in clinical neurophysiology outcomes for CSP, QSART, or NCS. Furthermore, both CSP and QSART can show inherent intertest variability (23).

In rodent models, improvements in corneal nerve structure and function have been associated with functional improvements in motor and sensory NCS and thermal nociception in the limbs (10). However, this association has not been demonstrated in human studies. An open-label study of exenatide and pioglitazone, or insulin, in type 2 diabetes did not find a coincident change in sudomotor function despite an increase in CNFL after 1 year (29). Similarly, another single-arm, open-label study found no improvement in quantitative sensory testing or NCS with n-3 PUFA supplementation despite CNFL increasing from baseline (12). DSP involves initial damage to peripheral small nerve fibers (i.e., unmyelinated C fibers and thinly myelinated Aδ-fibers, including in the cornea) that precede large nerve fiber involvement (3). Corneal subbasal nerve plexus parameters are a valid surrogate endpoint for evaluating the efficacy of DSP interventions (5). As DSP progresses in a length-dependent pattern, it is possible that improved CNFL is an early indicator of peripheral nerve recovery, preceding potential later improvements in sensory and/or motor function. A longer treatment duration may be required to observe changes in traditional clinical measures of small and large fiber function, such as neuropathy symptoms and neurophysiology outcomes. In a population with type 1 diabetes, although corneal nerve regeneration was evident 6 months after pancreas and kidney transplantation (30), improved neurophysiology outcomes, relative to an untreated group, were not observed until 36 months posttransplantation (5). A challenge with assessing neurophysiological outcomes is that most standard tests have poor diagnostic yield for early small nerve fiber damage (31). These factors indicate a need to carefully consider the duration of therapeutic treatment and selection of endpoints in future DSP trials.

The effect of oral n-3 PUFA supplementation on improving corneal nerve density suggests it may also benefit ocular surface health, in addition to influencing tear function (15). In diabetes, corneal nerve loss disrupts corneal epithelial integrity and reduces basal tear production (32). These changes can impair corneal wound healing and lead to persistent epitheliopathy (33). In the current study, enhanced tear stability was observed in the n-3 PUFA group at day 180 in comparison with placebo. This improvement occurred in a study population without overt ocular surface disease, as indicated by baseline tear osmolarity and dry eye symptom scores within physiological ranges. These findings suggest that n-3 PUFAs may have a role in limiting DSP-associated corneal keratopathy.

The current study used a moderate dose of long-chain n-3 PUFAs (1,800 mg/day). At the endpoint, the systemic Omega-3 Index in the n-3 PUFA group exceeded 8%, the therapeutic threshold in cardiovascular disease (34). Diabetes can create a prothrombotic state (35), and n-3 PUFAs have anticoagulant effects. The dose of n-3 PUFAs in the current study was selected to avoid potential increased bleeding risks associated with higher doses (>2,000 mg/day) (36). There were no unexpected AEs deemed related to the study interventions, and no participant had a worsening in their diabetic retinopathy grade. Nevertheless, the long-term effects of n-3 PUFA supplementation on vascular function in individuals with diabetes are not clear, warranting further study.

This is, to our knowledge, the first double-masked RCT to evaluate the efficacy of n-3 PUFA supplementation for modulating peripheral nerve health in people with type 1 diabetes. Analyses were specified a priori in a statistical analysis plan, including an ITT protocol. There was high participant retention (95%) and excellent compliance, confirmed with the Omega-3 Index. Treatment allocation was concealed, and double masking was assessed as effective. A comprehensive suite of neurophysiology parameters was evaluated. For evaluation of the primary outcome measure, the assessment of 12 randomly selected images with <20% overlap from the central cornea has been shown to produce an estimate that is within 10% of the true mean corneal CNFL 95% of the time (37).

Potential considerations in interpretation of the current findings include that additional assessments of small nerve fiber function, such as quantitative sensory testing and intraepidermal nerve fiber density, were not incorporated because of lack of availability. This study had a sample size based on the primary outcome. Given that peripheral nerve function tests are prone to greater variability, a larger sample size may be required to detect a difference in small and large fiber functional measures. Multiplicity adjustment was not performed for variables outside of primary and key secondary analyses, increasing risks of false positive findings for secondary and exploratory outcomes; these should therefore be interpreted with caution. Although one-third of our study population was diagnosed with SFN (classified with QSART and CSP), most had no large fiber neuropathy (classified with the MDNS). The primary outcome measure, CNFL, was selected because it is a validated surrogate end point for small fiber structure (29,38). However, it is an objective, rather than a patient-reported, outcome, and whether improvements in corneal small nerve fibers translate to meaningful improvements in peripheral sensory function has not been established.

In conclusion, a moderate daily dose of oral n-3 PUFA supplements for 6 months improved corneal subbasal nerve structure in people with type 1 diabetes, relative to placebo, consistent with a corneal neuroregenerative effect. There were no safety concerns with the intervention. Confirmatory trials in larger, well-defined populations, potentially over longer duration, are indicated to ascertain whether n-3 PUFA supplements confer similar benefits to peripheral nerve function.

Acknowledgments. The authors sincerely thank the participants involved in this study and the Departments of Endocrinology and Neurology/Neurophysiology at St. Vincent’s Hospital Melbourne for support with participant recruitment. The authors thank Linda Seiderer and Joanne Zahra, St. Vincent’s Hospital Melbourne, for assistance with organizing participant neurophysiology study visits. The authors thank Associate Professor Peter Keller, University of Melbourne, for assistance as the independent data manager, including generating the participant randomization code. The authors thank Alexander Britten-Jones for assistance with the reconciliation of returned study supplements. The authors acknowledge the assistance of Associate Professor Elif Ekinci, Austin Health, and Diabetes Victoria for their assistance in participant recruitment. Some of the participants in the study were recruited as registrants of the National Diabetes Services Scheme (NDSS). The NDSS is an initiative of the Australian Government administered by Diabetes Australia.

Funding. A.C.B.-J. was supported by an Australian Government Research Training Program Scholarship for her PhD studies. This project was supported by a University of Melbourne Neuroscience Interdisciplinary (MNI) seed grant (to J.T.K., L.J.R., R.J.M., L.E.D.) and a Rebecca L. Cooper Medical Research Foundation 2019 Project Grant (to L.E.D.). R.J.M. has received research grants from the Rebecca L. Cooper Medical Research Foundation, St. Vincent’s Research Foundation, JDRF, the Diabetes Australia Research Trust/Program, and the National Health and Medical Research Council of Australia.

The study funders had no role in the study design, data collection, data analysis, data interpretation, or report writing.

Duality of Interest. R.J.M. has received research grants from Novo Nordisk, Servier, Medtronic, and Grey Innovation. He has also received honoraria for lectures from Eli Lilly, Novo Nordisk, Sanofi, AstraZeneca, Merck Sharp & Dohme, Novartis, and Boehringer Ingelheim and is on the advisory boards for Novo Nordisk, Boehringer Ingelheim–Eli Lilly Diabetes Alliance, and AstraZeneca. Travel support has been supplied to R.J.M. by Novo Nordisk, Sanofi, and Boehringer Ingelheim; he has been a principal investigator for industry-sponsored clinical trials run by Novo Nordisk, Bayer, Cilag (subsidiary of Johnson & Johnson), and AbbVie; and he is also a council member for the Australian Diabetes Society. L.E.D. has previously received research funding from CooperVision for a clinical trial investigating oral n-3 PUFA supplements for treating contact lens discomfort. L.E.D.’s University research laboratory has also previously received funding from Alcon, Allergan, Azura Ophthalmics and Kedalion Therapeutics for anterior eye research, and she has acted on an advisory board for Santen. J.P.C. has received consultancy, speaker or travel funding from Alcon, Azura Ophthalmics, E-Swin, Johnson & Johnson Vision, Novartis and Théa Laboratoires, as well as University laboratory funding from Alcon, Azura Ophthalmics, E-Swin, Manuka Health NZ, Resono Ophthalmic, Théa Laboratoires and Topcon for research in the area of dry eye and ocular surface disease. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. A.C.B.-J. wrote the first draft of the manuscript with input from L.E.D. A.C.B.-J., J.T.K., L.J.R., R.J.M., and L.E.D. developed the trial protocol. A.C.B.-J., J.T.K., L.J.R., and L.E.D. implemented the trial procedures and acquired data. A.C.B.-J. and S.B. performed the statistical analysis and independently verified the data analysis. A.C.B.-J., S.B., and L.E.D. developed the statistical analysis plan. A.C.B.-J. and L.E.D. verified the underlying data and had the final responsibility for the decision to submit the manuscript for publication. J.T.K., L.J.R., and R.J.M. provided administrative and technical support. J.T.K., L.J.R., R.J.M., and L.E.D. acquired the funding. J.P.C., R.J.M., and L.E.D. supervised the study. L.E.D. created the study concept. All authors revised the manuscript. A.C.B.-J. and L.E.D. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Australasian Diabetes Congress 2020 (virtual platform), 11–13 November 2020.