Loss of neuronal membranes and synaptic integrity are major factors that contribute to the development of cognitive impairment in individuals with Alzheimer’s disease (AD). Membrane phospholipid metabolism can be investigated non-invasively using phosphorus Magnetic Resonance Spectroscopy (³¹P-MRS), and this was used previously to investigate the effect of a specific intervention on the brain¹. The use of ³¹P-MRS imaging makes it possible to investigate several regions of interest (ROIs) simultaneously. The medical food Souvenaid^®, containing the specific nutrient combination Fortasyn^® Connect^# (docosahexaenoic acid, eicosapentaenoic acid, uridine, phospholipids, choline, selenium, folic acid, and vitamins B6, B12, C and E), has been formulated to counteract synaptic loss and reduce membrane-related pathology in AD by promoting neuronal membrane formation and function². This study investigates whether Souvenaid affects brain phospholipid metabolism. The main outcomes of interests are the relative levels of phosphomonoesters (PME) (i.e. membrane phospholipid precursors) and phosphodiesters (PDE) (i.e. membrane phospholipid breakdown products) and their ratio. A total of 34 drug-naïve patients with mild AD (Mini-Mental State Examination [MMSE]≥20); were enrolled in this double-blind randomised controlled study (NTR3346) to receive either the test product (Souvenaid) or an iso-caloric control product for 4 weeks. At baseline and after 4 weeks, ³¹P-MRS imaging of the whole brain was performed at 3T (Siemens Magnetom Trio) with a dual-tuned ¹H/³¹P volume head coil (Rapid) and a 3D pulse-acquire sequence with the following parameter values: TR=2000 ms, TE=0.10 ms, 40° flip-angle, NA=4, WALTZ4 proton decoupling, nominal voxel size=16.25x16.25x16.25mm, TA=13:08 min. Resonances in spectra from four ROIs (left and right hippocampus, HL and HR; anterior cingulate cortex, ACC; and retrosplenial cortex, RSC) were fitted using Metabolite Report (Siemens) (Figure 1) and metabolite levels as a percentage of the total phosphorus signal were calculated. Both a quantitative evaluation of the fitting results (Cramer–Rao lower bound ≤30%) and a visual quality control (independently by two spectroscopists) were performed. Only metabolites that passed both quality checks were included in the final statistical analyses. PME was calculated as the sum of phosphocholine (PCh) and phosphoethanolamine (PEth) and PDE as the sum of glycerophosphocholine (GPCh) and glycerophosphoethanolamine (GPEth). In addition, blood samples were collected at both time points. Statistical analysis to determine group differences was performed using mixed models. One primary model consisted of a mixed model with the values measured at week 4 as outcome, with intervention group, sex, and brain region as fixed factors, considering brain region as a within subject factor, and adjusting for (imputed) baseline values. The other primary model was identical, but only observed baseline values were used.A total of 33 subjects (16 test; 17 control) were included in the modified ITT data analysis (one randomised subject dropped out at baseline due to claustrophobia). The mean MMSE score was 23.1 (SD=1.9), mean age was 73.7 years (SD =6.8), and 14 subjects were men. Quality control revealed that most valid phospholipid fits were from the RSC (2 subjects missing at week 4). Most phospholipid data were missing from the ACC, followed by HL and HR (15-42% and 9-30% missing at week 4, respectively). Main intervention effects are reported, since no significant interactions were observed between intervention and brain region. The PME/PDE ratio was higher after 4 weeks of test product compared with control product (primary model using [imputed] baseline values without one extreme outlier: LS mean difference = 0.14, SEM=0.07 p=0.061; primary model using observed baseline values: LS mean difference = 0.18, SEM=0.06, p=0.005). No significant intervention effects were found for PDE, PME, PCh, PEth, GPCh or GPEth. Compliance, based on diaries, was very high (96.4%, SD=4.8 [test] and 99.0%, SD=1.6 [control]), and this was confirmed by significantly increased plasma levels of DHA, EPA, choline, uridine and vitamin E and significantly decreased plasma homocysteine levels in the test group compared to the control group.

These results show it is feasible to perform 3D ³¹P-MRSI in a reasonable time in this mild AD population. However, spectral quality varies between brain regions, and this variation is possibly exacerbated by disease-specific pathology. The observed effects suggest that Souvenaid not only raises circulating levels of nutrients and fatty acids, but also affects brain phospholipid metabolism in patients with mild AD. The increase in PME/PDE may indicate enhanced phospholipid formation compared to breakdown of membrane phospholipids in the brain by 4 weeks intervention with Souvenaid.

^#Souvenaid and Fortasyn are registered trademarks of N.V. Nutricia.