Acute nicotinamide riboside supplementation increases human cerebral NAD⁺ levels in vivo

Abstract

Purpose:

The purpose of this study was to determine the effect of acute nicotinamide riboside (NR) supplementation on cerebral nicotinamide adenine dinucleotide (NAD⁺) levels in the human brain in vivo by means of downfield proton magnetic resonance spectroscopy (DF ¹H MRS).

Methods:

Downfield ¹H MRS was performed on ten healthy volunteers in a 7 Tesla MRI scanner with spectrally selective excitation and spatially selective localization to determine cerebral NAD⁺ levels on two back-to-back days: once after an overnight fast (baseline) and once 4 hours after oral ingestion of nicotinamide riboside (900mg). Additionally, two more baseline scans were performed following the same paradigm to assess test-retest reliability of the NAD⁺ levels in the absence of NR.

Results:

NR supplementation increased mean NAD⁺ concentration compared to the baseline (0.458±0.053 vs 0.392±0.058 mM; p<0.001). The additional two baseline scans demonstrated no differences in mean NAD⁺ concentrations (0.425±0.118 vs 0.405±0.082 mM; p=0.45), and no difference from the first baseline scan (F(2,16)=0.907; p=0.424).

Conclusion:

These preliminary results confirm that acute NR supplementation increases cerebral NAD⁺ levels in healthy human volunteers and shows the promise of DF ¹H MRS utility for robust detection of NAD⁺ in humans in vivo.

INTRODUCTION

Nicotinamide adenine dinucleotide (NAD⁺) is an important coenzyme involved in many redox reactions and plays a vital role in cellular metabolism. It also serves as a substrate for many of the regulatory proteins involved in key processes including DNA repair, aging, apoptosis, mitochondrial function (1–6). In humans, there are three primary pathways through which NAD⁺ is synthesized from its precursors L-tryptophan (TRP), nicotinic acid (NA), nicotinamide (NAM), nicotinamide riboside (NR) and nicotinamide mononucleotide (NMN), as shown in Figure 1 and these pathways are described in detail elsewhere (7–20).

Figure 1. Three major pathways for NAD⁺ synthesis in vivo.

The three pathways are (i) Kynurenine or de novo pathway, where the dietary TRP is the starting precursor; (ii) Preiss-Handler pathway, which uses dietary NA as the starting precursor; and (iii) Salvage pathway, where recycled NAM generated as byproduct or the upstream NAD⁺ vitamin precursor NR or NMN, all serve as precursors.

Since NAD⁺ concentrations in the brain and in various tissues have been found to decline with age, (21–23) lower NAD⁺ levels likely correlate with various aging-associated neurodegenerative and neurological disorders, and are under investigation for their potential to contribute to pathology (24, 25). Due to the complex nature of the neural network system, the brain has high energy demands for proper functioning, and maintaining the brain’s NAD⁺ levels by appropriate diet or through nutritional supplementation may have functional consequences. This is one of the reasons NAD⁺ precursors such as TRP, NAM, NMN, and NR have been proposed as supplements. While some of them are already in use, NR supplements have been steadily gaining popularity in recent years. Several preclinical studies with NR supplementation in rodents have shown increases in NAD⁺ levels in the brain and other tissues and improvements in brain function in preclinical models including Alzheimer’s disease, neuroinflammation, mitochondrial dysfunction, and in peripheral insulin resistance (26–28). Thus, improving brain NAD⁺ levels through NR supplements may be a non-invasive strategy to treat neurodegenerative, metabolic and neuropsychiatric disorders. However, human studies have largely been limited to studying the effects of NR on whole blood, plasma and peripheral blood mononuclear cells (29), but peripheral NAD⁺ measures might not reflect cerebral NAD⁺ levels. There remains a need for new methods to non-invasively detect and quantify the cerebral NAD⁺ levels in humans.

Two non-invasive techniques for in vivo detection of NAD⁺ levels are applicable to human studies: (i) phosphorous magnetic resonance spectroscopy (³¹P MRS), and (ii) downfield proton magnetic resonance spectroscopy (DF ¹H MRS). However, only a few studies have measured cerebral NAD⁺ levels in human participants with ³¹P MRS (23, 30, 31). One study showed an age-dependent decrease in cerebral NAD⁺ and NAD⁺:NADH redox ratio in healthy volunteers (23). Another study assessed the effects of a 30-day NR treatment (1000 mg daily) on brain NAD⁺ levels using ³¹P MRS in individuals with Parkinson’s disease and found increased brain NAD⁺ (based on the NAD⁺:ATP ratio, without absolute quantitation) and related metabolites compared to a 30-day placebo treatment (30). Shortcomings of ³¹P MRS are that it has low sensitivity resulting in longer scan times to achieve meaningful signal-to-noise ratio, and that quantification is complicated by the fitting of multiple overlapping resonances.

More recently, in vivo downfield spectroscopy with ¹H MRS has gained traction as a method to quantify brain NAD⁺ levels (32–35). DF ¹H MRS acquires the proton spectrum of metabolites resonating down-field to the water resonance, as opposed to the traditional up-field ¹H MRS spectrum. Another major difference compared to the up-field spectroscopy and especially in the acquisition of the NAD⁺ spectra is the use of pulse sequences without water suppression and selective excitation at the region of interest. These are needed as NAD⁺ resonances have cross-relaxation with water and any traditional water suppression scheme would lead to a loss of the metabolite signals on the spectrum. Using spectrally selective excitation and spatially selective localization, without water suppression, our group recently developed and established the reproducibility of a new DF ¹H MRS acquisition of human brain NAD⁺ levels. In the same study, a decrease in cerebral NAD⁺ concentration was shown over the course of aging in healthy volunteers (21). Here, we utilized the DF ¹H MRS method to quantify the effects of NR supplementation on cerebral NAD⁺ concentration in healthy human subjects.

METHODS

Participants:

A total of 10 healthy volunteers (5 males; 5 females; mean age = 32.1 ± 10.28 years old, range = 21 – 54 years) participated in the study. Written informed consent was obtained under an approved protocol from the Institutional Review Board of the University of Pennsylvania.

Study design:

All volunteers underwent a total of four magnetic resonance imaging (MRI) scans, and all MRI acquisitions were performed between 9:00 to 11:00 am. For all scans, volunteers were instructed to have dinner before 8 pm the night before scans, to fast overnight, and to refrain from having any exercise or breakfast/drinks (except water) before the scans. The first two scans were performed on back-to-back days. For the second scan, participants consumed 900 mg of nicotinamide riboside (NR) supplements (TRUNIAGEN 300 mg capsules; purchased from https://www.truniagen.com) orally four hours before the scan time (i.e., at 5:00 am in the morning).

Test-retest reliability:

The same volunteers underwent two additional scans on back-to-back days following the same fasting/no exercise instructions, without NR supplementation. The schematic of study protocol is shown in Figure 2A.

Figure 2. Schematic of the study protocol and representative DF ¹H MRS showing NAD⁺ resonances.

A. Each subject fasted overnight and without any breakfast to participate in the DF ¹H MRS at 9–10 AM. Subjects took NR supplement 4 hours before the MRI scan only for visit 2 (red color box). For the visit 3 there were variable time delays from visit 2 ranging for each subject ranging from 8 days to ~ 5 months (Subject 1: ~ 3 months; Subjects 2,3,4 & 6: ~ 5 months; Subject 7 & 10: ~ 2 months; Subject 8: 8 days; Subject 9: 10 days). During visits 3 and 4, subjects did not take any NR supplementation. B. Spectroscopy voxel position overlayed on the anatomical images is shown in red color. C. Downfield ¹H MRS of NAD⁺. Representative DF ¹H MRS showing the NAD⁺ resonances in red color at 9.33 ppm (H2), 9.15 ppm (H6), 8.84 ppm (H4). The NAD⁺ chemical structure is also displayed in the inset and the corresponding resonances in DF ¹H MR spectrum from nicotinamide moiety is listed in red color. The H5 resonance from the nicotinamide is buried under the huge signal comprising of a mixture of NAA, ATP, adenosine and purine nucleotides as shown in the spectral region from 7.9 to 8.5 ppm.

While the first two scans aimed to test the effect of NR supplementation on cerebral NAD⁺ levels compared to baseline, the latter two scans aimed to elucidate information about the basal fluctuation and test-retest reliability of cerebral NAD⁺ levels, thereby reducing the bias in measurements.

MRI acquisition protocol:

All the MRI and DF ¹H MRS acquisitions were done using a single-channel volume transmit, and 32-channel receive proton phased array head radiofrequency coil (Nova Medical, Wilmington, MA, USA) on a 7T scanner (MAGNETOM Terra, Siemens Healthcare, Erlangen, Germany). The study protocol consisted of a localizer, anatomical image acquisition with MPRAGE (36), and single voxel STEAM sequence without water saturation for reference voltage calibration followed by DF ¹H MRS acquisitions. MPRAGE images were acquired in the axial plane with the following parameters: voxel size = 1×1×1 mm³, slices per slab = 160, TR/TE/TI = 2300/3.69/1100 ms, FOV read = 256 mm, acquisition matrix = 192×256, acceleration factor PE = 2, and total acquisition time for MPRAGE alone was 4 min 6 s. A range of transmit B₁ voltages was acquired with a non-water-suppressed STEAM sequence and the resulting water signal amplitudes were fit to a sin³(𝚹) curve to determine the optimal transmit reference voltage.

To determine the cerebral NAD⁺ levels, DF ¹H MRS was acquired with a 90° E-BURP pulse (37) that has a selective spectral excitation centered at 9.1 ppm and an excitation bandwidth of 600Hz (1ppm on either side of the excitation center) followed by three narrow-band spatially selective refocusing 180° Shinnar-Le Roux pulses for spatial localization as described in our previous study (21). Repetition time (TR) and time to echo (TE) were 1000ms and 18ms, respectively. To maximize the signal-to-noise ratio for DF ¹H MRS acquisition, a large voxel was placed within the brain to cover the maximal area as shown in Fig. 2B. The volume of this voxel varied from 204 to 300 mm³ across the individuals. The total number of averages was 256. For water reference acquisition on the same voxel, TR was 10000ms, TE was 18ms and the number of averages were 16. The total time of the DF ¹H MRS acquisition alone including the water reference was 7 min 5 s.

To reduce bias in the DF ¹H MRS scans, we used the in-house written program ‘ImScribe’ (https://www.med.upenn.edu/CAMIPM/imscribe.html) (38) on repeat scanning days to place the voxel in the same brain area as the first scan based on participants’ anatomical scans. ImScribe utilizes the high-resolution T₁-weighted images and the spectroscopy voxel information as a target template from the first scan and performs within-subject rigid-body co-registration with the T₁-weighted images of the subsequent scan to provide the new spectroscopic voxel placement information for the later scans.

Brain Tissue Segmentation:

Voxel tissue composition was obtained by automated segmentation of the MPRAGE images, with the following steps: 1) bias field correction using SPM8 (39); 2) brain masking using HD-BET (40); 3) segmentation for gray matter (GM), white matter (WM), and cerebral spinal fluid (CSF) using FSL FAST (41). A spatial mask for the volume of the spectroscopy voxel in MPRAGE space was obtained using ImScribe (38) and applied to the whole brain segmentation results to derive the absolute and relative tissue components of the voxel.

A representative anatomical image with spectroscopy voxel showing the GM, WM and CSF along with the bar plots of GM, WM and CSF from all the scans for each volunteer is shown in Figure 3 (individual fractional values for all scans are given in Supplementary materials, Table S1), indicating the consistency in voxel placement for multiple scans of each subject.

Figure 3. Consistency of voxel placement measures.

Representative anatomical image showing the GM (green), WM (blue) and CSF (red) fractions within the spectroscopy voxel location. Bar plots of GM, WM and CSF fractions from all the scans for each volunteer showing the consistency of the voxel location for multiple scans is shown in the bottom of the figure.

Spectral Quantification:

A custom program written in MATLAB was developed for spectral fitting of down-field data, using the Hankel singular value decomposition (HSVD) method for estimating components of the time domain signal (42, 43). The analysis steps were: 1) raw data receive coil channel combination using the water reference scan to determine coil channel weightings; 2) application of the weightings for coil combination of the down-field metabolite signal; 3) 5 Hz line-broadening; 4) decomposition with HSVD to determine candidate components for water or NAD⁺ resonances (8 and 60 components for the water and NAD⁺, respectively); 5) assignment of components to water or NAD⁺ resonances based on tolerances for frequency and linewidth; 6) estimation of absolute NAD⁺ concentration by scaling the NAD⁺ component amplitudes to the water amplitude. The following assumptions were made in the quantification step: water T₁ = 1939 (GM), 1130 (WM), 3000 (CSF) ms; water concentration = 44 (GM), 38.5 (WM), 55 (CSF) M; water T₂ = 60ms; NAD⁺ T₁/T₂ = 147/31 ms (GM, WM, and CSF) (44). In all scans, the NAD⁺ resonances were fitted by a single component. The water signals were fitted to between 3 and 5 components. NAD⁺ concentrations from 9.33ppm (H2) are reported in this study.

Statistics:

A paired samples t-test was conducted for our primary analysis on the comparison of cerebral NAD⁺ concentrations before and after NR supplementation (i.e., scan 1 versus 2). Secondarily, we used paired samples t-test to compare the NAD⁺ levels from the second set of baseline scans (i.e., scan 3 versus 4). The Shapiro-Wilk test (45) was used to assess normality of the data. Furthermore, we conducted an exploratory one-way repeated measures ANOVA to assess whether the mean cerebral NAD⁺ concentrations differ significantly across the three baseline scans (scan 1, 3, and 4), in order to determine basal fluctuation of cerebral NAD⁺ levels. The Intraclass Correlation Coefficient (ICC) was used as an additional measure of the inter-scan reliability of NAD⁺ measurements. ICC estimates and their 95% confidence intervals (CI) were calculated based on an absolute-agreement, two-way random effect model. Bland-Altman plots are also included in the supplementary materials, Figure S1, to visually evaluate the agreement between the baseline scans. All analyses were conducted in R (version 4.3.1) with two-sided tests and a significance level of 0.05.

RESULTS

The data presented here are from nine volunteers because the data from one of the ten planned volunteers were degraded by motion and not usable. Subject demographics and measured NAD⁺ concentrations from each subject and timepoint are provided in Table 1. A representative anatomical image showing the voxel position from which the DF ¹H MR spectra was acquired is shown in Figures 2B & 2C. A zoomed in spectrum showing NAD⁺ resonances with their individual residual fits after baseline correction, along with a representative baseline corrected NAD⁺ spectra before and after NR supplementation, are also shown in Figures 4A & 4B. Note that H4, H6 resonances of the NAD⁺ spectra are affected by the baseline correction since these are riding on the wing of the larger resonance peak (8.25 ppm) whereas the H2 resonance is less affected. Since the H2 resonance is a singlet peak (9.33 ppm) and less affected by the baseline correction we choose this for the quantification. A bar plot of the mean NAD⁺ concentrations for all the four scans is shown in Figure 4C. The mean NAD⁺ concentration for the baseline scan was 0.392 ± 0.058 mM, and mean NAD⁺ after NR supplementation was 0.458 ± 0.053 mM. There was a consistent increase in NAD⁺ concentration with NR supplementation in all the volunteers scanned as shown in Figure 5A, ranging from 7% to 40%. On average, cerebral NAD⁺ levels were 0.065 mM (~16%) higher after taking the NR supplement, which was significant compared to baseline (t(8)=5.93, p<0.001). The difference in NAD⁺ concentration before and after NR supplementation is normally distributed according to the Shapiro-Wilk test (W=0.86, p=0.10)

Table 1. Demographics and summary of data for all volunteers.

Individual NAD⁺ concentrations from localized downfield ¹H MRS for all the scans from each volunteer along with demographic information is given the table below.

Subject No.	Gender	Age (Y)	Voxel size (cc)	NAD+ (mM)
				Scan 1		Scan 2		Scan 3		Scan 4
1	F	37	240	0.391	~24 hrs	0.432	82 days	0.532	~24 hrs	0.459
2	F	27	204	0.299	~24 hrs	0.422	153 days	0.307	~24 hrs	0.356
3	F	23	275	0.404	~24 hrs	0.440	159 days	0.415	~24 hrs	0.393
4	M	21	270	0.479	~24 hrs	0.512	158 days	0.412	~24 hrs	0.454
6	F	24	275	0.403	~24 hrs	0.503	159 days	0.507	~24 hrs	0.441
7	F	54	240	0.417	~24 hrs	0.458	40 days	0.444	~24 hrs	0.395
8	M	27	300	0.406	~24 hrs	0.503	8 days	0.404	~24 hrs	0.492
9	M	29	270	0.300	~24 hrs	0.352	10 days	0.207	~24 hrs	0.215
10	M	38	284	0.431	~24 hrs	0.497	50 days	0.601	~24 hrs	0.445
	Mean	31.11	262	0.392		0.458		0.425		0.405
	SD	10.39	29	0.058		0.053		0.118		0.082

Figure 4. NAD⁺ peak fits, its measures and statistics pre- and post-NR supplementation.

A. The zoomed in NAD⁺ spectrum (shown in solid black) along with the individual fits for NAD⁺ resonances H2 (brown), H6 (orange) and H4 (purple) after baseline (broken black lines) removal. B. A representative NAD⁺ spectra from one of the volunteers after baseline removal showing an increase in H2 and H6 resonances with NR supplementation (red) compared to no supplementation (black). Note that since H2 resonance is a singlet and less affected by baseline correction, it is used for quantitation. C. A statistical significance (p<0.001) between scan 1 (no supplement) and scan 2 (NR supplement) was observed. No statistical significance between scan 3 (no supplement) and scan 4 (no supplement) (p=0.45) was found.

Figure 5. Individual NAD⁺ measures for all the scans from each volunteer.

A. Plot of cerebral NAD⁺ levels from all the volunteers pre- and post-NR supplementation (scan1 & scan 2). Please note that the scan 2 indicated in red color font is with NR supplement while the scan 1 is with no supplement. B. Plot of repeated cerebral NAD⁺ measures (scan 3 & scan 4) without any NR supplementation from all the volunteers. C. Mean NAD⁺ levels for the scans 1, 3 & 4 with no NR supplementation. There was no statistical significance between either scan 1 & 3, scan 1 & 4 or scan 3 & 4.

To assess test-retest reliability of the brain NAD⁺ signal, the same volunteers underwent two additional scans (scan 3 and scan 4) without NR supplementation on back-to-back days. The time period between first set of scans (scan 1 and scan 2) and last set of scans (scan 3 and scan 4) was variable for each volunteer and varied between ~ 8 days to 5 months (mean ± sd: 92 ± 67 days). Cerebral NAD⁺ levels from the latter set of baseline scans for each volunteer are shown in Figure 5B. Mean NAD⁺ concentration for these two additional scans were 0.425 ± 0.118 (scan 3) and 0.405 ± 0.082 mM (scan 4), respectively. There was no significant difference in NAD⁺ levels between these two additional scans (t(8)=0.79, p=0.45) as shown in Figure 5C, and there was also no significant difference among all three baseline scans (F(2,16)=0.907, p=0.424). Overall, there was moderate reliability between scan 3 and 4 (ICC=0.73, 95% CI:0.21, 0.93) and also among the three baseline scans (ICC=0.66, 95% CI: 0.29, 0.90).

DISCUSSION

Previous human studies have indicated that NR increased blood cell NAD⁺ levels in human participants (46, 47), and that brain NAD⁺ levels covaried with blood and CSF markers (23, 30). Specifically, the NAD⁺ metabolome was elevated in the peripheral blood mononuclear cells within 4.1 hours post oral ingestion of 1000 mg of NR (48). Based on these findings, in the current study, we chose to scan the subjects 4 hours post ingestion of NR. One of the NR supplementation studies has shown that an increase in blood NAD⁺ levels might not necessarily reflect the changes in tissue NAD⁺ levels as demonstrated from the unaltered NAD⁺ levels in the skeletal muscle biopsy samples after NR supplementation (49). This necessitates the need for alternate techniques to detect the tissue specific NAD⁺ levels in vivo, such as the down-field ¹H MRS technique that was utilized here.

Here, we demonstrated for the first time that a single dose of 900 mg NR supplementation increased brain NAD⁺ levels in healthy human volunteers within four hours after the ingestion. While one previous ³¹P MRS study (using a 3D CSI grid with an 8×8 matrix and a nominal voxel size of 30×30×80 mm³) in individuals with Parkinson’s disease demonstrated an increase in brain NAD⁺ after a 30-day NR treatment (30), our study is unique in that it is the first study that assessed the effects of a single dose of NR on brain NAD⁺ levels and the first to detect effects of NR in the brains of healthy volunteers. Moreover, we quantified brain NAD⁺ levels with our recently developed DF ¹H MRS scanning sequence, which has previously shown excellent intra-subject and inter-subject coefficients of variation and a robust correlation with age in n=16 healthy subjects (21).

DF ¹H MRS has three main advantages over ³¹P MRS for the detection of NAD⁺: (1) it can detect the oxidized form of NAD⁺ independently of NADH and without spectral overlap at 9.3 ppm, (2) it has higher sensitivity, resulting in shorter scan times to achieve meaningful signal-to-noise ratio of the resonances, (3) it is not dependent on one-half of the NAD⁺ resonance of quartet for fitting and quantification. With DF ¹H MRS, we thus developed a non-invasive, robust measure of brain mitochondrial health that covaries with age and is sensitive to acute NR supplementation. The assessment of brain NAD⁺ by means of DF ¹H MRS can be utilized to assess mitochondrial brain function in clinical populations in vivo, including in patients with Alzheimer’s disease and related dementias, Parkinson’s disease, and neuropsychiatric disorders such as alcohol use disorder; for whom low brain NAD⁺ levels may serve as a biomarker of impaired brain mitochondrial health and impaired cognitive functioning. Brain NAD⁺ can also be utilized as a neurobiological outcome in clinical trials that assess clinical efficacy of NR or other NAD⁺ supplements and can provide meaningful insight into the neural effects of treatment responses.

The main limitations of the study were, first, the lack of blood or other peripheral NAD⁺ levels, second, a small sample size and lastly, the limited age range. However, our study was aimed to assess the sensitivity of the NAD⁺ signal with DF ¹H MRS to a single dose of NR. Future studies are necessary to additionally assess correspondence between the NR-induced NAD⁺ in brain and peripheral measures of NAD⁺. The current study was performed in healthy volunteers with ages ranging from 21 years to 54 years old to optimally detect effects of acute NR supplementation. Future studies with a larger age range including both male and female subjects are necessary to further explore effects of age, sex, BMI, and nutritional status on the NR-induced changes in cerebral NAD⁺ levels.

CONCLUSION

In summary, using the ultra-high-field DF ¹H MRS, we demonstrated that a single dose of NR produces a measurable and significant increase in brain NAD⁺ levels in healthy human volunteers. This pilot study paves the way for future studies that can further assess the effects of NR supplementation protocols on NAD⁺ concentrations in both healthy subjects and patient populations.

DATA AVAILABILITY STATEMENT

The data from this study will be provided upon reasonable request to the corresponding author.