Identification of metabolites associated with preserved muscle volume after aneurysmal subarachnoid hemorrhage due to high protein supplementation and neuromuscular electrical stimulation

The INSPIRE randomized clinical trial demonstrated that a high protein diet (HPRO) combined with neuromuscular electrical stimulation (NMES) attenuates muscle atrophy and may improve outcomes after aneurysmal subarachnoid hemorrhage We sought to identify specific metabolites mediating these effects. Blood samples were collected from subjects on admission prior to randomization to either standard of care (SOC; N = 12) or HPRO + NMES (N = 12) and at 7 days. Untargeted metabolomics were performed for each plasma sample. Sparse partial least squared discriminant analysis identified metabolites differentiating each group. Correlation coefficients were calculated between each metabolite and total protein per day and muscle volume. Multivariable models determined associations between metabolites and muscle volume. Unique metabolites (18) were identified differentiating SOC from HPRO + NMES. Of these, 9 had significant positive correlations with protein intake. In multivariable models, N-acetylleucine was significantly associated with preserved temporalis [OR 1.08 (95% CI 1.01, 1.16)] and quadricep [OR 1.08 (95% CI 1.02, 1.15)] muscle volume. Quinolinate was also significantly associated with preserved temporalis [OR 1.05 (95% CI 1.01, 1.09)] and quadricep [OR 1.04 (95% CI 1.00, 1.07)] muscle volume. N-acetylserine and β-hydroxyisovaleroylcarnitine were associated with preserved temporalis or quadricep volume. Metabolites defining HPRO + NMES had strong correlations with protein intake and were associated with preserved muscle volume.


Subjects
Subjects were selected among the 25 enrolled in the INSPIRE phase 2 randomized controlled trial (NCT03201094) who had available plasma samples 15 .Of those subjects, one in the SOC group had not consented to plasma collection, resulting in plasma samples available from 24 subjects.Detailed methods are available in the published trial results 15 .INSPIRE was approved by the institutional review board (IRB) at the University of Maryland School of Medicine (HP-00074174).All methods were performed in accordance with the Declaration of Helsinki.Prior to enrollment, informed consent was obtained from all patients or their legal guardians.All patients enrolled in INSPIRE had a diagnosis of aSAH, underwent aneurysm repair within 48 h of ictal hemorrhage, were at least 18 years of age, and had a Hunt Hess score (HHS) ≥ 2 and modified Fisher Scale (mFS) score > 1. Patients with subarachnoid hemorrhage for etiologies other than a ruptured aneurysm (trauma, arteriovenous malformation, neoplasm) were excluded.Patients were excluded if they were unlikely to survive one-week post hemorrhage or unlikely to remain in the intensive care unit (ICU) for more than 7 days.Patients with a body mass index < 15 or > 40 kg/m 2 , protein allergy, premorbid modified Rankin Scale (mRS) score > 1, or who were currently pregnant or diagnosed with a malignancy, inflammatory disorder, neuromuscular disorder, or renal failure were also excluded.
All subjects were randomized to either SOC or HPRO + NMES.Subjects in the HPRO + NMES group were administered a bolus of a whey protein power dissolved in water (8-10 oz) three times daily with a dose of 3 g leucine/feeding for a goal of 1.75 g/kg/d.SOC subjects received 1.2-1.4g/kg/day of protein delivered via enteral nutritional formulas or specific oral diets with no additional protein supplementation permitted.The NMES device was a L300 Plus® system (Bioness, Inc, Valencia, CA) with thigh cuffs applied bilaterally to stimulate with stimulator pads across the quadricep muscles.The NMES intervention included two 30-min session per day.All subjects underwent study interventions until post-bleed day (PBD) 14.Each subject was followed up at 90-days in person for outcome measurements.Nitrogen balance was calculated as described 16,17 .

Image acquisition
Included patients were scanned using one of two scanners: a 64-section CT unit (Brilliance; Philips Healthcare, Andover, MA) or a 128-section dual-source CT unit (SOMATOM Force; Siemens, Erlangen Germany).Images were archived at 3 mm section thickness.Baseline and follow-up series also were saved as Neuroimaging Informatics Technology Initiative (NifTI) files for voxel-wise labeling and volumetric analysis using 3D slicer (version 4.11.20210226,slicer.org).

Image Analysis and muscle volume measurement
Temporal muscle volumes were measured from 120 kV images using label masks of the temporal muscle that was created using a 3D threshold paint tool initially in the axial plane and editing in coronal and sagittal planes.Thresholding was set to a range of 30-80 units to minimize noise and avoid neighboring hyperdense blood and calvarium.Labeling was performed with small (1-3 mm diameter) spherical ROI to carefully delineate the interface of the membrane with adjacent structures.Labeling was performed by a resident and all scans were subsequently reviewed and edited by a radiology attending with 14 years of experience.Both reviewers were blinded to clinical data.Muscle volume was measured from the origin to the coronoid process of mandible (first slice depicting the coronoid process).Quantitative values (cm 3 ) were derived for each label class using the Segment Statistics slicer module.Quadricep muscle volumes were assessed by CT with scan conducted from the patella to the femoral head.Cross-sectional areas were calculated at the thigh by manually outlining the regions of interest using MIPAV (Medical Image Processing, Analysis, and Visualization v10) as described 21 .Cross-sectional areas were multiplied by the number of slices and distance between slices in order to obtain volumetric measurements in cm 3 .

Bioinformatics
All bioinformatics analyses were performed in R (R Foundation for Statistical Computing).Fold changes (FC) were calculated for each metabolite comparing either late (7-days) vs early (within 24 h of admission) or between HPRO + NMES and SOC at day 7. Changes in metabolites were considered to be significant increased at FC > 2 and decreased at FC < 0.5 with false discovered rate (FDR) corrected P-values < 0.05.Sparse partial least squared discriminant analysis (sPLS-DA) was conducted using the mixOmics library in R (http:// mixom ics.org).Adding discriminant analysis to the PLS algorithm allows for classification of large datasets [22][23][24] .We used sPLS-DA to select the most discriminative metabolites to classify groups.The perf function was used to determine the optimal number of components to use in order to maintain a classification error rate less than 0.1.The tune.splsda function was used to determine the number of metabolites in order to minimize the balanced error rate.

Statistics
All statistical analyses were performed in R (R Foundation for Statistical Computing).Distribution of each variable was assessed with a Shapiro-Wilk test 25 .Demographic variables were compared using a t-test (age, protein per day, nitrogen balance, muscle volume), Mann-Whitney U test (mRS), or χ-squared test (sex, race, HHS, DCI) based on the distribution of each.After confirming normal distributions for each metabolite, Pearson's correlation coefficients were calculated between metabolites and protein per day as well as nitrogen balance.Multivariable linear regression models were developed evaluate the relationship between selected metabolites (change in levels from admission to 7 days) and muscle volume (temporalis and quadricep muscles).All models were adjusted for age, sex, and aSAH severity (HHS < 4 vs ≥ 4).

Demographics
Demographics for all subjects are shown in Table 1.A total of 24 subjects were included with 12 in both the SOC and HPRO + NMES groups.The groups were well matched with no significant differences in age, sex, race, Vol:.( 1234567890

Metabolites detected
A total of 1370 metabolites were detected from plasma samples with 261 of these metabolites unable to be identified.These metabolites were excluded resulting in a total of 1109 metabolites available for analysis.Metabolites consisted of amino acids, carbohydrates, cofactors and vitamins, energy related pathways (glycolysis, gluconeogenesis, and TCD cycle), lipids, nucleotides, peptides, as well as partially characterized molecules (Supplementary Fig. 2).

Metabolite FC
Paired FC and corresponding FDR corrected P-values were calculated for each metabolite.Figure 1 shows paired FC between metabolite levels measured at PBD 7 compared to metabolite levels measures within 24 h of ictus (before randomization) in the SOC group (Fig. 1A), as well as those in the HPRO + NMES group (Fig. 1B).Data are displayed as volcano plots with metabolites considered to be significantly different at a FC of > 2 or < 0.5 [represented as log2(fold change) of 1 or − 1 in each plot] and FDR P-value < 0.05.No metabolite fold changes were significantly different in the SOC group, while 8 metabolites were significantly increased and 1 significantly decreased in the HPRO + NMES group.All differentially expressed metabolites are shown in Supplementary Table 1.

sPLS-DA analysis
sPLS-DA analysis was performed to determine metabolites contributing to differences between the HPRO + NMES group pre and post randomization (Fig. 2A).The top 10 metabolites that differentiated the HPRO + NMES group pre and post randomization are shown in Fig. 2B and are also listed in Supplementary Table 2.We also compared each metabolite between the SOC and HPRO + NMES groups.sPLS-DA was performed comparing SOC and HPRO + NMES groups using changes in each metabolite (Fig. 2C).The top 10 metabolites distinguishing SOC from HPRO + NMES are shown in Fig. 2D and also listed in Supplementary Table 2. Metabolites identified to have significant changes attributable to HPRO + NMES protocol were often amino acid metabolic intermediates.Considering both comparisons (1.HPRO + NMES pre and post randomization, and 2. SOC vs HPRO + NMES, Fig. 2B,D, respectively), a total of 18 unique metabolites were identified.
Changes from baseline to 7 days in each of the 18 metabolites comparing SOC and HPRO + NMES are shown in Supplementary Fig. 3.
Correlations between each of the 18 metabolites and temporalis and quadricep muscle preservation as well as those with significant negative correlations are shown in Supplementary Table 4.

Multivariable models for muscle volume
Separate multivariable models were developed to assess the association between individual metabolites and temporalis and quadricep muscle volume (Table 2).Models were developed for each of the 18 metabolites, with those models discussed below having significant associations with either temporalis or quadricep volume.Each model was adjusted for age, sex, and aSAH severity (HHS 1-3 vs 4-5).In the first model, the amount of protein intake per day was associated with increased volume of both temporalis [OR 1.13 (95% CI 1.07, 1.20), P = 5.69 × 10

Discussion
The INSPIRE trial demonstrated that a high protein diet delivered enterally combined with neuromuscular electrical stimulation reduces muscle atrophy in the quadricep muscle after aSAH 15 .Herein, we identify the effects of HPRO + NMES on circulating plasma metabolites and examine the associations between changes in metabolite levels and muscle volume in both quadricep and temporalis muscles.
Using untargeted metabolomics, we identified the effects of a combined treatment (HPRO + NMES) on metabolite concentrations.Specifically, 18 metabolites that were prominently affected by the treatment.Our findings support the concept that nutritional and muscle stimulation interventions change the metabolome and these changes can contribute to overall muscle preservation in the acute stages of aSAH.Not surprisingly, among HPRO + NMES patients, there were notable increases in amino acid derivatives in plasma.Many of these amino acid derivatives, especially as related to glutamine metabolism are established important building blocks to preserving muscle mass.We also found carnitine esters (β-hydroxyisovaleroylcarnitine, tigylcarnitine), a tryptophan catabolite (quinolinate), a monocarboxylic acid (6-oxopiperidine-2-carboxylate), a phosphatidylethanolamine (1-oleoyl-2-arachidonoyl-GPE (18:1/20:4), a monophosphoglycerate involved in glycolysis and the calvin cycle (3-phosphoglycerate), and urea to be higher after HPRO + NMES treatment.Other metabolites including the isoflavones genistein and daidzein were lower in the HPRO + NEMS group.These metabolites are present in their sulfonic acid conjugates, which is their most common circulating form after undergoing hepatic metabolism by sulfotransferase enzyme 26 .The furoic acid 3-carboxy-4-methyl-5-pentyl-2-furanpropionate (3-CMPFP) was also decreased in the HPRO + NMES group.Classification and suspected biological function of each metabolite are shown in 5.
Our analysis showed that as expected, half of the 18 that were affected by the treatment were significantly positively correlated with protein intake per day (Supplementary Table 3), namely the acetylated amino acids as well as other metabolites, such as the carnitine ester β-hydroxyisovalerocylcarnitine, which plays a role in leucine catabolism.Urea, the end product of the urea cycle that plays a pivotal role in metabolizing excess nitrogen was also correlated with increased protein intake per day 27 .Additionally, an acylcarnitine (tiglylcarnitine) was correlated with protein intake per day.Although not a direct amino acid derivative, tiglyl containing compounds play a role in the metabolism of isoleucine 28 .Consistently, metabolites that were most negatively correlated with protein intake per day consisted of fatty acids and ketones (Supplementary Table 3) suggesting a shift away from fatty acid and ketone metabolism in those subjects receiving more dietary protein.While a shift away from fatty acid metabolism may attenuate inflammation 29 , some fatty acids such as omega-3 polyunsaturated fatty acids as well as ketones are thought to play a beneficial role 30 .Future studies will be necessary to understand the functional significance of decreased fatty acids and ketones on functional recovery.
In our previous study, atrophy of the quadricep muscle correlated with protein intake.In this analysis, we add to our findings by confirming in a multivariable model adjusting for age, sex and aSAH severity that higher protein intake per day as well as higher nitrogen balance were associated with not only perseveration of the quadricep muscle but temporalis muscle as well.Recent reports indicate that differences in temporalis muscle volume may be a marker of disease severity and prognosis after aSAH 31 .Our findings would suggest that this muscle may also be a sensitive marker of nutritionally-driven metabolomic changes after aSAH.
Increased levels the acylated amino acid N-acetylleucine were associated with both temporalis and quadricep muscles preservation, while other acetylated amino acids were only significantly associated with temporalis muscle preservation.Although the quadricep muscle was exposed to facilitated exercise with NMES during the study period, the smaller size of the temporalis may have made smaller changes in muscle mass easier to detect.The branched chain amino acid leucine and its metabolic derivatives are increased in paradigms of amino acid supplementation combined with exercise.Leucine in particular impacts the mTORC1 pathway and plays an integral role in energy homeostasis 32 .N-acetylleucine (NAL) has therapeutic potential, with studies showing that it may improve motor function in cerebral ataxia 33 as well as lysosomal storage disease including Niemann Pick 34,35 and GM2 Gangliosidosis 36,37 .More recently, NAL treatment has been shown to improve motor and cognitive outcomes after TBI in a mouse model 38 .NAL therefore may be a promising treatment for functional recovery after aSAH.
Increased levels of quinolinate were also shown to be associated with muscle preservation.Quinolinate plays an important role in tryptophan metabolism via the kynurenine pathway 39,40 .While upstream metabolites such as kynurenine have been associated with muscle wasting during critical illness 41,42 , quinolinate does not have this effect 43 .Quinolinate can be used to synthesize nicotinamide adenine dinucleotide (NAD + ), which is depleted in response to proinflammatory stimuli 44 .Numerous reactions rely on NAD + such as DNA repair via poly-ADP ribosylation and sirtuins, which have regulatory roles in cellular metabolism 45,46 .Although it is unclear whether quinolinate has a benefit on functional outcomes after aSAH, it is possible that the higher protein intake provides more tryptophan as a substrate to produce quinolinate resulting in the ability to restore NAD + depleted after aSAH and the subsequent robust inflammatory response.Although we measured circulating quinolinate in plasma, quinolinate is also well known to have neurotoxic effects in the CNS as it is an agonist of the N-methyl-D-aspartate (NMDA) receptor and acts as an excitotoxin 47 .Therefore, additional studies will be needed to determine whether circulating quinolinate plays a beneficial role or is primarily a bioproduct of higher protein intake.
Although we have identified circulating metabolites that are increased after the HPRO + NMES intervention, our results are primarily correlative regarding which metabolites are associated with preserved muscle volume.It www.nature.com/scientificreports/ is well known that dietary protein intake stimulates muscle synthesis 48 .While we cannot exclude the possibility that some metabolites identified herein are byproducts of protein metabolism rather than being directly responsible for muscle synthesis, we believe that a number of these metabolites likely have beneficial roles.For instance, acetylated amino acids may play a role in muscle structure and function 49 .Furthermore, the acetylated amino acid NAL modulates metabolism broadly through its effects on mammalian target of rapamycin (mTOR) and activation of autophagy 38,50,51 .Dietary protein intake acts in concert with exercise to increase muscle synthesis 52 , and NMES acts as a surrogate for exercise 53 .Although many of the 18 metabolites that we have identified are strongly correlated with protein intake, NMES also contributes to their levels.For example, muscle contraction induced by NMES likely increases levels of acetylcarnitines 54 .We therefore suspect preserved muscle volume as well as increases in the metabolites identified in this manuscript are due to a combination of both HPRO and NMES.High protein diets have been extensively evaluated in critically ill patients.Data from patients with traumatic brain injury (TBI) have suggested that a higher protein intake (1.5-2.0 g/kg/day) may be beneficial for recovery 55 .However, the recent EFFORT Protein trial found no significant survival benefit of a high protein diet alone with an increased risk of acute kidney injury (AKI) in at risk patients 8 .This large, randomized trial was conducted in mechanically ventilated patients without acquired brain injury, limiting its applicability to our study.In this study, we have identified specific metabolites that are increased by HPRO + NMES treatment which in turn were also associated with intermediaries of energy homeostasis and muscle preservation.This suggests that a more tailored intervention including a combination of metabolites such as acetylated amino acids like N-acetylleucine along with facilitated exercise with NMES may be able to preserve muscle mass while avoiding the potentially deleterious effects of a high protein diet.
This study has several important limitations.First, the small sample size (12 subjects in each group) as well as numerous comparisons being made for a broad panel of metabolites may have resulted in some associations been found by random chance.We attempted to mitigate this by utilizing false discovery rate corrected P-values.This small samples size also precluded being able to make associations with functional outcomes.Second, patients in the intervention arm received a combination of HPRO and NMES making it difficult to deconvolute the effects of these two interventions, as discussed above.Although we determined associations between metabolites and protein per day, overlapping effects of NMES likely contributed the changes in metabolites observed.Third, this study was conducted at a single tertiary care facility, therefore given variability in practices and patient populations, these results may not be broadly generalizable.Finally, although we assessed changes in metabolites from admission to 7 days after aneurysm rupture, we are not able to determine the exact mechanisms by which certain metabolites may mitigate muscle atrophy.Nevertheless, this is one of the first studies to better understand the systemic metabolic effects of HPRO and NMES and provides an important foundation for additional studies.Further, the study groups were randomized, there was careful implementation of the study intervention, and metabolites were analyzed with statistical rigor.

Conclusions
The HPRO + NMES intervention results in a distinction profile of circulating metabolites including a variety of amino acid derivatives.Further larger studies are required to elucidate the mechanisms by which amino acid metabolic intermediates such as N-acetylleucine prevent muscle atrophy and evaluate their potential therapeutic use after aSAH.

Figure 1 .
Figure 1.Volcano plots comparing paired metabolites at early and late time points.Paired fold changes (FC) were calculated for each subject in the SOC group (A) and HPRO + NEMS group (B) comparing the early and 7-day timepoints.Volcano plots were created using log 2 (FC) and − log 10 (FDR corrected P-values).Changes were considered to be significant if FC was greater than 2 or less than 0.5 and FDR corrected P-value was < 0.05.Metabolite that significantly increased from admission to 7-day are shown in red, and those that decreased are shown in blue.SOC (standard of care), HPRO + NMES (high protein plus neuromuscular electrical stimulation), FC (fold change), FDR (false discovery rate), 3-CMPFP (3-Carboxy-4-methyl-5-propyl-2-furanpropionate3-Carboxy-4-methyl-5-propyl-2-furanpropionate).

Figure 2 .
Figure2.Sparse partial least squared discriminant analysis (sPLS-DA).sPLS-DA was performed to determine metabolites driving differences between groups.Groups consisted of paired changes in metabolites for subjects in the HPRO + NMES group from baseline to 7-days (A, B) and SOC versus HPRO + NMES (C, D).Loadings plots detail the top 10 metabolites accounting for differences between groups.

Table 1 .
Demographics and intervention characteristics.SOC standard of care, HPRO + NMES high protein plus neuromuscular electrical stimulation, HHS Hunt Hess scale, DCI delayed cerebral ischemia, mRS modified Rankin scale, SBBP short performance physical battery, IQR interquartile range.*P-values reflect comparisons between SOC and HPRO + NMES.† Muscle volumes are represented as percent changes, however original volumetric measurements were in cm 3 .