High-Resolution Plasma Metabolomics and Thiamine Status in Critically Ill Adult Patients

doi:10.21203/rs.3.rs-3597052/v1

Background and Aim:

Thiamine (Vitamin B1) is an essential micronutrient and a co-factor for metabolic functions related to energy metabolism. We determined the association between whole blood thiamine pyrophosphate (TPP) concentrations and plasma metabolites using high resolution metabolomics in critically ill patients.

Methods

Cross-sectional study performed in Erciyes University Hospital, Kayseri, Turkey and Emory University, Atlanta, GA, USA. Participants were ≥ 18 years of age, with an expected length of ICU stay longer than 48 hours, receiving furosemide therapy for at least 6 months before ICU admission.

Results

Blood for TPP and metabolomics was obtained on the day of ICU admission. Whole blood TPP concentrations were measured using high-performance liquid chromatography (HPLC). Liquid chromatography/mass spectrometry was used for plasma high-resolution metabolomics. Data was analyzed using regression analysis of TPP levels against all plasma metabolomic features in metabolome-wide association studies. We also compared metabolomic features from patients in the highest TPP concentration tertile to patients in the lowest TPP tertile as a secondary analysis. We enrolled 76 participants with a median age of 69 (range, 62.5–79.5) years. Specific metabolic pathways associated with whole blood TPP levels, using both regression and tertile analysis, included pentose phosphate, fructose and mannose, branched chain amino acid, arginine and proline, linoleate, and butanoate pathways.

Conclusions

Plasma high-resolution metabolomics analysis showed that whole blood TPP concentrations are significantly associated with metabolites and metabolic pathways linked to the metabolism of energy, amino acids, lipids, and the gut microbiome in adult critically ill patients.

critical illness

energy metabolism

metabolomics

micronutrients

thiamine

vitamin

Thiamine, an essential water-soluble vitamin, was the first B vitamin to be discovered and thus was named vitamin B₁(Whitfield, Bourassa et al. 2018). Thiamine pyrophosphate (TPP) is the active form of thiamine in the human body (Frank 2015). TPP is involved in a wide variety of biochemical pathways required to

maintain normal tissue and organ function, including as a co-factor in glycolysis, TCA cycle and pentose phosphate pathway metabolism. For example, the pentose phosphate pathway has a fundamental role in glucose metabolism and is also a major route for the synthesis of many neurotransmitters, nucleic acids, lipids, amino acids, steroids, and glutathione (Polegato, Pereira et al. 2019).

Critical illnesses are severe, life-threatening disease states with heterogeneous clinical presentations(Ferrario, Cambiaghi et al. 2016). Critically ill individuals exhibit variable levels of inflammation due to infection, organ dysfunction, and other factors that increase counter-regulatory hormones and cytokines which contribute to a hypercatabolic state associated with insulin resistance and muscle wasting (Cyr, Zhong et al. 2021, Rousseau, Prescott et al. 2021, Wendt, Castro-Pearson et al. 2021). Disturbed metabolic pathways associated with such hypercatabolic conditions in critically ill patients may contribute to and/or result in mitochondrial/energetic dysregulation, cellular damage, and organ failure (Puthucheary, Rawal et al. 2013, Christopher 2018, McClave, Wischmeyer et al. 2019, Gourd and Nikitas 2020, Siddiqui, Pandey et al. 2020, Flower, Page et al. 2021). Several reports, including our previous study, have shown that thiamine depletion may be common in intensive care unit (ICU) patients, especially those receiving chronic diuretic therapy prior to ICU admission (Attaluri, Castillo et al. 2018, Gundogan, Akbudak et al. 2019). Since thiamine is involved in energy metabolism and the metabolic processing of other essential nutrients, low thiamine status may be associated with significant alterations in key metabolic pathways in ICU patients.

High-resolution metabolomics (HRM) analysis is a rapidly developing science that utilizes liquid chromatography and ultra-high-resolution mass spectrometry (LC-MS) to detect small molecules, including nutrient-related metabolites, in plasma, urine, tissue, and other biosamples (Jones, Park et al. 2012, Jones 2016). Current HRM methods use advanced data extraction and bioinformatics methods to detect tens of thousands of metabolic features in plasma derived from endogenous and exogenous sources, including dietary nutrients, intermediates of macro- and micronutrient metabolism, the gut microbiome, environmental exposures, commercial products, and drugs(Jones, Park et al. 2012, Uppal, Walker et al. 2016). As examples, our recent studies have utilized plasma HRM to determine metabolites and linked metabolic pathways (e.g., amino acid metabolism, TCA cycle intermediates) associated with lean mass in working adults and in individuals with active pulmonary tuberculosis infection, respectively (Bellissimo, Jones et al. 2021, Collins, Jones et al. 2021). These and other studies have focused on a hybrid approach using targeted and untargeted (discovery)-based HRM detection of plasma metabolites and specific metabolic pathways enriched in significant metabolites to provide new insight into human metabolism and disease-associated pathophysiology(Alvarez, Chong et al. 2017). High-resolution metabolomics can be important for providing personalized and precision nutrition support for critical illness(Christopher 2018, Siddiqui, Pandey et al. 2020). Metabolomics data can be useful for assessing the risk related to nutritional deficiencies(Lasky-Su, Dahlin et al. 2017), clarifying the metabolic mechanism behind nutritional treatments (Amrein, Lasky-Su et al. 2021), comparing the use of enteral and parenteral nutrition (Gonzalez-Granda, Seethaler et al. 2021), and discovering new targets for nutritional interventions (Viana, Becce et al. 2021, Stolarski, Young et al. 2022).

Nutritional metabolomics using HRM thus represents a novel tool to explore nutrition-related physiology and pathophysiology in catabolic states such as critical illness.

The aim of this pilot study was to determine metabolites and metabolic pathways linked to whole blood TPP concentrations in adult ICU patients on chronic diuretic therapy, who may be at particular risk for thiamine depletion. We compared patients with lower versus higher TPP concentrations using both targeted plasma HRM (pathways known to be dependent upon thiamine as a co-factor, such as the pentose phosphate pathway and the TCA cycle) and untargeted plasma HRM analytical approaches.

The current study was conducted in the Erciyes University Medical ICU center in Kayseri, Turkey and in the Clinical Biomarkers Laboratory at Emory University, Atlanta, GA, USA.

The present clinical research was performed in accordance with the ethical standards of the responsible committees on human experimentation and with the Helsinki Declaration of 1975 and approved by the Erciyes University Ethics Committee (Date:15.01.2020, Number: 2020/35). Written informed consent was obtained from all patients or their legal representatives.

Study Participants

Participants were included if they were at least 18 years of age, were deemed to require ≥ 48 hours of ICU treatment and received furosemide therapy for 6 months or longer before ICU admission. Patients receiving high-dose oral thiamine (≥ 50 mg/day within 14 days prior to ICU admission) were excluded.

Demographic and Clinical Data

At ICU admission

Demographic characteristics (age, gender, body mass index (BMI)), reason for ICU admission, presence of comorbidity, and Acute Physiology and Chronic Health Evaluation II (APACHE II) score of the 76 study participants were recorded. The type, dose, and duration of diuretic use of all participants prior to admission to the Medical ICU were noted. The type of nutrition support, and the intake of carbohydrate, energy and insulin administration were documented on the ICU admission day.

Sample Preparation and Analysis

Whole blood thiamine concentrations

Non-fasted whole blood samples were obtained on the day of ICU admission. TPP concentrations were measured using high-performance liquid chromatography (HPLC; Shimadzu, 8040, Immuchrom, Japan) at Erciyes University. The reference range for normal TPP values was 28–78 ng/Ml (Evliyaoglu, van Helden et al. 2019).

Plasma for metabolomics analysis

Blood samples obtained on the day of ICU admission were centrifuged at 1500 g with a cooled centrifuge for 10 minutes. EDTA plasma samples were separated and stored at − 80ºC until all participant samples were collected. Samples were shipped on dry ice to Atlanta, GA for metabolomics batch analysis in the Clinical Biomarkers Laboratory at Emory University, Atlanta.

High-resolution metabolomics

Plasma was analyzed by high-resolution mass spectrometry as previously described (Soltow, Strobel et al. 2013, Liu, Walker et al. 2016). Plasma samples (50\(\mu\)l) were mixed with 100 µl acetonitrile containing 1.25µl internal standard solution with eight stable isotopic chemicals representing multiple small-molecule classes (Soltow, Strobel et al. 2013). Samples were then equilibrated on ice for 30 minutes, and centrifuged at 16,000 x g for 10 minutes to remove precipitated proteins. The supernatant was added to an autosampler vial and maintained at 4^◦C until analysis. Liquid chromatography was performed on triplicates of each sample using both C18 (Higgins C18 stainless steel column, 2.1 × _50 mm) and hydrophilic interaction liquid chromatography (HILIC, Waters XBridge BEH Amide XP HILIC column, 2.1 × 50 mm) liquid chromatography columns followed by negative (C18) or positive (HILIC) electrospray ionization (ESI) and high-resolution mass spectrometry (Thermo Orbitrap Fusion Tribrid).

Mass spectral data were collected over a 5-minute run period at a resolution of 120,000 and mass-to-charge (m/z) scan range of 85 -1275. Each batch of 40 samples also included six triplicates of a pooled normal human plasma reference standard (Qstd3) (Go, Walker et al. 2015). The raw data files were extracted and aligned using apLCMS (Yu, Park et al. 2009) and xMSanalyzer (Uppal, Soltow et al. 2013). The resulting feature table consisted of unique metabolic features (metabolites) defined by accurate mass m/z, retention time, and ion intensity. Data was filtered to exclude m/z features having a median coefficient of variation within technical replicates ≥ 75%, and only samples with Pearson correlation within technical replicates ≥ 0.7 were used for downstream analysis (Go, Walker et al. 2015). Triplicates were median summarized with the condition that at least two replicates had non-missing values.

Metabolite annotation was performed using xMSannotator, an R program that uses a clustering algorithm to provided tentative annotations of features using publicly available databases, including the Kyoto Encyclopedia of Genes and Genomes (KEGG) and the Human Metabolome Database (HMDB) (Uppal, Walker et al. 2017).

Statistical Analysis

Prior to the analysis, the feature table was filtered to retain the features having non-zero values in at least 50% of the samples overall and in at least 80% of the samples in a group. After filtering, the feature intensities were log2 transformed and quantile normalized. Multiple linear regression analysis was used to conduct metabolome-wide association analysis (MWAS) to determine the associations of metabolic features (metabolites) with TPP concentrations, adjusting for age, sex, BMI, and Apache II score as a priori covariates.

Given the relatively low number of participants with TPP levels below the lower limit of normal, we also performed a separate secondary regression analysis to compare participants with TPP concentrations in the lowest versus the highest tertile of whole blood TPP concentrations. This tertile analysis was also adjusted for age, sex, BMI, and APACHE II score. We adjusted for APACHE II score in HRM analysis given that inflammation during critical illness may alter blood concentrations of several micronutrients, with redistribution from blood to tissues (Berger, Shenkin et al. 2022).

Pathway enrichment analysis was performed using mummichog (v2.0.6), which uses permutation analysis to map features based on both m/z and retention time to specific pre-identified human metabolic pathways (Li, Park et al. 2013). For the MWAS, all features associated with thiamine with a raw p-value < 0.05 were included in the pathway analysis. For the tertile analysis, all features that discriminated between tertiles with a raw p-value < 0.05 were included. Using a raw p-value < 0.05 (compared to the much more stringent false discovery rate correction) protects against type II error and prevents information loss when performing pathway analysis. The mummichog algorithm maps the significant features to pathways and determines the p-values for these enriched pathways after adjusting for the null distribution of pathway p-values established using 500 permutations, which protects against type I error. Thus, running mummichog using metabolomic features with p < 0.05 effectively provides a compromise to handle both type I and type II error(Uppal, Walker et al. 2016). As additional protection against type I error, only the pathways enriched with at least four overlapping metabolites and one Level 1 confirmed metabolite (see below) were considered as significantly relevant to thiamine status.

Metabolites within specific pathways were identified and confirmed by comparison to an in-house library of metabolites previously validated using ion-dissociation tandem mass spectrometry (MS/MS) and co-elution with authentic standards(Liu, Nellis et al. 2020). Identification scores were assigned for metabolites based on an adaptation of the criteria proposed by Schymanski et al (Schymanski, Jeon et al. 2014): Level 1: confirmed by matching to an in-house library of MS/MS validated metabolites established using authentic standards Level 2: confirmed by matching to validated adducts from our in-house library and matches with online databases; Level 3: matches online databases and correlated with metabolites in pathways via mummichog; Level 4: computationally assigned annotation using the biostatistical program xMSannotator (medium or high confidence); Level 5: accurate mass (m/z) match only.

The median age of study patients was 69 (range, 62.5–79.5) years and 65% were female. The median APACHE II score on ICU admission was 14 (range, 11–20). The most common reasons for ICU admission were acute respiratory failure (62.3%) and metabolic disorders, including renal and hepatic failure, severe hyperglycemia, and hepatic encephalopathy (63.6%). Some participants had more than one of these conditions upon ICU admission. Chronic liver cirrhosis (53.2%) and congestive heart failure (50.6%) were the most frequent co-morbid diseases of the study participants and were the primary reason for chronic diuretic use prior to ICU admission. Their median duration of ICU stay was 5 days (range, 3–8) and ICU mortality rate was 23.4%. The median TPP concentration was 41.4 (range, 32.2–57.5) with a mean value of 48.5 ± 24.8 (SD) ng/mL. The whole blood TPP normal reference value: 28–78 ng/mL. The proportion of patients with below normal TPP levels was 14.3%. Additional demographic and clinical characteristics of the participants are listed in Table 1.

Table 1. Study Subject Demographic and Clinical Characteristics

Variable

Total (n=76)

Age (years, median (IQR))

mean ±SD

69.0 (62.5.0-79.5)

69.3± 13.3

Gender, n (%)

Male

Female

27 (35.0)

50 (65.0)

Body mass index (kg/m², median (IQR))

mean ±SD

27.5 (24.4-33.1)

29.0± 6.2

Admit to ICU, n (%)

From emergency department

From general floor

From outside hospital

41 (53.2)

31 (40.3)

5 (6.5)

Reason for admission in ICU*, n (%)

Acute respiratory failure

Metabolic disorder (renal failure, hyperglycemia and/or

and hepatic failure)

Cardiovascular disorders

Neurological disorders

GI bleeding

Trauma

Sepsis/Septic shock

48 (62.3)

49 (63.6)

11 (14.2)

8 (10.4)

7 (9.1)

Comorbidities*, n (%)

Liver cirrhosis

Heart failure

Renal failure

COPD

41 (53.2)

39 (50.6)

28 (36.4)

26 (33/8)

APACHE II score (median (IQR))

mean ±SD

14.0 (11.0-20.0)

15.6± 6.1

SOFA score (median (IQR)), mean ±SD

4.0 (3.0-6.0)

4.7± 2.7

Glasgow coma score (median (IQR))

mean ±SD

15.0 (12.0-15.0)

12.8± 3.7

Treatment with mechanical ventilation, n (%)

12(15.6)

Treatment with vasopressors, n (%)

18 (23.4)

Total calorie intake on ICU admission day (median (IQR)

(n=52)

625 (400-1000)

TPP levels (ng/mL, median (IQR) and mean ± SD

41.4 (32.2-57.5)

48.5± 24.8

* Some participants had multiple comorbidities and primary reasons for ICU admission.

Abbreviations: APACHE II, The Acute Physiology and Chronic Health Evaluation; COPD, chronic obstructive pulmonary disease, CRP, C-reactive protein, ICU, intensive care unit; SOFA, Sequential Organ Failure Assessment, TPP, Thiamine pyrophosphate

Whole blood TPP normal reference value: 28-78 ng/mL

IQR, Interquartile range (25^thquartile-75^thquartile).

HRM Analysis

A total of 18,189 plasma metabolomic features were identified from the C18/ESI negative column after the LC/MS run, with 16,598 features used in downstream analysis after pre-processing. There were 21,737 total plasma metabolomic features identified from the HILIC/ESI positive column after LC/MS, with 19,576 features used in downstream analysis after pre-processing.

Type 1 (m/z) and Type 2 (retention time) Manhattan plots were used to visualize the metabolomic features associated with TPP status for both MWAS and tertile analysis, respectively (Fig. 1, tertile analysis not shown). In the C18/ESI-, data, there were 672 features (368 negatively, and 304 metabolites positively) associated with TPP levels at raw P < 0.05 based on MWAS analysis, and in the HILIC/ESI + data, 1033 features were associated with TPP levels (505 negatively associated, and 528 positively associated) (Supplemental Tables 1 and 2, respectively). When comparing the highest TPP tertile to the lowest tertile, there were 718 features in the C18/ESI- data and 1063 features in the HILIC/ESI + data that varied between tertiles at raw P < 0.05 (Supplemental Tables 3 and 4, respectively).

Features associated with whole blood TPP concentrations were plotted using one-way Hierarchical Cluster Analysis (HCA). Features that were increased or decreased in relation to TPP levels reveal distinct clustering and regulation (Fig. 2A). Metabolomic feature clustering was more distinct in the tertile analysis shown in Fig. 2B.

Pathway HRM Analysis

Metabolites associated with whole blood TPP concentrations according to MWAS analysis were significantly enriched in nine metabolic pathways (Fig. 3, Panel A). The enriched pathways included glucose metabolism (fructose and mannose, pentose phosphate pathway), amino acid and nitrogen metabolism (BCAA degradation, arginine, and proline), lipid metabolism (fatty acid β-oxidation and linoleate metabolism), bile acid metabolism (bile acid biosynthesis), microbiome (butanoate metabolism), and vitamin metabolism (niacin metabolism), respectively.

Metabolites that differed between the lowest and highest TPP tertiles were enriched in 20 metabolic pathways (Fig. 3, Panel B). Most of the significant pathways were related to amino acid/nitrogen metabolism (e.g., aspartate and asparagine, BCAA degradation, arginine and proline, lysine, and others), glucose metabolism (pentose phosphate, hexose phosphorylation, fructose, and mannose), and vitamin/cofactor metabolism (biotin, lipoate and vitamin C).

The relationship between whole blood TPP concentrations and selected significant metabolites from enriched pathways found using MWAS analysis are shown in Table 2. There is a significant negative correlation between TPP concentrations and plasma gluconate, deoxyuridine and linoleate levels (p = 0.043, p = 0.005 and p = 0.034 respectively). We also focused on the TCA cycle-related metabolites pyruvate, citrate/isocitrate, succinate, α-ketoglutarate, and malate, which we have previously validated using our HRM workflow(Liu, Nellis et al. 2020). None of these TCA cycle metabolites in plasma were linked to TPP levels. Selected metabolites that differentiate patients in the lowest TPP tertile from those in the highest TPP tertile are listed in Supplemental Table 5.

Table 2

C18/ESI- and HILIC/ESI + MWAS analysis in association with TPP concentrations
Metabolic Pathway	m/z	Time (sec)	Adduct	Putative Annotation	Metabolite p-value	Association with TPP levels	ID score
Pentose phosphate pathway	195.0511	19	M-H[-]	Gluconate	0.043	negative	1
Pentose phosphate pathway	115.0399	269	M-H2O-H[-]	Deoxyribose	0.030	positive	3
Linoleate metabolism	295.2267	23	M + H[+]	OxoODE	0.028	negative	3
Linoleate metabolism	261.222	214	M-H20-H[-]	Linoleate	0.034	negative	2
Bile acid biosynthesis	401.3423	28	M + H[+]	α-Hydroxy-cholestenone	0.028	positive	3
Pyrimidine metabolism	263.0455	19	M + Cl[-]	Deoxyuridine	0.005	negative	2
Raw p-values < 0.05
ESI = electrospray ionization
HILIC = hydrophilic interaction liquid chromatography
ID = identification score
MWAS = metabolome-wide association study
Octenoyl-CoA: Octenoyl-Coenzyme A
OxoODE: Oxooctadecadienoic acid

Plots in Fig. 4A and 4B show the relationship between TPP and two metabolites involved in the pentose phosphate pathway, gluconate (p = 0.043) and xylose/ribose (p = 0.047), which are both negatively associated with TPP levels. The association between gut microbiome metabolites hippurate and aminobutyrate and TPP are shown in Fig. 4C and 4D (p = 0.02 and p = 0.017, respectively). Both metabolites were positively associated with TPP (increased in the highest TPP tertile). Metabolites that were correlated with each other across multiple pathways can be seen in the network/module plot in Supplemental Fig. 1. The metabolites that were significantly linked as a function of whole blood TPP levels were largely carbohydrates, but the module also included ascorbic acid and glycosylated nucleosides (e.g., uridine, cytidine).

Thiamine plays an essential role in the energy metabolism in the human body(Frank 2015, Whitfield, Bourassa et al. 2018). It acts as a co-factor for pyruvate dehydrogenase in glycolysis, alpha-ketoglutarate dehydrogenase in the TCA cycle and transketolase in the pentose phosphate pathway, among its more important functions(Frank 2015, Whitfield, Bourassa et al. 2018). We found that thiamine status, as assessed by whole blood TPP concentrations were significantly related to two key metabolites involved in function of the pentose phosphate pathway, namely, the hydrogen adducts (M-H) of gluconate and xylose/ribose, which were both increased in individuals with lower TPP concentrations. The pentose phosphate pathway is critical for a variety of key biochemical functions, including maintenance of carbon homoeostasis, generation of precursors for nucleotide and amino acid biosynthesis, and protection from oxidative stress via production of NADPH (Krüger, Grüning et al. 2011, Nalos, Parnell et al. 2016, Sigurdsson, Kobayashi et al. 2022) from the reduced form of nicotinamide adenine dinucleotide phosphate (NADP+). Thus, our data suggests the possibility that low TPP concentrations may contribute to changes in transketolase activity within the pentose phosphate pathway that, in turn, may disrupt energy generation via the tightly linked pathways of glycolysis and the TCA cycle (Frank 2015, Whitfield, Bourassa et al. 2018).

Thiamine deficiency has been previously found to disrupt energy metabolism in rats by affecting glucose transport and fatty acid β-oxidation in mitochondria (Gralak, Dębski et al. 2019). In addition, the peroxisomal α-oxidation of 3-methyl fatty acids has been shown to be dependent on TPP (Casteels, Foulon et al. 2003). However, to our knowledge, no data to date has associated thiamine status with fatty acid oxidation or lipid metabolism in humans. In the current study, pathway analysis and targeted assessment of specific metabolic features (see Table 2 and Supplemental Table 5, Figs. 3 and 4) identified fatty acid β-oxidation, linoleate metabolism, and squalene and cholesterol biosynthesis as lipid-related metabolic pathways linked to thiamine status. These unexpected hypothesis-generating data suggest the possibility that thiamine nutriture may broadly influence lipid metabolism in critically ill adults and should be further explored in larger, prospective cohorts.

Thiamine also plays an important role in amino acid metabolism. TPP is a critical coenzyme for branched-chain α-ketoacid dehydrogenase (BCKDH), which is essential for the catabolism of branched-chain amino acids (BCAA) and subsequent utilization in the TCA cycle, among other functions (Duran and Wadman 1985, Depeint, Bruce et al. 2006). Our study revealed significant associations of TPP levels with numerous amino acid metabolic pathways, including BCAA metabolism, arginine and proline metabolism, aspartate and asparagine metabolism methionine and cysteine metabolism, and the urea cycle (Fig. 3). As shown in Supplemental Table 5, the TPP concentrations were negatively associated with asparagine, lysine and valine.

Our data expand the metabolomic results of several studies in critically ill patients with or without sepsis where a variety of alterations in amino acids and amino acid pathways have been identified, but in whom thiamine status was not determined (Rogers, McGeachie et al. 2014, Chen, Liang et al. 2022, Ohlstrom, Sul et al. 2022). Given the critical role of amino acid-derived metabolites in the TCA cycle (e.g., via α-ketoglutarate, oxaloacetate, succinate, etc.), it is possible that the broad impact of TPP status on amino acid metabolism we observed may impact energy generation indirectly via the TCA cycle. Further translational studies are needed to confirm such an effect, particularly since we did not observe any direct impact of TPP status on the TCA cycle metabolites pyruvate, citrate/isocitrate, succinate, α-ketoglutarate or malate. It is also possible, though speculative, that depletion of TPP has adverse effects on skeletal muscle or other tissues which utilize amino acids or are highly involved in amino acid metabolism, a hypothesis that requires further study.

It is known that the gut microbiota can generate TPP in mice, although the nutritional significance of this is unclear (Sabui, Romero et al. 2021). Studies have shown that the gut microbiome composition and diversity is disrupted in human critical illness (Lamarche, Johnstone et al. 2018, Haak, Argelaguet et al. 2021). Therefore, it is possible that these disruptions may contribute to circulating TPP concentrations. Another study found that hypoxia in human colonic epithelial cells inhibited colonic uptake of gut microbiota generated TPP (Sabui, Ramamoorthy et al. 2022). We did not study the gut microbiome directly in our participants, but we observed that two major gut microbiome-derived metabolites, hippurate and aminobutyrate, were each decreased in patients in the lowest TPP tertile, and the butanoate (butyrate) metabolic pathway was significantly affected by TPP status (Figs. 3 and 4C-D). Thus, it is possible that thiamine status may be related to gut microbiome dysregulation by currently unknown mechanisms. Future studies should further explore the possible link between TPP status and gut microbiome in humans.

Limitations of this pilot study are the relatively small sample size and the cross-sectional study design, which precludes cause and effect relationships between thiamine status and the metabolic associations observed. Our study population was derived from a single Turkish medical center and did not include non-critically ill control participants or a control group of critically ill patients not receiving furosemide treatment. However, our pilot data will inform future prospective studies on the specific impact of furosemide treatment in ICU patients. We believe that our data, as presented, are valid to define the relationship of thiamine status on the plasma metabolome in our cohort. Future larger studies would ideally obtain fasted blood samples, but this is difficult in many ICU patients who are receiving continuous nutrition support.

Another limitation is that the sample size did not allow comparison between different primary reasons for ICU admission or nutritional status of participants. However, to our knowledge, this is the first study in human critical illness to link thiamine nutritional status with systemic metabolism using metabolomics analysis. Our rigorous plasma HRM methods are state-of-the art and all samples are analyzed in triplicate with internal standards and are well validated in the literature(Jones 2016). Data were also adjusted for illness severity (APACHE II score) on the day plasma for HRM was obtained. We found that whole blood TPP levels are linked to metabolic pathways and metabolites (e.g., lipid metabolism, gut microbiome) not generally considered in thiamine metabolism and therefore are hypothesis-generating for subsequent confirmatory and translational mechanistic studies.

Using a hybrid approach of targeted and untargeted metabolomics, we found changes in numerous metabolites and metabolic pathways associated with energy metabolism, amino acid metabolism, lipid metabolism and gut microbiome metabolism as a function of whole blood TPP concentrations in critically ill adults. Plasma HRM as utilized in this study provides new understanding of how the status of a single micronutrient, TPP, may impact whole body metabolism. These results will hopefully stimulate larger prospective studies to determine the impact of thiamine status and thiamine repletion on metabolic homeostasis in individuals with critical illness.

Conflict of interest: None

Acknowledgements

Funding: Erciyes University Scientific Research Committee (TSG-2021-10078) and TUBITAK 2219 program (1059B-192000150), each to KG, and National Institutes of Health grant P30 ES019776, to DPJ and TRZ.

Alvarez, J. A., E. Y. Chong, D. I. Walker, J. D. Chandler, E. S. Michalski, R. E. Grossmann, K. Uppal, S. Li, J. K. Frediani, R. Tirouvanziam, V. T. Tran, V. Tangpricha, D. P. Jones and T. R. Ziegler (2017). "Plasma metabolomics in adults with cystic fibrosis during a pulmonary exacerbation: A pilot randomized study of high-dose vitamin D(3) administration." Metabolism 70: 31-41.
Amrein, K., J. A. Lasky-Su, H. Dobnig and K. B. Christopher (2021). "Metabolomic basis for response to high dose vitamin D in critical illness." Clin Nutr 40(4): 2053-2060.
Attaluri, P., A. Castillo, H. Edriss and K. Nugent (2018). "Thiamine Deficiency: An Important Consideration in Critically Ill Patients." Am J Med Sci 356(4): 382-390.
Bellissimo, M. P., D. P. Jones, G. S. Martin, J. A. Alvarez and T. R. Ziegler (2021). "Plasma high-resolution metabolomic phenotyping of lean mass in a United States adult cohort." JPEN J Parenter Enteral Nutr 45(8): 1635-1644.
Berger, M. M., A. Shenkin, A. Schweinlin, K. Amrein, M. Augsburger, H. K. Biesalski, S. C. Bischoff, M. P. Casaer, K. Gundogan, H. L. Lepp, A. M. E. de Man, G. Muscogiuri, M. Pietka, L. Pironi, S. Rezzi and C. Cuerda (2022). "ESPEN micronutrient guideline." Clin Nutr 41(6): 1357-1424.
Casteels, M., V. Foulon, G. P. Mannaerts and P. P. Van Veldhoven (2003). "Alpha-oxidation of 3-methyl-substituted fatty acids and its thiamine dependence." Eur J Biochem 270(8): 1619-1627.
Chen, Q., X. Liang, T. Wu, J. Jiang, Y. Jiang, S. Zhang, Y. Ruan, H. Zhang, C. Zhang, P. Chen, Y. Lv, J. Xin, D. Shi, X. Chen, J. Li and Y. Xu (2022). "Integrative analysis of metabolomics and proteomics reveals amino acid metabolism disorder in sepsis." J Transl Med 20(1): 123.
Christopher, K. B. (2018). "Nutritional metabolomics in critical illness." Curr Opin Clin Nutr Metab Care 21(2): 121-125.
Collins, J. M., D. P. Jones, A. Sharma, M. Khadka, K. H. Liu, R. R. Kempker, B. Prideaux, K. Maner-Smith, N. Tukvadze, N. S. Shah, J. C. M. Brust, R. P. Sékaly, N. R. Gandhi, H. M. Blumberg, E. A. Ortlund and T. R. Ziegler (2021). "TCA cycle remodeling drives proinflammatory signaling in humans with pulmonary tuberculosis." PLoS Pathog 17(9): e1009941.
Cyr, A., Y. Zhong, S. E. Reis, R. A. Namas, A. Amoscato, B. Zuckerbraun, J. Sperry, R. Zamora, Y. Vodovotz and T. R. Billiar (2021). "Analysis of the Plasma Metabolome after Trauma, Novel Circulating Sphingolipid Signatures, and In-Hospital Outcomes." J Am Coll Surg 232(3): 276-287.e271.
Depeint, F., W. R. Bruce, N. Shangari, R. Mehta and P. J. O'Brien (2006). "Mitochondrial function and toxicity: role of the B vitamin family on mitochondrial energy metabolism." Chem Biol Interact 163(1-2): 94-112.
Duran, M. and S. K. Wadman (1985). "Thiamine-responsive inborn errors of metabolism." J Inherit Metab Dis 8 Suppl 1: 70-75.
Evliyaoglu, O., J. van Helden, M. Imöhl and R. Weiskirchen (2019). "Mining the Age-Dependent Reference Intervals of B Vitamins from Routine Laboratory Test Results." Lab Med 50(1): 54-63.
Ferrario, M., A. Cambiaghi, L. Brunelli, S. Giordano, P. Caironi, L. Guatteri, F. Raimondi, L. Gattinoni, R. Latini, S. Masson, G. Ristagno and R. Pastorelli (2016). "Mortality prediction in patients with severe septic shock: a pilot study using a target metabolomics approach." Sci Rep 6: 20391.
Flower, L., A. Page and Z. Puthucheary (2021). "Should nutrition therapy be modified to account for mitochondrial dysfunction in critical illness?" JPEN J Parenter Enteral Nutr 45(S2): 60-65.
Frank, L. L. (2015). "Thiamin in Clinical Practice." JPEN J Parenter Enteral Nutr 39(5): 503-520.
Go, Y. M., D. I. Walker, Y. Liang, K. Uppal, Q. A. Soltow, V. Tran, F. Strobel, A. A. Quyyumi, T. R. Ziegler, K. D. Pennell, G. W. Miller and D. P. Jones (2015). "Reference Standardization for Mass Spectrometry and High-resolution Metabolomics Applications to Exposome Research." Toxicol Sci 148(2): 531-543.
Gonzalez-Granda, A., B. Seethaler, M. Haap, R. Riessen and S. C. Bischoff (2021). "Effect of an intensified individual nutrition therapy on serum metabolites in critically ill patients - A targeted metabolomics analysis of the ONCA study." Clin Nutr ESPEN 43: 267-275.
Gourd, N. M. and N. Nikitas (2020). "Multiple Organ Dysfunction Syndrome." J Intensive Care Med 35(12): 1564-1575.
Gralak, M. A., B. Dębski and M. Drywień (2019). "Thiamine deficiency affects glucose transport and β-oxidation in rats." J Anim Physiol Anim Nutr (Berl) 103(5): 1629-1635.
Gundogan, K., I. H. Akbudak, K. Bulut, S. Temel, M. Sungur, M. Guven, N. J. Dave, D. P. Griffith and T. R. Ziegler (2019). "Thiamin Status in Adults Receiving Chronic Diuretic Therapy Prior to Admission to a Medical Intensive Care Unit: A Pilot Study." Nutr Clin Pract 34(4): 565-571.
Haak, B. W., R. Argelaguet, C. M. Kinsella, R. F. J. Kullberg, J. M. Lankelma, M. Deijs, M. Klein, M. F. Jebbink, F. Hugenholtz, S. Kostidis, M. Giera, T. B. M. Hakvoort, W. J. de Jonge, M. J. Schultz, T. van Gool, T. van der Poll, W. M. de Vos, L. M. van der Hoek and W. J. Wiersinga (2021). "Integrative Transkingdom Analysis of the Gut Microbiome in Antibiotic Perturbation and Critical Illness." mSystems 6(2).
Jones, D. P. (2016). "Sequencing the exposome: A call to action." Toxicol Rep 3: 29-45.
Jones, D. P., Y. Park and T. R. Ziegler (2012). "Nutritional metabolomics: progress in addressing complexity in diet and health." Annu Rev Nutr 32: 183-202.
Krüger, A., N. M. Grüning, M. M. Wamelink, M. Kerick, A. Kirpy, D. Parkhomchuk, K. Bluemlein, M. R. Schweiger, A. Soldatov, H. Lehrach, C. Jakobs and M. Ralser (2011). "The pentose phosphate pathway is a metabolic redox sensor and regulates transcription during the antioxidant response." Antioxid Redox Signal 15(2): 311-324.
Lamarche, D., J. Johnstone, N. Zytaruk, F. Clarke, L. Hand, D. Loukov, J. C. Szamosi, L. Rossi, L. P. Schenck, C. P. Verschoor, E. McDonald, M. O. Meade, J. C. Marshall, D. M. E. Bowdish, T. Karachi, D. Heels-Ansdell, D. J. Cook and M. G. Surette (2018). "Microbial dysbiosis and mortality during mechanical ventilation: a prospective observational study." Respir Res 19(1): 245.
Lasky-Su, J., A. Dahlin, A. A. Litonjua, A. J. Rogers, M. J. McGeachie, R. M. Baron, L. Gazourian, D. Barragan-Bradford, L. E. Fredenburgh, A. M. K. Choi, K. M. Mogensen, S. A. Quraishi, K. Amrein and K. B. Christopher (2017). "Metabolome alterations in severe critical illness and vitamin D status." Crit Care 21(1): 193.
Li, S., Y. Park, S. Duraisingham, F. H. Strobel, N. Khan, Q. A. Soltow, D. P. Jones and B. Pulendran (2013). "Predicting network activity from high throughput metabolomics." PLoS Comput Biol 9(7): e1003123.
Liu, K. H., M. Nellis, K. Uppal, C. Ma, V. Tran, Y. Liang, D. I. Walker and D. P. Jones (2020). "Reference Standardization for Quantification and Harmonization of Large-Scale Metabolomics." Anal Chem 92(13): 8836-8844.
Liu, K. H., D. I. Walker, K. Uppal, V. Tran, P. Rohrbeck, T. M. Mallon and D. P. Jones (2016). "High-Resolution Metabolomics Assessment of Military Personnel: Evaluating Analytical Strategies for Chemical Detection." J Occup Environ Med 58(8 Suppl 1): S53-61.
McClave, S. A., P. E. Wischmeyer, K. R. Miller and A. R. H. van Zanten (2019). "Mitochondrial Dysfunction in Critical Illness: Implications for Nutritional Therapy." Curr Nutr Rep 8(4): 363-373.
Nalos, M., G. Parnell, R. Robergs, D. Booth, A. S. McLean and B. M. Tang (2016). "Transcriptional reprogramming of metabolic pathways in critically ill patients." Intensive Care Med Exp 4(1): 21.
Ohlstrom, D. J., C. Sul, C. U. Vohwinkel, L. Hernandez-Lagunas, A. Karimpour-Fard, P. M. Mourani, T. C. Carpenter, E. S. Nozik and C. C. Sucharov (2022). "Plasma microRNA and metabolic changes associated with pediatric acute respiratory distress syndrome: a prospective cohort study." Sci Rep 12(1): 14560.
Polegato, B. F., A. G. Pereira, P. S. Azevedo, N. A. Costa, L. A. M. Zornoff, S. A. R. Paiva and M. F. Minicucci (2019). "Role of Thiamin in Health and Disease." Nutr Clin Pract 34(4): 558-564.
Puthucheary, Z. A., J. Rawal, M. McPhail, B. Connolly, G. Ratnayake, P. Chan, N. S. Hopkinson, R. Phadke, T. Dew, P. S. Sidhu, C. Velloso, J. Seymour, C. C. Agley, A. Selby, M. Limb, L. M. Edwards, K. Smith, A. Rowlerson, M. J. Rennie, J. Moxham, S. D. Harridge, N. Hart and H. E. Montgomery (2013). "Acute skeletal muscle wasting in critical illness." Jama 310(15): 1591-1600.
Rogers, A. J., M. McGeachie, R. M. Baron, L. Gazourian, J. A. Haspel, K. Nakahira, L. E. Fredenburgh, G. M. Hunninghake, B. A. Raby, M. A. Matthay, R. M. Otero, V. G. Fowler, E. P. Rivers, C. W. Woods, S. Kingsmore, R. J. Langley and A. M. Choi (2014). "Metabolomic derangements are associated with mortality in critically ill adult patients." PLoS One 9(1): e87538.
Rousseau, A. F., H. C. Prescott, S. J. Brett, B. Weiss, E. Azoulay, J. Creteur, N. Latronico, C. L. Hough, S. Weber-Carstens, J. L. Vincent and J. C. Preiser (2021). "Long-term outcomes after critical illness: recent insights." Crit Care 25(1): 108.
Sabui, S., K. Ramamoorthy, J. M. Romero, R. D. Simoes, J. M. Fleckenstein and H. M. Said (2022). "Hypoxia inhibits colonic uptake of the microbiota-generated forms of vitamin B1 via HIF-1α-mediated transcriptional regulation of their transporters." J Biol Chem 298(2): 101562.
Sabui, S., J. M. Romero and H. M. Said (2021). "Developmental maturation of the colonic uptake process of the microbiota-generated thiamin pyrophosphate." Am J Physiol Gastrointest Liver Physiol 320(5): G829-g835.
Schymanski, E. L., J. Jeon, R. Gulde, K. Fenner, M. Ruff, H. P. Singer and J. Hollender (2014). "Identifying small molecules via high resolution mass spectrometry: communicating confidence." Environ Sci Technol 48(4): 2097-2098.
Siddiqui, M. A., S. Pandey, A. Azim, N. Sinha and M. H. Siddiqui (2020). "Metabolomics: An emerging potential approach to decipher critical illnesses." Biophys Chem 267: 106462.
Sigurdsson, M. I., H. Kobayashi, K. Amrein, K. Nakahira, A. J. Rogers, M. Pinilla-Vera, R. M. Baron, L. E. Fredenburgh, J. A. Lasky-Su and K. B. Christopher (2022). "Circulating N-formylmethionine and metabolic shift in critical illness: a multicohort metabolomics study." Crit Care 26(1): 321.
Soltow, Q. A., F. H. Strobel, K. G. Mansfield, L. Wachtman, Y. Park and D. P. Jones (2013). "High-performance metabolic profiling with dual chromatography-Fourier-transform mass spectrometry (DC-FTMS) for study of the exposome." Metabolomics 9(1 Suppl): S132-s143.
Stolarski, A. E., L. Young, J. Weinberg, J. Kim, E. Lusczek, D. G. Remick, B. Bistrian and P. Burke (2022). "Early metabolic support for critically ill trauma patients: A prospective randomized controlled trial." J Trauma Acute Care Surg 92(2): 255-265.
Uppal, K., Q. A. Soltow, F. H. Strobel, W. S. Pittard, K. M. Gernert, T. Yu and D. P. Jones (2013). "xMSanalyzer: automated pipeline for improved feature detection and downstream analysis of large-scale, non-targeted metabolomics data." BMC Bioinformatics 14: 15.
Uppal, K., D. I. Walker and D. P. Jones (2017). "xMSannotator: An R Package for Network-Based Annotation of High-Resolution Metabolomics Data." Anal Chem 89(2): 1063-1067.
Uppal, K., D. I. Walker, K. Liu, S. Li, Y. M. Go and D. P. Jones (2016). "Computational Metabolomics: A Framework for the Million Metabolome." Chem Res Toxicol 29(12): 1956-1975.
Viana, M. V., F. Becce, O. Pantet, S. Schmidt, G. Bagnoud, J. J. Thaden, G. A. M. Ten Have, M. Engelen, A. Voidey, N. E. P. Deutz and M. M. Berger (2021). "Impact of β-hydroxy-β-methylbutyrate (HMB) on muscle loss and protein metabolism in critically ill patients: A RCT." Clin Nutr 40(8): 4878-4887.
Wendt, C. H., S. Castro-Pearson, J. Proper, S. Pett, T. J. Griffin, V. Kan, J. Carbone, N. Koulouris, C. Reilly and J. D. Neaton (2021). "Metabolite profiles associated with disease progression in influenza infection." PLoS One 16(4): e0247493.
Whitfield, K. C., M. W. Bourassa, B. Adamolekun, G. Bergeron, L. Bettendorff, K. H. Brown, L. Cox, A. Fattal-Valevski, P. R. Fischer, E. L. Frank, L. Hiffler, L. M. Hlaing, M. E. Jefferds, H. Kapner, S. Kounnavong, M. P. S. Mousavi, D. E. Roth, M. N. Tsaloglou, F. Wieringa and G. F. Combs, Jr. (2018). "Thiamine deficiency disorders: diagnosis, prevalence, and a roadmap for global control programs." Ann N Y Acad Sci 1430(1): 3-43.
Yu, T., Y. Park, J. M. Johnson and D. P. Jones (2009). "apLCMS--adaptive processing of high-resolution LC/MS data." Bioinformatics 25(15): 1930-1936.

No competing interests reported.

SupplementalDigitalContent.pdf

High-Resolution Plasma Metabolomics and Thiamine Status in Critically Ill Adult Patients

Status:

Journal Publication

Version 1

Abstract

Background and Aim:

Methods

Results

Conclusions

Figures

Introduction

Materials and Methods

Study Participants

Demographic and Clinical Data

At ICU admission

Sample Preparation and Analysis

Whole blood thiamine concentrations

Plasma for metabolomics analysis

High-resolution metabolomics

Statistical Analysis

Results

HRM Analysis

Pathway HRM Analysis

Discussion

Conclusions

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1