Authors: Hassan S. Dashti (1Department of Anesthesia, Critical Care and Pain Medicine, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA; 2Division of Sleep Medicine, Harvard Medical School, Boston, MA, USA; 3Division of Nutrition, Harvard Medical School, Boston, MA, USA; 4Programs in Metabolism and Medical & Population Genetics, The Broad Institute of M.I.T and Harvard, Cambridge, MA, US), Magdalena Sevilla-Gonzalez (4Programs in Metabolism and Medical & Population Genetics, The Broad Institute of M.I.T and Harvard, Cambridge, MA, US; 5Clinical and Translational Epidemiology Unit, Mongan Institute, Massachusetts General Hospital, Boston, MA, USA; 6Department of Medicine, Harvard Medical School, Boston, MA, US), Kris M. Mogensen (7Department of Nutrition, Brigham and Women’s Hospital, Boston, MA USA), Marion F. Winkler (8Department of Surgery, Rhode Island Hospital, Alpert Medical School of Brown University, Providence, RI, USA), Charlene Compher (9Biobehavioral Health Sciences Department, University of Pennsylvania School of Nursing, Philadelphia, PA, USA)
Categories: Article, Home parenteral nutrition, metabolomics, chronobiology, circadian rhythms, nutrition
Source: Clinical nutrition ESPEN
Authors: Hassan S. Dashti, Magdalena Sevilla-Gonzalez, Kris M. Mogensen, Marion F. Winkler, Charlene Compher
Home parenteral nutrition (HPN) is often cycled nocturnally and is expected to result in glucose intolerance and sleep disruption partly due to circadian misalignment. This study aimed to define the metabolic response when HPN is cycled during the daytime compared to overnight.
This analysis leveraged samples from a clinical trial in adults with short bowel syndrome consuming HPN (ClinicalTrials.gov: NCT04743960). Enrolled patients received 1 week of HPN overnight followed by 1 week of HPN during the daytime. Fasting blood samples were collected following each study period and global metabolic profiles were determined from plasma. Differential metabolite abundance was determined from normalized and scaled data using adjusted Linear Models for MicroArray Data models followed by pathway enrichment analysis.
Nine patients (mean age, 52.6years; 78%female; mean BMI 20.7kg/m^2^) provided samples. Among 622 identified metabolites, changes were observed in 36 metabolites at Punadj<0.05 with higher abundance of fatty acids, long-chain and polyunsaturated fatty acids (Dihomo-gamma-linolenic acid, arachidonate (20:4n6), docosahexaenoate (DHA;22:6n3)) and glycerolipids with daytime infusions. Enrichment analysis identified pathways related to the biosynthesis of unsaturated fatty acids, D-arginine, and D-ornithine metabolism, and linoleic acid metabolism (Punadj<0.05).
Daytime infusions of HPN may result in changes in circulating lipids and amino acid composing metabolic pathways previously implicated in circadian rhythms. As this is the first untargeted metabolomics study of HPN, larger studies are needed.
Parenteral nutrition is the intravenous administration of nutrition for individuals with compromised enteral access and/or absorption where oral dietary intake alone may be inadequate[1,2]. Advancements in catheter technology and the establishment of home parenteral nutrition (HPN) clinical programs in the early 1970s allowed patients to continue receiving therapy at home. HPN is currently prescribed to over 25,000 patients in the US experiencing acute gastrointestinal dysfunction and for individuals living with chronic intestinal failure. Patients with chronic intestinal failure are defined as those that are metabolically stable but live with a chronic condition requiring intravenous supplementation over months or years such as short bowel syndrome (SBS), intestinal dysmotility, and inflammatory bowel disease[3–5]. Infusions typically last 10–16 hours to limit hepatic injury and metabolic derangement such as dysglycemia, and there currently is no explicit guidance on when infusions should be administered. HPN is typically cycled nocturnally coinciding with sleep and the biological evening to limit interference with mobility and daily activities[6].
Circadian misalignment occurs when there is a mismatch between the central circadian system and 24-hour behaviors, such as eating during the nighttime[7]. Animal studies and controlled human experiments indicate that circadian misalignment has systemic health consequences including adverse cardiometabolic effects, inflammation, and impaired mood[8,9]. Countermeasures, such as appropriate meal timing and timed exposure to light and darkness, can limit circadian misalignment[10,11]. For example, in night shift workers, restricting food intake to the daytime limits circadian misalignment and improves glucose tolerance[11]. Metabolic profiling techniques (i.e., the systematic profiling of cellular metabolites including sugars, amino acids, organic acids, and lipids) have been employed to identify biological pathways related to circadian misalignment, and circadian rhythms in general. For example, the rhythmic profile of 75% of metabolites, mainly amino acids, were influenced by changes in behavioral cycles as a result of night shift work, providing insight into the response of the metabolome to circadian misalignment[12].
For HPN consumers, nocturnal infusions are known to result in glucose intolerance and sleep disruption and are also expected to contribute to chronic circadian misalignment as infusions likely coincide with the biological evening[13,14]. It is unclear what metabolic changes would result from modifying HPN timing from nighttime to daytime. The objective of the present exploratory study was to provide insight into metabolites that differ when HPN is cycled during the daytime compared to overnight in adult patients living with SBS. Findings from this translational research may provide preliminary data into biological pathways that are influenced when the timing of nutritional intake is manipulated in a high-risk clinical population and provide insight into clinically relevant biomarkers that are altered if patients choose to cycle during the day.
The present exploratory analysis leveraged biosamples from a completed quasi-experimental, single-arm, controlled, 2-week pilot and feasibility trial. The overall objective of that trial was to determine the feasibility, safety, and efficacy of daytime compared to overnight infusions of HPN in adults with SBS receiving habitual HPN (ClinicalTrials.gov: NCT04743960)[14]. Patients enrolled in the trial cycled HPN infusions for 1 week overnight (days 1–7) followed by daytime for another week (days 8–14). Duration (i.e., hours per day), frequency (i.e., days per week), and composition of infusions remained identical during the two study periods.
As previously described[14], eligible patients were adults (18—79 years) living with SBS, without any diabetes, and receiving overnight infusions of HPN for at least the past 6 months. Patients were ineligible 1) with known diabetes diagnosis or taking or intending to take any diabetes medication or medications influencing blood glucose; 2) with known diagnosis for circadian and sleep disorders or using sleep medication other than melatonin and unable or unwilling to stop during the study period; 3) with blindness, deafness, or inability to speak English; 4) pregnant or lactating; 5) have skin conditions; 6) within the last year had bariatric surgery or pregnancy; 7) within the last month had acute infections or illnesses requiring medical attention, hospitalizations, emergency department visits, blood transfusions, blood loss, or blood donations; 8) within the past 3 months had major changes in diet or physical activity level; and 9) unable or unwilling to give consent or comply with study procedures. Written informed consent was obtained from all study participants prior to study procedures. The protocol was approved by the Mass General Brigham Institutional Review Board (protocol #2020P003741).
To facilitate enrollment in the trial during the COVID-19 pandemic, 50% of participants enrolled in an in-person protocol (at Massachusetts General Hospital, Boston, MA) whereas the remaining 50% of participants enrolled in a comparable home-based protocol. Only participants enrolled in the in-person protocol had a venous blood sample collected at clinical visits following each 1-week study period (days 8 and 15). As no blood draws were collected from participants enrolled in the home-based protocol, they were excluded from this analysis. The blood draw was scheduled after at-least an 8-hour fast and at-least 8 hours from the end of an HPN infusion cycle. Upon drawing blood, samples were processed, aliquoted, and stored at −80°C.
Untargeted global metabolomics profiling on plasma samples was conducted using Ultrahigh Performance Liquid Chromatography-Tandem Mass Spectroscopy at Metabolon (Metabolon IncMorrisville, NC, United States), including multiple quality control steps and comparison to Metabolon’s well-validated metabolomics compound library[15]. A total of 1,015 metabolites were identified and quantified, of which 622 metabolites had non-missing data across all samples and were included in the present analysis. Metabolite values were normalized to the volume of sample extracted and then log-transformed and scaled with Z-score prior.
Differential metabolite abundance (outcome measure) between the two study periods (daytime vs. overnight) was determined using standard Linear Models for MicroArray Data (LIMMA) models adjusted for dietary fasting duration and time since the end of the last HPN infusion to account for any differences in time lapsed since last oral intake or intravenous nutrition[16]. Metabolites were considered at an unadjusted Punadj value and at a more stringent FDR threshold. For metabolites with Punadj value <0.05, pathway enrichment analysis was then conducted using MetaboAnalyst’s pathway enrichment tool[17]. Similarly, pathways were considered at an unadjusted Punadj value and at a more stringent FDR threshold.
A total of 20 patients completed the trial, of which 10 patients completed the in-person protocol and provided two fasting blood samples at clinical visits. Insufficient blood was drawn from one patient, and thus their data were excluded from this blood biomarker analysis. Nine patients (mean age, 52.6 years; 78% female; mean BMI 20.7 kg/m^2^) were included the present analysis (Table 1). Patients had a median dependency on HPN for 4 years, 56% of patients continue to eat by mouth, and 33% receive infusions daily. Blood draws following each 1-week period were conducted at the same time-of-day (median [quartile 1, quartile 3]; 45 [10:47, 45]; 20 [11:09, 45]; P =1.000). There was no difference in dietary fasting duration (overnight: 13.7 [12.9, 14.0] hours; 13.3 [10.9, 14.0] hours; P =0.938) and the time since the end of the most recent HPN cycle (overnight: 25.3 [8.0, 27.8] hours; 14.9 [14.5, 17.8] hours; P =0.734), however, it is worth noting that a shorter time has lapsed since the end of the HPN cycles in the second study period.
Among the 622 investigated plasma metabolites, 36 metabolites differed between the two weeks at Punadj <0.05 (Figure 1A; Supplementary Table 1). Most metabolites (n =30) increased in abundance following daytime compared to overnight infusions. Conversely, five metabolites—2-oxoarginine, tryptophan, 2R,3R-dihydroxybutyrate, threonine, and 1-stearoyl-2-docosahexaenoyl-GPC (18:0/5n)—showed a reduction in their abundance following daytime compared to overnight infusions. The top metabolites were primarily (69%) lipids, including long-chain and polyunsaturated fatty acids (e.g., Dihomo-gamma-linolenic acid, arachidonate (20:4n6), and docosahexaenoate (DHA; 6n3)), fatty acids, and diacylglycerol (Figure 1B). Additional metabolites included amino acids (e.g., threonine and tryptophan) and cofactors and vitamins (e.g., alpha-tocopherol and bilirubin). No individual metabolites were significant at the stringent FDR threshold.
Pathway enrichment analysis of the top 36 metabolites identified 3 pathways associated with the timing of HPN infusions at Punadj <0.05: biosynthesis of unsaturated fatty acids, D-arginine, and D-ornithine metabolism, and linoleic acid metabolism (Figure 1C). Of these pathways, only biosynthesis of unsaturated fatty acids, which included 7 identified metabolites, was significant at the stringent FDR threshold.
The role of food timing, or chrononutrition, has garnered promising findings and public interest, yet remains understudied in high-risk patient populations including those receiving HPN. We employed an unbiased molecular phenotyping approach to understand the complex relationship between the timing of HPN and changes in the metabolic response. Our analysis suggested 36 potential plasma metabolites associated with the timing of HPN infusions, of which the majority were unsaturated fatty acids and glycerolipids showing higher abundance during the daytime compared to the nighttime. The identified metabolites compose lipids involved in the biosynthesis of unsaturated fatty acids pathway. These metabolites and pathways, particularly mono and polyunsaturated fatty acids, may mediate favorable changes in glucose control and sleep patterns and may provide insight into clinically relevant biomarkers that may differ in light of growing interest from patients to transition from overnight to daytime infusions of HPN to achieve better sleep[6,18,19].
The identified circulatory biomarkers and related biological pathways have been previously implicated in circadian rhythms and circadian misalignment in animal studies and controlled human experiments. All three identified pathways (biosynthesis of unsaturated fatty acids, D-arginine, and D-ornithine metabolism, and linoleic acid metabolism) show impaired 24-hour oscillations in BMAL1 knockout murine muscle[20]. Furthermore, a mouse study found that ablation of the circadian clock in adipocytes is associated with a reduction in circulating polyunsaturated fatty acids[21]. In humans, simulated shifts in work schedules result in changes in metabolites implicated in food metabolism, including D-arginine and D-ornithine metabolism[22]. Furthermore, L-ornithine supplementation may influence the circadian clock[23]. We also identified a reduction in 3 amino acids with daytime HPN infusions, including tryptophan and threonine. Dietary tryptophan and threonine have been identified to control circadian behavioral rhythms and their depletion is found to alter the amplitude and activity period[24]. Among our findings is also plasma bilirubin, which is controlled by the circadian clock[25]. As circadian clocks influence a broad range of metabolic processes, including glucose uptake, insulin signaling, lipogenesis and lipolysis[26], further verification of the role of the circadian clock is necessary.
While differential expression of the described metabolites may be suggestive of changes in circadian misalignment with changes in the timing of HPN, our trial is not designed to measure misalignment in this population. It also remains unclear whether – and to what extent – nutrition provided intravenously influences circadian rhythms. Studies have focused on oral food intake when showing that feeding times aligned with the circadian clock appropriately entrains endogenous rhythms and prevents clock reprogramming[27]. It is likely that intravenous nutrition confers entrainment effects through insulin signaling; insulin acts as a systemic signal entraining circadian clocks across multiple organ systems[28]. Our study is part of an ongoing global effort aimed at reevaluating the historical practice of nighttime administration of nutrition support in hospital and home-based parenteral and enteral (i.e., tube feeds) nutrition. The first of these published trials demonstrated the safety and feasibility of daytime infusions of parenteral and/or enteral nutrition with an overnight fast in hospitalized children[29,30]. Integrating multiomic approaches (e.g., proteomics, transcriptomics, epigenomics, and metabolomics) in the analysis of these trials can help elucidate biological mechanisms involved in circadian-based interventions in high-risk patients receiving nutrition support.
This is the first metabolomics study examining HPN. Thus, it is not possible to compare our study findings with previous efforts to discern whether changes in metabolites identified in this analysis reflect time-of-day or are merely biomarkers of HPN content. In general, few small studies have examined metabolites associated with parenteral nutrition. A study of pediatric SBS patients found that amino acids, particularly glutamine, is associated with parenteral nutrition[31] and a study of surgical adults identified increases in long-chain fatty acids, phospholipids, ketone bodies, and branched-chain amino acids metabolites with early parenteral nutrition[32]. None of these metabolites were identified in our pilot study suggesting that our findings are unlikely specific to HPN composition. In our study, whereas the blood draws occurred at the same time-of-day, it was not possible to achieve identical “fasting durations” for both draws. Our study design inadvertently resulted in shorter elapsed time since the end of the HPN cycles for the second draw compared to the first. For enrolled patients who continue to eat by mouth, fasting duration from oral intake was similar between the two blood draws. It is also possible that the observed metabolites indicate aspects of oral intake such as fat or protein often encouraged in this population or changes in the timing of medication use[33]. Nonetheless, these metabolites may be informative in identifying clinically actionable biomarkers that change if a patient chooses to cycle infusions during the day instead of the night to achieve better sleep.
Important limitations should be acknowledged. Our study design allowed for within-patient comparison allowing for the control of interindividual differences across patients such as differences in HPN dependency and composition and medication use, however residual confounding for example by oral diet or changes in medication timing may still exist. Some pertinent medical information, including etiology of short bowel syndrome, remaining bowel length, and the presence or absence of a colon, were missing from the study. We had limited statistical power because of our modest sample size, and no associations remained significant at the stringent FDR threshold. For that reason, all findings should be considered preliminary and hypothesis generating. Because of our small sample size, it was not possible to conduct subgroup analyses, for example, analyses restricted to patients on exclusive HPN or with oral intake receiving supplemental HPN, as well as analyses restricted to patients with or without lipids included in the HPN. The role of oral intake should be considered in future studies because of the significant role of diet on reprogramming the circadian clock in peripheral tissues[34]. Further research in larger cohorts is necessary to verify whether these signals pertain to circadian rhythms or reflect acute changes in sleep and inflammation.