https://orcid.org/0000-0002-0464-5629
Abstract
To characterize longitudinal changes in blood biomarkers, leukocyte composition, and gene expression following laparoscopic sleeve gastrectomy (LSG).
LSG is an effective treatment for obesity, leading to sustainable weight loss and improvements in obesity-related comorbidities and inflammatory profiles. However, the effects of LSG on immune function and metabolism remain uncertain.
Prospective data were collected from 23 enrolled human subjects from a single institution. Parameters of weight, comorbidities, and trends in blood biomarkers and leukocyte subsets were observed from preoperative baseline to 1 year postsurgery in 3-month follow-up intervals. RNA sequencing was performed on pairs of whole blood samples from the first 6 subjects of the study (baseline and 3 months postsurgery) to identify genome-wide gene expression changes associated with undergoing LSG.
LSG led to a significant decrease in mean total body weight loss (18.1%) at 3 months and among diabetic subjects a reduction in hemoglobin A1c. Improvements in clinical inflammatory and hormonal biomarkers were demonstrated as early as 3 months after LSG. A reduction in neutrophil-lymphocyte ratio was observed, driven by a reduction in absolute neutrophil counts. Gene set enrichment analyses of differential whole blood gene expression demonstrated that after 3 months LSG induced transcriptomic changes not only in inflammatory cytokine pathways but also in several key metabolic pathways related to energy metabolism.
LSG induces significant changes in the composition and metabolism of immune cells as early as 3 months postoperatively. Further evaluation is required of bariatric surgery’s effects on immunometabolism and the consequences for host defense and metabolic disease.
Obesity is recognized as a chronic and systemic inflammatory disease (1). A positive correlation between obesity and cellular immune dysregulation has been widely documented over the past decade, characterized by an increase in circulating levels of cytokines and interleukins (2-5).
Since the original discovery by Hotamisligil et al (2) in 1993 demonstrating that cytokine levels are elevated in the adipose tissue of diabetic obese mice and its neutralization improves glucose handling, evidence is accumulating that immune pathways regulate metabolic homeostasis. This chronic inflammation in obesity, due to oxidative stress in adipocytes, facilitates the macrophage infiltration cascade and atypical cytokine production, leading to increased acute-phase proteins (4,5). The activation of immune signaling pathways upsets multiorgan homeostasis especially in liver, brain, and pancreas and is responsible for the subsequent metabolic dysfunctions (3,6).
With the incidence of obesity becoming pandemic (7), there is a growing interest to investigate the relationship between bariatric surgery and its beneficial effects on immunometabolism. It is now widely accepted that laparoscopic sleeve gastrectomy (LSG) is a safe and effective procedure (8,9). While it is rapidly increasing in popularity both as a metabolic and weight loss surgery of choice, LSG also improves systemic inflammatory profile (10). A prospective study composed of 22 participants with impaired glucose homeostasis undergoing LSG observed an improvement in leptin, C-reactive protein (CRP), and interleukin-6 (IL-6) 6 months following surgery (11). Moreover, an increase in clinical inflammatory markers has been found to be an independent risk factor for diabetes incidence with levels of white cell counts significantly ameliorated following LSG (11,12).
To explore mechanisms underlying the beneficial actions of bariatric surgery, global transcriptional profiling has been employed to identify genes whose expression is altered by surgery. After bariatric surgery, changes in inflammatory gene expression in adipose tissue tracked with sustained effects on postsurgical adipose morphology (13,14). Whole-blood gene expression has been investigated in 2 small bariatric cohorts, where metabolic pathways, particularly of lipids, in the blood were altered by surgery and correlated with weight loss and improvements in glucose metabolism (15,16). However, these studies examined exclusively or primarily Roux-en-Y gastric bypass patients and at single, later postoperative time points when weight loss had already occurred.
To date, there is a lack of evidence demonstrating the impact and timing of LSG on immune cell metabolism and function. The goal of this prospective study was to characterize longitudinal changes in immunometabolic blood biomarkers over 12 months in human subjects with obesity prior to and following LSG. We also examined the relationship of surgical weight loss with genome-wide gene expression changes in peripheral blood leukocytes. We hypothesized that LSG improves immune function by inducing changes in immune cell metabolism and composition.
Methods
Study Population
Subjects were identified and recruited prospectively from patients undergoing LSG between January 2017 and July 2019 at Brigham and Women’s Hospital (Boston, MA, USA). A multidisciplinary clinical team was involved and endorsed the indication for bariatric surgery. Decisions to undergo sleeve gastrectomy and study participation were achieved after full informed consents and written agreements between patients and clinicians in the surgical clinic. Entry criteria of the study included (1) both women and men (2), patients 18 to 70 years of age (3), with all patients (subjects) meeting the criteria for LSG following American Society for Metabolic and Bariatric Surgery guidelines (17), a body mass index (BMI) of >40 kg/m² or of >35 kg/m² with at least 1 comorbidity [eg, type 2 diabetes (T2D), hypertension, or dyslipidemia]. Patients with preexisting autoimmune diseases or on immunomodulatory treatments were excluded.
Approval for the study was obtained from the Partners Human Research Committee, the Institutional Review Board of Partners HealthCare (IRB protocols 2015P000500 and 2015P000787).
Study Design
All subjects were recruited preoperatively within 1 month of surgery and followed up postoperatively at 3, 6, 9, and 12 months (T3, T6, T9, and T12, respectively). Prior to surgery, all subjects were required to complete a baseline (T0) questionnaire on clinical information and changes of parameters were monitored at subsequent time points. Standard LSG comprised of a vertical stapled resection of the greater curvature of stomach over a 36- to 40-Fr bougie based on surgeon’s preference. Whole-blood samples were collected perioperatively (fasting) and at 3 (fasting), 6, 9, and 12 months (fasting) following surgery. Patient demographics and clinical characteristics were recorded in an electronic database (REDCap, version 8.10.20) (18) housed at Enterprise Research Information Science and Computing at Partners HealthCare.
Measurement of Clinical and Biochemical Data
At each time point, the following clinical information was ascertained by the means of administering a study questionnaire or consulting the electronic medical record; age, sex, race, past medical history, and BMI. All blood samples were collected on ice, centrifuged, and separated within 2 h of collection and subsequently processed. Biochemical measurements were analyzed within the local hospital-accredited laboratory (Laboratory Corporation of America Holdings, Morrisville, NC, USA) and Brigham Research Assay Core (Boston, MA, USA). A portion of the blood samples were stored in PAXgene Blood RNA Tubes (QIAGEN, Hilden, Germany) for later RNA sequencing (RNA-seq).
Statistical Analysis
Categorical data were presented as counts or percentages. Continuous variables with normal distribution were reported as means and bootstrapped nonparametric 95% CIs with 1000 resamples or SD. Post hoc pairwise comparisons of each follow-up time point vs baseline were performed using a paired Welch’s t-test. All tests were 2-sided with the level of significance set as P < 0.05 and were corrected for multiple comparisons using the Holm method (19). Statistical analyses were performed using GraphPad Prism (version 7.00) for Windows (GraphPad Software, La Jolla, CA, USA) or using R (version 3.6) (20) and Bioconductor (version 3.10) (21,22). Phenotype data were accessed programmatically using the R package “REDCapR” (version 0.9.8).
Gene Expression Profiling
An RNA-seq pilot was performed on pairs of whole blood samples from the first 6 subjects of the study (baseline and 3 months postoperatively). Sequencing was performed at Partners HealthCare Personalized Medicine (Boston, MA, USA). Three Illumina sequencing batches of 75 base-pair, paired-end RNA-seq data in the FASTQ file format were generated across multiple lanes. The lanes were merged to create single-sample FASTQ files, which were trimmed with the Skewer software (version 0.1.118) (23) and aligned to the GRCh38 genome reference assembly using the Spliced Transcripts Alignment to a Reference (STAR) software (version 2.5.2b) (24). The aligned BAM files were sorted and quantified using the Python library “HTSeq” (version 0.6.1p1) (25). The raw read counts were normalized using the R package “DESeq2” (version 1.26) (26).
Differential gene expression (DGE) modeling was performed using DESeq2. In total, 65 986 distinct gene transcripts were assayed based on the reference transcriptome, but only 36 593 remained after filtering out transcripts with mean normalized read counts of zero across the 12 samples. Given the small sample size for this pilot analysis (n = 6), the paired-sample DGE model included only study time point as a predictor. Correction for multiple testing was made using independent hypothesis weighting as implemented in the Bioconductor package “IHW” (version 1.14) (27, 28). DGE results were visualized using the R package “EnhancedVolcanoPlot” (version 1.4). Transcript counts were transformed by variance stabilization in DESeq2 and then bidirectional hierarchical clustering by Euclidean distance was performed on a subset of transcripts with significant DGE results using the R package “pheatmap” (version 1.0). Gene set enrichment analysis (GSEA) of the DGE results was performed using the Bioconductor package “fgsea” (version 1.12) (29). Hallmark gene sets from the Molecular Signatures Database (version 6.0) were tested as GSEA hypotheses (30). GSEA results were visualized using the Bioconductor package “clusterProfiler” (version 3.14) (31).
For a set of candidate genes, targeted quantitative polymerase chain reaction (qPCR) assays were performed to determine the messenger RNA expression over time following LSG (32). Six genes (SLCO2A1, PDPK2P, IGF2, IFNE, MMP15, ID1) were analyzed at baseline, 3, 6, and 12 months postsurgery using 12 patient blood samples. Briefly, RNA extraction and complimentary DNA synthesis were performed using Human PAXgene blood RNA extraction kit (Qiagen, Germany) and Superscript IV Vilo master mix (Life Technologies, Carlsbad, CA, USA), respectively. SYBR green (Life Technologies) was used for qPCR. Ubiquitin C was used as a housekeeping gene for sample normalization. Results were analyzed using the ΔΔCT method, and differences in expression between baseline and follow-up time points were tested using a paired t-test implemented in R. For a given candidate gene, P-values were then adjusted for multiple-testing using the Holm method.
Results
Baseline Demographic Characteristics
The baseline demographic data and preoperative comorbidities are summarized in Table 1. Out of the 24 enrolled subjects, 1 participant dropped out of study prior to baseline (T0). A total of 23 patients reached T3, 18 reached T6, 13 reached T9, and 12 patients completed the full study at T12. Loss to follow-up and/or missing data meant that sample size decreased by later time points (T6 to T12). The mean age ± SD of the 23 subjects was 44.2 ± 12.3 years, 78% were female, and the subjects had a mean weight of 124.6 ± 21.1 kg and a BMI of 45.2 ± 7.2 kg/m2.
Preoperative baseline demographics and obesity-related comorbidities of 23 study subjects undergoing sleeve gastrectomy
| Variables | Values |
|---|---|
| Age, years | 44.2 ± 12.3 |
| Weight, kg | 124.6 ± 21.1 |
| BMI, kg/m2 | 45.2 ± 7.2 |
| Female | 18 (78) |
| Self-reported racial/ethnic background: nonwhite(%)a | 9 (39) |
| Comorbidities at baseline | |
| Type 2 diabetes mellitus | 8 (35) |
| Prediabetes | 2 (9) |
| Hypertension | 11 (48) |
| Dyslipidemia | 13 (57) |
| Obstructive sleep apnea | 8 (35) |
| Variables | Values |
|---|---|
| Age, years | 44.2 ± 12.3 |
| Weight, kg | 124.6 ± 21.1 |
| BMI, kg/m2 | 45.2 ± 7.2 |
| Female | 18 (78) |
| Self-reported racial/ethnic background: nonwhite(%)a | 9 (39) |
| Comorbidities at baseline | |
| Type 2 diabetes mellitus | 8 (35) |
| Prediabetes | 2 (9) |
| Hypertension | 11 (48) |
| Dyslipidemia | 13 (57) |
| Obstructive sleep apnea | 8 (35) |
Data are reported as means ± SD or n (%).
aIncludes 2 African American subjects, 5 Hispanic/Latinx subjects, 1 Asian American subject, and 1 subject of “other” specified background.
Preoperative baseline demographics and obesity-related comorbidities of 23 study subjects undergoing sleeve gastrectomy
| Variables | Values |
|---|---|
| Age, years | 44.2 ± 12.3 |
| Weight, kg | 124.6 ± 21.1 |
| BMI, kg/m2 | 45.2 ± 7.2 |
| Female | 18 (78) |
| Self-reported racial/ethnic background: nonwhite(%)a | 9 (39) |
| Comorbidities at baseline | |
| Type 2 diabetes mellitus | 8 (35) |
| Prediabetes | 2 (9) |
| Hypertension | 11 (48) |
| Dyslipidemia | 13 (57) |
| Obstructive sleep apnea | 8 (35) |
| Variables | Values |
|---|---|
| Age, years | 44.2 ± 12.3 |
| Weight, kg | 124.6 ± 21.1 |
| BMI, kg/m2 | 45.2 ± 7.2 |
| Female | 18 (78) |
| Self-reported racial/ethnic background: nonwhite(%)a | 9 (39) |
| Comorbidities at baseline | |
| Type 2 diabetes mellitus | 8 (35) |
| Prediabetes | 2 (9) |
| Hypertension | 11 (48) |
| Dyslipidemia | 13 (57) |
| Obstructive sleep apnea | 8 (35) |
Data are reported as means ± SD or n (%).
aIncludes 2 African American subjects, 5 Hispanic/Latinx subjects, 1 Asian American subject, and 1 subject of “other” specified background.
Weight Loss and Related Improvements in T2D
Weight and BMI were recorded every 3 months up to 1 year postoperatively (Fig. 1A). Significant reductions were observed in percentage total body weight loss (TBWL) by T3 (mean = 18.1%, 95% CI: 16.6, 19.6; P < 0.001), T6 (24.0%, 95% CI: 20.6, 27.2; P < 0.001), T9 (29.2%, 95% CI: 24.4, 34.3; P < 0.001), and T12 (29.6%, 95% CI: 23.7, 35.4; P < 0.001) (Fig. 1B).
Weight change (A) and percentage total body weight loss (B) for each subject over 12 months following sleeve gastrectomy. Dark blue dots indicate values measured at baseline (preoperatively) and dark red dots indicate values measured during follow-up (postoperatively). Dashed lines between dots indicate missing data for a given subject. Data for each time point are summarized as means and 95% CIs. Significant differences adjusted for multiple testing were marked with the following thresholds: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Abbreviations: T0, baseline; T3, 3 months postoperatively; T6, 6 months postoperatively; T9, 9 months postoperatively; T12, 12 months postoperatively.
In a subset of 8 diabetic subjects who had baseline preoperative mean hemoglobin A1c (HbA1c) values of at least 6.5% and/or were actively using T2D medications, we examined 2 relevant biomarkers. Among these diabetic subjects, mean fasting plasma insulin levels were trending lower following LSG, from 53.6 µIU/mL (95% CI: 17.0, 96.5) at T0 to 13.1 µIU/mL (95% CI: 3.1, 30.7) at T12 (33), although this difference was not significant due to a large baseline variance (P ≥ 0.05). They also had a significant reduction and normalization of HbA1C by T3, changing from a preoperative mean of 6.7% (95% CI: 5.9, 7.5) to a postoperative mean of 5.4% (95% CI: 5.1, 5.9; P < 0.05) (33).
Leukocyte Profiles and Clinical Markers of Inflammation
As a sign of chronic inflammation, plasma clinical inflammatory markers were elevated at baseline prior to LSG, including white blood cells (9.4 K/µL, 95% CI: 8.3, 10.6), neutrophils (6.7 K/µL, 95% CI: 5.6, 7.7), CRP (7.0 mg/L, 95% CI: 5.2, 9.1), and IL-6 (5.4 pg/mL, 95% CI: 3.9, 7.2). This obesity-related chronic inflammatory state was ameliorated throughout the 12 months with significant changes observed as early as T3. White blood cells levels were reduced to 6.7 K/µL (95% CI: 6.3, 7.2; P < 0.001) at T3 and 5.9 K/µL (95% CI: 5.2, 6.7, P < 0.05) at T12 (Fig. 2A). A decrease in levels from T0 was observed at T3 for CRP (4.7 mg/L, 95% CI: 3.3, 6.1) (Fig. 2B) and IL-6 (4.3 pg/mL, 95% CI: 3.2, 5.7) (Fig. 2C), although the differences were not significant (both Ps ≥ 0.05). Further reductions occurred at T12 for CRP (2.4 mg/L, 95% CI: 1.2, 3.6; P < 0.05).
Changes in key serum clinical inflammatory markers and leukocytes composition: white blood cells (A), C reactive protein (B), interleukin-6 (C), neutrophil counts (D), lymphocyte counts (E), and neutrophil to lymphocyte ratio (F) from baseline to 12 months after undergoing sleeve gastrectomy. Dark blue dots indicate values measured at baseline (preoperatively) and dark red dots indicate values measured during follow-up (postoperatively). Dashed lines between dots indicate missing data for a given subject. Data for each time point are summarized as means and 95% CIs. Significant differences adjusted for multiple testing were marked with the following thresholds: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Abbreviations: T0, baseline; T3, 3 months postoperatively; T6, 6 months postoperatively; T9, 9 months postoperatively; T12, 12 months postoperatively.
Leukocyte composition was also altered post-LSG. A significant reduction in neutrophil counts in plasma was observed from T0 (6.7 K/µL, 95% CI: 5.6, 7.7) to T3 (3.8 K/µL, 95% CI: 3.4, 4.2; P < 0.05) (Fig. 2D). This early decrease in neutrophil counts was paired with a steady level of lymphocyte counts across the entire study from T0 (1.9 K/µL, 95% CI: 1.6, 2.3) to T12 (2.0 K/µL, 95% CI: 1.6, 2.5; P ≥ 0.05) (Fig. 2E). Correspondingly, a decreasing trend in neutrophil-lymphocyte ratio (NLR) was also observed from T0 (4.0, 95% CI: 3.1, 4.9) to T12 (1.8, 95% CI: 1.6, 2.0; P ≥ 0.05) (Fig. 2F). Importantly, the first significant reduction in NLR was observed as early as T3 (2.0, 95% CI: 1.7, 2.3; P < 0.05).
Plasma Hormonal Biomarkers
Having observed a marked reduction in systemic inflammation following LSG, we evaluated the impact of the LSG on several biomarkers of adipose tissue-related inflammation (adiponectin and resistin) and hormonal biomarkers of metabolism (leptin and ghrelin). Although small improvements following LSG were observed for the measured hormonal plasma markers of adipose tissue-related inflammation, the changes were nonsignificant (all Ps > 0.05) and were not sustainable (34).
However, in contrast to the patterns observed for the tested inflammatory hormones, LSG was associated with marked and sustained improvements in the metabolic hormonal biomarkers. At T3, mean plasma concentrations of both leptin and ghrelin were reduced at all time points. Leptin concentrations fell from 70.3 ng/mL (95% CI: 57.1, 84.4) at T0 to 30.4 ng/mL (95% CI: 21.5, 39.8) by T12 (P < 0.01) (Fig. 3A). Active ghrelin concentrations fell from 56.4 pg/mL (95% CI: 46.0, 67.9) at T0 to 37.4 pg/mL (95% CI: 33.2, 42.4) at T3 (P < 0.01), 42.4 pg/mL (95% CI: 35.7, 49.8) at T6 (P < 0.05), and 45.7 pg/mL (95% CI: 38.4, 53.7) at T12 (P < 0.05) (Fig. 3B).
Changes in serum levels of leptin (A) and ghrelin (B) from baseline to 12 months after undergoing sleeve gastrectomy. Blue bars indicate values at baseline (preoperatively). Dark blue dots indicate values measured at baseline (preoperatively) and dark red dots indicate values measured during follow-up (postoperatively). Dashed lines between dots indicate missing data for a given subject. Data for each time point are summarized as means and 95% CIs. Significant differences adjusted for multiple testing were marked with the following thresholds: *P < 0.05,**P < 0.01, ***P < 0.001, ****P < 0.0001. Abbreviations: T0, baseline; T3, 3 months postoperatively; T6, 6 months postoperatively; T9, 9 months postoperatively; T12, 12 months postoperatively.
Blood Gene Expression Changes in Several Key Metabolic Pathways
To better understand how LSG improved immune and metabolic profiles, we evaluated whether any genome-wide changes in whole blood transcriptome profiling were associated with post-LSG weight loss. In the first 6 subjects from this cohort (4 women and 2 men), we performed a pilot differential gene expression analysis to identify genes whose transcript counts were altered between T0 and T3. This subset of subjects had weight loss comparable to the rest of the cohort (TBWL= 19.4% vs 17.6%). Of 36 593 gene transcripts detected, 2557 (7.1%) demonstrated a significant change in expression 3 months after LSG compared to the baseline preoperative state [false discovery rate (FDR) < 5%] (Fig. 4), 337 of which showed more than 1 unit of log2 fold change (35).
Differential expression of 36 593 gene transcripts in whole blood from 6 subjects at 3 months following sleeve gastrectomy. The dashed horizontal line corresponds to a false discovery rate (FDR) of 5% (corresponding to nominal P = 0.0093). The dashed vertical line corresponds to a log2 fold-change (FC) = 1.0. Transcripts showing significant and substantive differential expression (FDR < 5%, log2 FC > 1.0) are plotted in red. All other transcripts not meeting these 2 criteria are plotted in light gray.
We subsequently sought to explore whether LSG affects these transcriptomes in a similar manner by performing unsupervised hierarchical clustering on sample-specific transcript counts. We chose as input a set of 112 transcripts measured in these 6 individual subjects that showed strong enough differential expression to clearly differentiate preoperative T0 samples from the T3 samples (log2 FC > 1.635, FDR < 5%). While 5 of the 6 individual subjects clearly clustered together based on their baseline samples, we identified an outlier subject (LSG0006) whose baseline transcript expression levels were more similar to the set of follow-up samples (Fig. 5). Interestingly, LSG0006 was also an outlier in weight loss response to LSG, achieving maximal weight loss at T3 (14.6% TBWL) and then experiencing early weight regain at T6 leading to the least weight loss in the cohort at T12 (5.1% TBWL) (Fig. 1B).
Heatmap of 112 individual transcript counts from strongest differential expression results in whole blood from 6 subjects at 3 months following sleeve gastrectomy (T3). Transcripts showing significant and substantive differential expression (false-discovery rate < 5%, log2 fold-change > 1.635) were chosen for display. All expression values are centered on the mean values across all 6 baseline (T0) samples. Red coloring in the heatmap indicates increased expression compared to the baseline mean (beige) across all subjects and blue indicates decreased expression. Samples (columns) are also annotated according to their time point: T0 (dark blue) and T3 (dark red).
To assign broad biological meaning to these differential gene expression results, a set of 50 hallmark gene sets (30), corresponding to distinct and coherent biological pathways, were tested for enrichment. In total, 28 gene sets were enriched at FDR < 5% (Fig. 6). The gene sets represented a variety of biological pathways that again suggested a close relationship between immunity and cell metabolism in the context of surgical weight loss. Among the pathways most enriched for increased expression following surgery were related to oxidative phosphorylation (OXPHOS), fatty acid oxidation (FAO), and Myc signaling—pathways that are essential to cell metabolic processes and cell proliferation.
A set of 28 hallmark gene sets significantly enriched for differential expression in whole blood from 6 subjects at 3 months following sleeve gastrectomy at a false discovery rate < 5%. Negative normalized enrichment scores indicate decreased expression of genes in the set following surgery, and positive score indicate increased expression. Abbreviation: MSigDB, Molecular Signatures Database.
Concordantly, the gene sets most enriched for downregulated expression following surgery were also closely related to immune cell metabolic function. Pathways related to estrogen response and heme metabolism, both essential biological processes involved in signal transduction and immune cell survival were found to be downregulated post-LSG. Furthermore, a downregulation in proinflammatory pathways are once again reflected in our pathway analysis, with reduced gene expressions for transforming growth factor beta signaling and several cytokine pathways.
To extend these biological findings of differential expression beyond the initial subjects and time points, we assayed gene expression for a small set of candidate genes of biological interest across the full set of subjects and time points as available (36). In particular, these genes showed significant large FCs in expression at T3 and were part of the core leading edges driving the gene set enrichments we observed. We demonstrated a significant reduction in expression of IGF2 at T3 (2.5-fold, P < 0.001) and also at T6 (2.4-fold, P < 0.001) following LSG, concordant with our RNA-seq data. Expression of IFNE (3.5-fold, P < 0.05) and of MMP15 (1.5-fold, P < 0.05) were found to be upregulated at T3, T6, and T12 and at T3 and T12, respectively. These individual expression results were incongruent at T3 compared with the RNA-seq data. However, these findings confirm that gene expression changes early and durably after LSG and should motivate continued investigation of the surgically regulated metabolic pathways identified.
Collectively, our differential expression results demonstrated that the genes most enriched for both upregulation and downregulation postsurgery represent metabolic pathways, suggesting that changes in cellular metabolism following bariatric surgery may influence immune cell activation, maintenance, and differentiation.
Discussion
The physiology and molecular underpinnings of the effects of LSG on weight, metabolism, and immune function are likely far more complex than anatomical modifications alone. In this prospective, longitudinal study, we (1) characterized the temporal changes in serum biomarkers over a year (2), characterized concurrent temporal changes in white blood cell composition, and (3) identified near-term gene expression alterations in several key immune and metabolic pathways in whole blood following LSG.
We demonstrated profound changes in hormonal markers including insulin, HbA1C, ghrelin, and leptin. Rapid improvements in fasting insulin concentrations and signs of diabetes remission has been reported within days following LSG (37). Mounting evidence suggests that the significant improvement in insulin resistance is a result of a combination of weight-independent effect secondary to metabolic hormones as well as subsequent weight loss. Gut hormones, such as ghrelin and leptin, play an integral role in the appetite-signaling process have also been shown to significantly decrease following LSG (38). Although we did not report any significant changes in serum adiponectin levels, a similar pattern has been observed in other weight loss studies (39). Serum adiponectin levels have been shown to be persistently low in insulin-resistant subjects while levels in insulin-sensitive subjects are more pronounced following weight loss (40, 41). Moreover, LSG is known to induce a smaller increase than Roux-en-Y gastric bypass in circulating adiponectin, and our smaller sample size may have not been able to demonstrate this change in view of our inclusion of insulin-resistant subjects (42).
Additionally, we observed the recovery of the immune system from its chronic inflammatory state. Our findings are in concordance with previous studies demonstrating improvements in metabolic hormones and circulating inflammatory cytokines following bariatric surgery (11,43,44) and weight loss (45). Nonetheless, our present work offers a unique perspective by sampling patients at 5 evenly spaced perioperative time points, which gave us a more complete appreciation of the temporal changes associated with LSG. Intriguingly, most of the improvements in clinical metabolic and immune function biomarkers dramatically took place by 3 months following LSG, with sustainable results at 1 year. We acknowledge that these modifications may have even started earlier postoperatively as suggested in several studies (37,44); however, the addition of earlier time points may be disruptive to patients’ clinical care.
The composition of leukocytes was also altered following surgery, with a reduction in absolute neutrophil counts. The differentiation of these leukocyte subpopulations may vary based on their role in the inflammatory response, as neutrophil activation is aligned with an active innate immune system, while lymphocytes are associated with the adaptive immune system (46). The intricate rebalance of neutrophil and lymphocytes following LSG amount to a reduced NLR, signifying a decrease in the detrimental effects of neutrophilia and a transition to a more lymphocyte-mediated adaptive immunity. Along with other clinical inflammatory markers, there is an emerging interest in NLR for its prognostic value. NLR has been proposed as a novel biomarker for proinflammatory pathologies such as T2D, atherosclerosis, and coronary artery disease (47-49). Moreover, a high postoperative day-1 NLR is shown to a predictor for 30-day complications including longer hospital stay, major complications, and reoperation rates (50).
We then performed RNA-seq on human whole blood prior to and 3 months after LSG to comprehensively characterize gene expression changes that may underlie LSG effects. Pathway analysis revealed robust changes in gene expression influencing both immune function and immune metabolism in nearly all subjects. At baseline, gene expression patterns in the subjects were also overall quite similar, with the exception of a single subject. This outlier, LSG0006, was a poor responder to LSG, having experienced both weight regain and failure to achieve diabetes remission. Interestingly, their baseline gene expression profile was more closely related to the post-LSG expression profile of the other subjects. The power of whole-blood gene expression profiles for predicting response to LSG merits further study.
In terms of possible biological mechanisms, we found evidence of substantial upregulations of OXPHOS and FAO metabolic pathways. Cellular metabolism has been suggested to play a critical role in immune cell reprogramming affecting B- and T-cell activation and proliferation (51-53). In particular, a shift toward OXPHOS and FAO is linked to a regulatory, anti-inflammatory phenotype in multiple cellular immune subsets (54). Emerging evidence has shown that both glucose and fatty acid metabolism are essential for the function and survival of lymphocytes. In particular, B1a cells, an effective member of the humoral immune response, have been shown to have an elevated rates of OXPHOS, lipid uptake, and fatty acid synthesis (53). Moreover, FAO is also recognized to have a crucial role in regulating T-cell responses. It has been observed to promote the generation of T regulatory cells and long lived memory T cells that are necessary for sustained adaptive immune function (55). As lymphocyte dysfunction is observed in diabetes and obesity (55), an immunometabolic shift to OXPHOS and FAO is a potential mechanism by which LSG improves immune function and insulin resistance.
In addition, there is a complex interplay between immune markers following surgery. We observed a general downregulation in cytokine pathways along with an upregulation in gene expressions of several immunoglobulins. This is in line with a number of human and animals studies that have also demonstrated a rise in circulating and adipose tissue immunoglobulins along with a change the regulatory function of B lymphocytes following bariatric surgery (56,57). These modifications in the adaptive immune system may explain the improvements in autoimmune or immune-mediated diseases detected following bariatric surgery (37,58).
Lastly, we have also identified several metabolic-centric pathways, such as estrogen and heme metabolism, that are downregulated post-LSG. Estrogen receptors play a pivotal role in the regulation of innate immune signaling pathways and myeloid cell development (59). Overexpression of heme metabolism may contribute to reactive oxygen species production, cellular hypoxia and DNA damage leading to cell death, a phenomenon often observed in chronic inflammation, obesity, and T2D (60,61).
We further extended our expression analysis of candidate genes using qPCR across the full cohort across the T0, T3, T6, and T12 time points. We anticipated that the results would demonstrate longer term changes in blood gene expressions and whether these changes are in line with weight loss trajectories. Concordant with our RNA-sequencing data, we demonstrated a significant reduction in IGF2, part of a family of insulin-like growth factors that assumes major roles in growth and development associated with obesity (62), reiterating changes observed in immune cell metabolism following LSG.
Taken together, our data suggest that LSG may have a profound effect on immune cells through the modulation of their own metabolic processes. We have previously published data that describes changes within metabolic and immune pathways in visceral adipose tissue following LSG using rodent model (56). This longitudinal study, however, provided us an initial evaluation into this complex post-LSG relationship between immunity remodeling and cell metabolism within human immune cells temporally over 5 equally divided time points. This study design has allowed us to further focus our gene expression analysis on the first 3 months following LSG, the time period whereby the most profound clinical and biochemical changes were demonstrated.
Our study has several limitations. While the study was prospective in nature, we only analyzed subjects who elected to undergo a single type of bariatric surgery procedure. The sex distribution of the cohort was skewed toward female, characteristic of a bariatric surgery cohort (8). We have a smaller data set for longer time points; however, this was not due to poor follow-up rate but rather reflects the ongoing development of a prospective study cohort. Despite having a satisfactory follow-up rate, several data points are missing due to technical issues. Our pilot transcriptomic study was also relatively modest in size, comprising 6 subjects, and was performed using whole blood, which is both a complex mélange of cell types and an indirect readout of metabolic programming throughout the body. We cannot definitively say how much of the gene expression changes following surgery reflect changes in immune cell proportions rather than altered immune cell programming. Nonetheless, our data suggest several noteworthy pathways meriting future investigations in obese human subjects undergoing bariatric surgery.
In summary, we found sleeve gastrectomy induces marked early changes in immune cell composition, function, and metabolism, as measured by biomarkers and gene expression. Further evaluation is required to determine bariatric surgery’s effects on immunometabolism and the consequences for vaccine response, host defense against pathogens, and allergy development. These studies may benefit from measurement of circulating immune cell function and health, together with their impact on metabolism. Identifying and understanding these associations will allow for the implementation of specific therapeutic strategies for treating obesity and its metabolic comorbidities.
Acknowledgments
The authors thank the study subjects for their generous participation in this work.
Financial Support: D.C.C-C. was supported by a grant from the National Heart, Lung, and Blood Institute (NHLBI) at the U.S. National Institutes of Health (NIH) (K01 HL127265). B.A.R. was supported by an NIH NHLBI grant (R01 HL086601). E.G.S. was supported by an appointed KL2 award from Harvard Catalyst (The Harvard Clinical and Translational Science Center) and the NIH National Center for Advancing Translational Sciences (NCATS) (KL2 TR002542). D.C.C.-C. and E.G.S. each received research coordinator support from the BWH Center for Clinical Investigation via a grant from Harvard Catalyst and the NIH NCATS (UL 1TR002541).
Author Contributions: A.T. and E.G.S. performed the surgeries; D.C.C-C., B.A.R., A.T., and E.G.S. designed this work; D.C.C-C., B.A.R., and E.G.S. funded this work; D.C.C-C. and E.G.S. cosupervised this work; D.C.C-C., E.M.L., A.M.W., O.J.I.-B., C.O.I., E.J.B., and E.G.S. performed the study recruitment and clinical phenotyping; T.L., E.J.M.R., R.S., K.H., S.T.M., L.S., and L.F. processed the samples and/or assayed the biomarkers; T.L., R.S.H., and D.C.C-C. performed the statistical analyses; R.P.C. and D.C.C-C. developed and ran the genome-wide gene expression analysis pipeline; R.S.H. performed the assaying of the targeted follow-up gene expression data; T.L., D.C.C-C., and E.G.S. drafted the manuscript; and all coauthors read and approved this work.
Additional Information
Disclosures: Ali Tavakkoli is a cofounder and consultant for AltrixBio. Eric Sheu is a consultant for Vicarious Surgical, Inc.
Data Availability
Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality. The corresponding author will on request detail the restrictions and conditions under which access to some data may be provided.
Publication History
A version of this manuscript was previously deposited in medRxiv (https://doi.org/10.1101/2020.07.31.20161687).
References
Author notes
Cosupervised this work.
