Is AMP‐activated protein kinase activation a central mechanism of cardio‐metabolic outcomes after metabolic and bariatric surgery?

Abstract

Metabolic and bariatric surgery (MBS) is the most effective treatment for severe obesity. It usually results in spectacular weight loss, associated with improvements of obesity‐associated comorbidities. The mechanisms underlying these benefits are not fully understood but could involve a postoperative activation of the enzyme 5′ AMP‐activated protein kinase (AMPK). Hence, as AMPK is largely expressed in insulin‐sensitive cells, it acts as a key regulator of cardio‐metabolic homeostasis, and its activity is down‐regulated in tissues from obese, insulin resistant patients. This narrative review aims to summarise the available clinical data regarding changes in AMPK activity following MBS and to discuss the potential relevance of these changes in postoperative physiology. The eight studies reporting specifically changes in AMPK activity following MBS in humans were analysed. They all showed increases in AMPK activity in tissues or blood cells, with effects observed as early as 3 months and persisting beyond 12 months post‐surgery. However, the data does not allow us to conclude on (i) the potential specificity of effects depending on the surgery procedure, (ii) the mechanisms involved in the AMPK activation, and (iii) its role in postoperative metabolic outcomes, highlighting that further investigations are warranted to address these issues. Understanding AMPK changes in postoperative physiology could establish its relevance as a potential prognostic marker of surgery metabolic outcomes and as a new target to improve the benefits of MBS. More generally, it may provide insights into the development of novel therapeutic strategies for obesity and associated comorbidities.

1. INTRODUCTION

Obesity is defined in adults by a body mass index (BMI) higher than 30 kg/m². Its worldwide prevalence has doubled between 1990 and 2022, to reach 16%, ¹ and it is expected to increase further over the next few years, reaching almost 50% by 2030 in the United States. ² Health consequences of obesity are significant, since at least 18 different complications or comorbidities have been identified, highlighting the widespread effects of excess body fat on overall health. ³ As a consequence, obesity causes 3.4 million deaths per year, resulting in a 4% reduction in disability‐adjusted life years. ³

The most effective treatment for severe obesity is metabolic and bariatric surgery (MBS). It usually results in a rapid and spectacular weight loss, with up to 75% loss of excess weight at 1 year and around 50% at 20 years after surgery, ⁴ along with significant improvements in various obesity‐associated comorbidities and complications. ⁵ Of note, benefits of MBS are not confined to physiological effects, but extend to mental health, with notable relief of psychiatric troubles like anxiety, depression, and eating disorders. ⁶ As a consequence, MBS improves quality of life to a significantly greater extent than anti‐obesity medications and lifestyle interventions ⁷ and significantly reduces obesity morbi‐mortality. ⁵ The mechanisms underlying cardio‐metabolic benefits of MBS are not fully understood; they may partly involve anatomical and/or physiological rearrangements of the gastrointestinal tract, irrespective of weight loss. ⁸ Further, few studies reported higher activity and/or expression of the 5′ AMP‐activated protein kinase (AMPK) following MBS compared to pre‐operative levels.^{9, 10, 11, 12, 13, 14, 15, 16}

AMPK is a ubiquitous enzyme, largely expressed in insulin‐sensitive cells. Its activation aims at increasing the cellular ATP/AMP(ADP) ratio, through stimulation of catabolic and inhibition of anabolic pathways. ¹⁷ At the body scale, it plays a key protective role in cardiovascular and metabolic regulation. AMPK activity is decreased in adipose tissues from insulin resistant obese subjects compared to insulin sensitive subjects.^{18, 19} Similar alterations were found in hepatocytes and peripheral blood mononuclear cells (PBMCs) from obese patients.^{12, 13} In the latter study, a significant correlation was found between AMPK activity and BMI. Still, there is no consensus on such a direct relationship, since it was not confirmed using AMPK activity in adipocytes isolated from human subcutaneous white adipose tissue. ²⁰ In a context of obesity, AMPK inhibition can occur as a result of the closely associated low‐grade chronic inflammation and excessive and/or unbalanced diet. ²¹ The deleterious consequences are multiple and may largely contribute to the development of obesity comorbidities.^{21, 22} This explains the ever‐growing search for pharmacological AMPK activators for the treatment of obesity and its comorbidities^{22, 23, 24} and resulted in the synthesis of several molecules with promising effects in preclinical models of obesity and metabolic syndrome.^{24, 25, 26}

The properties of AMPK raise the hypothesis that its activation following MBS may play a central role in the postoperative metabolic improvements. Therefore, the objective of this narrative review was to conduct the first critical analysis of the currently available clinical studies regarding AMPK changes after MBS. We then addressed the putative causal role of these changes in postoperative physiology and finally, the relevance of AMPK as a therapeutic target or prognostic marker.

2. METHODOLOGY

A literature search was conducted up to 14 March 2025, primarily using PubMed, Europe PMC and Google Scholar databases. The search terms included “bariatric surgery,” “metabolic surgery,” “obesity,” “AMPK,” “AMPK activation” and “pAMPK.” Different combinations of these keywords were also used. Additional articles were identified using the reference lists of publications dealing with MBS and associated AMPK changes.

We selected original research articles involving human subjects undergoing MBS and in which AMPK activity or expression was quantitatively assessed. Animal studies and reviews were excluded from the primary analysis but used for mechanistic discussion. Based on these selection criteria, only eight articles published between 2011 and 2022 were identified. Their main characteristics and results are summarised in Table 1.

TABLE 1.

Summary of the characteristics and main results of the eight studies measuring AMPK activity and/or expression post‐MBS.^{9, 10, 11, 12, 13, 14, 15, 16}

Study	Number of patients, sex and age	Type of MBS	Delay post‐MBS	Cardio‐metabolic/biochemical changes	Method of evaluation of AMPK activity and changes compared to pre‐surgery
R. M. Holmes et al. 9 (USA, 2011)	6; 38 ± 4 years	RYGB	6 months	↓ body weight and BMI ↓ plasma glucose and insulin, HOMA‐IR ↑ plasma adiponectin	$\frac{\text{pAMPK}}{\text{AMPK}}$, protein expression (WB) 1.7‐fold increase in AMPK activity in skeletal muscle
P. H. Albers et al. 10 (Denmark, 2015)	10 NGT (70% women) 10 T2DM (60% women) 40.1 ± 2.8 years	RYGB	1 week	T2DM: ↓ body weight ↓ plasma glucose and insulin NGT: ↓ plasma insulin	pAMPK and AMPK, protein expression (WB) No change in AMPK activity in skeletal muscle and scAT
			3 months	T2DM: ↓ body weight and fat mass ↓ plasma glucose and insulin, HbA1c NGT: ↓ body weight and fat mass ↓ plasma glucose and insulin, HbA1c ↑ plasma adiponectin	pAMPK and AMPK, protein expression (WB) Increase in AMPK activity in skeletal muscle (pAMPK 1.1‐fold a in NGT and 1.4‐fold a in T2DM, with no change in AMPK) No change in AMPK activity in scAT
			12 months	T2DM: ↓ body weight and fat mass ↓ plasma glucose and insulin, HbA1c ↑ plasma adiponectin NGT: ↓ body weight and fat mass ↓ plasma glucose and insulin, HbA1c ↑ plasma adiponectin	pAMPK and AMPK, protein expression (WB) Increase in AMPK activity in scAT (pAMPK 1.9‐fold a in NGT and 1.35‐fold a in T2DM, with no change in AMPK) No change in AMPK activity in skeletal muscle
X. J. Xu et al. 11 (USA, 2015)	11 (73% women) 46 ± 4 years	RYGB	3 months	↓ body weight and BMI ↓ plasma insulin ↓ CRP ↑ plasma adiponectin ↓ T2DM and metformin users	$\frac{p\text{AMPK}}{\text{AMPK}}$, protein expression (WB) 3.5‐fold increase in AMPK activity in scAT
G. Angelini et al. 12 (Italy, 2019)	15 with NAFLD 39.7 ± 2.1 years	SG	12 months	↓ body weight and BMI ↓ plasma glucose and insulin, HOMA‐IR ↓ triglycerides and LDL‐Cs, ↑ HDL‐Cs	$\frac{p\text{AMPK}}{\text{AMPK}}$, protein expression (WB) 3.4‐fold increase in AMPK activity in monocytes
C. F. García‐Prieto et al. 13 (Spain, 2019)	17 (65% women) 48.6 ± 2.5 years	Laparoscopic bypass (9/17) or SG (8/17)	12 months	↓ body weight and BMI ↓ plasma glucose and insulin, HOMA‐IR, HbA1c ↓ CRP ↓ T2DM and metformin users ↓ hypertension and ACEi/ARB users ↓ dyslipidaemia and statins users	$\frac{p\text{AMPK}}{\text{AMPK}}$, protein expression (WB) 1.25‐fold a increase in AMPK activity in PBMC
B. M. Varela‐Rodríguez et al. 14 (Spain, 2020)	4 (75% women) 45 ± 10.9 years	RYGB (1/4) or SG (3/4)	BMI <30 kg/m2 (22 to 33 months)	↓ BMI and fat mass (all) ↓ plasma insulin and HOMA‐IR (3/4), HbA1c (all) ↓ CRP (all) ↓ triglycerides (3/4), LDL‐Cs (2/4), ↑ HDL‐Cs (all) ↓ T2DM and hypertension (2/2)	pAMPK and AMPK, protein expression (WB) 6.5‐fold a increase in AMPK and 7‐fold a increase in pAMPK in scAT
R. Ferraz‐Bannitz et al. 15 (Brazil, 2021)	13 (100% women) 37.7 ± 8.2 years	RYGB	3 months	↓ body weight, BMI and fat mass ↓ HOMA‐IR ↓ DBP and SBP ↑ adiponectin (gene expression) ↓ IL‐6 (gene expression)	AMPK gene expression (RT‐qPCR) 4.8‐fold increase in AMPK in scAT
			6 months	↓ body weight, BMI and fat mass ↓ plasma insulin, HOMA‐IR ↓ DBP and SBP ↑ adiponectin (gene expression) ↓ IL‐6 and TNFα (gene expression)	AMPK gene expression (RT‐qPCR) 3‐fold increase in AMPK in scAT
Z. Abad‐Jiménez et al. 16 (Spain, 2022)	43 (84% women) 45.1 ± 11.4 years	RYGB	12 months	↓ body weight and BMI ↓ plasma glucose and insulin, HOMA‐IR, HbA1c ↓ DBP and SBP ↓ triglycerides, total‐Cs and LDL‐Cs, ↑ HDL‐Cs ↓ CRP, IL6 and IL1β ↓ T2DM, hypertension and dyslipidaemia	pAMPK and AMPK, protein expression (WB) 2‐fold a increase in AMPK and 1.5‐fold a increase in pAMPK in leukocytes

Note: An increase in the ratio of AMPK phosphorylated at Thr172 over total AMPK (pAMPK/AMPK; protein expression measured by Western blot, WB) reflects enzyme activation. Studies are sorted by year of publication.

Abbreviations: ACEi, angiotensin converting enzyme inhibitors; _scAT, subcutaneous adipose tissue; ARB, angiotensin receptor blockers; BMI, body mass index; DBP, diastolic blood pressure; CRP, C‐reactive protein; Cs, cholesterol; HbA1c, haemoglobin A1C; HDL, high density lipoprotein; HOMA‐IR, homeostasis model assessment of insulin resistance; LDL, low density lipoprotein; NAFLD, non‐alcoholic fatty liver disease; NGT, normal glucose tolerant; PBMC, peripheral blood mononuclear cells; RYGB, Roux‐en‐Y bypass; SBP, systolic blood pressure; SG, sleeve gastrectomy; T2DM, type 2 diabetes mellitus; WB, Western blot.

^a Graphically determined values.

3. OVERVIEW OF THE MAIN SURGICAL PROCEDURES AND THEIR HEALTH EFFECTS

According to the guidelines of the American Society for Metabolic and Bariatric Surgery and the International Federation for the Surgery of Obesity and Metabolic Disorders (IFSO), MBS is indicated for patients with a BMI >35 kg/m² regardless of the presence of comorbidities or for patients with a BMI ranging 30 to 34.5 kg/m² associated with comorbidities such as dyslipidemia or type 2 diabetes mellitus (T2DM); these BMI reference values may be adjusted according to ethnicity. ²⁷ From the first surgery MBS performed in 1953 to nowadays, numerous surgical procedures have been developed. The technique used varies greatly from one geographical area to another and must be adapted to the patient's profile. ²⁸ The most common are schematically illustrated in Figure 1; these are:

• Gastric band (GB): A band is placed around the stomach, reducing its capacity and slowing the passage of food through the organ. ²⁹ This procedure was widespread in the early 2000s but is gradually disappearing, currently accounting for only 1% of bariatric surgeries. ²⁸

• Gastric bypass, also called Roux‐en‐Y bypass (RYGB): In the stomach, a small pouch is created, diverting ingested nutrients from the stomach. ²⁹ This procedure represented 90% of all bariatric surgeries in 2003, but its use has gradually declined since then: 29% of MBS performed in 2018 (second most common technique), ²⁸ and only 18% in 2019. ³⁰

• Sleeve gastrectomy (SG): The stomach size is reduced by 80%. It is the most frequent surgery with 380 000 operation each year worldwide, ³¹ and represents 55% of MBS since 2014.

FIGURE 1.

Schematic representation of the three main MBS procedures (adapted from Wolfe et al. ²⁹ ). Gastric band (left) consists of placing a band around the stomach, the position of which is adjustable; the stomach banding reduces its volume and slows the passage of food through the organ. The Roux‐en‐Y bypass (RYGB) or gastric bypass (middle) procedure connects a small upper stomach pouch to the middle small intestine in a Y shape, diverting ingested nutrients from most of the stomach, the duodenum, and part of the jejunum. In the case of vertical sleeve gastrectomy (right), the stomach size is irreversibly reduced by 80%, resulting in a drastic restriction of ingested food. Historically, gastric band and vertical sleeve gastrectomy were referred to as predominantly restrictive procedures, as they decrease the functional volume of the stomach; the RYGB procedure was considered both restrictive and malabsorptive since the stomach bypass reduces nutrient absorption.

It is crucial to note that MBS does not guarantee sustained weight loss. Indeed, 20%–35% of patients undergoing RYGB experience suboptimal clinical responses, defined by the IFSO as a body weight or BMI loss of less than 20%. ³² Clinical response depends on the surgical procedure and patient‐specific factors like age, psychological distress, pre‐operative lean mass and BMI, and postoperative dietary adherence and medical follow‐up. ³³ Additionally, recurrent weight regain is a common concern. In a cohort study involving 1406 RYGB patients, weight regain from the lowest postoperative weight was progressively observed over time, reaching 15% at 5 years. ³⁴ Finally, surgery‐induced weight loss differs according to patient ethnicity; for example, RYGB resulted in significantly longer weight loss than SG after 1, 2, or 5 years in Hispanic and African American subjects, but not in the general population. ³⁵

As mentioned in Introduction section, the benefits of MBS on obesity morbi‐mortality are well established. Notably, surgical procedures that generate anatomical and/or physiological rearrangements of the gastrointestinal tract seem to have a higher impact on T2DM and dyslipidemia than those based on gastric volume reduction.^{36, 37} The difference may rely on greater changes in the secretory profile of gut hormones (e.g., GLP‐1) and adipokines (e.g., adiponectin, leptin), independently of weight loss.^{38, 39, 40} Of note, these procedure‐specific hormonal shifts may be of particular relevance as GLP‐1, adiponectin and leptin act as upstream regulators of AMPK. Last, the kinetics and intensity of beneficial effects depend on the type of surgical procedure. For example, adjustable GB induces slow effects on cardio‐metabolic regulation in parallel with weight loss. RYGB results in a higher rate of T2DM remission than SG after 1 year, but the difference between both procedures disappears 2–5 years after surgery. ⁴¹

4. AMPK ACTIVATION FOLLOWING MBS

From the eight studies reporting AMPK changes following MBS,^{9, 10, 11, 12, 13, 14, 15, 16} most of them assessed AMPK activity by evaluating AMPK phosphorylated at Thr172 (pAMPK), that is, the active form of the enzyme, over total AMPK (AMPK). All were performed in obese populations, but in addition, Albers et al. ¹⁰ distinguished normal glucose tolerant (NGT) from diabetic (T2DM) patients and Angelini et al. ¹² included patients with non‐alcoholic fatty liver disease (NAFLD). The number of subjects was generally low (<20), ranging from 4 ¹⁴ to 43. ¹⁶

4.1. Tissue specificity of AMPK activation

4.1.1. Skeletal Muscle

Two studies sought for AMPK changes in this tissue: Holmes et al., ⁹ with six participants with metabolic syndrome and a mean pre‐operative BMI of 48.5 ± 4.8 kg/m², and Albers et al., ¹⁰ who studied 10 NGT patients with a mean pre‐operative weight of 116.9 ± 4.9 kg and 10 T2DM patients with a mean pre‐operative weight of 121.5 ± 8.9 kg. Holmes et al. ⁹ observed a significant 1.7‐fold increase in the pAMPK/AMPK ratio, reflecting AMPK activation, 6 months postoperatively. Albers et al. ¹⁰ also reported increases in the phosphorylated form of AMPK in muscle biopsies from NGT and T2DM patients sampled 3 months postoperatively (x1.1 and x1.4 vs. pre‐surgery, respectively; p‐value < 0.05 for both). Since no major change in total AMPK was observed in parallel, higher pAMPK possibly reflected AMPK activation; however, glycogen synthase and acetyl‐CoA carboxylase (ACC), two downstream targets of AMPK, were unaltered.

4.1.2. Adipose tissue

According to Albers et al., ¹⁰ AMPK activation in abdominal subcutaneous adipose tissue is delayed compared to skeletal muscle. Hence, significant enzyme activation was observed only at 12 months post‐surgery, with a 1.9‐fold increase (p‐value < 0.001 vs. pre‐surgery) in pAMPK levels in NGT patients and a 1.3‐fold increase (p‐value < 0.05 vs. pre‐surgery) in T2DM patients. This delayed timeline stands in contrast to other findings. Xu et al., ¹¹ who analysed 11 patients with a mean pre‐operative BMI of 41.7 ± 1.4 kg/m², reported a significant 3.5‐fold increase in adipose AMPK activity (pAMPK/AMPK) as early as 3 months post‐MBS, accompanied by a large fall in malonyl‐CoA cell content reflecting AMPK‐mediated ACC inhibition. Similarly, Ferraz‐Bannitz et al., ¹⁵ who included 13 non‐diabetic patients with a mean pre‐operative BMI of 42.2 ± 4.2 kg/m², found a higher AMPK expression at the same early time point (i.e., 3 months), which persisted at 6 months post‐surgery, although to a lesser extent (x3 vs. x4.8; p‐value <0.0001 and <0.001 vs. pre‐surgery, respectively). Of note, however, authors only measured AMPK gene expression, which may not strictly reflect enzyme activation. Finally, Varela‐Rodriguez et al. ¹⁴ (four patients with a mean pre‐operative BMI of 50.0 ± 11.5 kg/m²) observed marked elevations in pAMPK (7.0‐fold) and total AMPK (6.5‐fold) levels. Although the specific pAMPK/AMPK ratio was not calculated, the proportionally greater rise in the phosphorylated form suggests a slight enhancement of AMPK activity post‐MBS.

4.1.3. Peripheral blood mononuclear cells

Owing to the invasiveness of tissue biopsies, most recent studies measured AMPK phosphorylation in blood cells.^{12, 13, 16} They consistently reported increases of phosphorylated AMPK at 12 months post‐surgery. Specifically, García‐Prieto et al., ¹³ who studied 17 patients with a mean pre‐operative BMI of 41.8 ± 0.9 kg/m² observed a 1.25‐fold increase in the pAMPK/AMPK ratio (p‐value<0.001 vs. pre‐surgery). Abad‐Jiménez et al. ¹⁶ who studied 43 patients with a mean pre‐operative BMI of 39.6 ± 4.9 kg/m², reported increases in total AMPK and pAMPK levels in the same range (1.5 to 2‐fold). Although the pAMPK/AMPK ratio was not provided, authors concluded on enhanced postoperative enzyme activity. Finally, Angelini et al. ¹² who selected 15 subjects with NAFLD and a mean pre‐operative BMI of 43.0 ± 0.9 kg/m², observed a more pronounced increase in AMPK activity (x3.4 vs. pre‐surgery; p‐value = 0.0001).

4.2. Temporal dynamics of AMPK activation

4.2.1. Immediate (<1 month) post‐operative changes

Earlier post‐MBS changes were assessed only once. ¹⁰ Clinically, this early stage corresponds to a remission period, with an adapted diet without solid food. In these patients, total AMPK and pAMPK in skeletal muscle and adipose tissue biopsies were unchanged. This suggests that the acute caloric restriction and surgical stress immediately after MBS do not trigger the AMPK pathway instantly. However, simultaneous inhibitory stress signals like inflammation might mask a moderate AMPK activation.

4.2.2. Mid‐term (1 to 12 months) post‐operative changes

Most studies evaluated AMPK changes between 1 and 12 months post‐MBS. The majority reported a significant upregulation of AMPK expression or activity during this time window in adipose and skeletal muscle tissues and in PBMC. Albers et al. ¹⁰ showed distinct tissue‐specific kinetics. In skeletal muscle, pAMPK activation was transient, occurring at 3 months but disappearing at 12 months post‐surgery. Inversely, adipose tissue exhibited a delayed response, showing AMPK activation only at 12 months. This observation was however not confirmed by Xu et al., ¹¹ in which AMPK activity in adipose tissue was already significantly enhanced at 3 months post‐MBS. Mean increases in AMPK regulation were of the same range, whatever the time of measurement (x2.3, x2.4 and x1.8 at 3, 6, and 12 months post‐MBS). The same tendency was observed while considering only changes in the most frequently studied tissue, that is, adipose tissue. However, the variability from one study to the other is high, due to the heterogeneity of (i) surgery procedures, (ii) included subjects and (iii) methods of assessment of AMPK activity. Thus, further studies are needed to clearly demonstrate a time‐dependency in AMPK postoperative activation, if any.

4.2.3. Long‐term (>12 months) post‐operative changes

To date, there is no data on AMPK activation over extended periods post‐surgery. The only study that adapted the endpoint was conducted by Varela‐Rodriguez et al., ¹⁴ which defined the measurement endpoint based on patient‐specific outcomes. Hence, AMPK levels were measured when patients could be considered as non‐obese, achieving a BMI <30 kg/m². As a result, the measurement time points varied from 22 to 33 months post‐surgery, representing a significantly longer follow‐up period compared to all other studies. Enhanced levels of phosphorylated (7.0‐fold, p‐value <0.05 vs. pre‐surgery) and total (6.5‐fold, p‐value <0.05 vs. pre‐surgery) forms of AMPK were found in adipose tissue, exceeding the protein levels measured in non‐obese control subjects and suggesting higher enzyme overall activity. The pAMPK/AMPK ratio was, however, not calculated. The study also suffered from a small number of patients (n = 4), among which half displayed T2DM and hypertension, undergoing either RYGB or SG. Such a long‐term follow‐up of AMPK changes is mandatory to clarify whether AMPK activation can influence the outcome of MBS, especially regarding late clinical deterioration (recurrent weight gain and metabolic complications) that often occurs after 5 years or more post‐MBS.

4.3. AMPK activation according to the surgical procedure

Most studies included patients undergoing either RYGB or SG, except for two studies^{13, 14} in which both procedures can be found. In the latter, no comparison between the two MBS was made regarding AMPK regulation. From the others, no difference between RYGB and SG can be evidenced, as a 3.4‐fold increase in AMPK activity is reported upon SG in Angelini et al., ¹² while increases following RYGB range from 1.1‐ to 4.8‐fold.^{9, 10, 11, 12, 15, 16} Thus, the available data does not allow to clarify whether the type of MBS can influence AMPK modulation. Here again, the scarcity of direct comparative studies represents a significant knowledge gap. Addressing this issue is important because RYGB and SG differently affect the secretion profile of intestinal hormones (e.g., GLP‐1, PYY) and adipocytokines, which may impact postoperative AMPK activation (see below). Interestingly, RYGB and SG also differently modify cardio‐metabolic regulation. For example, adjustable GB induces slow effects on cardio‐metabolic regulation, in parallel with weight loss. RYGB results in a higher rate of T2DM remission than SG after 1 year, but this difference tends to disappear after 2–5 years. ⁴¹ Similarly, hyperlipidemia is improved in 99% of patients with biliopancreatic diversion with duodenal switch, but in only 44% of patients with SG. ⁴² Deciphering the impact of MBS procedure on AMPK may thus help link AMPK changes to postoperative physiology.

4.4. Limits of the studies

Overall, studies suggest AMPK activation following MBS. Still, the findings also highlight tissue‐specific differences in the regulatory dynamics of AMPK, as well as the possible modulatory action exerted by metabolic adaptations. As an example, the enhancement of the Thr172‐phosphorylated form of AMPK following MBS was mostly blunted under insulin stimulation, while the inhibitory phosphorylation on Ser485 increased, ¹⁰ reflecting the deleterious influence of the insulin‐AKT signalling on AMPK activity. This probably represents a major limit of studies assessing AMPK activity at fixed time points following MBS. Individualised assessment of AMPK activity considering the patient's body weight (loss, stabilisation or regain) and metabolic status (improvement, remission, or relapse of metabolic comorbidities) could prove more relevant to evaluate more precisely AMPK changes and comprehend their involvement in the metabolic adaptations following MBS.

Another main issue is heterogeneity within cohorts regarding obesity comorbidities and pharmacological treatments prior to surgery. For instance, the group of 11 patients studied by Xu et al. ¹¹ included five patients with T2DM and five treated with metformin, a drug known to activate AMPK. ²³ Similarly, the cohort in Garcìa‐Prieto et al. ¹³ displayed significant clinical heterogeneity, including hypertension (10/17), hyperlipidemia (11/17), and T2DM (14/17, among which 76.5% treated with metformin), all complications associated with AMPK dysregulation. Same holds true in Abad‐Jiménez et al., ¹⁶ where 13 out of 43 patients were diagnosed with T2DM, 15 with hypertension and 10 with dyslipidemia, as well as in Varela‐Rodriguez et al. ¹⁴ (2/4 subjects with T2DM and hypertension). In an attempt to overcome this issue, Albers et al. ¹⁰ identified two subgroups of patients, according to glucose regulation. Still, the T2DM group included subjects without (2/10) or with various antidiabetic medications, among which metformin (4/10); although drugs were discontinued 3 days before MBS, long‐lasting adaptations in AMPK regulation cannot be ruled out. Further, diet‐induced weight loss is mandatory prior any MBS and practices may differ from one country to the other. Consequently, preoperative body weight and BMI values reported in the publications may not truly reflect the severity of obesity and the baseline AMPK activity may be flawed.

Last, a significant factor to consider in the context of obesity and MBS is a likely difference between men and women. Women are reported to undergo MBS at a rate approximately five times higher than men, ⁴³ a trend that is confirmed in the studies analysed herein. If biological sex does not seem to have a clear influence on clinical responses to MBS (excess weight loss, resolution of obesity comorbidities), ⁴⁴ its influence on AMPK expression and regulation pre‐ and post‐surgery remains largely unexplored. Of note, high fat diet‐induced obesity was associated with AMPK downregulation, insulin resistance, and liver and renal damages in male mice, while in contrast, female animals displayed preserved AMPK activity and were better protected against obesity comorbidities, despite similar weight gain. ⁴⁴

To summarise, assessment of AMPK activation pre‐ and post‐MBS may be affected by several confounding factors, probably accounting for the variability of the currently available data and limiting their comparability. Future studies using a standardised evaluation of AMPK activity and including homogenous cohorts followed up over longer periods are absolutely warranted to address in depth the MBS‐triggered AMPK adaptations.

5. IS AMPK ACTIVATION A CAUSE OR A CONSEQUENCE OF IMPROVED METABOLIC PARAMETERS FOLLOWING MBS?

5.1. MBS‐evoked physiological adaptations as triggers of AMPK activity

Several early and/or sustained physiological changes associated with MBS could account for postoperative AMPK activation, but no study attempted to specifically address this issue.

Ferraz‐Bannitz et al. ¹⁵ noted a positive significant correlation (Pearson's r = 0.947, p‐value = 0.015) between early (i.e., 3 months post‐surgery) changes in BMI and AMPK gene expression, but this correlation was no longer significant after 6 months (Pearson's r = 0.692, p‐value = 0.0846). In Albers et al., ¹⁰ the very early (i.e., 1‐week post‐surgery) slight but significant weight reduction in T2DM patients was not accompanied by AMPK changes. Last, by combining all the data extracted from the eight studies analysed herein, no correlation between weight loss and AMPK activity can be evidenced (Pearson's r = 0.166); even so, AMPK activity remains quite stable with time following MBS, as stated above, while weight loss increases (≈20 kg and ≈30 kg at 3 and 12 months, respectively). This all suggests that neither early nor sustained weight loss is sufficient to trigger AMPK. In agreement, a direct relationship between AMPK activity and body weight/BMI is still controversial.

Reduction of food intake is an immediate consequence of MBS whatever the surgical procedure. ⁴⁵ This is important as an excessive and/or unbalanced diet may reduce AMPK activity. Elevated levels of saturated free fatty acids, particularly palmitic acid, are known to inhibit AMPK. ⁴⁶ Inhibition is mediated by the accumulation of lipid intermediates, of which ceramides and diacylglycerol that activate respectively Protein Phosphatase 2A and a μ isoform of protein kinase C (PKD1), two AMPK inhibitors. ²¹ The reduced absorption of these dietary lipids following MBS may alleviate this down‐regulation. Furthermore, caloric restriction and reduced plasma insulin levels are well established stimuli for AMPK activation. Thus, undernutrition following MBS could progressively result in enhanced AMPK activity in cells. Regarding insulin concentrations, Albers et al. ¹⁰ noted significant falls in both NGT (77 ± 9 vs. 57 ± 6 pmol/L) and T2DM patients (97 ± 13 vs. 89 ± 18 pmol/L) already at 1‐week post‐surgery, but AMPK remained unchanged. Significant enzyme activation only appeared after 3 months, when insulin levels were further reduced (30 ± 3 and 51 ± 8 pmol/L, respectively). These temporal discrepancies suggest that postoperative insulin reduction, while likely a favouring factor, is not the sole or central driver of AMPK activation.

Another rapidly occurring effect of MBS is a gradual reduction of chronic inflammation. Pro‐inflammatory cytokines such as IL‐6 and TNFα promote Protein Phosphatase 2C and the suppressor of cytokine signalling 3, while downregulating the expression of Liver Kinase B1 (LKB1), a constitutively active AMPK upstream kinase ²¹ ; these all reduce AMPK phosphorylation and activity. Several studies investigated inflammatory markers at varying times post‐MBS; they consistently reported lower levels of circulating CRP (−50% at 3 months ¹¹ ; from −66% to −84% at 12 months^{13, 16}) and IL‐6 (−18% at 12 months ¹⁶ ). Down regulation of TNFα gene expression was also demonstrated. ¹⁵ However, the data does not allow decipher whether the reported higher pAMPK or pAMPK/AMPK values occurred as a consequence of, or in parallel with reduced inflammation. Even, whether bringing down global inflammation actually translates into AMPK activation remains questionable, since TNFα, a major AMPK inhibitor, was reported to decrease, ¹⁵ to keep unchanged ⁴⁷ or to increase ⁴⁸ postoperatively. Opposite results have also been reported regarding post‐surgery changes of IL‐6.

Last, changes in gut hormones and adipokines secretions following MBS may positively influence AMPK activity. Fasting and postprandial secretion of incretin hormones markedly increases following RYGB, ³⁹ contributing to the improvement in insulin sensitivity observed shortly after surgery, independently of significant weight loss. ⁴² Given that GLP‐1 can act as an AMPK activator, ⁴⁹ this early hormonal spike could theoretically stimulate AMPK. The relevance of such an effect of GLP‐1 in the context of MBS remains speculative since experimental evidence is lacking. In the studies analysed here, GLP‐1 levels were not measured, and no AMPK activation was detected at very early postoperative stage despite the presumed elevation in GLP‐1. Several other gut hormones may be affected by MBS, but changes vary depending on the surgical procedure and a possible link with AMPK is unexplored. For instance, peptide YY, an anorexigenic hormone ⁵⁰ secreted by L enteroendocrine cells in the distal small intestine and colon, is increased following RYGB or SG, but not following GB, although effects may be reversible with time. Zhou et al. ⁵¹ also reported increased obestatin levels following RYGB in mice, but conversely, decreased following SG. Stimulation of peptide YY and obestatin secretion may help reduce food intake, resulting in enhanced AMPK activity in cells. Of note, however, the effect of obestatin on satiety remains controversial, and few studies have investigated its involvement in obesity. Last, higher ghrelin levels were observed after 6 and 12 months in patients undergoing RYGB or SG. ⁵² The consequences of ghrelin changes are less clear cut, since the hormone activates AMPK in the hypothalamus by promoting CaMKK2 activation, but inhibits the enzyme in adipose tissue and in the liver. ²¹ As summarised above, opposite effects have been observed to date regarding AMPK changes in peripheral tissues, and data on central AMPK are lacking.

MBS increases circulating adiponectin, concomitant with a decrease in leptin secretion.^{39, 40} Here again, modifications in adipocytokine levels closely depend on the type of MBS. Patients undergoing RYGB experience larger increases in adiponectin levels than those with SG or single‐anastomosis sleeve ileal procedures. ⁵³ Parallel increases in AMPK activity and adiponectin levels were reported in most of the studies analysed herein.^{9, 10, 11, 15} Starting from 3 months post‐surgery, Ferraz‐Bannitz et al. ¹⁵ reported a 3.8‐fold (p‐value<0.01 vs. pre‐surgery) increase in ADIPOQ expression, which rose to 6.5‐fold (p‐value<0.0001 vs. pre‐surgery) at 6 months post‐surgery. While Albers et al. ¹⁰ confirmed increased plasma levels, they also noted persistent elevations up to 12 months with increases from 6.4 ± 0.8 pre‐surgery to 12.5 ± 1.4 mg/L in NGT and from 5.1 ± 0.7 to 8.5 ± 1.5 mg/L in T2DM patients (p‐value<0.001 vs. pre‐surgery for both). Changes in circulating adiponectin significantly and positively correlated with increases in pAMPK and its downstream target pACC in adipose tissues at 12 months post‐surgery (Pearson's r = 0.314 and 0.593, respectively; p‐value <0.05). ¹⁰ In addition, neither plasma adiponectin nor pAMPK was enhanced at 1‐week post‐surgery, suggesting that changes align with each other. Adiponectin binding to its receptor AdipoR1 activates AMPK by (i) favouring the action of the two up‐stream AMPK kinases, LKB1 and CaMKK2, ⁵⁴ and (ii) up‐regulating the expression of the downstream protein SIRT1. ⁵⁵ MBS‐associated changes in adiponectin signalling proteins, for example, receptors ⁹ and SIRT1,^{13, 15} have also been reported.

5.2. Are cardio‐metabolic improvements related to AMPK changes?

When assessed, the main obesity comorbidities were largely improved in all studies analysed herein, and the medication needs were lowered. For instance, the percentage of hypertensive subjects was reduced by 50 to 73% 1 year following MBS; similar ranges of decreases were reported regarding T2DM and dyslipidemia.^{13, 16} Notable improvements were already noticed earlier, ¹⁵ despite BMI values >30 kg/m². Improved cardio‐metabolic homeostasis is in line with the expected effects of AMPK activation and appears concomitantly; still, no direct correlation between AMPK and cardio‐metabolic parameters can be evidenced from the data.

AMPK exerts its effects through phosphorylation of many enzymes and transcription factors. ²¹ Only a few AMPK molecular targets have been studied in parallel with enzyme activation post‐MBS. One of these is ACC, the activity of which is reduced by AMPK upon phosphorylation on Ser79 and Ser221. ⁵⁶ Inhibition of ACC lowers malonyl‐CoA production, thus limiting fatty acid synthesis and helping Carnitine O‐Palmitoyl Transferase 1 (CPT‐1) activity and fatty acids β‐oxidation, ⁵⁷ reducing in turn total lipid content. This is in accordance with previous results showing that MBS restores β‐oxidation in β‐pancreatic cells in patients with T2DM ⁵⁸ and that an increase in β‐oxidation is negatively correlated with BMI. ⁵⁹ Paralleling AMPK changes, higher levels of ACC phosphorylation on Ser221 were observed in adipose tissues 12 months post‐MBS ¹⁰ and later, ¹⁴ but not earlier. ¹⁰ This is at odds with the 80% lower levels of malonyl‐CoA reported by Xu et al. ¹¹ at 3 months post‐MBS (12.5 vs. 2.5 pmol/mg adipose tissue, p‐value <0.05 vs. pre‐surgery). The reasons for this discrepancy are unclear. Of note, AMPK activity was also unchanged in Albers et al. at this time point, while it increased in Xu et al.

Effects of MBS on the AMPK‐dependent perilipin 2 (Plin2)‐LAMP2A lipolytic pathway have been addressed by Angelini et al. ¹² Authors reported reduced levels of Plin2 1 year after SG (2.3‐fold decrease, p‐value = 0.01 vs. pre‐surgery), together with higher LAMP2A levels (2.7‐fold increase, p‐value = 0.003 vs. pre‐surgery), promoting the binding of Plin2 to LAMP2A and its catabolism, thereby facilitating lipolysis. Although Plin2 levels remained higher than in the non‐obese population, the authors concluded that such an effect may reduce ectopic fat deposits and contribute to the improvement of non‐alcoholic steatohepatitis or fatty liver disease triggered by MBS. Accordingly, pharmacological stimulation of AMPK with metformin in human primary hepatocytes increased LAMP2A levels, reduced Plin2 levels, and decreased lipid droplets in the cells. ¹² Opposite results were, however, reported regarding changes in Plin1, another member of the perilipin family that plays a major role in controlling lipolysis in adipocytes and correlates to obesity.^{60, 61} Hence, no statistically significant difference in Plin1 gene expression was found before and 6 months after RYGB. ⁶² Additionally, it is important to note that lipolysis in subcutaneous adipose tissue normalises 2 years post‐MBS but returns to preoperative levels 5 years after the surgery. ⁶³

Last, pharmacological AMPK activation stimulates adiponectin secretion by cultured adipocytes,^{64, 65} although the physiological relevance of this effect is still debated.^{66, 67} As mentioned above, parallel increases in AMPK activity and adiponectin levels have been reported in most of the studies analysed herein. Adiponectin acts as a major insulin‐sensitizer. Accordingly, Glut4 protein expression was enhanced in both adipose tissue and skeletal muscle at 12 months post‐surgery. ¹⁰ In Varela‐Rodriguez et al., ¹⁴ restored insulin sensitivity following MBS was attested by normalisation of AKT signalling and Glut4 protein expression, associated with a profound structural and functional remodelling of adipose tissue. Normalisation of AMPK and pAMPK levels was also noted in this study. Increased adiponectin levels are expected to alleviate T2DM and more generally, glucose toxicity. However, the available data does not allow to decipher whether AMPK activation post‐MBS triggers adiponectin secretion, since adiponectin itself can activate AMPK.

To summarise, AMPK activation likely contributes to the improved physiology following MBS. Whether it causes this improvement remains an open question. Beyond weight loss, various effects of MBS may stimulate AMPK at an early postoperative stage, reflecting the complexity of the enzyme regulation. AMPK would then directly and rapidly improve glucose and lipid homeostasis but also induce more profound and long‐lasting effects through re‐balancing caloric intake and energy expenditure, limiting low‐grade inflammation and remodelling adipose tissues. Assuming that this mechanism holds true implies that AMPK activation occurs prior to cardio‐metabolic changes. To date, our knowledge on immediate AMPK changes (1 week postoperatively) stands on a single study ¹⁰ and there is a huge gap of knowledge on what happens within the 3 first months post‐MBS. Exploring the kinetics of changes is mandatory to understand the molecular mechanisms driving AMPK activation. Investigating postoperative AMPK and cardio‐metabolic adaptations using murine AMPK knock‐out models may also help address this issue. The fact that enzyme activation is already maximal at 3 months post‐MBS and does not increase further speaks for a causal role of AMPK; of note, AMPK activity is about three times lower in obese subjects, and post‐surgery increases are of the same range, suggesting that MBS does not amplify but rather restores AMPK activity to normal levels. The striking differences between MBS outcomes in young and elderly patients could also provide a starting point for answering this question. Hence, postoperative weight loss and improvements of obesity comorbidities are more pronounced in young patients^{68, 69} and the difference increases with age. ⁷⁰ Children and adolescents eligible to MBS exhibit remarkable resolution rates following intervention (95% for TD2M, 80% for hypertension and 66% for dyslipidemia), that largely surpass those reported in comparable adult cohorts.^{71, 72} In parallel, AMPK responsiveness to different stimuli is altered with age. ⁷³ Puzzling out whether age‐dependent variations in AMPK reactivity correlate with the clinical outcomes observed between the younger and older patients merits some attention.

Conversely, AMPK activation may develop merely as a consequence of post‐surgery metabolic improvements. If so, its multiple properties may serve to maintain and fuel profound metabolic adaptations and favor sustained clinical responses.

6. CONCLUSION AND PERSPECTIVES

With its high and ever‐growing prevalence, obesity remains a major global health challenge with limited efficient pharmacological treatment options. MBS stands out as the most effective and safest approach, although restricted to Class III obese patients. In many—but not all—patients, it facilitates significant loss of excess weight and improves various cardiovascular and metabolic obesity comorbidities. Few studies demonstrate concomitant increases in AMPK activity in insulin‐sensitive cells and blood monocytes. The mechanisms involved in this AMPK activation and its role in the improved post‐MBS physiology remain, however, essentially speculative and further investigations are clearly needed to address these issues.

Answering these questions may open promising avenues for improving clinical outcomes and developing innovative therapies for obesity and related comorbidities. Since AMPK activity is reduced in obese patients prior to MBS, one may assume that insufficient AMPK activation postoperatively could account for suboptimal initial responses or postoperative deterioration occurring a few years following surgery in many patients. Consequently, AMPK activity, easily measurable in blood monocytes, could serve as a prognostic biomarker for post‐MBS outcomes. Further, helping the normalisation of AMPK's activity through pharmacological agents could favor sustained weight loss and long‐term metabolic remissions. Such therapies could also prove efficient in reducing the risk of late clinical deterioration that occurs a few years following surgery.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

DATA AVAILABILITY STATEMENT

Data sharing not applicable to this article as no datasets were generated.