Effect of various weight loss interventions on serum NT-proBNP concentration in severe obese subjects without clinical manifest heart failure

Obesity is associated with a “natriuretic handicap” indicated by reduced N-terminal fragment of proBNP (NT-proBNP) concentration. While gastric bypass surgery improves the natriuretic handicap, it is presently unclear if sleeve gastrectomy exhibits similar effects. We examined NT-proBNP serum concentration in n = 72 obese participants without heart failure before and 6 months after sleeve gastrectomy (n = 28), gastric bypass surgery (n = 19), and 3-month 800 kcal/day very-low calorie diet (n = 25). A significant weight loss was observed in all intervention groups. Within 6 months, NT-proBNP concentration tended to increase by a median of 44.3 pg/mL in the sleeve gastrectomy group (p = 0.07), while it remained unchanged in the other groups (all p ≥ 0.50). To gain insights into potential effectors, we additionally analyzed NT-proBNP serum concentration in n = 387 individuals with different metabolic phenotypes. Here, higher NT-proBNP levels were associated with lower nutritional fat and protein but not with carbohydrate intake. Of interest, NT-proBNP serum concentrations were inversely correlated with fasting glucose concentration in euglycemic individuals but not in individuals with prediabetes or type 2 diabetes. In conclusion, sleeve gastrectomy tended to increase NT-proBNP levels in obese individuals and might improve the obesity-associated “natriuretic handicap”. Thereby, nutritional fat and protein intake and the individual glucose homeostasis might be metabolic determinants of NT-proBNP serum concentration.


Methods
The intervention study includes data from 72 severely obese individuals (20 men, 52 women, BMI > 37 kg/m 2 ) who were treated at the interdisciplinary center of obesity medicine of the University Hospital Schleswig-Holstein (Kiel, Germany), and who were selected to either receive a surgical or non-surgical weight loss therapy based on international guidelines and clinical judgement of the interdisciplinary obesity board. Of the total cohort, n = 28 underwent sleeve gastrectomy, n = 19 gastric bypass surgery, and n = 25 the VLCD. Only patients with complete follow-up data and without clinical heart failure diagnosis were included in this present analysis (Supplemental Fig. 1). Before inclusion into the study, a written confirmed consent was obtained from each subject. The study was conducted according to the guidelines laid down in the Declaration of Helsinki and was approved by the ethic committee of the Medical Faculty of the University of Kiel (Germany).
All patients underwent a fasting blood draw and a body weight and height measurement at baseline, e.g. before surgery and before the start of the VLCD (see study design in Supplemental Fig. 2). Patients in the surgery groups underwent laparoscopic sleeve gastrectomy or Roux-en-Y gastric bypass surgery. For gastric bypass surgery, a pouch length of 5-6 cm with an 18 mm bougie was used, a side-to-side anastomosis was performed using a 45-mm linear stapler, and the achieved biliopancreatic and alimentary limb lengths were 100 cm and 150 cm, respectively, as previously described 34,35 . For sleeve gastrectomy, a 14 mm bougie was used. Out of all 47 individuals who underwent bariatric surgery, n = 5 from the sleeve gastrectomy cohort and n = 3 from the gastric bypass cohort received a structured VLCD program using a liquid diet for 6 to 10 weeks pre-surgery. The decision was based on the surgeon's clinical judgement. Regardless of receiving a VLCD, all patients were advised to remain weight stable and consume a protein-rich, low-carbohydrate diet.
All patients were examined by a surgeon 1 month post-surgery. After 6 months, blood draws and body weight and height measurements were repeated. The VLCD intervention is presented in detail elsewhere 36 . Briefly, patients in the VLCD group were given an approximately 800 kcal/day formula-based and micronutrientbalanced diet for 3 months with a subsequent transition and weight maintenance phase of further 3 months. After 6 months, blood draws and body weight and height measurements were repeated.
In a second part of this paper, we explored determinants of NT-proBNP concentration in a cross-sectional study population. These exploratory analyses are based on data of a Northern German study population which was recruited as part of the project Food Chain Plus (FoCus, funded by the German Federal Ministry of Education and Research) by the PopGen biobank in Kiel, Germany. At time of analysis, n = 470 individuals aged between 18 and 80 years of the adiposity outpatient clinic of the University Hospital Schleswig-Holstein (UKSH) and from the general population of Kiel using the local population registry or word of mouth were enrolled into the study and analyzed in this present study. Baseline examinations were conducted between 2011 and 2015. Participants were asked to fill-in questionnaires with regard to demographic, lifestyle, and medical history, including the EPIC 12 month food frequency questionnaire 37,38 . Participants with incomplete questionnaire data (n = 22), unavailable NT-proBNP values due to technical issues with the ELISA kit (n = 31), and a preexisting diagnosis of heart failure (n = 30) were excluded (Supplemental Fig. 3). The study was conducted according to the guidelines laid down in the Declaration of Helsinki and was approved by the ethic committee of the Medical Faculty of the University of Kiel (Germany). All participants had given their informed consent prior to participation. In  www.nature.com/scientificreports/ Laboratory analyses. Blood was drawn in the morning in a fasted state. Serum was stored immediately at − 80 °C. Blood samples underwent routine laboratory analyses at the central laboratory of the University Hospital Schleswig-Holstein (Kiel, Germany), e.g., for measuring C-reactive protein (CRP) and interleukin 6 (IL-6). After determination of fasting insulin by an electro-chemiluminescence immunoassay (ECLIA; Elecsys system; Roche, Basel, Switzerland) and of fasting glucose by a glucose-hexokinase UV test (Hitachi Modular; Roche), the Homeostasis Model Assessment (HOMA) index was calculated as glucose (mg/dl) × insulin (μU/ml) / 405. In the intervention cohort, complete and valid fasting blood glucose data of both pre-and post-time points were available from n = 14 (sleeve gastrectomy), n = 9 (gastric bypass), and n = 25 (VLCD) participants. NT-proBNP was measured using a commercially available ELISA kit (Biomedica Medizinprodukte GmbH, Wien, KAT.NR. SK-1204) with an intra-and inter-assay CV of ≤ 4% and ≤ 7%. In the intervention group, 58% of NT-proBNP values were below the assay's detection limit of 25.4 pg/mL (n = 2 values were 0 pg/mL, n = 38 were 8.5 pg/mL [which corresponds to 1 pmol/L], and n = 2 values were between 8.5 and 25.4 pg/mL). In the cross-sectional group, 57% of NT-proBNP values were below detection limit (n = 35 values were 0 pg/mL, n = 167 were 8.5 pg/ mL, and n = 18 values were between 8.5 and 25.4 pg/mL). Tryptophan serum levels were measured by liquid chromatography and tandem mass spectrometry (Agilent 1100 HPLC/CTC-PAL Autosampler/Sciex API 4000 Triple Quadrupole) in an external specialized laboratory (Medizinisches Labor Bremen).

Anthropometric analyses.
Statistical analysis. Statistical data analysis was performed using the SAS statistical software package (SAS Enterprise Guide Version 7.15; SAS Institute, Cary, NC). Power calculations were performed prior to analyses to calculate the minimum detectable significant increase in NT-proBNP after bariatric surgery, based on a previous study with similar design 25 . Assuming a power of 0.80 and a 2-sided alpha level of 0.05, the sample size required to detect a significant increase in NT-proBNP after bariatric surgery was n = 19, which is comparable to the sample sizes of the intervention groups of this paper.
Unless otherwise specified, continuous parameters are presented as mean ± SD and were compared using paired or unpaired student t tests. Continuous parameters not normally distributed are presented as median (25th -75th percentiles) and were compared using Wilcoxon signed rank tests. For multiple comparisons of quantitative variables with normal and non-normal distributions we used one-way ANOVA and Kruskal-Wallis tests, respectively. The χ 2 test was used to compare categorical data. All individual changes (∆) in weight and NT-proBNP serum levels were calculated as the difference between post-minus pre-treatment and analyzed by Student's paired t-tests when data was normally distributed and by Wilcoxon signed rank test when data was not normally distributed. Changes between post-minus pre-treatment are presented as mean with 95% confidence interval (CI) when data was normally distributed. The Pearson's correlation coefficient was used to quantify associations between continuous variables.
Due to the skewed distribution of NT-proBNP values in the interventional cohort (n = 42 below the ELISA detection limit of 25.4 pg/mL), only NT-proBNP values above the detection limit (n = 30) were used in the main analysis. Sensitivity analyses were performed using all available NT-proBNP values including those below the ELISA detection limit (n = 72).
Due to the skewed distribution of NT-proBNP values in the cross-sectional FoCus cohort (n = 220 values below the ELISA detection limit), we created three groups representing 1) NT-proBNP concentration below detection limit (n = 220) as well as 2) low and 3) high NT-proBNP concentration. The latter two groups were created based on the median (158.9 pg/mL) of all detectable values. Individuals with an NT-proBNP concentration above 25.4 pg/mL but below 158.9 pg/mL were assigned to the low NT-proBNP group (n = 83), while individuals with an NT-proBNP concentration equal or above 158.9 pg/mL were assigned to the high NT-proBNP group (n = 84). To determine associations between NT-proBNP groups and other variables, p trend analyses were performed using a linear regression model with NT-proBNP group as ordinal covariate and the variable of interest as dependent variable. In addition, sensitivity analyses were performed in the whole cohort (n = 387) by using non-parametric Spearman correlation analyses which do not rest upon an assumption of normality.

Change in NT-proBNP serum concentration after three different weight loss interventions.
Baseline characteristics of the intervention groups are presented in Table 1. Briefly, all three intervention groups mainly consisted of females (76.0-78.9%) and mean age ranged from 43.3-50.6 years. The intervention groups were not statistically different in terms of baseline body weight (mean range 134.3-155.5 kg, p = 0.07), BMI (mean range 46.2-50.8 kg/m 2 , p = 0.07), fasting blood glucose (mean range 111.6-147.2 mg/dL, p = 0.12), and NT-proBNP concentration (median range 94.4-189.7 pg/mL, p = 0.72).
In univariate correlation analyses, there were no associations between changes in weight, BMI, or fasting blood glucose and concomitant changes in NT-proBNP concentration after all three weight loss interventions (all p > 0.50).

Exploratory analyses of NT-proBNP determinants in a cross-sectional study population.
In exploratory analyses, we investigated determinants of NT-proBNP serum concentration in a larger population using data from the cross-sectional FoCus cohort to gain potential mechanistic insights. Characteristics of the FoCus cohort are shown in Table 2. Briefly, the cohort consisted of 387 individuals (72.9% females) without clinical diagnosis of heart failure. Mean age was 53.4 ± 11.7 years, mean BMI was 32.1 ± 12.0 kg/m 2 , and mean systolic/diastolic blood pressure were 129.0 ± 13.7 and 80.7 ± 7.5 mmHg, respectively. The cohort comprised non-diabetic individuals with normal fasting glucose (euglycemic, n = 218) as well as prediabetic (n = 108) and diabetic (n = 61) individuals. Accordingly, mean fasting glucose was 108.5 ± 39.6 mg/dL and ranged from 67.0 to 387.0 mg/dL. Median NT-proBNP was 8.5 (8.5-131.0) pg/mL and significantly higher in females compared to males (17.4 [8.5-153.1] pg/mL vs. 8.5 [8.5-50.7] pg/mL, p = 0.0003).
However, high NT-proBNP concentration was associated with less fat intake (p trend = 0.02, Fig. 4C) and less protein intake (p trend = 0.03, Fig. 4D). These results remained significant after further adjustment for BMI (both p ≤ 0.05). There was no trend for decreased carbohydrate and total energy intake with increasing NT-proBNP concentration (both p trend ≥ 0.10, Figs. 4A,B).
There was no association between NT-proBNP serum concentration and fasting glucose serum concentration (p trend = 0.14). However, when grouping participants based on normal (< 100 mg/dL), prediabetic (between 100 and 126 mg/dL) and diabetic (> 126 mg/dL) fasting glucose, in euglycemic individuals, a higher NT-proBNP concentration was associated with lower fasting blood glucose (p trend = 0.005). In a between-group analysis, euglycemic individuals with high NT-proBNP values had a 3.1 mg/dL (CI: -5.6, -0.5, p = 0.01) lower fasting blood glucose compared to individuals with no detectable NT-proBNP (Fig. 5A). This association remained significant after further adjustment for BMI (p trend = 0.01). The association was not found in prediabetic or diabetic individuals (both p trend > 0.8, Fig. 5B,C). There were no associations between NT-proBNP serum concentration and fasting insulin or HOMA-IR in the whole cohort (all p trend > 0.14) and in euglycemic, prediabetic, and diabetic subgroups (all p trend > 0.07). There was also no association between NT-proBNP concentration and measures of inflammation such as CRP, IL-6, and tryptophan (all p trend ≥ 0.30).
For all presented results, sensitivity analyses were performed by using Spearman correlation analyses in the whole cohort and similar results were obtained (data not shown) except for tryptophan, were an inverse association has been found (r = -0.11, p = 0.03). Table 1. Baseline characteristics of intervention groups. Values are presented as mean ± SD for continuous variables or number (frequency) for categorical variables with minimum and maximum in parentheses. 1 Only complete and valid fasting blood glucose values were included in the analysis (sleeve gastrectomy: n = 14, gastric bypass n = 9, VLCD: n = 25). 2 Skewed values are expressed as medians with interquartile ranges separated by a hyphen and with minimum and maximum in parentheses. 3 Only NT-proBNP values above the ELISA detection limit were included in the analysis (sleeve gastrectomy: n = 14, gastric bypass n = 7, VLCD: n = 9). BMI, body mass index; NT-proBNP, N-terminal pro brain natriuretic peptide; VLCD, very low-calorie diet. www.nature.com/scientificreports/ www.nature.com/scientificreports/  1 Skewed values are expressed as medians with interquartile ranges and with minimum and maximum in parentheses. Statistical significance between males and females was determined with unpaired t test for normally distributed data, with Wilcoxon test for skewed data, and with χ 2 test for categorical data. An asterisk (*) denotes significant differences between both groups (p < 0.05). BMI, body mass index; CRP, C-reactive protein; HOMA-IR, homeostatic model assessment of insulin resistance; IL-6, interleukin 6; NT-proBNP, N-terminal pro brain natriuretic peptide.   (3) high NT-proBNP concentration. The latter two groups were created based on the median (158.9 pg/mL) of all detectable values. Individuals with an NT-proBNP concentration above 25.4 pg/ mL but below 158.9 pg/mL were assigned to the low NT-proBNP group (n = 83, middle bar), while individuals with an NT-proBNP concentration equal or above 158.9 pg/mL were assigned to the high NT-proBNP group (n = 84, right bar). Bars denote median NT-proBNP concentration and error bars denote 95% CI of the median. Y axis is formatted in log 10 to account for skewed distribution of NT-proBNP values. CI; confidence interval; NT-proBNP, N-terminal pro brain natriuretic peptide.

Discussion
In this present study including 72 severely obese individuals with relatively low NT-proBNP serum concentration, we found that sleeve gastrectomy tended to increase NT-proBNP after 6 months of follow-up, while NT-proBNP concentration remained unchanged after gastric bypass surgery and VLCD. Blood glucose was reduced during all 3 interventions. In exploratory, cross-sectional analyses in a cohort of n = 387 individuals, we found that NT-proBNP was higher in females and that higher NT-proBNP concentration was associated with less fat intake as well as with lower fasting glucose in individuals with normal glucose regulation. There was no association between NT-proBNP and measures of obesity and inflammation markers.
Sleeve gastrectomy might improve the "natriuretic handicap" of severely obese individuals. Natriuretic peptides, including BNP (and its N-terminal counterpart NT-proBNP) can be considered as antagonists of the renin-angiotensin-aldosterone system (RAAS) as they decrease blood pressure, fluid volume and exert antifibrotic and antihypertrophic effect in the heart 1 . Previous studies reported that obesity is associated with decreased BNP and NT-proBNP concentrations 6-14 , possibly due to attenuated natriuretic peptide secretion 16 and/or increased clearance as a result of obesity-associated glomerular hyperfiltration 39 . Therefore, individuals with obesity seem to have a "natriuretic handicap" 17 systolic blood pressure. Bars denote mean and error bars denote 95% CI of the mean. P trend analyses were performed using a linear regression model with NT-proBNP group as ordinal covariate and the variable of interest as dependent variable. Definition of NT-proBNP groups is shown in Fig. 2. BMI, body mass index; CI; confidence interval; NT-proBNP, N-terminal pro brain natriuretic peptide. www.nature.com/scientificreports/ which exposes them to greater risk for developing hypertension 16,18,19 , hypertension-related cardiovascular disorders 6 , and type 2 diabetes 20,21 . Weight loss interventions can reverse this "natriuretic handicap" [22][23][24][25][26][27] , although some studies also reported decreased BNP and NT-proBNP concentrations after weight loss [29][30][31] . However, these latter studies may be confounded by concomitant reductions in plasma volume and blood pressure, themselves being strong determinants of natriuretic peptide levels 40 .
Interestingly, in one study, weight loss induced by Roux-en-Y gastric bypass surgery increased BNP and NT-proBNP concentrations to a greater extent when compared to weight loss induced by a VLCD 23 , suggesting a weight-loss independent mechanism that improves natriuretic peptide release, possibly by yet unidentified alterations of the neuronal/hormonal balance due to the "disconnected" stomach and duodenum 25 . In contrast, another bariatric surgery technique, gastric banding, has been shown to reduce NT-proBNP in obese individuals 31 .
To date, banding surgery techniques are less and less used in favor of sleeve gastrectomy which reduces the stomach to about 15% of its original size to achieve weight loss and achieves better outcomes and has less complications 32 . So far, no study investigated the change in NT-proBNP after sleeve gastrectomy. Association between NT-proBNP concentration and (A) energy intake, (B) carbohydrate intake, (C) fat intake, and (D) protein intake. Information about the nutritional intake was assessed by a self-reported 12-month nutritional, retrospective (food frequency) questionnaire. Bars denote mean and error bars denote 95% CI of the mean. P trend analyses were performed using a linear regression model with NT-proBNP group as ordinal covariate and the variable of interest as dependent variable. Definition of NT-proBNP groups is shown in Fig. 2. CI; confidence interval; NT-proBNP, N-terminal pro brain natriuretic peptide. www.nature.com/scientificreports/ Therefore, in this present study, we compared sleeve gastrectomy to gastric bypass surgery and VLCD and found that sleeve gastrectomy tended to increase NT-proBNP by a median of 44.3 pg/mL (p = 0.07) in severely obese individuals after 6 months of follow-up and an average weight loss of 38 kg. These data support the idea that the "natriuretic handicap" of obese individuals can be improved with surgically-induced weight loss, especially sleeve gastrectomy. The median increase of 44.3 pg/mL is comparable to previous data from gastric bypass surgery showing an increased NT-proBNP concentration by ~ 24-46 pg/ml 6 months after surgery 23,25,27 .
Gastric bypass surgery and VLCD did not significantly increase NT-proBNP concentration (all p ≥ 0.50). This is different from previous studies showing increased NT-proBNP after gastric bypass surgery 23,25,27 and a hypocaloric diet [22][23][24] , although NT-proBNP did not change following gastric bypass surgery in another study 41 . As possible confounding variables like sex, age, and BMI were not significantly different between the interventional groups, we hypothesize that other, not measured, determinants (left ventricular hypertrophy, systolic blood pressure, pulse rate) might have contributed to the divergent results between the interventional groups. Furthermore, our finding might have been influenced by a selection bias, that is, patients with higher BMI were more likely to receive sleeve gastrectomy while patients with lower BMI were more likely to receive gastric bypass or VLCD treatment. Lastly, weight loss tended to be greater in the sleeve gastrectomy cohort compared to the gastric bypass cohort, which is different from some 42,43 but not all 44,45 studies of similar duration and might have contributed to the greater increase in NT-proBNP in the sleeve gastrectomy cohort.
While we cannot provide a physiological explanation why NT-proBNP increased with sleeve gastrectomy only, we hypothesize that concomitant changes in vasoactive hormones might have contributed to the increase in NT-proBNP after sleeve gastrectomy. Accordingly, one recent study reported that aldosterone, angiotensinogen, angiotensin II, bradykinin, endothelin, neprilysin, and renin were significantly reduced after sleeve gastrectomy while natriuretic peptides significantly increased 33 .
Our results merit further verification in future studies primarily designed to compare different surgical weight loss strategies with regard to changes in natriuretic peptides, including natriuretic peptide type A (ANP).
Interestingly, we also found that fasting blood glucose was reduced during all 3 weight loss interventions which confirms previous studies 25,28,46 . We also report that the VLCD-induced reduction in fasting blood glucose was associated with the concomitant amount of weight loss, which is comparable to a previous study 47 . Exploratory analyses: determinants of NT-proBNP concentration. In order to explore potential nutritional and metabolic factors contributing to the above-mentioned effects in this present study, we also investigated which demographic, anthropometric, and metabolic characteristics are associated with NT-proBNP concentration in a cross-sectional cohort of 387 individuals. While we found that NT-proBNP is higher in women compared to men-which is in line with previous studies 15,23,48 and mediated by lower circulating androgens 48 -we did not find an association between NT-proBNP and measures of obesity. This is different from previous studies which consistently reported this association [6][7][8][9][10][11]15 and might be due to, again, the relatively lower NT-proBNP concentration in our cohort (median 8,5 pg/mL) and the fact that our cohort did, by design, not include individuals with a BMI from 25 to 30 kg/m 2 which might have further influenced this association. Euglycemia was defined as fasting plasma glucose < 100 mg/dL, prediabetes was defined as fasting plasma glucose between 100 and 126 mg/dL, and diabetes was defined as fasting plasma glucose > 126 mg/dL. Bars denote mean and error bars denote 95% CI of the mean. P trend analyses were performed using a linear regression model with NT-proBNP group as ordinal covariate and the variable of interest as dependent variable. Statistical significance of between-group differences was calculated by unpaired t test. Definition of NT-proBNP groups is shown in Fig. 2. CI; confidence interval; NT-proBNP, N-terminal pro brain natriuretic peptide. www.nature.com/scientificreports/ Similarly, we did not find an association between NT-proBNP and inflammation markers. In previous studies, there is conflicting evidence whether natriuretic peptides are higher or lower in a proinflammatory state 17,49 . Interestingly, lower NT-proBNP concentration was associated with higher fat intake in our cohort. Studies investigating the associations between natriuretic peptides and macronutrient intake are scarce, however, one study reported that a traditional Mediterranean high-fat diet decreased NT-proBNP to a greater extent compared to a low-fat control diet 50 . Similarly, in another study, a high-fat diet resulted in lower BNP concentration due to increased circulating levels of neprilysin 51 , an enzyme implicated in BNP degradation 51 . These data-including ours-indicate that low-fat diets may be suitable to reverse the metabolic handicap in individuals with obesity and might explain the effects of weight loss interventions. Future longitudinal studies are warranted to investigate this hypothesis. We also found an inverse association between protein intake and NT-proBNP level in our cohort, e.g. lower protein intake was associated with a higher NT-proBNP concentration. We could not find studies investigating the association between protein intake and NT-proBNP. But higher NT-proBNP has been associated with sarcopenia in patients with type 2 diabetes which supports our data 52 . Carbohydrate intake had no effect on NT-proBNP level in our cohort, which has been reported previously 53 .
Interestingly, in euglycemic individuals, we found that higher NT-proBNP concentration was associated with lower fasting glucose, such that individuals with high NT-proBNP concentration (median 415.6 pg/mL) had a lower fasting glucose by 3.1 mg/dL compared to individuals with an NT-proBNP concentration below assay detection limit of 25.4 pg/mL. Our findings are supported by data from the general population 15 and by a study showing that BNP infusion lowers plasma glucose after a glucose load in healthy, euglycemic men 54 . In that latter study, the authors hypothesized that the BNP-mediated decrease in glucose is due to increased glucose distribution but not due to increased cellular glucose uptake 54 . In contrast to euglycemic individuals, higher BNP is not related to better glucose tolerance in individuals with diabetes 55 , which supports our data. We did not find an association between NT-proBNP and measures of insulin resistance, however, previous studies reported that NT-proBNP was inversely related to insulin resistance 13 and hyperinsulinemia 15 . Limitations. In our study we show for the first time that (i) sleeve gastrectomy is able to increase NT-pro-BNP concentration and (ii) that NT-pro-BNP concentration is related to nutritional fat and protein intake as well as glucose homeostasis in euglycemic individuals. However, some limitations have to be considered: We did not perform echocardiography in both the intervention and cross-sectional cohorts, and therefore, could not adjust NT-proBNP values for echocardiographic variables such as left ventricular mass and left atrial size. Additionally, we did not measure BNP in our cohort, which has been reported to increase after surgically induced weight loss 33 . Further, NT-proBNP concentration was relatively low in both of our cohorts with around 58% of values below the assay's detection limit of 25.4 pg/mL, which may be due to concomitant drug treatment (which was not assessed in this cohort) and which have influenced the results. However, sensitivity analyses were performed considering only values above the detection limit and similar results were found. In the interventional cohort, we did not find an association between changes in NT-proBNP and weight, BMI, or fasting glucose while some 22,23,28 but not all 24,25 interventional weight loss studies reported these associations.

Conclusion
Surgically induced weight loss by sleeve gastrectomy tended to increase NT-proBNP in individuals with severe obesity and thus might improve their "natriuretic handicap". In a cross-sectional analysis, higher NT-proBNP was associated with less fat and protein intake and with lower fasting glucose in euglycemic individuals, suggesting that nutritional fat and protein intake as well as the individual glucose homeostasis might be metabolic determinants in regulating serum NT-pro-BNP concentration.