Differential effects of bariatric surgery and lifestyle interventions on plasma levels of Lp(a) and fatty acids

Background Limited evidence suggests that surgical and non-surgical obesity treatment differentially influence plasma Lipoprotein (a) [Lp(a)] levels. Further, a novel association between plasma arachidonic acid and Lp(a) has recently been shown, suggesting that fatty acids are a possible target to influence Lp(a). Here, the effects of bariatric surgery and lifestyle interventions on plasma levels of Lp(a) were compared, and it was examined whether the effects were mediated by changes in plasma fatty acid (FA) levels. Methods The study includes two independent trials of patients with overweight or obesity. Trial 1: Two-armed intervention study including 82 patients who underwent a 7-week low energy diet (LED), followed by Roux-en-Y gastric bypass and 52-week follow-up (surgery-group), and 77 patients who underwent a 59-week energy restricted diet- and exercise-program (lifestyle-group). Trial 2: A clinical study including 134 patients who underwent a 20-week very-LED/LED (lifestyle-cohort). Results In the surgery-group, Lp(a) levels [median (interquartile range)] tended to increase in the pre-surgical LED-phase [17(7–68)-21(7–81)nmol/L, P = 0.05], but decreased by 48% after surgery [21(7–81)—11(7–56)nmol/L, P < 0.001]. In the lifestyle-group and lifestyle-cohort, Lp(a) increased by 36%[14(7–77)—19(7–94)nmol/L, P < 0.001] and 14%[50(14–160)—57(19–208)nmol/L, P < 0.001], respectively. Changes in Lp(a) were independent of weight loss. Plasma levels of total saturated FAs remained unchanged after surgery, but decreased after lifestyle interventions. Arachidonic acid and total n-3 FAs decreased after surgery, but increased after lifestyle interventions. Plasma FAs did not mediate the effects on Lp(a). Conclusion Bariatric surgery reduced, whereas lifestyle interventions increased plasma Lp(a), independent of weight loss. The interventions differentially influenced changes in plasma FAs, but these changes did not mediate changes in Lp(a). Trial registration Trial 1: Clinicaltrials.gov NCT00626964. Trial 2: Netherlands Trial Register NL2140 (NTR2264). Graphical abstract Supplementary Information The online version contains supplementary material available at 10.1186/s12944-022-01756-1.

1 2 an increase in plasma Lp(a) levels in adults with or without type 2 diabetes (T2D), while plasma Lp(a) levels showed a strong tendency to decrease in patients without T2D who underwent bariatric surgery(7).
A recent meta analysis showed that bariatric surgery signi cantly decreased circulating Lp(a) levels, and that the decrease in Lp(a) was not associated with change in body mass index (BMI)(8). Lp(a) levels may also be regulated, to some extent, by changes in plasma fatty acids (FAs). A positive, novel association between plasma levels of the n-6 FA arachidonic acid (AA) and Lp(a) in patients with familial hypercholesterolemia has recently been shown(9). Other studies have shown that an increased intake of total-and saturated fat is accompanied by a decrease in Lp(a) levels(10-13), and that supplementation of conjugated LA lead to increased Lp(a) levels (14). The composition of the plasma FA pool may be altered both by caloric restriction(15) and bariatric surgery (16)(17)(18), where possible contributing factors are the reduced dietary intake, changes in the dietary composition, malabsorption of lipids(19) and release of FAs from the body fat deposits during weight loss. Whether changes in plasma Lp(a) levels following caloric restriction or bariatric surgery are mediated by changes in plasma FA levels is not known. It is important to identify opportunities to reduce adverse changes to Lp(a) during weight loss dieting, through intervening on plasma FAs. In this study, the effects of Roux-en-Y gastric bypass surgery (RYGB) and an intensive lifestyle intervention , including caloric restriction and exercise, on plasma Lp(a) and FA levels in patients with obesity were compared, and it was also examined whether possible effects on plasma Lp(a) levels were mediated by changes in plasma FA levels. The effects of a lifestyle intervention on plasma Lp(a) and FA levels in an independent cohort of patients with T2D and overweight or obesity were also examined. Methods Study subjects and design This study includes two independent trials. Trial 1 is a two-armed non-randomized study which compared the 1-year effects of RYGB (surgery-group) with intensive lifestyle intervention (lifestyle-group) (Clinicaltrials.gov NCT00626964), conducted at the Morbid Obesity Centre, Vestfold Hospital Trust, Tønsberg, Norway between February 2008 and February 2011. Inclusion criteria were BMI ≥ 40 kg/m2, or ≥ 35 kg/m2 and at least one obesity related comorbidity. The primary outcome (arterial stiffness) and data on weight-loss and changes in metabolic biomarkers have previously been published (20,21). Trial 2 includes individuals who participated in the Prevention Of Weight Regain (POWER) cohort study [Netherlands Trial Register NL2140 (NTR2264)] (lifestyle-cohort)(22). Participants were recruited at the outpatient diabetes clinic of the Erasmus University Medical Centre, Rotterdam, The Netherlands, between March 2010 and April 2015. The inclusion criteria were BMI >27 kg/m² and T2D. The primary outcome (Lp(a) levels) and data on weight-loss and changes in metabolic biomarkers have previously been published(7). Trial 1 was approved by the The Regional Committees for Medical and Health Research Ethics in Norway (code: S-05175) and trial 2 was approved by the Medical Ethics Committee of the Erasmus Medical Center (reference numbers MEC-2009-143, MEC-2014-090 and MEC 2016. Both trials were conducted according to the principles in the Declaration of Helsinki, and written informed consent was provided by all the participants. Interventions Trial 1: The participants in the surgery-group followed a low energy diet (LED) (<900 kcal per day) for 7 weeks prior to surgery (pre-surgery phase), and were followed for 52 weeks after surgery (post-surgery 1 1 1 phase) where they received standard follow-up care at the Morbid Obesity Centre -a total follow-up of 59 weeks. The participants in the lifestyle-group underwent a dietary and physical activity intervention which lasted for a total of 59 weeks(20, 21). They received nutritional counseling according to Norwegian nutritional guidelines and every participant's energy intake was reduced by 1000 kcal/day, they also underwent 90 min supervised training sessions, including weight bearing and aerobic exercise, 3 days/week during the rst 12 weeks. Thereafter, the participants received monthly follow-ups, and were advised to maintain physical activity for 60-90 min per day throughout the study period (59 weeks). Trial 2: The participants underwent a dietary intervention which lasted for a total of 20 weeks. During the rst 8 weeks, the participants followed a very LED of approximately 750 kcal per day , which consisted of two meal replacements (Glucerna SR, Abbott Nutrition, Lake Forest, Illinois, USA ), plus a small dinner , providing a total of 67 g carbohydrates, 11.5 g of bre, 54 g protein and 32 g fat (of which 16 g monounsaturated FAs) daily and micronutrients according to RDA recommendations. Thereafter, energy intake was slowly increased up to approximately 1300 kcal per day over 12 weeks. In addition, 30-60 minutes of daily exercise was encouraged during the entire intervention. Outcomes The main outcomes were plasma levels of Lp(a) and FAs. Plasma levels were measured at baseline (trial 1 and 2), 7 and 59 weeks (trial 1), and at 20 weeks (trial 2). Laboratory analyses In trial 1, plasma Lp(a) concentrations were measured using a particle-enhanced immunoturbidimetric method, by Roche Diagnostics at an accredited medical laboratory, Oslo University Hospital, Rikshospitalet, Oslo, Norway (NS-EN ISO 15189:2007). The samples were stored for 6-9 years at -80?C, and had not been thawed prior to the Lp(a) analysis. In trial 2 Lp(a) concentrations were measured using the Diagnostic System #171399910930 7 (DiaSys Diagnostic System, GmbH, Holzheim, Germany). The samples were stored for 5-10 years at -80?C before analysis, and had not been thawed prior to the Lp(a) analysis. Plasma Lp(a) levels were subsequently re-measured in a sub-group of participants from trial 2 using Roche Diagnostics to evaluate method agreement. Measurements of fasting serum blood glucose and lipoprotein pro les have been described previously (21,22). Plasma free FA pro les were determined https://app.ithenticate.com/en_us/report/92826333/similarity 4/13 3 11 9 5 7 14 by Gas Chromatography-Flame Ionization Detector analysis at the commercial laboratory Vitas Analytical

Services
. The serum samples were thawed and aliquoted to dried blood spot (DBS) paper (Whatman 903 paper) until GCanalysis. One 4.7 mm punch of human plasma DBS paper were methylated with sodium methoxide in methanol. After methylation, FA methyl esters (FAME) were extracted with hexane.
After thorough mixing and centrifugation , 3 µl of the hexane phase was injected into the

McNemar's test, paired T-test or Wilcoxon Signed Rank test were
used when investigating within-group changes. Statistical between-group comparisons were made between the lifestyle-group of trail 1 (baseline-59 weeks) and the surgery-group of trial 1 (week 7-59). Between-group differences in changes from baseline (lifestyle-group) and from week 7 (surgery-group) to end of 8 intervention were estimated using a robust linear regression approach which is a non-parametric iterative method using weights from absolute residuals.
The results are expressed as means (95% CI), and STATA version 15.0 was used to perform the analyses.
https://app.ithenticate.com/en_us/report/92826333/similarity 5/13 8 6 10 Mediation analyses were performed using the PROCESS macro (version 3.3 ) for SPSS written by A F Hayes (23), with group (lifestyle vs. surgery) as the independent variable, change in Lp(a) level (7 weeks to 59 weeks for the surgery-group and baseline to 59 weeks for the lifestyle-group) as the dependent variable and change in FA level (7 weeks to 59 weeks for the surgery-group and baseline to 59 weeks for the lifestyle-group) as the mediator variable.
Hayes uses the three steps as originally suggested by Byron and Kenny.
In step 1 the independent variable is regressed on the mediator. In step 2 the independent variable is regressed on the dependent variable. In step 3 the nal model with the independent variable and the moderator as covariates is tted and the proportion of the association between the independent and dependent variable which is explained by the mediator can be calculated.  Table 1. In trial 1, more than 60% (n=102) of the participants in both arms were female, and 97% (n=155) were White. The participants in the surgery-group were younger (41 years vs. 47 years, p=0 .011), had a higher BMI (46 kg /m2 vs . 42 kg/m2, p<0.001), and were less often diagnosed with cardiovascular disease (2% vs. 16%, p=0.004), compared with participants in the lifestyle-group. A total of 26% (n=41) had T2D, and 19% (n=30) were prescribed a statin, with no difference between the groups. In trial 2, 60% (n=80) of the participants were women and 45% (n=73) were White, the median age was 55 years and the mean BMI 35 kg/m2. All participants were diagnosed with T2D, 49% (n=65) were on insulin treatment, 16% (n=22) were diagnosed with cardiovascular disease and 58% (n=77) received statin treatment.
Weight loss and changes in metabolic biomarkers In trial 1, the initial 7-week LED in the surgery-group led to a mean (95% CI) total body weight loss (TBWL) of 7 (6-7)%, followed by an additional 27 (25-28)% TBWL after surgery (week 7-59) ( Table 1). The lifestyle-group had a TBWL of 10 (8-12)% at 59-week follow-up. The participants in trial 2 (lifestyle-cohort) had a TBWL of 9 (8-10)% at 20 weeks follow-up. The serum levels of triglycerides, fasting glucose and C-reactive protein decreased signi cantly over time in both groups in trial 1 and also in trial 2 (Table 1). Serum total 10 cholesterol and LDL cholesterol remained unchanged in the lifestyle-group in trial 1, but decreased signi cantly after surgery in trial 1 and also in trial 2. High density lipoprotein (HDL) cholesterol levels increased in both groups in trial 1 and also in trial 2. Lipoprotein (a) In the surgery-group, the median (IQR) concentration of Lp(a) tended to increase during the 7-week pre-surgery LED-phase [from 17 (7-68) to 21 (7-81) nmol/L, p=0.05], but were decreased by 48% after surgery [from 21 (7-81) to 11 (7-56) nmol/L, p<0.001] at week 59 (Table 1).
There was also a signi cant 35% decrease in plasma Lp ( week follow-up. Changes in Lp(a) levels from baseline to 20 weeks for the participants in trial 2 are presented in Figure   2B.
There was no signi cant association between change in Lp(a) levels and change in body weight in trial 2 (data not shown). Fatty acids Saturated fatty acids In trial 1, plasma levels of total saturated FAs did not change after surgery (week 7-59), but decreased slightly after the 59-week lifestyle-intervention (Table 2) FAs changed signi cantly more after surgery than after the lifestyle intervention (Table 3). During the pre-surgical LED phase in trial 1 and during the lifestyle-intervention in trial 2, plasma levels of all saturated FAs decreased or remained unchanged (Table 2). Monounsaturated fatty acids In trial 1, plasma levels of total monounsaturated FAs, mainly oleic acid (C18:1 n-9), did not change after surgery (week 7-59) or after the 59-week lifestyle-intervention, while plasma levels https://app.ithenticate.com/en_us/report/92826333/similarity 7/13 of palmitoleic acid (C16:1 n-7) decreased and eicosenoic acid (C20:1 n-9) levels increased in both groups (Table 2).
Vaccenic acid (C18:1 n-7 cis) levels decreased after surgery and remained unchanged in the lifestyle-group, resulting in a signi cant between-group difference (Table 3). During the pre-surgical LED phase in trial 1 and during the lifestyleintervention in trial 2 (lifestyle-cohort), plasma levels of total monounsaturated FAs, oleic acid and eicosenoic acid did not change during follow-up, while plasma levels of palmitoleic acid decreased (Table 2). Vaccenic acid levels increased during the pre-surgery LED phase, but did not change during follow-up in trial 2. Polyunsaturated fatty acids n-6 fatty acids In trial 1, plasma levels of total n-6 FAs did not change after surgery (week 7-59) or after the 59-week lifestyleintervention (Table 2). By contrast, plasma levels of linoleic acid (LA; C18:2 n-6), eicosadienoic acid (EDA; C20:2 n-6) and dihomo-gamma-linolenic acid (DGLA; C20:3 n-6) increased after surgery, but did not change in the lifestyle-group.
Gamma-linolenic acid (GLA; C18:3 n-6) increased in the surgery-group and decreased in the lifestyle-group, while arachidonic acid (AA; C20:4 n-6) levels decreased in the surgery-group and increased in the lifestyle-group. Plasma levels of GLA, EDA, DGLA and AA changed more after surgery than in the lifestyle group (Table 3). During the presurgical LED-phase in trial 1 and during the lifestyle-intervention in trial 2 (lifestyle-cohort), plasma levels of total n-6 FAs and AA increased, while DGLA decreased (Table 2). Plasma levels of LA and EDA remained unchanged in the presurgical LED-phase, but increased in trail 2, while GLA levels decreased in the pre-surgical LED-phase, and did not change in trail 2. n-3 fatty acids In trial 1, plasma levels of total n-3 FAs, eicosapentaenoic acid (EPA; C20:5 n-3) and docosahexaenoic acid (DHA; C22:6 n-3) decreased after surgery (week 7-59), but increased in the lifestyle-group (Table   2). Alpha linolenic acid (ALA; C18:3 n-3) levels did not change during follow-up in the surgery-group, but decreased in the lifestyle-group, while docosapentaenoic (DPA; C22:5 n-3) levels increased in both groups (Table 2). Between-group differences in change were signi cant for ALA, DPA, EPA and DHA (Table 3) (27). Also, in a cohort overlapping with the trial 2 cohort, Lp(a) increased in patients with overweight, with and without T2D, undergoing a calorie restricted diet for 3-4 months (7).
However, other studies have showed no

change in Lp(a) levels after various dietary interventions aimed at weight loss
(28-30). Bariatric surgery and lifestyle-interventions did also differently in uence FA levels. After bariatric surgery, plasma levels of total saturated FAs did not change form baseline, however plasma levels of palmitic acid decreased, while myristic acid, pentadecylic acid and stearic acid increased. The increased proportions of the saturated FAs myristic acid, pentadecylic acid and stearic acid following RYGB have previously been shown(31). Total plasma saturated FAs decreased after the lifestyle interventions, which was mainly due to a reduction in palmitic acid. This nding is in accordance with results from a previous 12-week randomized controlled trial comparing mild-calorierestriction (minus 300 kcal/day) with a control diet(15). Calorie restricted diets typically include low levels of total fat and especially saturated fats, as was also the case for the participants in the lifestyle intervention groups in this study.
However, dietary intake of saturated and also mono-unsaturated fats may not necessarily correlate with plasma levels as these dietary FAs are endogenously synthesized and remodeled(32). Plasma levels of the polyunsaturated FAs, on the other hand, correlate more strongly with dietary intake, and may better re ect dietary intake. Plasma levels of a number of n-3 FAs as well as the n-6 FA AA decreased after surgery, but increased during lifestyle interventions.
Previous studies on the effect of calorie restriction and bariatric surgery on polyunsaturated FA levels showed somewhat con icting results. In patients undergoing RYGB, the proportions of circulating n-3(33) and n-6 FAs (16) increased from baseline to 1 year after surgery. In contrast, among 13 women undergoing RYGB, phospholipid FA composition was similar to baseline levels at 6 months post-surgery, except for a decrease in content of EPA (17 Bile acids also act as FXR agonists. Although bile acids were not measured in this study, previous studies have shown that circulating levels of bile acids are increased after RYGB(39, 40), which may partly explain the lowering of Lp(a) levels among the surgical patients. Interestingly, bile acid synthesis and levels have been shown to be increased in women with obesity, and to be normalized within 3 days on a caloric restriction diet(41). 18 Diet-induced lowering of bile acid production may therefore partly explain the increased Lp(a) levels observed during the lifestyle interventions.
Future studies should determine whether the differential effects of surgery and lifestyle intervention on Lp(a) are mediated by changes in bile acid levels. Furthermore, one could speculate that exercise may have in uenced the observed change in Lp(a) levels. A recent review reported that results from studies on the effect of exercise on Lp(a) levels have been inconsistent, with some reporting no effect while others have reported mildly increased or decreased levels(42). However, studies among younger individuals or patients with diabetes, showed more moderate Lp(a)lowering effects by exercise. In trial 1, the participants underwent a physical activity intervention, and 1 1 the majority of the patients reported that they completed >3 hours of light physical activity per week and >3 hours of vigorous physical activity per week during follow-up . Participants in the surgery group did not follow an exercise program prior to or following surgery. Exercise was also encouraged in trial 2, but the amount of physical activity performed did not change signi cantly from baseline. As Lp (a) increased during all lifestyle interventions even though only participants in the lifestyle-group of trail 1 followed an exercise program, it is less likely that the observed increase in Lp(a) was caused by exercise. Strenghts and limitations The strengths of the present study are its prospective design and the use of two independent trials with a relatively high number of patients with detailed analyses on both plasma FAs and Lp(a). Limitations include the non-randomized design, and plasma levels of Lp(a) and FAs being exploratory endpoints in both trials. Further, as the participants had been referred to a tertiary care center, these ndings may not be generalized to all individuals with overweight and obesity. Of note, plasma Lp(a) levels were higher in trial 2 compared with trial 1, which may be partly explained by differences in analytical methods between the trials. Plasma samples from trial 2 were measured using a particle-enhanced immunoturbidimetric method 19 by DiaSys Diagnostic System, but also later reanalyzed in a sub-group of patients using Roche Diagnostics, as applied in trial 1. The median Lp(a) value in trial 2 was 13% higher using the DiaSys Diagnostic System versus using the Roche Diagnostics method. Repeated freezing/thawing cycles may in uence Lp(a) levels in samples (43). However, the plasma samples were only frozen and thawed once before Lp(a) analyses in both trials, thus this is likely not an issue here. Another possible explanation may be differences in ethnicities between the trials. More than 55% of the participants in trial 2 were of non-White ethnicity, whereas 98% of the participants in trial 1 were White. Lp(a) levels are reported to vary across ethnicities, and people of non-White ethnicities are reported to have higher Lp(a) levels compared with those of White ethnicity(42). All participants in trial 2 had T2D compared with only 20% in trial 1. Previous studies have shown con icting results regarding whether patients with T2D having higher or lower plasma Lp(a) levels than patients without T2D(7, 44, 45).
Also, the participants in trial 2 were older than the participants in trial 1, and some studies suggest that Lp(a) increases with age, but the results are con icting(46-50). Polyunsaturated FAs are also susceptible to degradation through freezing/thawing cycles. In trial 2, the samples were frozen only once prior to the FA analysis, while in trial, the samples were frozen twice before analysis. There is thus a possibility that there may have been some degradation of the 188 189 190 191 192 193 194 195 196 197 198 199