The rapid growth of obesity in recent decades is a significant challenge for public health policies. According to the World Health Organization, approximately 650 million people older than 18 years were obese in 2016, equivalent to more than 13% of the world population.¹

Although an imbalance between caloric intake and energy expenditure is a fundamental aspect of the genesis of obesity, environmental, dietary, socioeconomic, genetic, psychological, metabolic, neuroendocrine, and lifestyle factors are also involved in this process.^2,3

In individuals with extreme obesity (body mass index [BMI] >40 kg/m²) and grade II obesity (BMI >35 kg/m²), which are associated with metabolic complications, surgical treatment is indicated and has been shown to be the most effective of the existing strategies; in other words, it promotes the most significant weight loss in the short and long term and provides early and significant improvement in metabolic alterations and obesity-related diseases.^2,4,5

The Roux-en-Y gastric bypass (RYGB) is one of the main surgical techniques for treating obesity.⁵ RYGB reduces the gastric volume by 90% to 95%, creating a small bag next to the lesser curvature with a capacity of approximately 30 mL. Then, the small intestine is segmented 30 to 50 cm distal to the angle of Treitz, and the distal jejunum is anastomosed to the gastric pouch, forming the Roux canal and excluding the rest of the stomach, duodenum, and part of the jejunum from the alimentary loop.⁵

Maximum weight loss after RYGB generally occurs in the first 12–24 months and can reach up to 70% of excess weight in the first year.^5,6 According to the Brazilian Association for the Study of Obesity and Metabolic Syndrome, surgical treatment is regarded as successful when more than 50% of excess weight loss (EWL) occurs, and the patient is no longer morbidly obese.⁶ There is no consensus on the best criterion for defining a successful response to bariatric surgery (BS),^7-10 but in a recent analysis, an EWL >50% demonstrated good sensitivity and specificity as a criterion for weight loss success from different bariatric techniques.⁷

Compared with clinical treatment and other surgical techniques, RYGB presents the best results in terms of weight loss and maintenance in the short and long term.^4,5,11 Nonetheless, 4% to 38% of patients undergoing RYGB surgery do not achieve successful weight loss (SWL) in the first 2 postoperative years.⁷

Several factors have been suggested as predictors of worse weight loss after RYGB: higher age and BMI, preoperative comorbidities, changes in taste and food reward mechanisms, and the presence of some genetic polymorphisms, particularly mutations in the brain-derived neurotrophic factor (BDNF) and lysophospholipase-like1 (LYPLAL1) genes.^12-21 For some of those factors, further studies are needed to elucidate their predictive effect. Therefore, we evaluated some predictive components of SWL 12 and 24 months after RYGB.

Study population

This retrospective longitudinal study analyzed individuals who underwent RYGB between 2004 and 2014 and were followed up by the multidisciplinary team at the Nutritional Therapy Outpatient Clinic for Extreme Obesity at the Hospital das Clínicas of the Federal University of Minas Gerais (HC-UFMG). Patient clinical and anthropometric parameters, family history, physical activity (PA), and food consumption were collected from medical records collected at the first consultation with the team (presurgical period), the immediate preoperative period, and the follow-up consultations at 12 and 24 months postoperatively.

The inclusion criteria were age between 18 and 65 years, follow-up at the clinic for at least 2 years postoperatively, no use of anti-obesity medication after surgery, no additional surgery during the follow-up period, and having DNA samples extracted from oral swabs and stored in the DNA bank of the laboratory of biochemistry and nutritional immunology at the Institute of Biological Sciences at UFMG. Patients with missing or incomplete data on the first consultation and those whose weight was not recorded both 12 and 24 months after surgery were excluded from the analysis. The initial sample was n=74. After applying the exclusion criteria, 63 patients were included in the study.

This study was approved by the Research Ethics Committee (COEP in Portuguese) of UFMG under technical report 5.183.414 (CAAE: 28392619.9.0000.5149). The data and biological materials analyzed came from research approved by COEP/UFMG under technical report 1.077.885 (CAAE: 13227713.0.0000.5149), and the terms of free and informed consent were collected from all individuals included in the study.

Data collection

Anthropometry

Weight and height were measured using a digital platform scale (Welmy) with a capacity of 300 kg and a coupled vertical millimeter anthropometer, and BMI was calculated using the weight (kg)/height² (m) equation. Total weight loss (TWL) was determined as the difference in weight between the two postoperative moments and the initial weight. The percent of total weight loss (%TWL) was determined using the equation %TWL=[(initial weight)–(postoperative weight)]/[(initial weight)]×100. Excess weight was calculated considering a BMI of 25 kg/m² as ideal, and the %EWL was determined using the equation %EWL=[(initial weight–postoperative weight)/(initial weight–ideal weight)]×100. According to the %EWL obtained after 12 and 24 months, participants were allocated into the SWL group (%EWL >50) and the insufficient weight loss (IWL) group (%EWL <50) (Fig. 1).

Physical activity

Data on the PA were collected at 12 and 24 months postoperatively and included modality, frequency, and length of exercise. Based on this information, the physical activity level (PAL) was classified in an adapted way using the criteria established by the short version of the International Physical Activity Questionnaire validated in Brazil by Matsudo et al.²²

Energy and macronutrient intake

Dietary intake was assessed using a 24-hour recall (24hR) collected during each consultation and recorded in the medical records as home measurements. For the analysis, household measurements were converted into grams or milliliters. The total intake of energy, carbohydrates, total fat, and protein and the percentage of consumption of each macronutrient were calculated using Avanutri software version 4.1 (Avanutri Co.).

Genotypic analysis

For genotyping, the stored DNA samples were quantified using a nanodrop and adjusted to a concentration of 50 ng/μL by dilution in Milli-Q water (Merck). Samples with insufficient DNA volume or a concentration below 15 ng/μL were excluded from the analysis.

The Val66Met (rs6265–BDNF) and rs4846567 LYPLAL1 polymorphisms were genotyped using the TaqMan (Thermo Fisher Scientific) single nucleotide polymorphism genotyping methodology, with two probes labeled with differential fluorescence to allow the detection of both alleles in a single reaction (Applied Biosystems Inc. [ABI]). The analyses were performed on an ABI 7900 HT fast polymerase chain reaction real-time system (ABI), and genotyping was conducted using TaqMan Genotyper Software 84 according to the manufacturer’s instructions.

Statistical analysis

Quantitative variables are expressed as the median (interquartile range [IQR]), and categorical variables are presented as the frequency. To compare groups, the Wilcoxon test was used to analyze quantitative variables, and Fisher’s exact test or the chi-square test was used for categorical variables. The allele frequency of the analyzed genes was calculated by allele count, and deviation from the Hardy-Weinberg equilibrium (HWE) was tested using Fisher’s exact test. Logistic regression models were used to verify variables predictive of surgery success. Univariate regression analyses were adjusted by incorporating all collected variables with a P-value less than 0.2 in the descriptive analysis. Independent variables with P<0.2 in the univariate model were included in the multivariate analysis. The significance level adopted was P<0.05, and all statistical analyses were computed using the R program version 4.0.5 (R Foundation for Statistical Computing).

The demographic, clinical, anthropometric, and energy and macronutrient intake characteristics of the participants at the time of surgery are shown in Table 1. In the studied population, the average age was 47 years (IQR, 38 to 54), and of the 63 participants included, 79% were female, and 73% had a family history of obesity. In the evaluation of comorbidities associated with obesity, those with the highest prevalence were hypertension (84%), type 2 diabetes mellitus (T2DM; 43%), dyslipidemia (40%), and apnea or dyspnea (38%).

Among the anthropometric characteristics before surgery, the median BMI, weight, and amount overweight were 53 kg/m² (IQR, 46 to 58), 137 kg (IQR, 114 to 156), and 72 kg (IQR, 51 to 88), respectively. In the analysis of food consumption, the average caloric intake was 1,543 kcal (IQR, 1,125 to 2,110), with 17.6% (IQR, 13.8 to 19.3) coming from protein, 58% (IQR, 51 to 65) from carbohydrates, and 26% (IQR, 20 to 32) from lipids.

Weight loss response after surgery

The postoperative anthropometric and dietary characteristics, comorbidity frequencies, and PALs of the participants are shown in Table 2. Twelve months after surgery, the medians of BMI, EWL, and TWL were 35 kg/m² (IQR, 31 to 39), 63% (IQR, 53 to 73), and 44 kg (IQR, 35 to 54), respectively. The median daily caloric intake was 864 kcal (IQR, 718 to 1,072), with 19.6% (IQR, 15.0 to 23.2) of the total caloric value coming from proteins, 54% (IQR, 48 to 59) from carbohydrates, and 26% (IQR, 20 to 33) from lipids. In the assessment of PA practice, 70% of the population had low PAL, 29% had moderate PAL, and 1% had high PAL.

Of the 63 participants, 12 (19%) had EWL <50% at 12 months postoperatively and were placed in the IWL group, and the remaining 51 composed the SWL group. The characteristics of each group in this period are shown in Table 1. The EWL in the SWL and IWL groups was 66% (IQR, 58 to 78) and 41% (IQR, 38 to 44), respectively, and the %TWL was 35 (IQR, 29 to 40) and 25 (IQR, 22 to 29), respectively.

In comparing the SWL and IWL groups after 12 months, significant differences were observed in preoperative weight (SWL 132 kg [IQR, 112 to 144]; IWL 167 kg [IQR, 139 to 197]; P=0.009), BMI (SWL 51 kg/m² [IQR, 45 to 56]; IWL 65 kg/m² [IQR, 57 to 70]; P=0.004), and excess weight (SWL 68 kg [IQR, 50 to 80]; IWL 106 kg [IQR,80 to 127]; P=0.008), with the highest medians observed in individuals with IWL. In the TWL analyses, the IWL and SWL groups did not differ significantly in absolute values (P>0.05), but higher values were observed in the SWL group (P<0.001) in the percent analysis. In the analysis of preoperative comorbidities, family history of obesity, PAL, and energy and macronutrient intake, the groups did not differ significantly (P>0.05).

After 12 months, different weight loss trajectories were observed, leading to a failure rate of 16% (n=10) at 24 months. Of the 12 participants with IWL in the first year, four continued to lose weight and achieved an EWL >50% by 24 months. Conversely, two participants in the SWL group experienced weight relapse and were thus included in the IWL group at 24 months.

The characteristics of each group 24 months after surgery are shown in Table 1. A significantly high percentage of individuals with SWL at 12 months was also in the SWL group at 24 months (SWL 92%, IWL 20%; P<0.001), demonstrating that participants with successful EWL in the first year generally maintained a good response to surgical treatment. Differences between the groups in TWL at 12 months were analyzed, but they were not significant (SWL 47 kg [IQR, 36 to 56]; IWL 36 kg [IQR, 25 to 49]; P=0.089). Furthermore, none of the anthropometric measurements, energy, macronutrient intake, frequency of comorbidities before surgery, or PAL differed significantly between groups (P>0.05).

Genetic aspects

The frequencies of the Val66Met (rs6265–BDNF) and rs4846567 (LYPLAL1) polymorphisms 12 and 24 months after surgery are shown in Table 3. For the Val66Met variant of BDNF, a low frequency of carriers of the mutated allele (GA and AA) was found in the sample (9.5%), with no significant difference between the groups in either of the analyzed periods (P>0.05). For LYPLAL1 rs4846567, the frequency of the mutated T allele carriers (GT+TT) was 38%, with no significant difference between groups in either of the periods analyzed (P>0.05).

Because previous studies demonstrated sexual dimorphism in the anthropometric characteristics of individuals with the rs4846567 variant, the hypothesis of a possible effect only in females was tested by analyzing only females (n=50). In this subpopulation, the general frequency of the polymorphism (GT+TT) was 42%, but the groups still did not differ in the two postoperative periods analyzed (P>0.05).

Factors associated with successful surgery

The results of the univariate and multivariate analyses at 12 and 24 months after surgery are shown in Tables 4 and 5. Among the studied variables initially associated with the success of surgery, a high initial BMI remained in the multivariate model as an independent predictor of the worst response in the two evaluated moments (Table 4). On the other hand, high TWL at 12 months predicted a better response at 24 months after surgery (Table 5).

SWL and weight maintenance after BS are of great importance for patient quality of life, providing improvement in metabolic complications and recovery of physical capacity, in addition to positively affecting social interactions, vitality, and the psychosocial aspects of obesity.²³

Although RYGB is an effective and successful treatment for most patients with severe obesity, variability in the trajectory of weight loss after surgery has been extensively reported in the literature, and a substantial number of patients do not obtain a successful response.²⁴ The consequences of unsuccessful weight loss during the first postoperative years include a greater probability of non-resolution, relapse, and incidence of comorbidities, impaired physical function and social relationships, reduced self-esteem, and increased concern with the body and the physical form.^25,26

In this retrospective longitudinal study, more than 80% of patients had SWL during the first and second postoperative years. The failure rates in our population (19% and 16% at 12 and 24 months, respectively) are in line with the results of other studies using the same criteria and cutoff point (%EWL <50), which demonstrated a failure rate between 9 and 26% in the first 2 years.^27-29

The relationship between some preoperative factors, such as the presence of comorbidities, BMI, food intake, PAL, and genetic polymorphisms, and the success of BS has received great attention in the literature, but few studies have evaluated all these factors simultaneously.

This study found that preoperative BMI was the principal predictor of the weight loss response during the first 2 years after surgery. Interestingly, we also found that more significant TWL 12 months after surgery was associated with a greater chance of EWL success after 2 years, suggesting that TWL at 12 months might be an early indicator of weight-related surgical outcomes.

Among the preoperative aspects associated with weight loss, a higher initial BMI has consistently been shown to predict a lower chance of successful EWL after BS.^14,21,27,30 Thus, greater attention to follow-up care is needed for these patients.

A negative correlation between initial BMI and %EWL (r=–0.28, P<0.001) was demonstrated by Al-Khyatt et al.²⁷ in 227 individuals who underwent RYGB and were followed up for 12 months. In that study, which analyzed three BMI ranges (<50, 50–60, and >60 kg/m²), individuals with values above BMI >60 kg/m² obtained the worst EWL results. A similar result was found by Sillén and Andersson,¹⁴ who analyzed 218 patients: a high initial BMI predicted a lower chance of SWL 12 and 24 months after surgery (odds ratio, 0.8; P<0.001).

Other presurgical factors have also been reported to influence weight loss trajectory. In the study by Al-Khyatt et al.²⁷, age and the presence of T2DM and hypertension were also associated with a worse response in terms of weight loss at 12 months. We did not observe that association in this study, and our results of no association between age or presurgical clinical comorbidities and the success of the surgery at 12 and 24 months are in line with those reported by Sillén and Andersson.¹⁴

EWL during the first postoperative year has also been shown to predict the likelihood of SWL in subsequent years. Ritz et al.¹³ reported different evolution profiles of %EWL over time in patients who, at 24 months postoperatively, had EWL greater than 50%, between 25% and 50%, and below 25%. In general, the results indicate that individuals with a higher %EWL at 6 months were more likely to progress and reach SWL by 24 months. That observation is corroborated by the results of this study. In our univariate analysis, %EWL >50 at 12 months predicted success at 24 months postoperatively (P<0.001). In the multivariate model, however, that relationship did not remain significant, and TWL at 12 months proved to be the best predictor of SWL at 24 months (P=0.008).

The dietary intake patterns of patients undergoing RYGB change significantly after surgery. Some explanations for those changes are the restriction of ingested volume, the high frequency of food intolerances and aversions, changes in taste, and a reduction in appetite that results from changes in the secretion of intestinal hormones such as glucagon-like peptide 1 (GLP-1) and peptide YY (PYY).^31-33

In a sample of 41 individuals with severe obesity who underwent RYGB, Molin Netto et al.³⁴ demonstrated significant changes in eating patterns and food choices 6 months after surgery. In that study, the authors found a significant reduction (P<0.05) in the intake of calories, lipids, lean meats, chicken breast, vegetables, fruits, and whole milk, as well as in the consumption of unhealthy foods such as soft drinks, sausages, hot dogs, hamburgers, pizza, breaded chicken, dulce de leche, chocolates, and truffles. At the same time, an increase in the intake of fish, yogurt, and natural juices was observed.

Because the intake of calories and nutrients can influence energy balance and thus body weight, it is assumed that food intake after surgery could affect the success of weight loss. Furtado et al.³⁵ compared food intake and PALs in individuals with weight loss success (EWL >50%) and failure (EWL <50%) 24 months after surgery. Using the 24hR method to assess dietary intake, the authors found an average caloric intake of 1,437.7 kcal in the success group (20% proteins, 52.36% carbohydrates, and 27.64% lipids) and an average intake of 1,299.9 kcal (18.8% proteins, 55.36% carbohydrates, and 25.29% lipids) in the failure group, with no significant differences between the groups. Similarly, we found no significant association between nutrient intake and surgery success (P>0.05 in the two postoperative periods analyzed).

Some factors could explain this lack of association. Although the 24hR is a simple, quick, and easy-to-apply dietary survey, as a single-day report, it might not represent the individual’s usual intake.³⁶ In addition, underreporting and recent changes in the quality and quantity of food consumed should be considered when analyzing the results. In a systematic review, Wehling and Lusher³⁷ confirmed that the frequency of underreporting of food consumption is significantly high among individuals with obesity, and some foods might deliberately go unreported during the interview, especially in situations of loss of control over eating.

PAL has a significant role in controlling body weight and is one of the strategies recommended for obesity treatment.² In a retrospective study, Nuijten et al.³⁸ assessed the effect of changes in PA on weight loss 2 years after BS (92% RYGB and 8% sleeve). In that study, habitual PA was evaluated using the Baecke questionnaire, and the participants were divided into three groups after surgery: reduced PA, maintained PA, and increased PA. After 2 years, bariatric patients who presented an increase in habitual PA had a higher %TWL (P<0.001) and %EWL (P<0.001) than those who reduced or maintained their PA.

Regarding the success of the surgery, Amundsen et al.²⁵ demonstrated that 5 years after the operation, the weekly walking time, total PA, and daily duration of PA was significantly lower in the group of participants with %EWL <50 than in those with SWL. Likewise, they observed a longer length of light PA and total daily PA in individuals who showed a weight regain of >15% during the 5 years after surgery.

In neuronal cells, the Met66Val (rs6265) polymorphism, a mutation that changes the valine residue in codon 66 of preproBDNF to a methionine residue, can interrupt cell traffic and the activity-dependent release of BDNF.^39,40 The influence of this variant on the success of BS was analyzed by Peña et al.¹⁹ in 158 individuals treated with the RYGB (n=99, 62.7%) and sleeve (n=59, 37.3%) techniques. In that study, frequencies of 65.82%, 29.11%, and 5.07% were found for the Val/Val, Val/Met, and Met/Met genotypes, respectively. To test their effects on weight loss, a prospective analysis performed by those authors demonstrated an interaction between the presence of the Met allele and T2DM at the time of surgery. Stratifying the genotypes according to the presence of T2DM, a significantly higher %EWL was observed after 24 months in carriers of the Met allele (Val/Met or Met/Met) who did not initially have diabetes (P=0.038).

The influence of the rs4846567 variant close to LYPLAL1 on the outcomes of RYGB was demonstrated by Bandstein et al.¹⁷ In their study, the TT genotype was associated with significant weight loss and less hunger 24 months after surgery. According to those authors, that variant in an intergenic region could lead to lower availability of the LYPLAL1 promoter region, reducing its expression.

In this study, no associations between the rs4846567 and rs626 variants and the analyzed weight loss outcomes were observed. However, it should be emphasized that in our population, the genotypic frequency did not conform to the HWE (P<0.001), a deviation that can be explained by the sample size, one of the limitations of this study.

Other limitations that should also be considered in the analysis of the results presented in this study are the retrospective data collection and the absence of precise instruments for evaluating the practice of physical exercises, food intake, and lifestyle.

Some positive points of this study deserve to be emphasized, such as the simultaneous comparison of multiple factors in patients treated with the same surgical technique and the evaluation of independent predictors of an SWL response.

In conclusion, a high preoperative BMI was an independent predictor of a worse weight loss response in the first 2 years after RYGB, and high TWL at 12 months was a predictor of SWL at 24 months. Regarding the other parameters analyzed, future studies with larger samples are necessary to elucidate the roles of PA, dietary intake, preoperative comorbidities, and the rs6265 and rs4846567 variants on the weight loss responses of individuals undergoing RYGB.

The authors declare no conflict of interest.

We thank the team from the molecular medicine laboratory and the team from the Nutritional Therapy Outpatient Clinic for Extreme Obesity for their collaboration in the development of this work and the student Hortênsia Ribeiro for her support with the food intake analysis.

Study concept and design: SHCM, JIAL, LM, and LBR; acquisition of data: SHCM, GCR, and CMMR; analysis and interpretation of data: SHCM; drafting of the manuscript: SHCM; critical revision of the manuscript: JIAL and LBR; statistical analysis: RPS; obtained funding: LM and LBR; administrative, technical, or material support: LM; and study supervision: JIAL and LBR.