Introduction

Weight regain after hypocaloric diet-induced weight loss has emerged as one of the most significant obstacles in obesity therapy [1, 2]. This fact perpetuates the obesity epidemic that affects more than 30 % of U.S. adults [3] and 23 % of Spanish adults [4]. Diverse therapeutic strategies have been adopted to decrease body adiposity. Nutritional interventions, such as caloric restriction diets, can be an efficient therapeutic approach to promote weight loss in obese patients [5, 6]. Although most therapeutic trials involving drugs [7] or dietary treatments show mean weight losses higher than 5 % of body weight [5], the long-term success of maintaining the weight loss is usually poor [8]. Therefore, the identification of factors associated with weight regain can enhance our understanding regarding energy homeostasis regulation and would have positive implications in the personalization of therapeutic approaches to obesity.

Several biological and psychological factors may contribute to weight regain after a weight loss treatment [9]. There is also evidence that some weight reduction therapies fail due to compensatory metabolic adaptations, which promote rapid and efficient weight regain [1, 8]. Among the mechanisms implicated, orexigenic and anorexigenic hormones appear to have a clinically relevant role [10].

Ghrelin is a circulating orexigenic hormone [11] implicated in both the short-term control of food intake at single meals and long-term body weight regulation [12]. Moreover, this hormone improves the cardiovascular function [13] and it exerts an anti-inflammatory effect that contributes to extend the life span [14]. Leptin is a key anorexigenic hormone that relays information about peripheral adiposity by signaling to the central nervous system to regulate food intake and energy expenditure [15]. There is a complex interaction between leptin and ghrelin, which are reciprocally regulated during the physiological response to food intake, inducing changes in body weight [16].

Nutrigenomic/nutrigenetic and proteomic studies have been performed to identify markers with the potential to predict successful weight maintenance [17–20]. We have previously demonstrated [21] that patients who regained ≥10 % of the lost weight after weight reduction induced by an 8-week-balanced hypocaloric diet (regainers) have higher leptin levels and lower ghrelin levels than those who maintained the lost body weight (non-regainers). Interestingly, this result suggests the existence of two different predetermined populations of patients according to the leptin and ghrelin levels conditioning the outcomes [21]. However, there are no studies that effectively demonstrate clinical usefulness in discriminating between regainers and non-regainers of diet-induced weight loss. In fact, a recent systematic review concluded that little evidence within the reviewed studies support pre- or post-weight loss leptin, ghrelin or insulin sensitivity as predictor of regain, suggesting that these hormones taken alone, are not sufficient predictors of weight regain in free-living humans [22]. Under these evidences, we suggest that the leptin/ghrelin ratio could be a better predictor of weight regain than either leptin or ghrelin alone.

The aims of this study performed in obese patients were the following: (1) to evaluate the ability of the leptin/ghrelin ratio as a non-invasive tool for the early identification of obese patients more likely to regain the lost weight after an energy restriction program and (2) to assess a leptin/ghrelin ratio cut-off that may help to differentiate weight loss regainers from non-regainers.

Methods

Study patients and design

The trial was based on a nutritional intervention monitored by trained dieticians from the Department of Nutrition, Food Science and Physiology of the University of Navarra, as previously described [21]. The protocol used for this study was approved by the Clínica Universidad de Navarra Ethical Committee (Ref. 54/2006) in agreement with the Helsinki Declaration and followed national and EU guidelines.

A group of overweight and obese patients (n = 104) were enrolled in a nutritional weight loss program that occurred over 2 months [21]. At 32 weeks, i.e., 6 months after dieting ended, the patients returned to the clinical research unit for further assessment (Fig. 1). The patients were then categorized as “regainers” or “non-regainers” depending on whether they regained more than 10 % of the lost weight. The circulating leptin/ghrelin ratio (L/G) was evaluated in 48 (29 men/19 women) obese patients that maintained the initial weight reduction (non-regainers) and 40 (20 men/20 women) obese patients that regained the lost weight (regainers). Sixteen patients were excluded due to the lack of completed data for leptin and ghrelin levels.

Fig. 1
figure 1

Schematic representation of the experimental design. Unsuccessful weight loss maintenance was defined as weight loss regain of at least 10 % of the lost weight after the energy restriction treatment. Based on these criteria, 40 patients (20 men/20 women) regained >10 % of the lost weight (regainers), and 48 patients (29 men/19 women) maintained the initial weight reduction (non-regainers). Preliminary results of this work were presented at the 53rd Congress of the Spanish Society of Endocrinology and Nutrition (SEEN; May 2011. Endocrinol Nutr 58:80)

Full size image

Weight loss was induced with a hypocaloric diet designed to produce a 30 % energy restriction (500–600 kcal/day) of the subject’s total energy expenditure. The nutritional intervention program consisted of an 8-week consumption of a balanced hypocaloric diet containing 55 % of the energy supply as carbohydrates, 15 % as protein and 30 % as fat, which was implemented by a food exchange system. The dietary compliance was assessed through 3-day weighed food records [23]. Upon completion of the dietary intervention (week 8), the patients were only instructed on a healthy lifestyle such as dietary intake and physical activity [21].

Anthropometric measurements were performed according to standardized procedures at baseline (week 0), the endpoint (week 8) and the follow-up period (week 32). Venous blood samples were collected after a 12-h overnight fast at the beginning of the restriction diet (baseline; week 0) and at the end of the diet intervention (endpoint; week 8). The EDTA-plasma and serum of patients were separated from whole blood and immediately frozen at −80 °C until assayed.

Calculation of the leptin/ghrelin ratio

Plasmatic ghrelin levels were measured with a radioimmunoassay (RIA) using a commercially available kit (Linco Research Inc., St Charles, MO, USA). This assay detects both acylated and deacylated ghrelin. The serum concentrations of leptin were assessed using a radioimmunoassay-based method (DPC, Los Angeles, CA, USA). To express the L/G ratio in arbitrary units, it was calculated as leptin in ng/mL multiplied by 103 and divided by ghrelin in pg/mL.

Because the function of these hormones is mediated by a molecule-receptor interaction, the L/G ratio was also studied on a molar basis to further evaluate its implication in weight loss regain at a more physiological level. Thus, the number of leptin and ghrelin molecules was determined through Avogadro’s constant (N A = 6.022 × 1023 mol−1).

Statistical analysis

The sample size of this interventional trial was estimated accounting for the weight loss after treatment (main variable) and calculated according to the equation reported by Mera et al. [24]. Thus, to detect differences, the sample size was established at a minimum of 85 obese/overweight patients who finished the nutritional intervention.

Weight regain was calculated (in percentage) as the difference in body weight between the follow-up period (week 32) and the endpoint (week 8) and related to the endpoint. Successful weight loss maintenance was considered with the criterion of 10 % weight regain, according to a previously published report [21].

The normal distribution of variables was performed using the Kolmogorov–Smirnov and the Shapiro–Wilk tests. Accordingly, the parametric Student’s t or nonparametric Mann–Whitney U test was applied to detect differences between regainers and non-regainers. The Spearman coefficient was used to evaluate the potential association between body weight regain and the leptin/ghrelin ratio (L/G). A logistic regression analysis was applied to assess the potential predictive capacity of the L/G ratio in weight loss regain.

The discriminatory ability of the L/G ratio at baseline (L/GB) and after 8 weeks of nutritional treatment (L/G8W) was estimated between regainers and non-regainers with the area under the receiver operating characteristic (ROC) curve (AUC) [25]. The AUCs of the L/GB and L/G8W differences were assayed according to the method of comparing the areas under ROC curves described by Hanley and McNeil [26].

Using the ROC curve analysis, the L/G cut-off value was defined as the level that yielded the maximum sensitivity and specificity for predicting weight regain propensity. Additionally, the diagnostic efficiency (percent of correctly classified) of the L/G ratio was calculated as the percent of agreement. These results are often interpreted as negligible efficiency (<20 %), minimal (20–40 %), moderate (41–60 %), good (61–80 %) and excellent (>80 %) [27]. The guidelines for the publication of studies of the diagnostic accuracy of medical tests were followed [28, 29].

Statistical analysis was performed with SPSS 15.0 software (SPSS Inc., IL, USA) for Windows XP (Microsoft, WA, USA). Results are presented as mean ± standard deviation and a p ≤ 0.05 was considered statistically significant.

Results

The analysis of the clinical characteristics of the patients categorized according to the 10 % of weight regain classification criterion showed no significant differences in age, body weight, body fat mass or waist circumference between regainer and non-regainer patients at baseline (Table 1). Significant differences were found between both groups for the BMI in women, the body weight decrease induced by the dietary treatment in men, and the weight loss regain in both genders, according to the study design.

Table 1 Pre-treatment clinical and biochemical features of non-regainer (n = 48) and regainer (n = 40) patients according to the 10 % of weight regain classification criterion at week 32
Full size table

As previously described [21], baseline leptin levels were statistically higher in regainer women and baseline ghrelin levels were statistically lower in regainer men compared with the same genders in the non-regainer group (Table 1). When the leptin/ghrelin ratio (L/G) was calculated, both the L/G ratios at baseline (L/GB) and after 8 weeks of hypocaloric treatment (L/G8W) were approximately twofold higher in regainers than non-regainers for both women and men (Fig. 2). On a molar basis, the L/G ratio difference was maintained at approximately twofold higher in regainers than non-regainers (p < 0.05). In fact, for women, the L/GB was 5.44 ± 2.49 for non-regainers vs. 9.03 ± 6.81 for regainers, and for men, the ratio was 3.36 ± 2.22 for non-regainers vs. 5.67 ± 4.44 for regainers. The L/G8W for female non-regainers was 4.03 ± 4.05 vs. 8.03 ± 7.87 for regainers, and in men, the ratio was 1.84 ± 3.00 for non-regainers vs. 2.88 ± 1.97 for regainers.

Fig. 2
figure 2

Box and whiskers plots comparing the circulating leptin/ghrelin ratio (L/G) between the non-regainer and the regainer group at baseline and after 8 weeks of an energy restriction treatment in men and women. The lower and upper end of the box represents the 25th and 75th percentile of the distribution, respectively. The center line inside the boxes indicates the median or 50th percentile, and the whiskers show the interquartile range. The L/G values are shown as leptin (ng/mL × 103)/ghrelin (pg/mL). a.u denotes arbitrary units

Full size image

Interestingly, a correlation analysis showed that higher weight regain (%) was associated with higher values of L/GB (r = 0.30; p = 0.005) and L/G8W (r = 0.43; p < 0.001) (Fig. 3).

Fig. 3
figure 3

Association between the leptin/ghrelin ratios at baseline (L/GB) and after 8 weeks of energy restriction treatment (L/G8W) with body weight regain 6 months after finishing dietary intervention (32 weeks). The patients who presented higher values of the L/G ratio at baseline and after energy restriction treatment regained the lost body weight in the 6 months of follow-up. a.u denotes arbitrary units

Full size image

To further investigate the potential predictive ability of the L/G ratio in weight regain, a logistic regression analysis was performed. Thus, after adjusting for age, gender and body weight at baseline, L/GB appeared to be a potential independent predictor of weight regain (odds ratio 1.051 (95 % CI 1.013–1.091); p = 0.008), whereas a non-statistically significant tendency was observed for L/G8W (odds ratio 1.031 (95 % CI 0.998–1.065); p = 0.067).

Receiver operating characteristic curves were used to evaluate the diagnostic power of L/G to discriminate regainer from non-regainer patients. The area under the ROC curve (AUC) of L/GB was 0.69 (95 % CI 0.52–0.86; p = 0.040) for women and 0.68 (95 % CI 0.53–0.84; p = 0.030) for men (Fig. 4). According to the combination of maximum sensitivity and specificity calculated with the ROC curve, the optimal cut-off value for L/GB was 26.2 units for women and 9.5 units for men (Fig. 4). Using this optimal L/GB cut-off value, the positive (PPV) and negative predictive values (NPV) were approximately 70 % in women. For men, the optimal L/GB cut-off value was associated with a PPV and NPV of 50 and 90 %, respectively. To further validate the usefulness of the L/G ratio as a novel marker of weight regain predisposition, the patients included in the initial sample of the current study were reclassified. In this analysis, 70 % of women and 60 % of men were correctly classified (efficiency) by the L/GB ratio (Fig. 4).

Fig. 4
figure 4

Receiver operating characteristic (ROC) curves for the circulating leptin/ghrelin ratio before (L/GB) calorie restriction treatment to discriminate between regainer and non-regainer women and men. The arrows indicate the location of the selected cut-off point with the best sensitivity and specificity. The data of sensitivity (S), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV) and the proportion of patients correctly classified (efficiency) are shown

Full size image

The circulating L/G8w also showed the capacity to discriminate between regainers and non-regainers in women (AUC = 0.73 (95 % CI 0.58–0.89); p = 0.012) and men (AUC = 0.77 (95 % CI 0.63–0.91); p = 0.002). However, although the AUC for L/G8w was slightly higher than the L/GB, this difference was not statistically significant in women (z = 0.353; p > 0.05) or men (z = 0.792; p > 0.05).

Discussion

This study demonstrates the clinical value of the circulating leptin/ghrelin ratio as a novel biological marker to discriminate between overweight/obese patients with a high predisposition to significantly regain weight after an energy restriction treatment and patients able to maintain the lost weight. Thus, a leptin/ghrelin ratio value before treatment higher than 26.2 units for women and 9.5 units for men was able to detect the majority of patients more likely to regain the lost weight. Thus, the leptin/ghrelin ratio appears to be a useful, non-invasive tool to avoid the efforts of both patients and health professionals in following a hypocaloric diet in those patients with a high probability of regaining weight.

Identifying robust biological markers to guide appropriate therapy and thereby improve clinical outcomes is an important task [29]. Classically, markers have been used as a clinical tool to facilitate the diagnosis, classification, prognostication and therapy of several diseases [30–32]. In obesity therapy, the prevention of weight regain after a nutritional intervention to induce weight loss is the most clinically relevant issue [1]. Observations of regainers and non-regainers in clinical studies focused on weight loss therapies are common. This important weight loss variability among individuals is due to several factors, including metabolic adaptations [33] and genetic predisposition [17, 34]. In this regard, several markers have been suggested as predictors of weight regain, including nutrigenomic/nutrigenetic [17, 18, 20] or proteomic [19] approaches. However, as far as we know, the current study shows for the first time the clinical usefulness of a marker to detect, at an early time, patients who are more likely to regain the diet-induced weight loss.

The most basic requirement for a novel risk marker is that the association between the marker and the outcome of interest be statistically significant when tested as a predictor of future events [29]. In fact, it was previously described that high leptin and low ghrelin levels were associated with weight loss regain [21]. These results are counterintuitive because ghrelin is a potent orexigenic signal and leptin is a satiety signal, suggesting a disruption in the sensitivity to these hormone signals, probably in the central nervous system of those subjects with a higher predisposition to regain body weight [21] which even could be determined by an epigenetic regulation [35]. Interestingly, this association was strengthened in the present study when the L/G ratio was evaluated, which was found to be associated with the percentage of weight regain. Those patients who regained the lost weight showed an L/G ratio approximately twofold higher than the weight maintenance group in both women and men. Reinforcing these results, this difference between regainers and non-regainers was conserved when the L/G ratio was evaluated with leptin and ghrelin on a molar basis.

The inverse, i.e., the G/L ratio was also evaluated, but the differences between regainers and non-regainers were not statistically significant (data not shown), and therefore the G/L was not employed. Since leptin levels are increased with increased body weight, it could be hypothesized that the lack of statistical significance in G/L ratio contrary to the L/G could suggests that the leptin resistance could be the main compound of the weight regain predisposition rather than the compensatory low levels of ghrelin observed in a state of excess body weight.

Based on these results, the diagnostic accuracy of the L/GB ratio for the detection of regain in women and men was demonstrated. Moreover, the leptin/ghrelin ratio after 8 weeks of energy restriction treatment (L/G8w) showed a similar ability to detect failure in weight loss maintenance. Leptin and ghrelin levels were also different between the two weight maintenance groups, but the diagnostic accuracy of these hormones evaluated separately showed no individual power to differentiate between both groups (data not shown).

Since the laboratory test provided good discrimination, the next step was to determine a cut-off point for clinical use [36]. In the current study, the optimal cut-off point, defined as the level with a maximum combination of sensitivity and specificity, appears to have an adequate efficiency to discriminate regainers from non-regainers, classifying more than 60 % of patients correctly [27].

Taken together, the calculation of the L/G ratio at baseline could allow health professionals to identify whether an obese patient should initially be treated with a conventional dietary program or with a personalized treatment, such as a special dietary program with the capacity to improve weight loss maintenance [37]. Moreover, this ratio could even help in making a decision on the advisability of bariatric surgery in obese patients with a detected resistance to maintain the weight loss after an energy restriction treatment.

A limitation of the current study could be the limited number of patients, which may affect the accuracy of the L/G ratio as a clinically relevant biomarker of weight regain predisposition. However, the results of this study add important information because it is common to find a low discriminative ability between case and control patients for a common heterogeneous disease, such as obesity [38], metabolic syndrome [39] or cancer [32]. Moreover, the current study is in the exploratory phase (Phase I), in which a new test is first evaluated in a clinical setting to determine whether the test has any ability to discriminate between disease and non-disease patients [25]. In this type of trial, the sample size estimation is based on an AUC >0.50, and at least ten disease patients and ten controls should be included in the study [25].

In conclusion, these results show for the first time that the determination of the fasting leptin/ghrelin ratio before beginning a weight loss therapy program could be a useful non-invasive approach to personalize obesity therapy. A baseline leptin/ghrelin ratio higher than 9.5 units for men and 26.2 units for women identified most of the patients who regained more weight after a successful initial weight loss. Further studies are needed to validate the leptin/ghrelin ratio as a useful marker of weight regain.