ANOVA: Analysis of variance; ANCOVA: Analysis of Covariance; BMI: Body Mass Index; CA 125: Cancer Antigen 125; COC: Combined Oral Contraceptives; CP: Cycle Phase; DIE: Deep Infiltrating Endometriosis; DF: Degrees of Freedom; ELISA: Enzyme-Linked Immunosorbent Assay; GnRHa: Gonadotrophin-Releasing Hormone agonist; IUD: Intrauterine Device; LNG: Levonorgestrel; MS: Mean Square; PF: Peritoneal Fluid; POP: Progestin-Only Pill; rASRM: Revised classification of the American Society of Reproductive Medicine; SD: Standard Deviation; SS: Sum of Squares.

Endometriosis is a chronic disease affecting approximately 10% of women in their reproductive years, often leading to pelvic pain and infertility. It is characterized by endometrial-like tissue, which grows outside the uterus [1]. A thorough medical history and gynecological examination including transvaginal ultrasound, are the primary diagnostic tools [2], although due to their low sensitivity, visualization during laparoscopy remains the gold standard. However, given its invasiveness, it is no longer recommended as a primary diagnostic procedure and according to previous guidelines hormonal therapy can be started empirically to treat endometriosis-associated pain [3]. Hormonal therapy in turn is not tolerated or inefficient in about a third of the patients [4,5]. Hence, endometriosis treatment may be delayed for up to ten years [6].

In the search for a more effective and less invasive diagnostic tool, many biomarkers have been proposed, but none is sufficiently effective to be adopted into clinical practice [7]. Some biomarkers have also been implicated in pathogenesis and represent potential therapeutic targets, as they are elevated in the peritoneal cavity in patients with endometriosis [8,9].

Leptin, a 16-kilodalton (kDa) protein, is an adipokine mainly produced in adipocytes and involved in the lipid metabolism, regulating hunger and food intake [10]. There is also increasing evidence that leptin is linked to immune function and plays a role in autoimmune disorders [11]. Furthermore, leptin may play a role in fertility. Higher serum leptin levels have been found in women with unexplained fertility [12]. Moreover, the leptin receptors have been shown to be expressed in endometrial cells and leptin is assumed to be produced by human ovarian follicles and stimulates Gonadotrophinreleasing hormone (GnRH)-secretion in the hypothalamus [ 13,14].

An association between leptin and endometriosis has also been previously suggested. A significantly elevated PF leptin concentration in women with endometriosis was identified in a study of 60 women with compared to 40 women without endometriosis. This study also showed leptin concentrations positively correlated with the stage of endometriosis [15]. Another study analyzing 60 patients with endometriosis and 18 controls confirmed elevated leptin levels in patients with endometriosis but found an inverse association between leptin concentrations and endometriosis stage, suggesting a role of leptin especially in the formation of peritoneal endometriosis lesions [16]. In contrast, a study using women with ovarian endometriosis found lower PF leptin levels in the endometriosis patients [17]. Other studies, however, did not reveal any difference in PF leptin concentrations between women with and without endometriosis [18,19].

Given the need to improve diagnostic options for women with endometriosis, we aimed to use a cohort of sufficient size and power to assess the difference in PF leptin concentrations between women with and without endometriosis and determine its potential as a diagnostic biomarker or therapeutic target for endometriosis.

The study followed the “Strengthening the reporting of observational studies in epidemiology” guidelines [20]. This is a retrospective analysis of prospectively collected data. All women of reproductive age undergoing laparoscopic intervention for endometriosis or other benign gynecological conditions in the Department of Obstetrics and Gynecology, University of Bern (Switzerland) between 2007 and 2018 were recruited for the study. Written informed consent, detailed information on hormonal treatment usage, cycle phase and demographic biomarkers were obtained from all participants prior to surgery. The relevant Ethical committee (Project-ID 2020-00937) approved the project. Exclusion criteria were patients suffering from other inflammatory diseases, pregnancy, malignancy and surgery performed in an emergency situation. Surgery took place after at least 8 hours (h) of fasting. The peritoneal fluid (PF), that was present in the abdominal cavity at the beginning of the operation (between 0.5-40 ml), was aspirated from the pouch of Douglas at the beginning of the laparoscopy before performing any surgical procedure. For the laparoscopy, Karl Storz instruments and devices were used (distributed by Anklin AG, 4153 Reinach, Switzerland). Intraoperative findings and revised American Fertility Society (rASRM) stage of endometriosis were assessed and documented. Menstrual cycle was determined by selfreported cycle day, or serum progesterone measurement in nanomols per liter (nmol/L). If the surgery took place between the 10^th and 19^th cycle day or if the date of the last menstruation was not available, the menstrual cycle phase was determined according to the progesterone concentration in PF (cut-off 25 nmol/L)). Patients taking hormonal treatments were considered amenorrheic. In patients with no hormonal therapy, the cycle phase was marked as unknown if the cycle day was between the 10th and 19th cycle day without available progesterone concentration or if the cycle day and progesterone concentration in PF were not available or provided conflicting results. The cycle phase was also marked as unknown if the history of hormonal treatment was unknown.

Peritoneal fluid samples were clarified by centrifugation (10 minutes (min) at 1800 times gravity) and stored undiluted at −70°C prior to assay. Hemolyzed PF or samples with insufficient PF volume were excluded. The total protein content in PF was determined using a micro-bicinchoninic assay (Quanti-Pro® BCA, Sigma-Aldrich, St. Louis, 140 Missouri, United States of America) to ascertain absence of dilution with abdominal flushing medium under the procedure. Samples with a protein concentration below 10 ng/ml were excluded. Peritoneal fluid leptin concentrations were measured by ELISA as previously described [17], using a Duoset® microplate sandwich protocol (DY398; R&D Systems, Abingdon, England). The diluent (Cat. No. DY995) was obtained from the same provider, the standard range was from 2000 to 31.25 pg/ml in serial 1:2 steps as recommended, and the sample dilution was 1:50. All incubations were run at a constant temperature of 25°C, in DY995 and for the duration of the recommended in the protocol.

Raw data were imported in IBM SPSS (Statistical Package for the Social Sciences) statistics for Windows (Version 25.0. IBM Corp: Armonk, New York, United States of America) for statistical analysis. Mean values and standard deviations (SD) were calculated for continuous variables and percentages for qualitative variables. Body mass index (BMI) and age between patients with and without endometriosis were compared by t test while the influence of cycle phase was assessed via a χ2 test. Data normality was tested using the Kolmogorov– Smirnov test. Due to a deviation from a normal distribution (p=0.000), we performed a logarithmic transformation log(10) to obtain log transformed leptin levels corresponding to a normal distribution (Figures 1 and 2).

Linear regression analysis was used to study the relationship between leptin concentrations and BMI in all patients as well as in patients based on endometriosis status. Differences between regression lines were analyzed using ANOVA. Differences in logarithmic leptin concentrations between patients with and without endometriosis (fixed factor, two levels: yes, no) and between menstrual cycle phases (fixed factor, four levels: proliferative, secretory, unknown, hormonal treatment) were tested using analysis of covariance (ANCOVA; general linear model, full factorial). The potential confounders age and Body Mass Index were included as covariates in the ANCOVA model to reduce sources of bias. The obligatory assumptions as a prerequisite for carrying out this analysis (Levene’s test for homogeneity of variance, homogeneity of regression, normality check, and independence) were checked and met.

A total of 1054 patients were included in the analysis after excluding cases with protein levels lower than 10 ng/ml and three extreme outliers with very low leptin levels of 0.1 ng/ ml. Body mass index was available for 1035 patients (98.2%). 19 patients had no BMI described. 653 patients (62.0%) were diagnosed with endometriosis, of which 276 (42.3%) patients with mild stage (rASRM stage I or II) and 377 patients (57.7%) with advanced stage (rASRM stage III or IV) endometriosis. 401 patients did not have any intraoperative finding of endometriosis and were counted as controls (38.0%). 433 patients (41.4%) had a hormonal treatment at the time of the surgery, of which 313 (72.3%) patients with endometriosis and 120 (27.7%) patients without endometriosis. The cycle phase was available in 424 (40.2%) patients when excluding patients with hormonal treatment from which 291 (27.6%) of the patients underwent surgery in the proliferative, 133 (12.6%) in the secretory cycle phase.

Body mass index was statistically significantly lower (p<0.001) in patients with endometriosis (23.0 ± 4.0 kg/m²) compared to patients without endometriosis (24.5 ± 4.8 kg/m²). Patients with endometriosis were significantly younger (33.0 ± 6.2 years) than control patients (35.1 ± 8.2 years), p≤0.000. The detailed patient`s characteristics are summarized in Table 1.

Table 1: Patient’s characteristics.

A simple linear regression analysis with leptin concentrations as the dependent variable and BMI as the explanatory variable showed that 31.3% of the variance of leptin was attributable to the patient’s BMI (β= 0.050, R²= 0.313; F (1,1033)=470.728, p<0.001) (Figure 3). Splitting patients by endometriosis status showed similar results (with endometriosis: β=0.055, R²=0.299; F (1,644)=274.471, p=0.000 (Figure 4); without endometriosis: β=0.043, R²=0.328; F (1,387)=188.787, p<0.001) (Figure 4). The same analysis performed with age as the explanatory variable resulted in a minimal effect on the leptin variance by the patient’s age (β=0.004, R²=0.005; F (1,1052)=5.175, p=0.023) ( Figure 5).

The mean leptin concentration in patients with and without endometriosis was Log(10) Leptin 1.020 ± .015 and 1.065 ± .017, respectively. A one-way ANCOVA was conducted to compare leptin concentrations between patients with and without endometriosis and between patients in proliferative and secretory menstrual cycle phases whilst controlling for BMI (covariate) and age (covariate) (Table 2). 1035 patients with available information for all variables were included in the model. No significant difference was found in the mean leptin levels between patients with and without endometriosis (F(1,1026)=3.828, p=.051) (Table 2a). Patients in the proliferative cycle phase without hormonal therapy had significantly lower leptin concentrations compared to patients under hormonal therapy (p=.003) (Table 2b). In patients with advanced stages of endometriosis (rASRM III-IV), leptin concentrations were significantly lower (Log(10)Leptin 1.016 ± .020) compared to patients without endometriosis (Log(10) Leptin 1.073 ± .017) (F(1,752)=4.733, p=0.030) (Tables 3a, 3b).

Table 2: ANCOVA covariance analyses on log(10) leptin levels. Independent variables: Endometriosis: patients with endometriosis (N=646) compared to patients without endometriosis (N=389). Cycle phase: patients in proliferative phase (N=284), secretory phase (N=125) and unknown cycle phase (N=195) compared to patients under hormonal treatment (N=431). Total N=1035.

Table 3: ANCOVA covariance analyses on log (10) leptin levels. Independent variables: Patients with high-grade endometriosis (N=372) compared to patients without endometriosis (N=389). Cycle phase: patients in proliferative phase (N=225), secretory phase (N=93) and unknown cycle phase (N=138) compared to patients under hormonal treatment (N=305). Total N=761.

In this study, we found that after correcting for BMI and age there was no significant difference in PF leptin concentrations between patients with and without endometriosis (Table 2a). On the contrary, leptin tended to be lower in patients with compared to without endometriosis with even statistically significantly lower concentrations in patients with high-stage endometriosis (Tables 3a, 3b).

Three systematic reviews and meta-analyses have previously reported elevated PF leptin concentrations in patients with endometriosis compared to controls. In the meta-analysis by Tian et al., elevated PF but not serum leptin concentrations were found in patients with endometriosis. In subgroup analyses, significantly elevated PF leptin levels were shown both for early and advanced stage endometriosis. In this meta-analysis 13 studies on PF leptin concentrations were included, consisting of 376 patients with endometriosis and 246 controls [21]. Kalaitzopoulos et al., assessed 18 studies on 531 women with endometriosis and 331 controls and reported similar results. Among them are all 13 studies included by Tian et al. as well as five additional studies [22]. Zhao et al. reported both elevated PF and serum leptin levels in endometriosis in their recently published meta-analysis [23]. They included 13 studies on PF leptin concentrations, whereby some of the studies included in the other meta-analyses were omitted but two additional studies were included. In these meta-analyses however, most studies included had sample sizes of less than 50 patients. To the best of our knowledge, the largest study on leptin levels in PF until now included 58 patients with endometriosis and 40 controls [17]. Studies with very small sample sizes are prone to be less representative of the population, have less statistical power and tend to publication bias. Our study has a larger sample size than each of the three mentioned meta-analyses, resulting in greater statistical power.

Five studies have shown lower or similar PF leptin levels in endometriosis, which is in line with our findings. Interestingly, three of these studies included a relatively high number of patients (56, 72 and 98 patients) compared to other studies evaluating PF leptin in endometriosis [17-19,24-26].

Elevated serum leptin concentrations have been reported in obese patients [11,27] and a correlation of BMI and leptin concentration has also been shown in follicular fluid [28]. In line with these findings, our study reports a strong influence of the patient’s BMI on the variance of PF leptin levels. Moreover, endometriosis has been associated with a lower average BMI than women without endometriosis [29,30], which was confirmed in our cohort (Table 1). Therefore, we included BMI as a covariate in the ANCOVA analysis. Although most studies included in the above-mentioned meta-analyses have taken the patients’ BMI into account, BMI was not included as a covariate. Instead, patient groups were either matched according to BMI or a leptin-BMI ratio was applied. Some studies have not considered the BMI in the methodology or statistical analysis at all [25,26]. A confounding bias has to be assumed in these studies. In the present study, we logarithmically transformed the initial leptin values in order to obtain a normal distribution ( Figures 3 and 4) and perform parametric statistical tests. To the best of our knowledge, this has not been performed by any of the previously published studies. Previous studies mostly used non-parametric tests, which apply to non-normally distributed data but have weaker statistical power. Certain studies nevertheless performed parametric tests but did not describe the distribution of the leptin values [25,31-33]. The normal distribution and consequently performed parametric tests in our study represent an advantage in our opinion.

Our observation of lower leptin values in the proliferative compared to the secretory cycle phase is consistent with previously published studies [34,35]. Pathophysiologically, higher leptin expression in the secretory endometrium is assumed, although this remains unclear in detail. Leptin receptors have been found expressed in human endometrium, whereas contradictory findings exist concerning leptin expression by endometrial cells [14,36,37]. Increased leptin expression in ectopic lesions of patients with endometriosis compared to eutopic endometrium has been reported [32,38,39], whereas another study showed no significant elevation in leptin concentration in ovarian tissue affected by endometrioma compared to normal ovarian tissue [24]. Therefore, due to the controversial findings, no conclusive statement regarding leptin expression in the ectopic endometrium could be made yet. Our results suggest no higher expression of leptin in ectopic endometrium.

If higher leptin levels were true in endometriosis, we would also expect higher leptin levels in higher stages of endometriosis and lower leptin levels in patients under hormonal therapy. This was however not shown in our study, which additionally speaks against a positive correlation between endometriosis and PF leptin levels.

Possible limitations also need to be elucidated. First, the possible batch effect due to the duration of the study (11 years), which could mask differences in leptin levels between groups. On the other hand, it is the duration of the study, which allowed a large sample size and sufficient statistical power. Moreover, the retrospective design of the study has to be mentioned. Due to some missing data on hormonal treatment or cycle day as well as progesterone value, some patients could not be included in the ANCOVA analysis. Finally, the determination of the cycle phase could be subject to a certain bias. Although we determined the cycle phase by means of cycle day and, in the mid-cycle period, by means of the progesterone value, as described under paragraph 4, it is possible that individual cases were assigned to an incorrect cycle phase, especially in the case of longer cycles of 35 days and more and short cycles of less than 24 days.

This study reports on peritoneal fluid leptin concentrations in patients with and without endometriosis in the largest sample size to date. In contrast to the existing literature, patients with endometriosis had no difference in PF leptin concentrations compared to patients without endometriosis. This contradictory finding questions whether leptin plays a role in endometriosis pathogenesis and its utility as a biomarker for the disease.

The authors declare no conflict of interest.

We would like to thank Marie Roumet, PhD (University of Bern) for statistical consulting.

The study was conducted according to the guidelines of the Declaration of Helsinki, and approved by the Ethics Committee of Bern (Kantonale Ethikkommission für die Forschung). Project-ID 2020-00937, date of approval: 07/23/2020.

Informed consent was obtained from all subjects involved in the study.