Blood tests for prediction of deep endometriosis: A case-control study

Deep endometriosis (DE) is the most aggressive subtype of the disease. The diagnosis of DE is challenging. No biomarkers have been identified for discriminating women with DE from those without DE. Developing new approaches for predicting DE before surgery is of crucial importance. It is unclear whether DE could be identified by routine hematological evaluation before surgery. To evaluate the role of blood hemostatic parameters and inflammatory indices in the prediction of DE before surgical intervention. A case-control study investigated the value of routine plasma hemostatic parameters and inflammatory indices in women with DE and without endometriosis. Univariate analysis and multivariate analysis following adjustment for potential confounding factors were performed. Receiver operating characteristic curves were generated, and the areas under the curve was calculated to assess the predictive values of the selected parameters. Elevated plasma fibrinogen (Fg) and decreased hemoglobin (HB) levels were found in women with DE compared with controls. Plasma Fg and HB levels were significantly associated with DE after adjusting for potential confounding factors. The diagnostic value of Fg or HB alone for DE detection before surgical intervention was limited, but the combination of Fg and HB had good predictive value for DE. It suggested that the combination of Fg and HB levels could be used as a reliable predictor of DE. Based on the model, a thorough assessment is recommended for suspected patients with DE. Further studies are required to investigate how Fg and HB contribute to the development of endometriosis, particularly DE.

A case-control study was performed at the Women’s Hospital, Zhejiang University School of Medicine between January 2015 and December 2016. Approval for this study was obtained from the Institutional Ethics Committee at Women’s Hospital School of Medicine, Zhejiang University (IRB-20200049-R). Data were retrospectively retrieved from an electronic database. Inclusion criteria for the DE group were: (1) 18 to 40 years of age[ 13 ]; and (2) DE defined as endometriotic lesions that infiltrated the uterosacral ligaments by > 5 mm and muscularis propria (bladder, intestine, ureters). Lesions were confirmed by pathology. Patients with DE who simultaneously had SUP or OMA were also included[ 2 , 14 ]. Exclusion criteria for the DE group were: (1) A history of abnormal uterine bleeding in the previous 3 mo; (2) A history of acute inflammation, suspected infectious disease, malignancy, metabolic diseases, and autoimmune disease in the previous 3 mo; (3) Pregnancy; (4) Hormonal therapy, including oral contraceptives, gonadotropin-releasing hormone analogs, or any other hormonal treatment, antithrombotic and hemostatic agents, and herbal compounds during the previous 3 mo; (5) Medical emergencies; and (6) With non-fasting lipid profiles. Women between 18 and 40 years of age with surgical treatment at the same time for benign gynecologic diseases, including benign ovarian tumors, tubal infertility, cervical intraepithelial neoplasia, and intrauterine adhesion, but without any evidence of endometriosis, were recruited as controls. Detailed histories, thorough physical examinations of the abdominal-pelvic cavity, and sonography screening were performed by designated experts. The exclusion criteria were the same as for the DE group. Women to be enrolled in the controls with suspected endometriosis presenting dysmenorrhea or tenderness in the pelvic cavity or were also excluded. During the study period, 698 patients with endometriosis were scheduled for surgery in the general gynecology department. However, 667 women were excluded because they did not meet the selection criteria, 154 who were > 40 years of age, 251 with non-fasted blood collection, 138 with no surgical treatment, with hormonal treatment, or complicated by other diseases, 124 with pathologically proven endometriosis but not DE phenotype. The remaining 170 eligible women without endometriosis were enrolled; 75 cases were excluded, 30 because of suspected endometriosis, 21 with no surgical treatment, and 24 complicated by inflammatory, or metabolic, or autoimmune diseases. All participants had routine peripheral blood tests before surgery. Blood samples with ethylenediaminetetraacetic acid (EDTA) as the anticoagulant were used to obtain a complete blood count, platelet count, HB level, neutrophil count, and lymphocyte count with a, autoanalyzer (Beckman, Coulter LH750). Coagulative parameters, including PT, TT, APTT, and Fg were determined with an automatic blood coagulation analyzer (STAGO, Evolution ISTA-R-IV, Germany). The neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), international normalized ratio (INR), platelet distribution width, neutrophil number, lymphocyte number, and mean platelet volume were also calculated. Considering crosstalk between lipid metabolism and coagulation function, serum lipid profiles including total cholesterol, total triglycerides, high-density lipoprotein, and low-density lipoprotein were tested and measured by enzymatic colorimetric assay after an overnight fast of 12 h. C-reactive protein (CRP) levels were simultaneously determined. The blood samples without EDTA were centrifuged at 3500 g , and serum supernatants were collected. CRP levels were assayed in fresh serum using an immunoturbidimetric assay (Abbott, Architect C16000 ). Intra- and interassay coefficients of variation for all measurements were 5% and 10%, respectively. Continuous data were reported as means ± SD for normally distributed variables, and variables that were not normally distributed were reported as medians and range. Between-group differences of variables with a normal distribution were tested by analysis of variance and Student’s t- test. For variables with a non-normal distribution, differences were compared with Kruskal–Wallis and Wilcoxon tests. Categorical variables were reported as n (%), and the χ 2 -square test was used to compare the distribution across different groups. Stepwise logistic regression was used to assess the association of hemostatic profiles with the presence of DE. All indices of interest such as age, body mass index (BMI), history of delivery and abortion, inflammatory indices, and serum lipid profiles that could cause a confounding bias were entered into the initial model as potential risk factors with SLE = 0.05 and SLS = 0.10. The final model was built using all significant variables in the multivariate analysis. Receiver operating characteristic (ROC) curves were constructed, and the area under curves was calculated to determine the predictive power of the independent risk factors. The statistical analysis were conducted with SAS, version 9.4 (SAS Institute, Cary, NC, United States). P values < 0.05 were considered statistically significant.

A total of 126 women were enrolled in this study, 31 with DE and 95 without DE (Figure 1 ). The indications for surgery in the DE patients were a pelvic mass, history of infertility, pelvic pain with failed analgesics. DE involved the uterosacral ligament in 26 (83.9%) patients, the colorectal septum in two (6.5%), the ureter in one (3.2%), and the sigmoid in two (6.5%). Adenomyosis was suspected in 13 women with no uterine fibroids on transvaginal ultrasound. The indications for surgery in controls were benign ovarian tumors and tubal infertility (54 cases), cervical intraepithelial neoplasia (28 cases), and intrauterine adhesion (13 cases). Baseline clinical characteristics are shown in Table 1 . There were no differences in age, BMI, parity, abortion, and lipid profiles between the study and the control groups. Flowchart of enrollment. DE: Deep endometriosis. Baseline clinical characteristics and serum lipid profiles of participants Data are mean ± SD for normally distributed variables, medians (Q1-Q3) for variables without a normal distribution, and n (%) for categories variables. BMI: Body mass index, DE: Deep endometriosis; HDL: High-density lipoprotein; LDL: Low-density lipoprotein; TC: Total cholesterol; TG: Total triglycerides. Plasma Fg ( P < 0.01), INR (P < 0.05), and CRP levels ( P < 0.01) of women with DE were significantly higher than those in controls. Plasma HB levels ( P < 0.05) and TT ( P < 0.05) of women with DE were significantly lower than those in controls. Differences between the other hematological parameters in the two groups were not significant (Table 2 ). Hemostatic parameters and inflammation indices of participants with or without deep endometriosis Data are mean ± SD for normally distributed variables and medians (Q1-Q3) for abnormal variables without a normal distribution. APTT: Activated partial thromboplastin time; CRP: C-reactive protein Fg: Fibrinogen; HB: Hemoglobin; INR: International standardized ratio; MPV: Mean platelet volume; NLR: Neutrophil-to-lymphocyte ratio; PCT: Platelet distribution width; PLR: Platelet-to-lymphocyte ratio; PLT: Platelet count; PT: Prothrombin time; TT: Thrombin time; WL: Lymphocyte number; WN: Neutrophil number. Multivariate analysis (Table 3 ) showed that plasma Fg levels [odds ratio (OR) 1.67, 95%CI: 1.13-2.46], PT (OR 1.63, 95%CI: 1.12-2.38), plasma HB levels (OR 0.63, 95%CI: 0.42-0.92), and TT (OR 0.69, 95%CI: 0.48-0.99) were significantly associated with the presence of DE (all P < 0.05). No relationships between the other hematological parameters and the presence of DE were found. APTT, INR, NLR, and PLR were not included in the multivariable logistic models, considering that they overlapped with other parameters. Unadjusted association of hemostatic parameters and inflammation indices with deep endometriosis by logistical regression CI: Confidence interval; CRP: C-reactive protein; Fg: Fibrinogen; HB: Hemoglobin; MPV: Mean platelet volume; OR: Odds ratio; PCT: Platelet distribution width; PLT: Platelet count; PT: Prothrombin time; SE: Standard error; TT: Thrombin time; WL: Lymphocyte number; WN: Neutrophil number. After adjusting for potentially confounding factors including age, BMI, history of delivery and abortion, and serum lipid profiles, plasma Fg levels [adjusted OR (aOR) 2.12, 95%CI: 1.31-3.75] and plasma HB levels (aOR 0.48, 95%CI: 0.29-0.78) remained significantly associated with the presence of DE (both P < 0.05, Table 4 ). The relationship between PT/TT and DE was no longer significant. Stepwise logistical regression analysis and receiver operating characteristic analysis for deep endometriosis, hemostatic parameters, and inflammatory indices P < 0.05. Age, BMI, history of delivery and abortion, and serum lipid profiles were entered into the stepwise regression model with SLE = 0.05, SLS = 0.10. AUC: Area under the curve; CI: Confidence interval; Fg: Fibrinogen; HB: Hemoglobin; OR: Odds ratio. The predictive performance of Fg and HB for DE was investigated using ROC analysis. The area under the curve (AUC) of Fg was 0.639 (95%CI: 0.524-0.755, sensitivity = 58.1%, specificity = 70.5%), and that of HB was 0.664 (95%CI: 0.552-0.776, sensitivity = 64.5%, specificity = 62.1%) for the diagnosis of DE. The AUC of the combination of both markers was 0.773 with improved sensitivity (67.7%) and specificity (78.9%) at a cutoff of 3.09 (g/L) and 126 (g/L), respectively (Table 4 , Figure 2 ). The combination of both markers as a dual marker index significantly improved the diagnostic accuracy. Receiver operating characteristic curves for fibrinogen and hemoglobin alone or their combination in the diagnosis of deep endometriosis. AUC: Area under the curve; Fg: Fibrinogen; HB: Hemoglobin.

A combination of Fg and HB could be used in routine clinical practice as a reliable predictor of DE before surgical intervention. Future studies with larger samples are needed to verify the findings and to investigate how Fg and HB contribute to the development of endometriosis, particularly DE.

This study examined the predictive values of hemostatic parameters and inflammatory indices for DE. Our data suggest that the combination of Fg and HB levels could be used as a reliable predictor of DE. To the best of our knowledge, this is the first report that evaluated hemostatic parameters and inflammatory indices for the prediction of DE. Fg is a known coagulation factor associated with hypercoagulation. As an acute-phase reaction and a hemostatic parameter, Fg has an important role in coagulation, inflammation, and the maintenance of hemostasis[ 22 - 23 ]. Fg is also a marker of inflammation and a major determinant of thrombosis and hemorheology[ 24 ]. In this study, we found elevated Fg levels in DE patients, which is consistent with previous findings[ 15 - 16 ]. Increasing evidence shows that pathophysiological changes in endometriosis have features in common with those observed during tissue injury and repair (TIAR)[ 3 , 25 - 26 ]. TIAR process may contribute to the development of endometriosis. The coagulation and fibrinolytic systems have important roles in TIAR[ 27 ]. An increase in plasma Fg is more likely caused by recurrent bleeding in the ectopic implantation of the endometrium and the impairment of the fibrinolytic system in endometriosis. Elevated Fg levels may reflect hemorheological disorders, a potential hypercoagulable status, and subclinical systemic inflammation in endometriosis. In this study, a shortened TT was found in women with DE, which is in line with previous studies[ 16 - 17 ]. TT reflects anticoagulation, and a shortened TT indicates hypo-fibrinolysis. In this study, an inverse relationship between TT and DE was initially detected by multivariate analysis. After adjusting for confounding factors, the association was no longer significant. Moreover, no differences in APTT and TT were found between the DE and the control groups. Decreased plasma HB levels were identified in women with DE in this study, which is in line with the findings in women with OMA[ 28 - 29 ]. Moreover, our results revealed an inverse relationship between plasma HB levels and the presence of DE. An inverse association between severity of endometriosis and plasma HB levels was reported in another study[ 30 ]. The exact cause of low plasma HB levels in patients with endometriosis is not clear. It may be associated with erythrocyte regulation of iron metabolism disorders or chronic systemic inflammation[ 31 - 32 ]. Low plasma HB levels may be associated with hypoxia, which has been reported to facilitating endometriosis development[ 33 ]. Further studies are required to investigate how HB contributes to the development of DE. Serum CA-125 antigen is the most frequently used biomarker in the diagnosis of endometriosis in clinic practice[ 11 , 34 - 35 ]. Santulli et al [ 36 ] reported that serum CA-125 antigen was significantly associated with the severity and the penetration depth of DE; but it is not widely used in the diagnosis of DE. In this study, we found that either plasma Fg levels alone or plasma HB levels were not powerful enough to predict DE. A good predictive value for DE was obtained when plasma Fg levels were combined with HB levels. The AUC of the combination was 0.773, and the specificity was 78.9% at cutoffs of 3.09 g/L and 126 g/L. Ding et al [ 37 ] investigated the predictive role of Fg for endometriosis and found that the combination of Fg and serum CA-125 had good predictive power for OMA. They showed that Fg had potential predictive value for endometriosis, which is consistent with our results. Our model for predicting DE with the use of plasma Fg and HB may have clinical implications. Using this model, patients suspected of DE should undergo a thorough preoperative assessment through pelvic examination and pelvic imaging to detect DE nodules. Nonetheless, further studies for optimized predictive tools for DE are warranted. Endometriosis is associated with an inflammatory response. In this study, increased CRP levels were found in the DE group, and there the difference in NLR between women with or without DE was not significant. The results of the value of inflammatory indices such as CRP and NLR in endometriosis in previous studies are not consistent[ 29 , 37 - 38 ]. This inconsistency may be associated with a different course, subtypes, and sample sizes used in those studies. In this study, multivariate analysis did not identify any association between CRP and NLR and the presence of DE. Low HB levels could be caused by other bleeding disorders such as adenomyosis and uterine fibroids. In this study, 13 patients in the DE group had suspected adenomyosis with no uterine fibroids. Nonetheless, they did not complain of abnormal menstrual bleeding, and the HB levels were still within the normal range. Thus, the decreased HB levels of the DE group could not be attributed to concomitant adenomyosis. In addition, the coexistence of adenomyosis and endometriosis is well known[ 39 ]. We could not exclude the women with both DE and adenomyosis from the DE group in this study. Our study has several limitations. First, the size of the DE group was relatively small. The low incidence of DE and strict criteria imposed in this study limited the enrollment. Additionally, this is the first study that evaluated the predictive role of hemostatic parameters for DE, and one of the aims of this study was to inspire future larger investigations. Second, as we did not include cases with only SUP or OMA subtypes, the results may not be applicable for all patients with endometriosis. Finally, there was no ideal control group for studying plasma Fg levels in DE. Our control group consisted of women with surgery for benign gynecological conditions, which permitted a thorough assessment of DE. However, some of the conditions, such as tubal infertility or ovarian cysts, might be associated with altered plasma levels. In addition to the limitations, the following strengths should also be pointed out. First, the results of Fg/HB and the presence of DE were consistent in both univariate and multivariate analysis. Second, we for adjusted those confounding factors to eliminate possible effects on coagulation function and inflammatory response, which could make the predictive value of Fg and HB more reliable. In addition, these effects were not investigated in the previous relevant studies on endometriosis. The study findings support the routine combination of Fg and HB as an essential part of the preoperative assessment of patients with suspected DE. The model can be adopted for use in clinical practice. Furthermore, this study suggests that an altered coagulation system may have key involvement in the development of endometriosis. The results also suggested that patients with DE may have a potential hypercoagulable state. Further studies are required to determine the anticoagulant therapy for these patients.

Endometriosis, which is characterized by the presence of endometrial glands and stroma at ectopic sites, affects approximately 10% of women of reproductive age. Up to 80% of women with endometriosis suffer from chronic pain, and up to 50% of women suffer from infertility. Endometriosis-related productivity loss and decreased quality of life lead to a heavy economic burden[ 1 ]. Endometriosis can be classified as superficial endometriosis (SUP), ovarian endometrioma (OMA), and deep endometriosis (DE)[ 2 ]. DE is the most aggressive of the three subtypes that constitute endometriosis. It is defined as an endometriotic lesion penetrating a depth of > 5 mm and showing aggressive behavior[ 3 ]. It can affect the uterosacral ligaments, parametrium, bladder, and bowel. Patients with DE usually present with severe pelvic pain and low fertility. The heterogeneity of the disease makes the diagnosis of DE a clinical challenge[ 4 - 6 ] that may be delayed for more than 8 years[ 7 ]. Accidental intraoperative diagnosis of DE is also common. DE often requires surgical therapy, and a high incidence of surgical morbidity of DE has been reported[ 8 ]. Therefore, developing new approaches for predicting DE before surgery is of crucial importance. In previous years, symptoms and clinical history, pelvic examination, blood tests, transvaginal ultrasound, and magnetic resonance imaging (MRI) have been proposed for the preoperative prediction of endometriosis[ 9 - 13 ]. However, the clinical presentation of DE tends to vary. Some women experience severe pain, while others remain asymptomatic[ 10 - 11 ]. Pelvic examination results and the accuracy of ultrasound or MRI diagnosis can significantly vary in relation to the location of DE[ 13 ], and assessment by vaginal examination or image diagnosis depends on the level of expertise. The benefit of blood test prediction of deep endometriosis would have the advantages of being noninvasive, no exposure to harmful radiation, rapid reporting, and low cost. Several studies have explored the predictive value of blood biomarkers such as serum CA-125 in DE[ 11 - 12 ]. Low sensitivity and specificity reduce the value of serum CA-125 as a single test in the diagnosis of DE. In fact, a reliable noninvasive marker for preoperative diagnosis of this disease has not yet been introduced. The most widely accepted etiologic mechanism of endometriosis is retrograde menstruation resulting in ectopic implantation of endometrium in the pelvic cavity. The ectopic implanted endometrium can lead to recurrent bleeding, subsequent repeated tissue injury, and inflammation[ 14 ]. Endometriosis has also been associated with increased activation of the coagulation system and fibrinolysis system, and elevated plasma fibrinogen (Fg) levels, shortened thrombin time (TT), shortened activated partial thromboplastin time (APTT), shortened prothrombin time (PT), and increased expression of urokinase-type plasminogen activator and plasminogen activator inhibitors[ 15 - 19 ]. The data indicate that women with endometriosis might have a potential hypercoagulable state. Additionally, a high concentration of proinflammatory cytokines was reported in women with endometriosis[ 20 - 21 ]. Nevertheless, whether these routine hemostatic parameters and inflammatory indices have any predictive value in terms of preoperative diagnosis of DE has not yet been determined. This study was conducted to assess whether DE could be identified by routine hematological parameters before surgery. The study objectives were to estimate the predictive values of routine hemostatic parameters and inflammatory indices for DE.

We thank Li M from Haining Center for Disease Control and Prevention of Zhejiang Province, China for reviewing the statistical methods used in this study.