Diagnostic Value of MRI for Pelvic Lymph Node Extracapsular Invasion in Early-Stage Endometrial Carcinoma

Introduction

As the most common malignant tumor of the female reproductive system,¹ endometrial carcinoma (EC) ranks as the fourth among cancers in women in developed countries.² There are close to 200000 new cases per year globally.³ Treatment options for EC include radiotherapy, hormone therapy, surgery, and chemotherapy, with the choice of treatment mainly relying on staging.⁴ About 70% of patients with EC are diagnosed as stage I, and the overall prognosis is optimal, with 5-year survival rates of 74%-91%, but pelvic lymph node metastasis (LNM) still occurs in about 10% of patients in the early clinical phase, and LNM is associated with a poorer prognosis in EC.⁵ A prior study has shown specific expression of cancer antigen 125 (CA125) in EC patients.⁶ Notably, preoperative CA125 can act as a risk stratification predictor for EC LNM and lymphovascular space invasion (LVSI).⁷ Meanwhile, several studies have evidenced that serum CA125 predicts LNM.^6,8 Lymph node extracapsular invasion (LNEI) refers to cancer cells breaking through the outer membrane of a lymph node and invading its surrounding tissues such as nerves, blood vessels, fat or muscle.⁹ It has been documented that LNEI is a common microscopic feature of LNM bound up with metastasis and recurrence in patients with tumors.^10,11 What’s more, LNEI has been implicated as an important factor in poor prognosis in many tumors, such as thyroid cancer, bladder cancer, head and neck cancer.^12–14 To date, there are relatively few studies on pelvic lymph node extracapsular invasion (PLNEI) in EC. Therefore, the diagnosis of PLNEI in patients with early-stage EC and accurate assessment of the disease are crucial for the prognosis of these patients.

Magnetic resonance imaging (MRI) is widely regarded as the most precise modality for the preoperative evaluation of EC, and is effective in assessing local extent and deep myometrial invasion of tumors in EC.^15,16 The capability of MRI to provide multiplanar imaging, combined with its superior soft-tissue contrast resolution, enhances the identification of critical prognostic factors in low-risk patients, which informs subsequent treatment decisions.¹⁷ MRI has been demonstrated to be accurate in evaluating various prognostic indicators, including parametrial invasion, tumor size, lymph node invasion, and pelvic sidewall involvement.¹⁸ MR volumetric measurement can be utilized as a tool for tumor response evaluation, surgical planning, and prognosis assessment.^19,20 Furthermore, quantitative parameters derived from MRI, such as volume index (VI), total tumor volume (TTV), tumor volume ratio (TVR), and mean apparent diffusion coefficient (ADC_mean), can be utilized to assess tumor grade and LVSI. Among which, VI refers to products of maximum anterior-posterior, transverse, and cranial-caudal tumor diameters, TTV is defined as the tumor volume measured on fat-saturated T1-weighted postcontrast images, TVR is calculated as (TTV/TUV) * 100, and ADC_mean is the mean of ADC value of the primary tumor, a normal, and a suspicious LN.^21,22 It has been documented that VI > 15.29 is linked with high-risk tumors, and VI > 25.45 and TVR > 21.1 predict LVSI in EC.²³ EC patients usually exhibit diffusion limitation and reduced ADC values.^24,25 Nakamura et al have revealed that patients with lower ADC values display shorter disease-free survival than those with higher ADC values, and it can be regarded as an independent prognostic factor for disease recurrence.²⁶ Moreover, a tumor diameter ≥ 2 cm suggests a high risk of type 1 tumor.²⁷ However, in EC patients, tumor volume measurements have shown a stronger correlation with LNM, parametrial involvement and tumor differentiation than the single diameter tumor analysis.²⁸ However, few studies have reported on the diagnostic value of MRI for PLNEI in EC patients. Therefore, the aim of this study was to probe the diagnostic value of MRI for PLNEI in patients with early-stage EC, which was of great significance for the prognosis and treatment of EC.

Materials and Methods

Ethics Statements

The study was conducted in accordance with the Declaration of Helsinki. The research protocol was approved by the Academic Ethics Committee of The First Affiliated Hospital of Fujian Medical University. All patients who participated in this study were fully informed of the purpose of the study and all signed an informed consent before MRI examination.

Sample Size Estimation

G*Power 3.0.10 software (Heinrich-Heine-Universität Düsseldorf, Nordrhein-Westfalen, Germany) was utilized for sample size estimation prior to this study (Supplementary Figure 1). The t test was was selected as the test method, with significance level (α) set at 0.05, the test efficacy (1-β) set at 0.80, and an anticipated effect size (Cohen’s d) of 0.5 (medium effect). Sample size calculations were performed using G*Power 3.1 software. The results indicated that a minimum sample size of 128 cases was necessary to ensure sufficient statistical power for detecting the findings.

Study Subjects

The 186 patients with early-stage EC admitted to The First Affiliated Hospital of Fujian Medical University from August 2017 to July 2022 were retrospectively and consecutively selected, and 142 patients were finally enrolled as the study subjects based on the inclusion and exclusion criteria. They were then arranged into the PLNEI_negative group (64 cases) and the PLNEI_positive group (78 cases) according to the histopathological results of radical resection of EC.

Inclusion and Exclusion Criteria

Inclusion criteria were as follows: (1) diagnosed as EC by histological examination; (2) International Federation of Gynecology and Obstetrics stage I–II; (3) with complete clinical information, pathological and follow-up data; (4) underwent MRI examination before surgery.

Exclusion criteria included (1) patients who have received other treatments (such as radiotherapy, chemotherapy) before the imaging examination; (2) complicated with the other malignant tumors, systemic inflammatory diseases and chronic diseases; (3) with lung, heart, liver, kidney dysfunctions or in the acute phase of cardiovascular and cerebrovascular adverse events; (4) whose parameter measurements were affected by the artifacts of the MRI.

Data and Sample Collection

By consulting the electronic medical records, the general clinical data of enrolled subjects were collected, including age (years), body mass index (BMI) (kg/m²), smoking history, drinking history, and menopausal status. Additionally, clinicopathological data were also recorded, such as histological type (EC, non-EC), histological grade (G1, G2, and G3 grade represented high, intermediate, and low differentiation, respectively, with the lower differentiation manifesting the higher malignancy), depth of myometrial invasion (< 1/2, ≥ 1/2), LVSI positive detection (negative, positive), LVSI type (conjoined, satellite), cervical involvement (negative, positive), progesterone receptor (negative, positive) and serum CA125 level.

Pelvic MRI

MRI were carried out utilizing a 1.5-T Siemens Avanto system. The patients experienced a fasting period for ≥ 4 h. No antispasmodic medication was used at the time of the examination, and the bladders of the patients were mildly distended. The imaging protocol consisted of T2-weighted axial, coronal, and sagittal images [repetition time (TR)/echo time (TE): 5190/108, 4450/108, 4290/108, respectively; section thickness: 5, 4, 4.5 mm; field of view (FOV): 420, 450, 280 mm]. In addition, diffusion-weighted imaging was performed at b-values of 50, 400, and 800 s/mm² (TR/TE: 6600/81; slice thickness: 5 mm; FOV: 420 mm). T1-weighted axial images (TR/TE: 716/10; section thickness: 5 mm; FOV: 420 mm) and pre- and post-contrast fat-saturated T1-weighted axial images (TR/TE: 716/10; section thickness: 5 mm; FOV: 420 mm) were obtained. Gadolinium-diethylenetriamine pentaacetic acid was intravenously administered at a dose of 0.1 mmol/kg body weight, followed by contrast-enhanced imaging in axial, coronal, and sagittal planes.

Image Analysis

All MRI images were uploaded to a GE AW4.6 workstation (GE Healthcare, Waukesha, WI, USA). A number of senior radiology physicians reviewed the films in a blind method. Osirix MD version 10 (Pixmeo SARL, Bernex, Switzerland) software was adopted for image analysis. VI was defined as the products of maximum anterior-posterior, transverse, and cranial-caudal diameters of the tumor. TVR was calculated using the formula TTV/total uterine volume (TUV) × 100, where TTV was measured on fat-saturated T1-weighted postcontrast images, which provides enhanced delineation of lesion boundaries, and TUV was measured on high-resolution T2-weighted sagittal images (excluding the cervix and fibroids that deformed the outer edge of the uterus). For the calculation of the ADC, regions of interest (ROIs) were delineated around the tumor on each section of the high b-value images, and these ROIs were subsequently copied on the ADC map. The ADC_mean of the primary tumor (ADC_tumor), as well as a normal lymph node and a suspicious lymph node, were computed.

Statistical Analysis

Data were statistically analyzed and graphed using the SPSS 21.0 (IBM, Armonk, NY, USA) and GraphPad Prism 8.0.1 (GraphPad Software, San Diego, CA, USA) software. The Shapiro–Wilk test was utilized for normal distribution, and normally distributed measures were expressed as mean ± standard deviation, with t test for comparisons between groups. Categorical data were expressed as frequency and percentage, using Chi-square test for comparisons between groups. Independent influencing factors affecting PLNEI in patients with early-stage EC were assessed using multivariate logistic regression. The individual influencing factors were first subjected to univariate analysis, and those with P < 0.05 were included in the multivariate logistic regression analysis (Enter method was chosen to screen the independent variables). The analysis of differences in the area under the receiver operating characteristic (ROC) curve (AUC) was performed by Delong test. t test was a two-sided test, with P < 0.05 and P < 0.01 considered statistically significant.

Results

Analysis on the Clinical Baseline Data of Early-Stage EC Patients

There were significant differences in histological type, histological grade, depth of myometrial invasion, LVSI positive detection, LVSI type, cervical involvement and serum CA125 level between the PLNEI_negative group and the PLNEI_positive group (all P < 0.05). No significant differences in age, BMI, smoking history, drinking history, menopausal status, and progesterone receptor were noted between the two groups (all P > 0.05) (Table 1).

Table 1 Analysis on the Clinical Baseline Data of Patients with Early-Stage EC

Comparative Analysis on VI, TTV, TVR, and ADCmean Between Early-Stage EC Patients with/without PLNEI

The differences in VI, TTV, TVR and ADC_mean were comparatively analyzed between early-stage EC patients without and with PLNEI. The results indicated that early-stage EC patients without PLNEI exhibited a VI value of 25.30 ± 7.47 cm³, a TTV value of 12.96 ± 5.32 cm³, a TVR value of 20.57 ± 3.85 cm³, and an ADC_mean value of 0.80 ± 0.07 (× 10⁻³ mm²/s). In contrast, early-stage EC patients with PLNEI demonstrated VI, TTV, TVR and ADC_mean values of 39.72 ± 9.90 cm³, 23.64 ± 7.19 cm³, 25.45 ± 3.72 cm³, and 0.75 ± 0.07 (× 10⁻³ mm²/s), respectively. Notably, the VI, TTV, and TVR values were higher in early-stage EC patients with PLNEI than those without PLNEI (P < 0.001) (Figure 1A–C). Conversely, the ADC_mean value was lower in early-stage EC patients with PLNEI than in those without PLNEI (P < 0.001) (Figure 1D).

Figure 1 Comparative analyses on MRI parameters (VI, TTV, TVR, ADCmean) between the two groups of early-stage EC patients with or without PLNEI. Comparisons of VI (A), TTV (B), TVR (C), and (D) ADCmean values between the two groups of patients. Data were expressed as mean ± standard deviation. Comparisons between the two groups of data were performed using t test. *** P < 0.001.

Analysis on the Relationships of VI, TTV, TVR and ADC_mean Values with Clinicopathological Characteristics

To further explore the relationships of VI, TTV, TVR and ADC_mean values with the clinicopathological characteristics of patients with early-stage EC, all patients were were allocated into the low expression groups (L-VI, L-TTV, L-TVR, L-ADC_mean) and the high expression groups (H-VI, H-TTV, H-TVR, H-ADC_mean) based on the median values of VI, TTV, TVR, and ADC_mean. A comparative analysis on the clinicopathological characteristics in early-stage EC patients was conducted. Among early-stage EC patients, significant differences were observed in the H-VI group regarding the histological type, histological grade, depth of myometrial infiltration, LVSI positive detection, LVSI type, cervical involvement, and serum CA125 level (all P < 0.05). Similarly, patients in the H-TTV group demonstrated significant differences in the histological grade, depth of myometrial infiltration, LVSI positive detection, LVSI type, cervical involvement, and serum CA125 level (all P < 0.05). Furthermore, patients in the H-TVR group showed significant differences in the histological grade, depth of myometrial infiltration, cervical involvement, LVSI type, and serum CA125 level (all P < 0.05). Additionally, patients with high ADC_mean value exhibited significant differences in the histological grade, depth of myometrial infiltration, cervical involvement, and serum CA125 level (all P < 0.05) (Table 2).

Table 2 The Relationships of VI, TTV, TVR, and ADCmean Values with Clinicopathological Characteristics of Early-Stage EC Patients

VI, TTV, TVR, and ADC_mean Were Independent Influencing Factors for PLNEI in Patients with Early-Stage EC

To further assess the independent influencing factors affecting PLNEI in patients with early-stage EC, combining with the results of the analysis in Table 1 (independent variables at P < 0.05), we included the presence or absence of PLNEI as the dependent variable, and the histological type, histological grade, depth of myometrial invasion, LVSI positive detection, LVSI type, cervical involvement, serum CA125, VI, TTV, TVR, and ADC_mean as the independent variables in a logistic univariate regression analysis model, which showed that all of the above parameters were influencing factors for the occurrence of PLNEI in patients with early-stage EC. Subsequently, logistics multivariate regression analysis was performed using the indicators with P < 0.05 in the logistics univariate regression analysis as the independent variables. As reflected by the results, serum CA125, VI, TTV, TVR, and ADC_mean were all independent influencing factors for PLNEI. The details were listed in Table 3.

Table 3 Logistic Univariate/Multivariate Regression Analyses on Influencing Factors for PLNEI in Patients with Early-Stage EC

VI, TTV, TVR and ADC_mean Had High Diagnostic Value for PLNEI in Patients with Early-Stage EC

ROC curve analysis was performed to evaluate the diagnostic value of VI, TTV, TVR and ADC_mean for PLNEI in early-stage EC patients. The results suggested that the AUC of VI value for diagnosing PLNEI in early-stage EC patients was 0.871, with the cut-off value of 33.94, the sensitivity of 71.43%, and the specificity of 92.19% (Figure 2A), that of TTV value was 0.875, with the cut-off value of 18.01, the sensitivity of 87.18%, and the specificity of 70.31% (Figure 2B), that of TVR value was 0.812, with the cut-off value of 24.80, the sensitivity of 55.13%, and the specificity of 89.06% (Figure 2C), and that of ADC_mean value was 0.683, with the cut-off value of 0.76, a sensitivity of 56.41%, and a specificity of 75.00% (Figure 2D). These findings indicated that VI value > 33.94, TTV value > 18.01, TVR value > 24.80, and ADC_mean value < 0.76 could aid in the diagnosis of PLNEI in early-stage EC patients. Furthermore, the AUC for their combination detection was 0.937, which was higher than the diagnostic value of VI, TTV, TVR or ADC_mean alone (all P < 0.05) (Figure 2E and F). Collectively, these data suggested that VI, TTV, TVR and ADC_mean had high diagnostic value for PLNEI in patients with early-stage EC. Please refer to Table 4 for more details.

Table 4 Independent Diagnostic Value of VI, TTV, TVR, and ADCmean and in Combination for PLNEI in Patients with Early-Stage EC

Figure 2 Diagnostic value of MRI parameters (VI, TTV, TVR, ADCmean) for PLNEI in patients with early-stage EC. ROC curve analysis to evaluate the diagnostic value of VI (A), TTV (B), TVR (C), and ADCmean values (D) and their combination (E) for PLNEI in patients with early-stage EC, respectively. Multiple ROC curves were compared (F).

Discussion

EC is the most frequent gynecologic malignancy in rich countries, with a mortality rate of about one-fifth.²⁹ EC is staged as per the International Federation of Obstetrics and Gynecology surgical system, and early diagnosis of the gynecologic malignancy stage is crucial for outcome prediction and treatment planning.³⁰ Among which, symptoms in patients diagnosed with early-stage EC tend to be postmenopausal vaginal bleeding, and EC in early-stage has a good survival rate and requires an adjunctive treatment in the light of the risk of recurrence.³¹ Notably, MRI occupies an important position in examining local staging of gynecologic tumors.³² Moreover, a growing number of studies have demonstrated that MRI has been applied to the study of clinical gynecologic malignancies.^33–35 Based on this context, our study probed the diagnostic value of MRI for PLNEI in early-stage EC patients. The experiment result unveiled that MRI was a good method to detect PLNEI in EC patients, and VI, TTV, TVR, and ADC_mean had high diagnostic value for PLNEI in early-stage EC patients.

Volume MRI-derived parameters, specifically VI, TTV, TVR have been shown to predict LVSI, and these MRI volume parameters can effectively stratify patients into high- and low-risk profiles prior to surgical intervention.²³ A prior study conducted by Nougaret et al has found that a TVR of 25% or greater indicates deep myometrial invasion, high-grade tumors, and LVSI.²⁴ The ADC value may play a significant role in the management of individualized treatment plans and in enhancing outcomes for patients diagnosed with EC, with a low ADC value usually representing the aggressive biological behavior of malignancies.^36,37 The ADC_mean is effective in characterizing the cellular properties of lesions, which exhibits latent intrinsic heterogeneity, and is the most commonly utilized metric for assessing the diffusion capacity of lesions.²¹ In line with the previous study, our study found that compared with EC patients without PLNEI, EC patients with PLNEI had higher VI, TTV and TVR values and lower ADC_mean value. Additionally, VI value > 33.94, TTV value > 18.01, TVR value > 24.80, and ADC_mean value < 0.76 had high diagnostic value for PLNEI in early-stage EC patients, with their combination manifesting higher value than them along. VI is recognized as a significant factor for predicting LNM in patients with EC.³⁸ Specifically, a VI > 17 cm³has been demonstrated to be correlated with diminished disease-free survival and overall survival rates, and the largest tumor volume exerts an independent effect on the recurrence and survival outcomes of EC patients.³⁹ The ADC_mean value of EC patients with stronger invasiveness and worse prognosis is lower.⁴⁰ Another key focus of our study, preoperative serum CA125, a pivotal indicator for predicting the prognosis of EC patients, comes into play in the selection of surgical treatment options.⁶ Serum CA125 level is distinctly higher in patients with worsening disease and with LNM in EC patients.⁴¹ Similarly, our study revealed that serum CA125 level, VI, TTV, TVR, and ADC_mean were independent influencing factors for PLNEI. To conclude, the evidence altogether unmasked that the MRI parameters played crucial clinical parts in diagnosing PLNEI in EC patients.

Various pathological parameters such as histological grade, histological type, depth of myometrial invasion, surgical pathological stage, and cervical involvement play a key part in the prognosis of EC.⁴² A prior study has demonstrated that histological grade, histological type, depth of myometrial invasion, cervical involvement, LVSI, parametrial involvement are distinctly relevant to LNM.⁵ Innovatively, our study unraveled that there were significant differences in histological type, histological grade, depth of myometrial invasion, LVSI positive detection, LVSI type, cervical involvement and serum CA125 level between early-stage EC patients with or without PLNEI. Moreover, distinct differences in histological type, histological grade, depth of myometrial infiltration, LVSI positive detection, LVSI type, cervical involvement, and serum CA125 level were found between patients with high or low VI values. Similarly, patients with high or low TTV values demonstrated significant differences in histological grade, depth of myometrial infiltration, LVSI positive detection, LVSI type, cervical involvement, and serum CA125 levels. Furthermore, prominent differences in histological grade, depth of myometrial infiltration, cervical involvement, LVSI type, and serum CA125 level were noticed between the patients with high or low TVR values. Additionally, patients with high or low ADC_mean values exhibited significant differences in histological grade, depth of myometrial infiltration, cervical involvement, and serum CA125 levels. Thus, it was persuasive to conclude that the MRI parameters had close associations with clinicopathological characteristics of patients with early-stage EC.

Taken together, this study found that MRI had a good effect on the detection of PLNEI in EC patients, and VI, TTV, TVR and ADC_mean values had high diagnostic value for PLNEI in early-stage EC patients. There were still limitations in our study. First, the numbers of cases and events included in this study for analysis were limited. Second, only the MRI parameters (VI, TTV, TVR and ADC_mean) were utilized for evaluation. In the future, we will conduct a larger multi-center study to expand the sample size and matched controls to increase the reliability of the results. In addition to MRI parameters (VI, TTV, TVR and ADC_mean), we will also combine MRI imaging, multifunctional MRI and positron emission tomography/MRI for differential diagnosis of PLNEI in patients with early-stage EC, thus further clarifying the diagnostic value of MRI for PLNEI in patients with early-stage EC.

Conclusion

All in all, our study highlighted that MRI parameters (VI, TTV, TVR, ADC_mean) were associated with clinicopathological characteristic of early EC patients, and were independent influencing factors for the occurrence of PLNEI. VI value >33.94, TTV value > 18.01, TVR value > 24.80, and ADC_mean value < 0.76 had high diagnostic value for PLNEI in early EC patients. These parameters assisted clinicians in the preoperative stratification of patients into high and low risk for PLNEI and played an important role in the planning of personalized treatment strategies and the selection of appropriate surgical interventions.