Dual-mode 3D shear wave elastography with qualitative and quantitative assessment improves diagnosis of inguinal lymph node metastasis

Abstract

Background

Inguinal lymph node (ILN) metastasis significantly affects prognosis and treatment strategies in patients with gynecological malignancies. Conventional ultrasound (US) provides morphological assessment but has limited sensitivity for detecting early metastatic changes. Three-dimensional shear wave elastography (3D-SWE) enables volumetric stiffness assessment with multiplanar visualization and may overcome the planar limitations of two-dimensional SWE (2D-SWE). This study aims to determine whether 2D-SWE or 3D-SWE improves the diagnostic performance of US for identifying ILN metastasis and to compare the complementary value of qualitative pattern analysis and quantitative elasticity metrics.

Methods

We retrospectively included 131 ILNs from 125 women with gynecological malignancies who underwent conventional US, 2D-SWE, and 3D-SWE. Qualitative color pattern analysis and quantitative elasticity parameters (Emax, Emean) were evaluated alone and in combination with US features. Histopathology served as the reference standard. Diagnostic performance was assessed using receiver operating characteristic (ROC) analysis and compared with DeLong tests.

Results

Of the 131 ILNs, 72 (55.0%) were malignant and 59 (45.0%) were benign. Among the three orthogonal 3D-SWE planes, the coronal plane demonstrated superior diagnostic performance across both qualitative and quantitative assessments. As standalone methods, qualitative coronal 3D-SWE achieved an area under the curve (AUC) of 0.846, while quantitative coronal-plane Emax achieved an AUC of 0.847, with no significant difference between them (P=0.920). When combined with conventional US, qualitative coronal 3D-SWE achieved the highest overall diagnostic performance [AUC 0.891, 95% confidence interval (CI): 0.834–0.947], with a sensitivity of 86.11% and a specificity of 88.05%, significantly outperforming conventional US alone (P<0.001). Among quantitative models, sagittal-plane Emax combined with US showed the best performance (AUC 0.858, 95% CI: 0.795–0.921), with a sensitivity of 87.50% and a specificity of 79.19%, also significantly better than conventional US alone (P<0.001). There was no significant difference between the best qualitative and quantitative combined models (AUC 0.891 vs. 0.858, P=0.331). Emax consistently outperformed Emean across all planes (AUC range, 0.809–0.847 vs. 0.749–0.795).

Conclusions

In this retrospective study, integrating 3D-SWE with conventional US significantly improved the detection of ILN metastasis through multiplanar qualitative and quantitative elasticity assessment. The coronal plane demonstrated superior diagnostic performance across both qualitative and quantitative assessments, and qualitative and quantitative 3D-SWE offered comparable incremental value when combined with US. These findings support 3D-SWE as a useful adjunct to conventional US for preoperative ILN evaluation in gynecological malignancies, particularly vulvar and vaginal cancers, in which ILN status is a key prognostic determinant.

Introduction

Inguinal lymph node (ILN) metastasis significantly impacts disease prognosis and treatment planning in gynecological malignancies, particularly in vulvar and vaginal cancers, in which ILN status is a critical prognostic factor. Accurate ILN staging enables safer selection of sentinel lymph node biopsy over systematic lymphadenectomy, reducing treatment-related morbidity such as lymphedema (1-3). While computed tomography (CT) and positron emission tomography-CT (PET-CT) are established for preoperative staging, ultrasound (US) demonstrates superior diagnostic performance for ILN evaluation. A Meta-analysis reported a pooled sensitivity of 0.85, specificity of 0.86, and negative predictive value (NPV) of 0.92, consistent with findings from individual studies (4-6). In comparison, CT demonstrates a lower NPV of approximately 75%, while PET-CT exhibits variable sensitivity (50–86%) (1,3). Accordingly, US is recommended as the first-line imaging modality for ILN staging in experienced centers, with CT/PET-CT reserved for assessing pelvic or distant metastases (7). Despite the good performance of conventional US, its diagnostic accuracy remains imperfect, and it primarily relies on morphological criteria. Shear wave elastography (SWE), a noninvasive technique that quantifies tissue stiffness, has demonstrated high diagnostic accuracy in detecting metastatic lymph nodes across various cancers, including breast, thyroid, and rectal malignancies (8-16). Studies suggest that SWE may provide higher specificity than conventional US in selected settings, thereby improving clinical decision-making for differential diagnosis, staging, and lymph node metastasis assessment (17-23). Recent advances integrating two-dimensional SWE (2D-SWE) with deep learning radiomics have further yielded noninvasive imaging biomarkers for assessing axillary lymph node metastasis in early breast cancer (24,25). The reproducibility of 2D-SWE in evaluating superficial ILNs has also been established, reinforcing its reliability for clinical use (26).

Three-dimensional SWE (3D-SWE), an advancement over its 2D counterpart, provides a noninvasive approach to tissue characterization by measuring and depicting the 3D distribution of tissue elasticity. This technique has been applied by Lee et al. and Youk et al. in the diagnosis of breast lesions and represents an important development in elastographic imaging technology (27,28). Subsequent studies by Tian et al. and Choi et al. demonstrated that SWE enhances the diagnostic performance of US in the evaluation and differentiation of breast tumors (29,30). Farghadani et al. reported that 3D-SWE had superior diagnostic value to magnetic resonance imaging (MRI) for the assessment of breast masses, further supporting its utility in clinical practice (31). Additionally, the combination of 3D-SWE and conventional US, as demonstrated by Xu et al., can reduce the incidence of unnecessary biopsies (32). As a diagnostic tool, 3D-SWE has also been proven beneficial in evaluating thyroid lesions, assessing liver volume post-ablation, and examining testicular pathology (33-36).

However, the comparative diagnostic performance of qualitative and quantitative 3D-SWE approaches for detecting ILN metastasis in gynecological malignancies has not been investigated. This retrospective study aimed to evaluate the diagnostic efficacy of dual-mode 3D-SWE, incorporating both qualitative and quantitative assessments, when combined with conventional US for identifying ILN metastases in patients with gynecological malignancies. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1314/rc).

Methods

Study population

We conducted a single-center retrospective diagnostic accuracy study of consecutive gynecological cancer patients who underwent ILN evaluation using conventional US, 2D-SWE, and 3D-SWE during the 18 months from January 1, 2019 to June 30, 2020. This retrospective study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Institutional Review Board of Sichuan Cancer Hospital (approval No. SCCHEC-03-2017-009). The requirement for informed consent was waived due to the retrospective nature of the study and the use of anonymized data.

Inclusion and exclusion criteria

From this database, lymph nodes were included if they met both criteria: (I) complete and diagnostically adequate conventional US, 2D-SWE, and 3D-SWE image sets; and (II) a definitive histopathological confirmation obtained via US-guided fine-needle aspiration cytology (FNAC) or surgical excision within 14 days of the SWE examination.

Lymph nodes were excluded if any of the following applied: (I) history of another active malignancy or prior radiotherapy/chemotherapy for the current gynecological cancer (to avoid confounding treatment effects on lymph node morphology and stiffness); (II) insufficient imaging data or poor image quality precluding reliable quantitative or qualitative analysis; or (III) lack of complete clinical follow-up data (Figure 1).

Figure 1.

Patient flow diagram. A retrospective study evaluating lymph node metastasis in gynecological tumor patients using conventional US, 2D-SWE, 3D-SWE. 2D-SWE, two-dimensional shear wave elastography; 3D-SWE, three-dimensional shear wave elastography; FNA, fine-needle aspiration; SWE, shear wave elastography; US, ultrasound.

Histopathological diagnosis

Histopathological confirmation was obtained via FNAC or surgical excision. For pathological diagnosis, cytological smears were stained with Papanicolaou stain, and histological sections were stained with hematoxylin and eosin (H&E). Immunohistochemistry was applied when necessary to confirm metastatic origins. Two pathologists with >10 years of experience independently reviewed all specimens. Benign lymph nodes demonstrated preserved nodal architecture with reactive changes; metastatic nodes contained malignant cells consistent with the primary tumor. Benign FNAC diagnoses were confirmed by at least 12 months of follow-up, demonstrating stability.

Image acquisition

US examination

All US examinations were performed using an AixPlorer US system (SuperSonic Imagine, Aix-en-Provence, France) with a 4–15 MHz linear-array transducer. Two board-certified radiologists (Y.S. and J.D., each with >12 years of experience in gynecologic oncology imaging) independently evaluated all ILNs, blinded to histopathologic results. For each lymph node, the following morphologic features were systematically assessed: calcification, liquefaction, echogenic enhancement, vascularity pattern on color Doppler imaging, fatty hilum status, cortical thickening, and border definition.

SWE examination

SWE was performed using the same AixPlorer US system. For 2D-SWE, an SL10-2 linear probe (2–10 MHz) was used; for 3D-SWE, an SL15-4 volumetric probe (4–15 MHz) was employed. The elasticity scale was fixed at 0–180 kPa for all measurements. Immediately following conventional US, both 2D and 3D SWE were performed by the same experienced radiologists using identical patient positioning and protocols to minimize variability. 2D-SWE was acquired in the transverse plane. The sampling frame was positioned to fully encompass the entire lymph node. After allowing the image to stabilise for 3–5 seconds, patients were instructed to briefly hold their breath while a dynamic image sequence was captured, thereby minimising motion artifacts. The frame exhibiting uniform elastic colour filling and optimal temporal stability was selected as the reference frame for analysis. Technical parameters including focal position, imaging depth, and gain, were standardized across examinations to maintain consistency. For 3D-SWE, the SL15-4 probe was positioned at the same transverse plane used for 2D-SWE acquisition to ensure spatial correspondence. A volumetric sweep was performed with the patient maintaining a breath-hold to minimize motion artifacts, generating spatially co-registered multiplanar data from a single-volume acquisition. All imaging data were systematically archived, with each individual lymph node serving as the analytical unit.

Image quality control and data curation

All original DICOM images were anonymised and stored immediately upon export. An experienced radiologist with over 5 years of experience in elastography, who was not involved in subsequent analysis, conducted quality screening of all SWE images. Acceptance criteria included: adequate colour filling with minimal inter-frame variation, absence of significant motion artefacts, and parameters conforming to preset specifications. Compliant images were included. Subsequently, all images were reassigned randomised identifiers and allocated to independent analysis datasets, ensuring analyses were conducted under complete blinding.

SWE image analysis

Quantitative SWE assessment

2D SWE analysis

Offline analysis was performed on anonymized images by two experienced radiologists independently. A standardized circular 2-mm region of interest (ROI) was manually placed at the visually stiffest cortical area, avoiding cystic, necrotic, or calcified regions. Maximum (Emax) and mean (Emean) elasticity values (kPa) were recorded. Each measurement was repeated three times, and the average was calculated.

3D SWE analysis

A structured “automatic initial screening-grid analysis-manual confirmation” workflow was implemented for objective ROI selection. The system automatically identified and highlighted the highest-stiffness region within the entire 3D volume of the lymph node and then generated three orthogonal planes (transverse, coronal, sagittal) centered on the identified stiffest region and created a 3×3 slice array for comprehensive spatial sampling (Figure 2). A radiologist subsequently reviewed the automatically selected regions on B-mode/elastography images to confirm anatomic validity, then selected the maximum-stiffness slice from each orthogonal plane, excluding cystic, necrotic, calcified, or perinodal tissues. On each confirmed plane, a 2-mm circular ROI was placed in the stiffest cortical region using the same principles as in 2D-SWE, yielding six quantitative parameters per lymph node: transverse plane (3D-SWE-A-Emax, 3D-SWE-A-Emean), coronal plane (3D-SWE-C-Emax, 3D-SWE-C-Emean), and sagittal plane (3D-SWE-S-Emax, 3D-SWE-S-Emean). Each measurement was repeated three times and averaged.

Figure 2.

Multi-slice 3D SWE assessment of an inguinal lymph node. (A) Conventional B-mode ultrasound image showing a lymph node with loss of fatty hilum and a longitudinal-to-transverse diameter ratio >0.5. (B-D) Multi-slice 3D SWE images derived from a 3×3 slice array centered on the stiffest region, displayed in three orthogonal planes: (B) transverse; (C) coronal; (D) sagittal. The mass exhibits heterogeneous stiffness with a characteristic “hard-rim sign” (peripheral stiffening relative to the center). 3D, three-dimensional; SWE, shear wave elastography.

Qualitative SWE assessment

2D SWE analysis

Using the methods of Luo et al., the qualitative assessment of lymph nodes was categorized into four color patterns, similar to those for the diagnosis of breast lesions (37-39). Color pattern 1: a homogeneous pattern characterized by uniform blue staining of the nodal region; color pattern 2: filling defect within the lymph node; color pattern 3: even distribution within the lymph node, with localized stained areas at the margins; color pattern 4: filling defect within the lymph node, accompanied by localized stained areas at the margins. These four mutually exclusive patterns were defined by the presence or absence of two features on the color elastogram: filling defects (i.e., ‘black hole’ or ‘hard-rim sign’) and marginal staining (i.e., locally stained regions), indicating increased stiffness (39) (Figure 3).

Figure 3.

3D-SWE images of benign and malignant inguinal lymph nodes. (A) A 48-year-old female patient diagnosed with stage I cervical cancer with benign lymph nodes. (B,C) A 55-year-old female patient diagnosed with stage IIIC hypo-differentiated squamous cell carcinoma of the cervix with ILN metastases. Color pattern 3 (homogeneity within the lymph node, with locally stained areas at the margins) is observed in the transverse section. Color pattern 4 (filling the defect within the lymph node, with locally colored areas on the margins) is observed in the sagittal section. In the coronal plane, color pattern 3 (homogeneous within the lymph node, with locally discolored areas at the margins) is observed, and a ring of high hardness (referred to as a ‘hard ring’) appears around the periphery of the lymph node, whereas the central part is relatively soft. Metastatic lymph nodes were confirmed via puncture biopsy of the lymph nodes in the inguinal region. Staining method: top row (two panels): hematoxylin and eosin (H&E), ×200; bottom row (two panels): papanicolaou, ×400. 3D, three-dimensional; ILN, inguinal lymph node; SWE, shear wave elastography.

3D SWE analysis

The same four-colour pattern classification system was applied independently to the three reconstructed planes of 3D-SWE (transverse, coronal, and sagittal). Each plane was interpreted independently.

All the US and SWE images were evaluated by two experienced radiologists, each with over 10 years of clinical expertise. In instances of disagreement, a consensus was reached through collaborative discussions.

Diagnostic strategy

US

A lymph node was classified as suspicious for metastasis if any of the following features were present: microcalcification, liquefaction, diffuse/focal echogenic enhancement, peripheral or mixed vascularity, absent hilum, ill-defined margins, or focal/diffuse cortical thickening (40-43).

Qualitative SWE

For 2D-SWE, Color pattern 1 indicated benign lymph nodes, whereas color patterns 2–4 were considered indicative of metastatic infiltration (39,42,44-46). For 3D-SWE, each plane was independently assessed. All qualitative assessments were performed independently by two experienced radiologists; discrepancies were resolved by consensus.

Quantitative SWE

Optimal diagnostic thresholds were determined via ROC analysis using the Youden index. Values exceeding the threshold were classified as malignant, whereas those below the threshold were classified as benign.

Construction and comparison of joint diagnostic models

Due to significant inter-planar multicollinearity among SWE parameters derived from the three orthogonal planes (Pearson correlation coefficients: 0.85–0.91, P<0.001; variance inflation factor: 9–14, all >5), simultaneous multi-planar modeling was precluded. Separate logistic regression models were constructed for each plane (transverse, sagittal, and coronal), integrating US features with either qualitative SWE scores or quantitative parameters (Emax or Emean) from the respective plane. The optimal model from each combination was identified by its area under the curve (AUC). The highest-performing qualitative model was designated Model A, and its quantitative counterpart was Model B.

Outcome measures

The primary outcome was the diagnostic performance (AUC, sensitivity, specificity) of 3D-SWE combined with US for detecting ILN metastasis.

Secondary outcomes included: (I) comparison of qualitative versus quantitative 3D-SWE; (II) comparison across different 3D-SWE planes; and (III) comparison of 2D-SWE versus 3D-SWE.

Statistical analysis

Histopathological results served as the gold standard. Continuous variables were assessed for normality using the Shapiro-Wilk test: normally distributed data were presented as mean ± standard deviation and compared using independent samples t-tests; non-normally distributed data were expressed as median [interquartile range (IQR)] and compared using the Wilcoxon rank-sum test. Categorical variables were presented as frequencies (percentages) and analyzed using the chi-square test or Fisher’s exact test. Diagnostic performance was evaluated using receiver operating characteristic (ROC) curve analysis, with optimal thresholds determined by maximizing the Youden index. The AUC with 95% confidence intervals (CIs), sensitivity, and specificity were calculated for each parameter, qualitative score, and combined model. DeLong’s test was used to compare AUCs between different diagnostic strategies. Logistic regression was used to construct joint diagnostic models combining conventional US with SWE parameters. Inter-observer agreement was assessed using the intraclass correlation coefficient (ICC) for quantitative parameters (<0.50: poor; 0.50–<0.75: moderate; 0.75–0.90: good; >0.90: excellent) and weighted kappa coefficients for qualitative scores (values <0.40: poor; 0.40–0.60: moderate; 0.61–0.80: substantial; 0.81–1.00: almost perfect) (47). Statistical analyses were performed using R (version 4.5.0), GraphPad Prism (version 9.7.0), and MedCalc (version 22).

Results

Study population and pathological findings

A total of 125 patients (all female; mean age, 54.7±10.50 years; range, 35–72 years) were retrospectively enrolled in this study between January 1, 2019 and June 30, 2020. No significant differences in baseline characteristics were observed between the benign and malignant groups (all P>0.05) (Table S1). The demographic characteristics are summarized in Figure 4.

Figure 4.

The basic demographic characteristics of the subjects. The study cohort comprised 125 patients, with data collected on age, menopausal status, BMI, internal medical comorbidities, primary tumor type, CA125 levels, and ultrasound characteristics. BMI, body mass index; CA125, cancer antigen 125; CDFI, color Doppler flow imaging; SD, standard deviation.

Among the 131 ILNs analyzed, pathological examination revealed 59 benign lymph nodes (45.0%) and 72 malignant lymph nodes (55.0%). All lymph nodes were confirmed by pathological examination, of which 124 (94.7%) were diagnosed via FNAC, and 7 (5.3%) were diagnosed via direct surgical excision biopsy.

Cervical cancer accounted for the majority of cases (n=80, 61.1%), followed by vulvar cancer (n=27, 20.6%), vaginal cancer (n=13, 9.9%), ovarian cancer (n=7, 5.3%), endometrial cancer (n=2, 1.5%), and fallopian tube cancer (n=2, 1.5%) (Table S1).

US findings

Conventional US demonstrated moderate diagnostic performance with an AUC of 0.772 (95% CI: 0.690–0.841), sensitivity of 76.39%, specificity of 78.00%, positive predictive value (PPV) of 80.88%, NPV of 73.02%, and accuracy of 77.10%. Among morphologic features, the malignant group exhibited a significantly higher prevalence of liquefaction [11.1% (8/72) vs. 1.7% (1/59); P=0.041] and loss of fatty hilum [47.2% (34/72) vs. 16.9% (10/59); P<0.001] compared with the benign group. No significant differences were observed in calcification (P=0.976), echogenic enhancement (P=0.092), color Doppler flow imaging patterns (P=0.059), cortical thickening (P=0.123), or border characteristics (P=0.184) (Table S2).

Elastography parameters

Both Emax and Emean values were significantly higher in metastatic lymph nodes compared with benign nodes across all imaging modalities (all P<0.001). For 2D-SWE, median Emax was 39.1 kPa (IQR, 24.4–79.3) in metastatic nodes versus 17.4 kPa (IQR, 13.8–26.4) in benign nodes. Among 3D-SWE planes, the coronal plane showed the greatest stiffness differential: median Emax of 58.3 kPa (IQR, 26.6–96.9) in metastatic nodes versus 21.3 kPa (IQR, 17.5–25.0) in benign nodes. The transverse and sagittal planes showed similar trends (transverse Emax: 58.1 vs. 21.3 kPa; sagittal Emax: 51.5 vs. 20.8 kPa). Emean values followed a consistent pattern, with metastatic nodes ranging from 32.5 to 38.5 kPa versus 15.5 to 18.1 kPa in benign nodes (Table S3) (Figure 5). The Inter-observer ICC was 0.87 (95% CI: 0.79–0.92) and the intra-observer ICC was 0.91 (95% CI: 0.85–0.95), confirming excellent reproducibility.

Figure 5.

Heatmap of SWE parameters differentiating malignant and benign inguinal lymph nodes. The heatmap illustrates the comparison of maximum (Emax) and mean (Emean) elastic modulus values between malignant and benign lymph nodes across 2D-SWE, and 3D-SWE in transverse, sagittal, and coronal planes. Malignant nodes showed significantly higher Emax and Emean values in all evaluated SWE modalities and planes. Color intensity represents relative parameter values (warmer colors = higher values). 2D, two-dimensional; 3D, three-dimensional; SWE, shear wave elastography.

Qualitative elastography analysis

Qualitative SWE color patterns 2–4 (indicating heterogeneous stiffness distribution with hard-rim signs or filling defects) were significantly more prevalent in malignant lymph nodes across all imaging planes P<0.001). The prevalence of patterns 2–4 in malignant nodes increased progressively across planes: 2D-SWE, 77.8% (56/72); 3D-SWE transverse, 83.3% (60/72); 3D-SWE sagittal, 81.9% (59/72); and 3D-SWE coronal, 86.1% (62/72). Conversely, benign nodes showed low rates of patterns 2–4: 2D-SWE, 22.0% (13/59); 3D-SWE transverse, 18.6% (11/59); 3D-SWE sagittal, 16.9% (10/59); and 3D-SWE coronal, 16.9% (10/59) (Table 1). The 3D-SWE coronal plane demonstrated the highest detection rate with the largest Z-statistic (−7.886). Agreement for qualitative pattern classification was excellent, with weighted kappa values of 0.85 (95% CI: 0.78–0.90) for inter-observer and 0.90 (95% CI: 0.86–0.94) for intra-observer assessment.

Table 1. Qualitative analysis of 2D-SWE and 3D-SWE planes for metastatic and benign ILN.

Imaging plane	Metastatic (n=72)	Benign (n=59)	Z	P value
2D-SWE			−6.333	<0.001
Present	56	13
Absent	16	46
3D transverse SWE			−7.365	<0.001
Present	60	11
Absent	12	48
3D sagittal SWE			−7.385	<0.001
Present	59	10
Absent	13	49
3D coronal SWE			−7.886	<0.001
Present	62	10
Absent	10	49

2D, two-dimensional; 3D, three-dimensional; ILN, inguinal lymph node; SWE, shear wave elastography.

Diagnostic performance of the quantitative findings

The diagnostic performance of individual SWE parameters is summarized in Table 2. Among quantitative measures, 3D-SWE parameters achieved superior diagnostic performance compared with 2D-SWE. For 2D-SWE, the AUCs of Emax and Emean were 0.809 (95% CI: 0.735–0.873) and 0.749 (95% CI: 0.666–0.821), respectively. Among 3D-SWE quantitative parameters, Emax consistently outperformed Emean across all planes. The Emax AUCs for the transverse, sagittal, and coronal planes were 0.840 (95% CI: 0.765–0.898), 0.830 (95% CI: 0.754–0.890), and 0.847 (95% CI: 0.774–0.904), respectively, with the coronal plane achieving the highest diagnostic efficacy. The corresponding Emean AUCs were 0.795 (95% CI: 0.716–0.860), 0.766 (95% CI: 0.684–0.836), and 0.784 (95% CI: 0.704–0.852), respectively. The differences in Emax AUCs between 2D-SWE and 3D-SWE planes did not reach statistical significance (all P>0.05, DeLong test). Conventional US yielded an AUC of 0.772 (95% CI: 0.690–0.841).

Table 2. Diagnostic performance of ultrasound and SWE parameters for differentiating metastatic from benign lymph nodes.

Parameter	AUC (95% CI)	Sensitivity (%)	Specificity (%)	Cut-off (kPa)	P value†
Quantitative SWE parameters
2D-SWE
Emax	0.809 (0.735–0.873)	80.56	72.88	22.50	0.287
Emean	0.749 (0.666–0.821)	78.47	70.34	17.20	0.323
3D-SWE
Transverse plane
Emax	0.840 (0.765–0.898)	77.78	76.27	25.00	0.019
Emean	0.795 (0.716–0.860)	75.00	72.03	18.50	0.052
Sagittal plane
Emax	0.830 (0.754–0.890)	73.61	76.27	27.30	0.054
Emean	0.766 (0.684–0.836)	72.22	73.73	20.00	0.078
Coronal plane
Emax	0.847 (0.774–0.904)	79.17	74.58	25.10	0.008
Emean	0.784 (0.704–0.852)	76.39	71.19	18.80	0.756
Qualitative SWE patterns‡
2D-SWE pattern	0.779 (0.698–0.847)	77.78	78.00	–	0.870
3D-SWE patterns
Transverse plane	0.823 (0.747–0.885)	83.33	81.36	–	0.227
Sagittal plane	0.825 (0.749–0.886)	81.94	83.05	–	0.217
Coronal plane	0.846 (0.772–0.903)	86.11	83.05	–	0.084
Conventional US features	0.772 (0.690–0.841)	76.39	78.00	–	–

^†, P values represent comparisons with conventional US features using DeLong’s test; ^‡, qualitative assessments based on elasticity distribution patterns (see Methods for grading criteria). 2D, two-dimensional; 3D, three-dimensional; AUC, area under the curve; CI, confidence interval; Emax, maximum elasticity; Emean, mean elasticity; SWE, shear wave elastography; US, ultrasound.

Compared with conventional US, 3D-SWE coronal Emax showed the largest incremental gain (ΔAUC =0.075, P=0.008), followed by the transverse plane (ΔAUC =0.068, P=0.019) and sagittal plane (ΔAUC =0.058, P=0.054). Complete diagnostic performance metrics, including PPV, NPV, accuracy, and Youden index based on optimal Emax cutoff values are detailed in Table S4.

Diagnostic performance of the qualitative findings

The AUC values for qualitative assessment were: 2D-SWE, 0.779 (95% CI: 0.698–0.847); 3D-SWE transverse, 0.823 (95% CI: 0.747–0.885); sagittal, 0.825 (95% CI: 0.749–0.886); and coronal, 0.846 (95% CI: 0.772–0.903). The 3D-SWE coronal pattern achieved the highest AUC with a sensitivity of 86.11% and specificity of 83.05%, approaching statistical significance compared with conventional US (P=0.084, DeLong test) (Table S5). No significant difference was observed between the qualitative and quantitative approaches in the coronal plane (AUC 0.846 vs. 0.847; P=0.920, DeLong test). Complete diagnostic performance parameters are detailed in Table 2.

Combined diagnostic models

To evaluate the incremental diagnostic value of SWE, we constructed 12 combined logistic regression models incorporating each SWE parameter with conventional US features (Table 3). Most combined models demonstrated improved diagnostic performance compared with conventional US alone (AUC 0.772). Emax-based and qualitative models showed the largest AUC increments (ΔAUC ranging from 0.057 to 0.119), while Emean-based models exhibited relatively modest gains (ΔAUC 0.026–0.052). The highest AUC among qualitative models was achieved by Model A (3D-SWE coronal qualitative pattern + US), with an AUC of 0.891 (95% CI: 0.834–0.947), sensitivity of 86.11%, and specificity of 88.05%, which was significantly higher than conventional US alone (P<0.001, DeLong test) (Table 3).

Table 3. Comparative diagnostic performance of conventional US and combined SWE models.

Model	AUC (95% CI)	Sensitivity (%)	Specificity (%)	P value†
Baseline model
Conventional US (reference)	0.772 (0.690–0.841)	76.39	78.00	–
Qualitative SWE combined models
2D-SWE qualitative + US	0.836 (0.768–0.904)	72.22	91.92	0.017
3D-SWE transverse qualitative + US	0.872 (0.812–0.933)	72.22	94.92	<0.001
3D-SWE sagittal qualitative + US	0.873 (0.812–0.934)	72.24	96.61	<0.001
3D-coronal-qualitative + US (Model A)	0.891 (0.834–0.947)	86.11	88.05	<0.001
Quantitative SWE combined models
2D-SWE Emax + US	0.829 (0.758–0.900)	83.33	72.88	0.004
2D-SWE Emean + US	0.798 (0.722–0.875)	77.78	77.97	0.180
3D-SWE transverse Emax + US	0.849 (0.783–0.914)	72.22	84.75	<0.001
3D-SWE transverse Emean + US	0.824 (0.752–0.896)	83.33	72.88	0.011
3D-sagittal-Emax+ US (Model B)	0.858 (0.795–0.921)	87.50	79.19	<0.001
3D-SWE sagittal Emean + US	0.819 (0.745–0.893)	79.17	77.97	0.025
3D-SWE coronal Emax + US	0.844 (0.777–0.911)	76.39	83.05	<0.001
3D-SWE coronal Emean + US	0.812 (0.736–0.887)	81.94	74.58	0.060

Model A, conventional US combined with 3D-SWE coronal plane qualitative score (the best-performing qualitative model); Model B, conventional US combined with 3D-SWE sagittal plane Emax value (the best-performing quantitative model). Combined models were developed using logistic regression; highly correlated variables were not entered simultaneously to avoid multicollinearity. ^†, compared with conventional US (DeLong test). 2D, two-dimensional; 3D, three-dimensional; AUC, area under the curve; CI, confidence interval; Emax, maximum elasticity; SWE, shear wave elastography; US, ultrasound.

Among quantitative models, Model B (3D-SWE sagittal Emax + US) showed the best performance, with an AUC of 0.858 (95% CI: 0.795–0.921), sensitivity of 87.50%, and specificity of 79.19%, significantly outperforming conventional US alone (P<0.001, DeLong test) (Figure 6). No statistically significant difference was found between Model A and Model B (ΔAUC =0.033, P=0.331), suggesting that qualitative and quantitative 3D-SWE approaches provide comparable incremental diagnostic value when combined with conventional US.

Figure 6.

ROC curves comparing diagnostic performance of US, SWE, and combined modalities for lymph node metastasis assessment. (A) ROC curves for lymph node assessment via US, 2D-SWE Emax, and 3D-SWE Emax in the transverse, sagittal, and coronal planes, as well as quantitative 3D-SWE combined with US. (B) ROC curves for the assessment of lymph nodes via US, qualitative 2D-SWE, qualitative 3D-SWE in the transverse, sagittal, and coronal planes; and qualitative 3D-SWE combined with US. 2D, two-dimensional; 3D, three-dimensional; 2D-SWE-Emax, the maximum elasticity modulus of 2D-SWE; 3D-SWE-A, transverse plane of 3D-SWE; 3D-SWE-A-Emax, the maximum elasticity modulus in the transverse plane of 3D-SWE; 3D-SWE-C, coronal plane of 3D-SWE; 3D-SWE-C-Emax, the maximum elasticity modulus in the coronal plane of 3D-SWE; 3D-SWE-S, sagittal plane of 3D-SWE; 3D-SWE-S-Emax, the maximum elasticity modulus in the sagittal plane of 3D-SWE; ROC, receiver operating characteristic; SWE, shear wave elastography; US, ultrasound.

Discussion

Accurate preoperative assessment of ILN metastatic status is essential for treatment planning in gynecologic oncology, particularly for vulvar and vaginal cancers, where ILN involvement directly influences surgical extent and prognosis. Although CT and PET-CT remain the standard imaging modalities for systemic staging, US proves superior for superficial ILN assessment, with reported pooled sensitivity and specificity values supporting its use as the primary imaging modality (4,48-50). However, conventional US relies primarily on morphological criteria, and its diagnostic accuracy remains imperfect. 3D-SWE captures spatially coregistered multiplanar elasticity data in a single volumetric acquisition, enabling comprehensive assessment of lymph node stiffness heterogeneity across all imaging planes simultaneously, a capability not achievable with conventional 2D-SWE. This retrospective study demonstrates that integrating 3D-SWE with conventional US significantly improves discrimination between metastatic and benign ILNs in patients with gynecological cancers, with the coronal plane providing the highest incremental diagnostic value.

Although 2D-SWE combined with US has shown promise in various clinical applications (51-60), evidence for its diagnostic utility in lymph node metastasis detection remains inconsistent and warrants further investigation. Chami et al. reported no diagnostic improvement for malignant lymph nodes using this combination, possibly due to the high proportion of lymphomas (45). This inconsistency underscores the potential advantage of 3D-SWE’s volumetric assessment over planar evaluation.

Qualitatively, 3D-SWE combined with US, particularly analysis of the coronal plane, proved the most effective approach for identifying metastatic lymph nodes. The observed “black hole” or “hard-rim sign” (color patterns 2–4) in metastatic lymph nodes aligns with prior descriptions, including “nodular pattern”, “markedly heterogeneous with some unfilled regions”, “a colored area at the margin”, and “hard-rim sign” (38,39,55,61-63). Chen et al. and Xue et al. demonstrated that the “crater sign” (also referred to as the hard-rim sign) of 3D SWE in the coronal plane plays a significant role in improving the specificity of the qualitative diagnosis for breast tumors (63,64). Hard-rim sign formation results from cancer cell invasion via afferent lymphatic vessels, leading to increased tumor cell density and stiffness, focal staining at the marginal sinuses, and ultimately structural disruption, marked stiffness elevation, and adhesion to surrounding tissues (39,65). Filling defects in metastatic lymph nodes result from three primary mechanisms: shear wave energy attenuation at the lymph node margin; extremely low echogenicity or stiffness exceeding the SWE measurement threshold; and obstruction of shear wave propagation by cancer cell growth-related features including heterogeneity, neovascularization, necrosis, and inflammation (66-68). Collectively, these factors impact the filling of SWE images and offer valuable imaging information for diagnosing metastatic lymph nodes. In our study, the qualitative SWE patterns, particularly color patterns 2–4, are more frequently observed in the coronal plane, which is less susceptible to acoustic attenuation and thus more accurately reflects the lymph node elasticity. Additionally, the coronal view is more sensitive in detecting features such as tumor margin invasion and tissue adhesion, which may be less apparent in other planes.

The pathophysiological basis for the diagnostic utility of SWE can be inferred by correlating elastographic findings with established US features of malignancy. The high sensitivity of qualitative patterns 2–4 (hard-rim sign, filling defects) for metastasis (Table 1) corresponds to the desmoplastic reaction and focal tumor deposits that increase tissue stiffness. This mechanistic interpretation is supported by our conventional US findings showing significant association between metastatic ILNs and loss of fatty hilum (P<0.001) and presence of liquefaction (P=0.041), as detailed in Table S2. Liquefactive necrosis, commonly seen in metastatic deposits, likely explains the “filling defect” pattern (soft areas within a stiff rim), a feature more comprehensively captured in 3D volumetric data than in single-plane 2D-SWE. The observed stiffness elevation in metastatic nodes reflects underlying extracellular matrix remodeling and tissue sclerosis induced by tumor invasion (69), a biomechanical alteration that may theoretically precede morphological changes detectable on conventional US (70). Consistent with this mechanistic interpretation, previous studies have demonstrated that integrating 3D-SWE with conventional US consistently enhances diagnostic performance across various clinical settings, including quantitative Emax assessment, combined BI-RADS evaluation, and reduction of unnecessary biopsies, supporting the clinical value of these biomechanical correlates (29,30,32,71).

The clinical significance of our findings is underscored by the enhanced performance of combined diagnostic models (Table 3). Integrating 3D-SWE coronal qualitative patterns with US features (Model A) increased the AUC from 0.772 to 0.891, with a sensitivity of 86.11% and specificity of 88.05%. This represents a clinically meaningful improvement that could reduce indeterminate diagnoses and guide clinical decision-making. In practice, a lymph node with suspicious US features (e.g., absent hilum, liquefaction) that also exhibits malignant stiffness patterns (color patterns 2–4) on 3D-SWE would have a high probability of metastasis, potentially prompt definitive histopathological confirmation, and obviating the need for short-interval follow-up. Conversely, a node with benign US morphology and a homogeneous soft elastogram (pattern 1) could be monitored with greater confidence, potentially reducing unnecessary invasive procedures.

In our study, metastatic ILNs exhibited significantly higher stiffness than benign nodes across all imaging modalities, consistent with established literature (20,21,72-74). Both qualitative pattern analysis and quantitative Emax measurement demonstrated high and comparable diagnostic efficacy when assessed individually in the coronal plane (AUC: 0.846 vs. 0.847; P=0.920, DeLong test), yet they capture distinct dimensions of elasticity information—spatial distribution versus peak mechanical properties—providing complementary diagnostic value. This finding aligns with breast imaging evidence demonstrating equivalent 2D-SWE performance between qualitative and quantitative approaches (51,75), yet contrasts with reports favoring qualitative assessment alone (39,46). This discrepancy likely reflects technical variability in ROI placement and threshold dependency inherent to single-plane 2D-SWE, limitations that 3D-SWE mitigates through spatially coregistered multiplanar data acquisition, yielding more reproducible peak stiffness measurements (39,76,77). The qualitative patterns map spatial stiffness heterogeneity through color-coded visualization, enabling identification of the “hard-rim sign” indicative of capsular invasion, whereas Emax quantifies peak stiffness at the mechanically stiffest focus. This dual-mode approach leverages qualitative assessment for spatial pattern delineation, while quantitative Emax measurement provides standardized, threshold-based metrics (optimal cutoff: ~25 kPa), ensuring objective reporting and inter-observer consistency.

The consistently larger AUC of Emax compared with Emean across all imaging planes (Table 2) has important implications for clinical practice. Maximum stiffness values capture the most pathologically altered tissue components, whereas mean values are diluted by heterogeneous tissue composition including surrounding normal parenchyma. This distinction is particularly relevant in partially infiltrated lymph nodes, where malignant foci may be focal rather than diffuse. Consistent with previous validation studies and meta-analyses (16,46,74,78,79), Emax more effectively identifies malignant nodes by targeting peak stiffness values, while the averaging effect of Emean reduces sensitivity for detecting focal areas of increased stiffness characteristic of early or partial metastatic infiltration.

In our study, the flat topography of the inguinal region enhances US probe contact and minimizes soft tissue strain, affording clearer elastographic images than the irregular surfaces of the neck and axilla. This anatomical advantage makes the inguinal region particularly suitable for elastography assessment. In contrast, Chami et al. argued that quantitative 2D-SWE is not capable of diagnosing malignancy in axillary lymph nodes, partly because of its deeper location and increased susceptibility to signal loss compared with other SWE locations (45).

Our findings demonstrate that elasticity parameters derived from multiple imaging planes exhibit significant predictive value for differentiating benign from malignant lymph nodes, with notable plane-specific differences. Among the three orthogonal planes, the coronal plane demonstrated superior diagnostic performance: qualitatively, coronal patterns achieved the highest sensitivity (86.11%) and optimal combined model efficacy (AUC 0.891); quantitatively, coronal Emax yielded the highest individual AUC (0.847). We hypothesize that the coronal plane’s superior performance reflects its ability to detect biomechanical changes along the longitudinal axis of the lymph node. Tumor infiltration via afferent lymphatic vessels induces characteristic changes—including capsular tension and structural distortions—that become detectable before morphological changes are evident on conventional imaging. However, the precise mechanistic basis for these plane-specific differences warrants further investigation.

While these findings support the diagnostic value of 3D-SWE in distinguishing ILN metastasis, several pathological and technical constraints warrant consideration. Pathologically, benign and malignant lesions may exhibit overlapping elasticity: reactive lymph nodes with extensive fibrosis can mimic metastatic stiffness through collagen deposition, whereas necrotic or cystic metastatic nodes may appear falsely soft due to reduced cell density in areas of necrosis. We mitigated such confounding through stringent exclusion criteria, including prior radiotherapy or chemotherapy. Technically, the core challenge in locally sampled elastography is whether the ROI accurately reflects the stiffness of a heterogeneous lesion. Our structured “automatic initial screening-grid analysis-manual confirmation” workflow systematically identifies the stiffest solid regions while excluding non-solid areas, enhancing measurement objectivity compared with 2D-SWE single-plane selection. Future efforts should integrate multi-parameter diagnostic models and artificial intelligence (AI)-assisted ROI segmentation to further enhance measurement objectivity and address the diagnostic ambiguity of moderate elasticity values, particularly in low tumor burden states such as micrometastases.

Limitations

Given the single-center study, our findings require validation through prospective, multicenter investigations to establish their generalizability across diverse patient populations, tumor types, and body mass index (BMI) distributions. Significant inter-planar collinearity precluded simultaneous multi-planar modeling, limiting the assessment of their combined diagnostic effects. Future research may further expand the sample size and explore the effective integration of multi-planar information. Additionally, as a retrospective design relying on routine H&E-based histopathology without systematic ultrastaging, we could not definitively classify micrometastases, and these findings require validation through prospective investigations with standardized ultrastaging protocols.

Conclusions

In this retrospective study, integrating 3D-SWE with conventional US significantly improved ILN metastasis detection through multiplanar qualitative and quantitative elasticity evaluation. The coronal plane demonstrated the highest diagnostic performance using either qualitative pattern scoring or quantitative Emax measurement. These findings support 3D-SWE as a valuable adjunct to conventional US for preoperative ILN assessment in patients with gynecological malignancies, particularly in vulvar and vaginal cancers, where ILN status is a key prognostic determinant.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This retrospective study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments, and was approved by the Institutional Review Board of Sichuan Cancer Hospital (approval No. SCCHEC-03-2017-009). The requirement for informed consent was waived due to the retrospective nature of the study and the use of anonymized data.

Data Sharing Statement

Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1314/dss

qims-16-06-480-dss.pdf ^{(142.5KB, pdf)}

DOI: 10.21037/qims-2025-1314