Photo Credit: Mohammed Haneefa Nizamudeen
Combining MR elastography with clinical radiological data improved the prediction of lymphovascular space invasion in patients with endometrial cancer.
In patients with endometrial cancer who are scheduled for surgical intervention, lymphovascular space invasion (LVSI) is an important characteristic to identify prior to surgery execution. The challenge is that identifying LVSI requires a pathologic examination after a hysterectomy has already been performed.
Jiuquan Zhang, MD, PhD, and colleagues developed a prospective single-center study at a university teaching hospital to explore noninvasive testing options that could offer predictive value regarding LVSI before surgical intervention. As written by Dr. Zhang and colleagues in Radiology. “This study hypothesized that MR [magnetic resonance] elastography–based mechanical characteristics could be used to predict LVSI status. Therefore, the purpose of this study was to evaluate MR elastography–based mechanical characteristics for noninvasive prediction of LVSI status in participants with endometrial cancer.”
MR Elastography & Tumor Stiffness
Studies have found that MR elastography, considered a magnetic form of palpation, provides viable and applicable information on tumor stiffness. This information can help assess tumor aggressiveness, which can help assign tumor grade, International Federation of Gynecology and Obstetrics stage, and myometrial invasion.
Participation in the study was contingent on the patient experiencing clinical symptoms that could be linked to endometrial cancer. Patient participants also underwent magnetic resonance imaging (MRI) and MR elastography between October 2022 and July 2023. After exclusions, 101 patients with endometrial cancer were included in the study. The median age of the participants was 53 years (interquartile range 49-61 years).
The MR elastography readings taken from each patient participant were analyzed on bioqic-apps.com, and T2-magnitude images along with c (ie, in meters per second representing stiffness) and φ (ie, in radians representing viscosity) maps were developed.
Clinical-Radiologic Model
The researchers developed a clinical-radiologic model that included measuring cancer antigen 125 (CA125) and tumor volume. They also examined the c and φ MR elastography models and then combined these two models.
Most patient participants (n=72) were negative for LVSI. The c value was found to be greater in the LVSI positive cohort compared to the LVSI negative cohort (median, 2.2 m/sec [interquartile range {IQR}, 2.0–2.8 m/sec] vs 4.1m/sec [IQR, 3.2–4.6 m/sec]; P<.001). In terms of φ value, there was no difference noted between the LVSI-positive cohort and LVSI-negative cohort (median, 0.96 radians [IQR, 0.74–1.06 radians] vs 0.93 radians [IQR, 0.78–1.05 radians]; P=0.77).
The LVSI-positive cohort was found to have higher CA125 levels and greater tumor volume (median, 48 U/mL [IQR, 19.8–85.3 U/mL] and 77.5 cm3 [IQR, 38.2–116.3 cm3], respectively) compared to the LVSI-negative cohort (median, 17.45 U/mL [IQR, 10.5–
29.5 U/mL] and 31.2 cm3 [IQR, 8.7–72.5 cm3], respectively; P=0.001 for both). The LVSI-positive cohort also had greater designations as high–International Federation of Gynecology and Obstetrics stage cancers than the LVSI-negative cohort (P=0.046).
The researchers examined interobserver reliability, finding that the interobserver intraclass correlation coefficients for c and φ were 0.97 (95% CI, 0.97-0.99) and 0.89 (95% CI, 0.85-0.92). Furthermore, the reproducibility of the MR elastography measurements was confirmed as satisfactory after the application of a Bland-Altman analysis, which showed good reproducibility between the two radiologists examining the results.
Through a univariable analysis it was found that LVSI positivity correlated with CA125 level (odds ratio [OR], 1.02 [95% CI, 1.01, 1.03]; P=0.005), tumor volume (OR, 1.01 [95% CI, 1.00-1.01]; P = 0.03), and c value (OR, 7.93 [95% CI, 3.59-17.51]; P<0.001). The researchers then made an adjustment of the multivariable model for the variables with P <0.05 in the univariable analysis; this revealed that CA125 level (OR, 1.02 [95% CI, 1.01-1.04]; P=0.009), tumor volume (OR, 1.01 [95% CI: 1.00-1.01]; P=0.04), and c value (OR, 9.06 [95% CI, 3.64-22.53]; P<0.001) were independent predictors associated with LVSI positivity.
The clinical-radiologic model attained an area under the curve (AUC) of 0.77 (95% CI: 0.68-0.85) for predicting LVSI positivity. The model that included only MR elastography attained an AUC of 0.89 (95% CI, 0.82-0.95); cutoff, 3.01 m/sec; P=0.07). When the clinical-radiologic model was combined with the MR elastography model, LVSI positivity was predicted with an AUC of 0.93 (95% CI: 0.86-0.97).
MR elastography data combined with clinical-radiologic model data achieved higher specificity (89%) than clinical-radiologic model data alone (69%). Adding MR elastography data to the clinical radiologic data also resulted in higher positive predictive value for determining LVSI positivity (52% clinical radiologic alone vs 77% combined).
