Precise staging  of preoperative 3.0-T MR imaging for esophageal carcinoma by using ex vivo MR imaging-matched pathologic findings as the reference standard
Yi Wei1,2, Shao-Cheng Zhu1,2, Sen Wu1,2, Da-Peng Shi1,2, and Dan-Dan Zheng3

1Radiology, Zhengzhou University People's Hospital, Zhengzhou, China, People's Republic of, 2Henan Provincial People's Hospital, Zhengzhou, China, People's Republic of, 3GE Healthcare,MR Research China, Beijing, China, People's Republic of

Synopsis

Magnetic resonance imaging (MRI) was reported to evaluate the esophageal layers invasion in vitro and demonstrated that high-resolution T2-weighted imaging can clearly depict 8 layers of esophagus. However, former studies were mostly carried on ultra-high-field scanner ex vivo, which can not satisfy the need of preoperative staging and provide essential information for clinic. In this study, an in vivo experiment was conducted on 3.0T clinical scanner to prospectively establish the MRI signal characteristics of the normal esophageal wall and to assess the diagnostic accuracy of high-resolution MR imaging for depicting the depth of esophageal wall invasion by corresponding to ex vivo MR imaging-matched certain histopathological slice.

Introduction

The prognosis of patients with esophageal carcinoma is heavily dependent on the histopathologic staging of the carcinoma. Accurate preoperative histopathologic staging is important for the treatment options [1, 2].However, the common examination modalities are extremely difficult to identify the preoperative staging precisely due to the deep location of the esophagus, artifacts caused by the respiratory and cardiac motion. Magnetic resonance imaging (MRI) was reported to evaluate the esophageal layers invasion in vitro and demonstrated that high-resolution T2-weighted imaging can clearly depict 8 layers of the esophagus which can provide essential information of the carcinoma invasion [3]. However, former experiments were carried on ultra-high-field scanner ex vivo and the experimental specimens were after 10% formalin fixed,which can not satisfy the normal physiological state and provide essential information of carcinoma invasion for clinic preoperation.The purpose of this study was to prospectively establish the MRI signal characteristics of the normal esophageal wall in vivo and to assess the diagnostic accuracy of high-resolution MR scanner for depicting the depth of esophageal wall invasion by corresponding to ex vivo MR imaging-matched certain histopathological slice.

Method

From January 2014 to October 2015, 73 patients with esophageal squamous cell carcinoma were potentially eligible for this prospective study. Finally 51 patients were recruited in this study, including 38 men (mean age, 63.61±8.16 years; range, 40-84 years) and 13 women (mean age, 63.08 ±8.79 years; range, 50-77 years) went through the experiment following the workflow shown in Fig1.MR imaging was performed by using a 3.0-T MRI system (Discovery MR750, GE Healthcare, Milwaukee, USA) with a 8 channel abdominal coil. High-resolution T2-weighted MR images were obtained with the TR = 7200 ms, TEeff = 82.0 ms, slice thickness = 3.0 mm with 1mm gap, field of view = 36 ×36 cm2, matrix size =320 ×320, NEX=2, resulting in a voxel volume around 1.125×1.125×3 mm3. The interval between the pre-operative MR scanging and the surgery was about 5-7 days. The specimens were fixed with formalin and stained with hematoxylin-eosin (H-E),sectioned in MR plane with a 5-um-slices. The MR images were interpreted by two experienced radiologists independently, who were blinded to the results of the histopathological examinations. When the radiologists could not fully agree on the findings, a consensus was achieved by discussion. The MR diagnostic staging of the 51 esophageal carcinomas were compared with the histopathologic examinations,which were served as the standard of reference. The correlation between the MR images and histopathologic slices were defined by the tumor’s contour or the adjacent lymph nodes. The Kappa test was used to evaluate the MR findings and the histopathological results agreement plus the interobserver agreement.

Results

Corresponding to the histological results, high-resolution MRI can clearly depict the normal esophagus as consisting of three layers,mucosa(hypointense),submucosa(hyperintense),muscularis propria(hypointense)(Fig.2).MR images correctly diagnosed 4 of 9 mucosa invasion, 7 of 10 submucosa invasion, 10 of 14 muscularis propria invasion(Fig.3) and 16 of 18 adventia,(Table 1). Table 2 showed the performance of high-resolution MR images with respect to the assessment of invasion of the different esophageal layers. The diagnostic accuracy of mucosa invasion, submucosa invasion, muscularis propia invasion and adventia invasion were 92%, 96%, 92% and 92%, respectively. And the sensitivity and specificity were 89-98% and 89-94%.A k value of 0.631 was obtained for the correlation between the high-resolution MRI and histopathological findings. Interobserver agreement between two radiologists was excellent (k value, 0.737).

Discussion

High-resolution MR images can achieve precise staging of esophageal carcinoma which provide essential information for preoperative options.The esophageal layers are accurately depicted with three different layers,mucosa(hypointense),submucosa(hyperintense),muscularis propria(hypointense),and the carcinomas demonstrate slightly hyperintense,basing on the difference of signal intensity,the depth of carcinoma invasion were precisely depicted.However,stagement for mucosa invasion was not quite satisfied,further study of MRI enhanced scan may help to improve the stagement accuracy of mucosa invasion.

Conclusion

High-Resolution MRI can provide precise staging of preoperation in vivo with good diagnostic sensitivity, specificity and accuracy for assessing esophageal carcinoma invasion.

Acknowledgements

Thanks for the recruited patients who devoted themselves to the medical research for the development of this discipline.Thanks for the doctors of pathological dept who supported us in histopathological examination.Thanks for the technicians who helped us in the protocol optimization.

References

[1]Kakuta T, et al.Prognostic factors and causes of death in patients cured of esophageal cancer.Ann Surg Oncol. 2014 ;21(5):1749-55.

[2]Lorenz D, Origer J, Pauthner M,et al.Prognostic risk factors of early esophageal adenocarcinomas. Ann Surg. 2014;259(3):469-76.

[3]Yamada , Hikishima K, Miyasaka N, et al.Esophageal carcinoma: ex vivo evaluation with diffusion-tensor MR imaging and tractography at 7 T.tractography at 7 T.Radiology. 2014 ;272(1):164-73.

Figures

Figure 1:Flow diagram shows patient selection process, with inclusion and exclusion criteria, and experiment workflow.

Figure 2:(2a)High-resolution MR image can clearly depict the esophageal layers as consisting of 3 layers,mucosa(hypointense) ,submucosa(hyperintense),muscularis propria(hypointense)(white arrow). (2b)The corresponding ex vivo image demonstrates 3 different layers of the esophagus(black arrow).(2c)Histopathological slice of the normal esophageal wall show mucosa,submucosa and adventia(black arrow)(H&E stain;original maginifaction,×4.)

Figure 3: (3a)High-resolution MR image depicts the irregular high signal intensity invading into the low signal intensity muscularis propria(red arrow),and the muscularis propria is disrupted. (3b)The corresponding ex vivo image demonstrates isointense carcinoma disrupting the muscularis propria (red arrow).(3c)histopathological slice shows carcinoma involving the muscularis propria(H&E stain;original maginifaction,×4.)

Table 1:Data are based on the 51 carcinoma from 51 patients,the histologic findings were standard as the reference.

Table 2:Data are the number of lesions. NA = not applicable



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1121