What’s New in Gastrointestinal Imaging? – July 2022

7 months ago

Extramural venous invasion (EMVI) revisited: a detailed analysis of various characteristics of EMVI and their role as a predictive imaging biomarker in the neoadjuvant treatment response in rectal cancer.

Altinmakas E, Dogan H, et al. Abdominal Radiology. (2022) 47:1975-1987.

A recently described parameter for assessing prognosis in rectal cancer is extramural venous invasion or EMVI. EMVI is defined as the presence of tumor cells in the veins beyond the muscularis propria. MRI is useful in detecting EMVI at initial diagnosis and after treatment. This study assessed certain features including size, diameter, and large vein (superior, middle or inferior rectal vein) involvement of MR- detected extramural venous invasion and their impact on neoadjuvant therapy response in rectal adenocarcinoma. The authors studied 57 patients with advanced rectal adenocarcinomas who underwent MRI before and after neoadjuvant therapy. On pre neoadjuvant therapy scans 17 patients were MR-EMVI negative and 40 were MR-EMVI positive. The authors found that complete/near-complete responders had less number and smaller diameter of MR-EMVI when compared with partial responders (mean of 1.45 for complete/near-complete responders vs mean of 2.54 for partial responders and mean of 1.8 mm for complete/near-complete responders vs mean of 3.3 mm for partial responders). Large vein involvement was not statistically associated with treatment response. Additionally, MRI was found to have a high sensitivity for EMVI detection and a moderate specificity for EMVI detection. These findings show that the size and number of EMVI are clinically significant in determining prognosis for patients with rectal cancer and should be incorporated into MRI evaluation of these patients.



Rectal MRI radiomics inter- and intra- reader reliability: should we worry about that?

Kwok HC, Charbel C, etal. Abdominal Radiology. (2022) 47: 2004-2013.

Advances in the treatment of locally advanced rectal cancer had created more options for patients. Patients who respond well to neoadjuvant chemoradiotherapy may be offered a ‘watch and wait’ approach rather than surgery. This approach can be limited by the poor accuracy of diagnosing clinical complete response. Digital rectal exam has a specificity of 56% and a sensitivity of 24%, and endoscopy has a sensitivity of 65-87% and a specificity of 39-78%. Rectal MRI can also have variability in its ability to diagnose clinical response as well as suffering from interobserver variability, with interobserver agreement ranging from 0.47 to 0.64.To address these shortcomings ongoing studies are aiming to find radiomics features to develop more reliable methods for predicting complete response. The authors selected 13 studies which explored imaging features to predict complete response to determine which features showed high inter reader reliability. Features including use of baseline versus restaging MRI, method of segmentation (3D VOI vs 2D VOI) were assessed. Intra- reader agreement varied from 0.73 to 0.93 and inter- reader agreement varied from 0.60 to 0.99. The authors found that features obtained from baseline MRI and from features extracted from the entire tumor volume using 3D VOI had higher agreement.



Accuracy of high resolution multidetector computed tomography in the local staging of rectal cancer.

Venkataramanan R, Munikrishnan V, et al. J Gastrointestinal and Abdominal Radiol ISGAR 2022; 5:1-15.

Recent advances in the treatment of rectal cancers has allowed for a subset of patients to be treated with neoadjuvant chemoradiotherapy instead of primary surgery. Accurate local staging of rectal cancers is crucial in differentiating which patients can be appropriately offered each option. MRI has served as the standard for local staging of rectal cancer due to its superior distinction between bowel wall layers. Recent literature has suggested that recent advances in CT such as CT perfusion may allow CT staging may perform equally well. The authors therefore aimed to evaluate the accuracy of high resolution CT for local rectal cancer staging, including patients with low rectal cancer as CT has previously been shown to be less accurate in this group. 93 patients underwent high resolution CT for staging of rectal cancer. 64 of these patients underwent primary surgery, while 29 underwent surgery post neoadjuvant chemoradiotherapy. The authors found that in differentiating stages T2 and less than T2 from T3 and T4 rectal cancer accuracy was overall 91%, sensitivity 87%, specificity 94%, and kappa score 0.8. In the primary surgery group accuracy, sensitivity, specificity, and kappa score were 89%, 76%, 94%, and 0.7. In the neo adjuvant chemoradiotherapy group accuracy, sensitivity, specificity, and kappa score were 97%, 100%, 94%, and 0.9. In the low rectal group accuracy, sensitivity, specificity, and kappa score were 94%, 89%, 97%, and 0.82. These results demonstrate that high resolution CT is an accurate tool for local staging of rectal cancer.



A multicenter study on the preoperative prediction of gastric cancer microsatellite instability status based on computed tomography radiomics.

Liang X, Wu Y, et al. Abdominal Radiology (2022) 47: 2036-2045.

Microsatellite instability is an important feature in gastric cancer and can help direct cancer treatment and change the options for treatment offered to a patient. Due to this important testing of newly diagnosed gastric cancer for Microsatellite instability is routinely performed. Currently testing for MSI is performed via immunohistochemical staining or molecular biology tests requiring biopsy or surgical pathology samples. The authors of this study aimed to construct and validate a radiomics feature model based on CT images and clinical characteristics of gastric cancer patients to predict MSI status prior to surgery. The pre-operative CT scans of 189 gastric cancer patients were divided into a training and validation group. Radiomics features including tumor morphology, and features that describe the texture of the tumor interior and surface were extracted from the images. The only clinical features found to be significantly correlated with MSI were tumor age and tumor site, and these features were included in the model. The authors found that their model predicted MSI status in the training data set with an accuracy, sensitivity, and specificity of 0.72, 0.63, and 0.77 respectively, and in the validation dataset with an accuracy, sensitivity, and specificity of 0.67, 0.79, and 0.60 respectively. These findings demonstrate that radiomics markers in CT images can be used as a non-invasive tool for gastric cancer MSI status.



Evaluation of radiological and temporal characteristics of acute appendicitis on the non-enhanced computed tomography images.

Gao X, S W, et al. Abdominal Radiology (2022) 47: 2279-2288. 

Acute appendicitis is generally diagnosed via CT in the emergency room, however assessing for complications of appendicitis is an important adjunct to the diagnosis. Some recent studies have argued for antibiotic treatment of uncomplicated appendicitis, increasing the importance of detecting complications such as perforation and gangrene. This study aimed to investigate the pre-operative features of appendicitis on non-contrast CTs, and the time evolution of acute appendicitis and whether these factors could be used to predict gangrene and perforation of inflamed appendices. Imaging features analyzed included appendix diameter, appendicolith, appendiceal intraluminal gas, periappendiceal gas and fat stranding/fluid, and short axis diameter of mesenteric lymph nodes. Time periods included pre-CT delay, preoperative delay, time between onset of symptoms and development of appendicitis complication, time between onset of symptoms and admission, and time between admission and surgery. With the exception of mesenteric lymph node diameter, all of the CT characteristics were found to be significantly different between simple, gangrenous and perforated appendicitis. The presence of appendicolith was found to be an independent risk factor for both gangrene and perforation when analyzed using pre-operative delay in a hazard model. The median time in these patients for gangrene was 104 hours if appendicolith was absent on CT versus 76.2 hours if appendicolith was present, and the median time for perforation was 103 hours if appendicolith was absent versus 77.5 hours if appendicolith was present. These findings demonstrate the interactions between timeline and radiologic features of acute appendicitis and can benefit the radiologists ability to evaluate for complications.



  • Share