Can diffusion-weighted imaging (DWI) magnetic resonance imaging (MRI) have an impact in monitoring the efficacy of neoadjuvant chemotherapy (NCT) for the treatment of rectal adenocarcinoma?
In a retrospective study, recently published in Radiology, researchers compared the use of MRI tumor regression grading (MR-TRG), DWI-MRI, DWI-modified MR-TRG (DWImodMR-TRG), MRI complete response and a radiologic neoadjuvant response score in 224 study participants who had post-NCT surgery for locally advanced rectal cancer without radiation.
The study authors found that the use of DWI-modified MR-TRG method for assessing tumor response had less sensitivity than the MRI complete response method (71 percent vs. 76 percent). However, the DWI-modified MR-TRG approach offered comparable specificity (89 percent for both) and negative predictive value (NPV) (93 percent vs. 94 percent) to that of the MRI complete response method, and a higher AUC (90 percent vs. 83 percent), according to the researchers.
The research findings revealed the use of DWI had more than double the effectiveness of MR-TRG in determining the efficacy of NCT.
“In our study, among 52 participants who were interpreted as having MRI complete response, MR-TRG (mainly based on T2 sequences) identified 22 cases (42.3%) while DWI identified 47 cases (90.4%), indicating that the addition of DWI can help avoid missing MRI complete response after NCT, which is also consistent with previous studies,” wrote lead study author Yu Shen, M.D., who is affiliated with the Colorectal Cancer Center within the Department of General Surgery and the Department of Radiology at the West China Hospital and Sichuan University in Chengdu, China, and colleagues.
The researchers also found that the specificity rates for the MRI-based methods ranged from 91 to 97 percent, and the NPVs ranged from 84 to 95 percent.
“ … All the tested MRI-based models showed remarkably high specificity and negative predictive value, suggesting that MRI can depict residual tumors. This is particularly important in determining whether an organ preservation strategy is applicable,” pointed out Shen and colleagues.
Three Key Takeaways
- Enhanced sensitivity with DWI. The addition of DWI to MRI-based assessments significantly improves the detection of complete responses after NCT. In the study, DWI identified 90.4 percent of cases with MRI complete response, compared to only 42.3 percent identified by MR-TRG.
- High specificity and negative predictive value (NPV): Both the DWI-modified MR-TRG and the MRI complete response methods demonstrated high specificity (89 percent) and NPV (93 percent or higher), suggesting their effectiveness in accurately depicting the presence of residual tumors, which is crucial for determining organ preservation strategies.
- Mitigating misclassification. Reviewing DWI sequences in conjunction with T2-weighted imaging helps reduce erroneous classifications. The DWI-modified MR-TRG method increased sensitivity from 43 percent to 71 percent, highlighting its potential to improve the accuracy of tumor response assessments.
Noting that diffusion restriction in tissues may suggest residual tumors, the study authors emphasized that reviewing DWI sequences in addition to T2-weighted imaging can help mitigate erroneous classifications.
“By recognizing the restricted diffusion on additional DWI sequences, misclassified MR-TRG 2 may be downgraded to DWImodMR-TRG 1 and reclassified as a complete response. In this way, DWImodMR-TRG increased the sensitivity from 43% (18 of 42) to 71% (30 of 42), with an AUC of 0.90,” added Shen and colleagues.
(Editor’s note: For related content, see “Hybrid MRI Deep Learning Model Shows Promise in Predicting Tumor Deposits with Rectal Cancer,” “Study Finds Questionable Adherence to Guidelines for Rectal MRI” and “Seven Key Considerations with Rectal Cancer MRI.”)
In regard to study limitations, the authors acknowledged the possibility of over-staging for some study participants due to a relatively long interval between response assessment and radical surgery. The researchers also noted a lack of non-operative strategies in the study, no assessment of inter-observer agreement for pathology nor assessment of possible correlations of different models with survival outcomes.
