Abstract 263P
Background
The aim of this study is to determine the relationship between the MR-derived parameters mrTRG, PET-CT SUV value, and the pathological tumour regression grade pTRG.
Methods
A prospective analysis from June 2019 to June 2020 satisfying the inclusion criteria was conducted, i.e., (i) histologically confirmed rectal carcinoma within 10 cm of the anal verge; (ii) locally advanced rectal cancer (T3–T4) as seen on pre-CRT MR imaging; (iii) performance of 1.5 Tesla (T) rectal MR imaging before and after CRT; and (iv) recipient of neoadjuvant CRT. Pre-CRT and post-CRT MR imaging were performed using a 1.5 Tesla MR unit by surface coil.
Results
A final sample of 50 patients had Grade 1 (n = 10, 20%), Grade 2 (n = 14, 28%), Grade 3 (n = 12, 24%), and Grade 4 (n = 14, 28%) mrTRG. Grade 1 was the HPE pTRG with the second-highest prevalence after NARCT (34%), then Grade 2 (24%), Grade 0 (22%), and Grade 3 (20%). MRI at baseline showed tumour stage t3 in 44% of patients, followed by t2 (22%), t4 (20%), and t1 (14%), with 76% concordance. NACRT response was best in patients with MRI lesion sizes of 2-3cm (75%), and lowest in those with 4-5cm (45.7%), without significant correlation (p = 0.214). Patients with partly circumferential tumor placement on baseline MRI responded best to NACRT (65.2%) without significant association (p = 0.189). SUVmax >64% successfully indicated responders among 60% patients with 96.8% (95%CI=90.6–103.0) sensitivity and 94.7% specificity.
Conclusions
The MRI-based CRT response assessment showed a high ability to predict SUV and a cutoff value that was clinically meaningful for diagnosis. Recommendations (a) Standardizing the modified Ryan's TRG method and studying its factors are needed. (b) TRG system accuracy, reliability, and validity need more study. (c) Subjects who had neoadjuvant CRT for ypStage II locally advanced rectal cancer were put into two groups: good responders and poor responders. This was done using a modified classification system that used ypStage and TRG as predictors of outcome. (d) This categorization is simple and useful for prediction. Clinical validation of a universal regression scoring system is a research goal.
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.