87 Comparison of MRI response evaluation methods in rectal cancer: a multicentre and multireader validation study 6 Introduction The standard treatment for locally advanced rectal cancer is neoadjuvant chemoradiation (CRT) followed by surgery [1]. Nowadays, there is a paradigm shift to opt for organ-preserving treatment alternatives in patients who respond very well to CRT. Patients with clinical evidence of a complete response after CRT may be entered into a Watch & Wait (W&W) program where patients are deferred from surgery and closely monitored using a combination of imaging and endoscopy. The International Watch & Wait Database (IWWD) recently published the oncologic outcomes of the first 1000 registered W&W patients, showing a good 5-year overall and disease-free survival of 85–94% [2]. The introduction of W&W and other organ-preservation strategies has urged the need for accurate response assessment after CRT to facilitate the patient selection. MRI has an important role in detecting the presence of extraluminal residual disease (e.g. remaining positive lymph nodes) that may render organ preservation unfeasible. MRI is also used as an adjunct to endoscopy to assess the response of the primary tumor in the bowel wall. The diagnostic performance of MRI in this setting is limited owing to difficulties in interpreting fibrotic changes of the tumor bed after CRT [3, 4]. Different methods have been published to address this issue and aid in visually classifying tumor response on MRI after CRT. One of the most well-known is the MRI tumor regression grade (mrTRG), derived from similar TRG scores used in histopathology [5]. The mrTRG can help radiologists classify the degree of fibrotic transformation of the tumor bed on T2-weighted (T2W) MRI to estimate the tumor response [6,7,8,9]. Since the introduction and recognition of diffusion-weighted imaging (DWI) sequences as a valuable adjunct to discern viable residual tumor from fibrosis, modified response systems have been reported that combine tumor regression on T2W-MRI with DWI findings [10,11,12]. Other published methods focus on specific MRI patterns or “signs”. These include the “DWI patterns” approach of Lambregts et al [13], which combines morphological patterns on pre- and post-CRT T2W-MRI with distinct DWI signal patterns post-CRT to differentiate complete responders, and the “split scar” sign published by Santiago et al [14] that describes a typical layered appearance of the tumor bed on T2W-MRI after CRT (referred to as the “split scar”) as a sign indicating a complete response. Most of these response methods were published fairly recently. So far they have mainly been tested by expert readers in single-center study settings. Little is known about how well these methods can be reproduced in daily clinical practice, using less curated datasets, and by radiologists with more general expertise. Therefore, this study aims to validate and compare the above-described methods to asses response after CRT on restaging MRI using a multicentre dataset of clinical MRIs derived from everyday practice, taking into account diagnostic performance, agreement among readers with different expertise levels, and reader preference.
RkJQdWJsaXNoZXIy MjY0ODMw