Determining whether an oesophageal tumour has reached the trachea or major vessels is crucial for planning treatment. After preoperative therapy, anatomy can be distorted and tissue planes hard to judge, so widely used contrast-enhanced CT may not always provide clear answers. A single-centre analysis of consecutive surgical cases assessed whether a structured MRI method could bring greater clarity at this challenging point in care. Two radiologists applied simple five-point scores on MRI to judge tracheal and vascular involvement and compared these readings with CT-based staging and final pathology. The approach aimed to offer a more reliable way to decide when disease is truly unresectable and when surgery remains a realistic option.
Structured MRI Scores and Practical Workflow
The MRI method used two short scales that mirror clinical questions. One scale assessed tracheal involvement, the other focused on vessels such as the thoracic aorta and right subclavian artery. Each scale progressed from preserved tissue planes to definite invasion, with diffusion-weighted imaging used to support the highest suspicion levels. Radiologists relied primarily on axial T2-weighted images to examine the interface between tumour and adjacent structures, then turned to diffusion-weighted imaging only when features suggested possible breach.
Must Read: AI and Hyperspectral Imaging for Early Oesophageal Cancer Detection
Imaging timing was aligned with the surgical pathway. Patients had CT before and after preoperative therapy, while MRI was performed after therapy and close to surgery. CT reads followed a recognised classification to separate clearly resectable, borderline and frankly invasive categories. The MRI reads were done independently and at a different time from CT to avoid recall, using the same concise criteria for every patient. This created a like-for-like comparison between day-to-day CT staging and short, teachable MRI rubric that focuses attention on visible tissue planes rather than indirect signs.
The cohort represented typical locally advanced disease moving towards possible conversion surgery. Pathology ultimately confirmed that tracheal invasion and vascular invasion were present in a minority of cases, underscoring the importance of distinguishing genuine spread from treatment-related changes. Surgical margins varied, reflecting the difficulty of operating in previously treated fields and reinforcing the value of imaging that can more confidently separate borderline contact from true invasion.
Clearer Discrimination and More Consistent Reads
Across readers and targets, MRI separated invasive from non-invasive cases more cleanly than CT. In practical terms, the structured MRI scores reduced the grey zone created by post-therapy scarring and altered fat planes. When the tracheal score was set to require firm signs of breach, classification became both accurate and repeatable. The vessel score behaved similarly, with the highest category reserved for smooth, close tumour contact along the vessel wall and supportive diffusion-weighted signal. By anchoring positive calls to a small number of visible features, the system limited false alarms that can occur when CT angles and loss of fat are interpreted in isolation.
When the two MRI scores were combined into a single decision rule for suspected T4 disease, performance improved in a way that matters clinically. The combined MRI rule maintained sensitivity while markedly increasing specificity compared with CT in the same patients. This pattern held across the two readers. The net effect is fewer patients labelled as invasive when they are not, without missing those who do have invasion. Statistical comparisons using resampling supported this overall advantage, including a clearly higher summary measure of discrimination for one reader and a positive trend for the other.
Reader agreement followed the same direction. Using the top categories on the MRI scales led to substantial or even perfect agreement for tracheal and vascular calls, whereas the CT categories produced only moderate agreement and less stable binary decisions. The five-point MRI scales also showed better weighted agreement than their CT counterparts. This matters for service delivery, as reproducible reads reduce downstream variability in multidisciplinary discussions and help align decisions across teams.
Impact on Preoperative Assessment and Planning
The reasons for MRI’s advantage are practical. MRI offers stronger soft-tissue contrast and direct visualisation of the interfaces that determine operability, which is especially helpful after therapy when CT findings can be ambiguous. CT remains essential for whole-thorax evaluation and surgical planning, yet its staging categories rely on indirect cues that are vulnerable to post-treatment change. The MRI rubric reframes the question around a few anatomic checkpoints and uses diffusion-weighted imaging only to reinforce the highest levels of suspicion, rather than as a primary driver of calls. This reduces over-calling in borderline cases and provides a shared language that is easy to teach, remember and audit.
For multidisciplinary teams, the combined tracheal and vascular scores support more confident decisions when CT suggests borderline invasion. A clearer negative call can preserve surgical options, while a clearer positive call can steer patients toward non-surgical strategies without delay. The higher reader agreement also means fewer disagreements to resolve and more predictable discussion outcomes. Within the limits of a single-centre experience and a modest sample, these features point to a practical role for MRI scoring when the question is not the extent of tumour along the oesophagus, but whether it has truly crossed into structures that determine T4 status.
A concise set of MRI scores for tracheal and vascular involvement provided clearer, more consistent staging than CT in patients assessed after preoperative therapy. By centring decisions on visible tissue planes and reserving diffusion-weighted support for the highest concern levels, the method reduced borderline calls, improved agreement between readers and increased specificity for pathological T4 while preserving sensitivity. These gains make the approach a useful adjunct to CT when operability hangs on whether tumour has breached the trachea or major vessels, helping teams choose the right course without introducing new assumptions.
Source: European Radiology
Image Credit: iStock
