Abstract 427P
Background
The identification of pathologic complete responder (pCR) patients with locally advanced rectal cancer (LARC) after neoadjuvant treatment (NAT) is of interest in terms of patient prognosis and subsequent therapy options. A four-pattern based evaluation combining diffusion weighted imaging (DWI) and tumor morphology on T2-weighted magnetic resonance has shown promising results in identification pCR. Our aim was to evaluate the performance and reproducibility of this approach in a retrospective series in our center.
Methods
Pre and post NAT MRI scans obtained in 120 patients with LARC between 2009 and 2014 were retrospectively evaluated by 2 radiologists with 20 and 3 years of experience respectively. Due to artifacts in DWI, tumoral mucine or absence of pathologic specimen, 79 patients were finally evaluated. Reviewers (R), blinded to pathologic final result, assigned 1 of 4 morphologic and DWI tumoral response patterns to every case, including assessment of the absence or presence of residual tumor.
Results
R1 and R2 identified 6 and 5 of 15 patients with pCR respectively. Assignments of patients to each group and data on sensitivity and specificity both global and per group are listed in the table. Correlation for pattern assignment between both reviewers showed a k=0.68. Table: 427P
| Global | A | B | C | D | ||||||
| Reviewer | 1 | 2 | 1 | 2 | 1 | 2 | 1 | 2 | 1 | 2 |
| n | 79 | 32 | 23 | 18 | 17 | 26 | 37 | 3 | 2 | |
| Sensitivity | 97 | 87 | 96 | 80 | 100 | 100 | 94 | 85 | 100 | 100 |
| Specificity | 40 | 33 | 100 | 100 | 0 | 0 | 12 | 20 | - | - |
Conclusions
The identification of residual tumor with DWI in LARC after NAT can be achieved with high sensitivity and modest specificity. Distinguishing morphologic and DWI patterns allow us to select a group with very good results (pattern A) and identify a group in which pCR rarely occur (pattern B). Group classification of patients into each group could be done with good correlation between reviewers. The majority of the discrepancies concentrate in patterns A and C assignments.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.