P-0216 - Evaluating the learning curve of Magnetic Resonance Imaging (MRI) as a predictive tool in locally advanced rectal cancer (RC) - a retrospective 12-y...

Date 28 June 2014
Event World GI 2014
Session Poster Session
Topics Rectal Cancer
Imaging, Diagnosis and Staging
Presenter Niamh Coleman
Citation Annals of Oncology (2014) 25 (suppl_2): ii14-ii104. 10.1093/annonc/mdu165
Authors N. Coleman1, A. Jayaram1, L. Connell2, F. Mohamad Rom2, N. Keegan2, E. Harrold2, C. Lee3, B. Mehigan2, J. Kennedy Michael2
  • 1St. James' Hospital, Dublin/IE
  • 2St James' Hospital, Dublin/IE
  • 3St James' Hosptial, Dublin/IE



MRI is increasingly used to stage RC and allow stratification of patients (pts) facilitating better targeting of preoperative therapy. Accurate preoperative staging is crucial to identify patients who will most benefit from neoadjuvant therapy (NAT) prior to curative surgical resection (Sx). We assessed our institution's experience of NAT versus upfront Sx in stage I-III RC, assessing the utility of MRI as a predictive tool in evaluation of LN status.


In a retrospective analysis, we assessed preoperative MRI staging & definitive histopathological confirmation of LN status collecting data from a prospectively maintained database for patients with a dx of stage I-III RC from Jan 2002–Jan 2014.


A total of 326 patients underwent curative Sx for RC. Of these, 151(46%) had upfront Sx (81 male, 70 female) with a median age 62 (range 33-91). LN staging by MRI was accurate in 51% cases (n = 77). LN status could not be evaluated (Nx) in 15.89% (n = 24). More advanced LN staging noted on histology in 21.85% (n = 33) while LN stage deemed more aggressive by MRI in 11.26%(n = 17). 39 patients received adjuvant 5FU-based chemotherapy. Relapsed disease was noted in 21 of Sx patients (13.9%), 20 systemic & 1 local recurrence. Of the 175 (64%) pts who proceeded to NAT, 117 were male, 58 female with a median age 56 (range 26-86). In this cohort, 79% pts (n = 138) had both a preoperative MRI & hx report available for comparison. MRI in this group matched hx in 51.4% cases (n = 90). 23 pts (13.1%) had more advanced LN staging on hx while LN stage was deemed more aggressive by MRI in 14.28% (n = 25). Relapsed disease in this cohort occurred in 40 pts, 37 systemic & 3 local recurrences. From 2007-2014, 111 pts proceeded to NAT. Nodal status on MRI could not be evaluated in n = 2 pts. Of those 98 pts with node positive disease (NPD), n = 51 remained NPD post-NAT. Relapsed disease occurred in 9.9% (n = 11) of these patients. 46 patients with NPD became node negative (NND) following NAT. Of this subset, relapsed disease was noted in 8.1% (n = 9). 12 patients remained NND following NAT, and relapsed disease was noted in 2.7% of this cohort (n = 3). For the first 150 of our patients treated, MRI sensitivity was 44.8%, compared with the next 150 patients with MRI sensitivity of 51%. Overall in our study, MRI accurately predicted LN status in 58.95% (n = 167).


MRI is a valuable tool in stratifying pts who may benefit from NAT, but its predictive value certainly has limitations. Further analysis of the NAT subset revealed modestly reduced recurrence risk in those patients who became node negative following NAT. Improved overall accuracy was noted in MRI over time, illustrating a distinct learning curve and coinciding with formation of RC Radiology MDT in our institution. Further analysis of the dataset will aim to identify other factors involved and create a predictive nomogram for the RC pt with locally advanced dx.