MERCURY
Diagnostic accuracy of preoperative magnetic resonance imaging in predicting curative resection of rectal cancer: prospective observational study (MERCURY, Magnetic Resonance Imaging and Rectal Cancer European Equivalence Study)
Phase observationalN = 408
- Population
- Consecutive patients presenting with all stages of rectal cancer who had MRI before total mesorectal excision with histopathological assessment of the specimen
- Design
- Prospective observational multicentre study in 11 colorectal units in four European countries (January 2002 to October 2003). Of 679 consecutive patients who consented, 408 with complete MRI and histopathology data were analysed (BMJ full text). Patients with rectal cancer of all stages had high-resolution MRI before total mesorectal excision; the MRI-predicted state of the circumferential resection margin was compared with histopathology of the surgical specimen. Not randomised. The follow-up reports analyse 374 patients (5-year outcomes by MRI-assessed circumferential resection margin), 111 patients treated with neoadjuvant therapy, and 122 patients with MRI good-prognosis tumours treated without radiotherapy.
- Primary endpoint
- Accuracy of MRI for predicting a curative resection
- Arms
- All patients (MRI before TME) (experimental, n = 408): High-resolution pelvic MRI before total mesorectal excision; histopathology of the specimen as the reference
- MRI-clear circumferential resection margin (control): Tumour more than 1 mm from the mesorectal fascia on MRI (follow-up cohort of 374 patients)
- MRI-involved circumferential resection margin (experimental): Tumour 1 mm or less from the mesorectal fascia on MRI (follow-up cohort of 374 patients)
- People
- MERCURY Study Group, MERCURY Study Group
OSHR 1.97 (95% CI 1.27–3.04)Δ 60-month OS−20 pointsDFSHR 1.65 (95% CI 1.01–2.69)Δ 60-month DFS−19.9 pointsLocal recurrence after rectal surgeryHR 3.50 (95% CI 1.53–8.00)
Show the numbers
| Endpoint | Analysis | All patients (MRI before TME) / MRI-involved circumferential resection margin | MRI-clear circumferential resection margin | Effect |
|---|---|---|---|---|
| Clear circumferential resection margin at surgery408 consecutive patients with MRI before total mesorectal excision | primaryMERCURY 2006 | 87% (83–90%) | — | 354 of the 408 patients had a clear circumferential resection margin on histopathology. |
| MRI specificity for a clear circumferential margin408 consecutive patients with MRI before total mesorectal excision | primaryMERCURY 2006 | 92% (90–95%) | — | 327 of the 354 patients with a clear margin were predicted clear by MRI. The abstract labels this 'specificity for prediction of a clear margin'. |
| Clear margin at surgery after an MRI-predicted clear margin408 consecutive patients with MRI before total mesorectal excision | primaryMERCURY 2006 | 94% (91–96%) | — | MRI predicted a clear margin in 349 patients; 327 of them had a clear margin at surgery. |
| OSFollow-up cohort of 374 patients with rectal cancer; analysis population otherwise not stated in the abstractFive-year overall survival by MRI-assessed circumferential resection margin (clear vs involved) | long term-updatemedian follow-up 62 monthsTaylor 2014 | 42.2% at 5 y | 62.2% at 5 y | HR 1.97 (95% CI 1.27–3.04)P<.01Surviving patients were followed for a median of 62 months. The abstract does not label the direction of the hazard ratio; because 5-year survival is lower with an MRI-involved margin and the ratio is above 1, it can only describe an MRI-involved relative to an MRI-clear margin. Observational association, not a treatment effect. |
| DFSFollow-up cohort of 374 patients with rectal cancer; analysis population otherwise not stated in the abstractFive-year disease-free survival by MRI-assessed circumferential resection margin (clear vs involved) | long term-updatemedian follow-up 62 monthsTaylor 2014 | 47.3% at 5 y (33.7–60.9%) | 67.2% at 5 y (61.4–73.0%) | HR 1.65 (95% CI 1.01–2.69)P<.05The abstract does not label the direction of the hazard ratio; because 5-year DFS is lower with an MRI-involved margin and the ratio is above 1, it can only describe an MRI-involved relative to an MRI-clear margin. MRI-involved margin was the only preoperative staging parameter that remained significant for OS, DFS and local recurrence on multivariate analysis (abstract). |
| Local recurrence after rectal surgeryFollow-up cohort of 374 patients with rectal cancer; analysis population otherwise not stated in the abstractTime to local recurrence by MRI-assessed circumferential resection margin; the abstract gives a hazard ratio only, no per-group rates | long term-updateTaylor 2014 | — | — | HR 3.50 (95% CI 1.53–8.00)P<.05Hazard ratio for an MRI-involved versus an MRI-clear margin. |
| Endpoint | Analysis | All patients (MRI before TME) / MRI-involved circumferential resection margin | MRI-clear circumferential resection margin | Effect |
|---|---|---|---|---|
| Clear circumferential resection margin at surgery408 consecutive patients with MRI before total mesorectal excision | primaryMERCURY 2006 | 87% (83–90%) | — | 354 of the 408 patients had a clear circumferential resection margin on histopathology. |
| MRI specificity for a clear circumferential margin408 consecutive patients with MRI before total mesorectal excision | primaryMERCURY 2006 | 92% (90–95%) | — | 327 of the 354 patients with a clear margin were predicted clear by MRI. The abstract labels this 'specificity for prediction of a clear margin'. |
| Clear margin at surgery after an MRI-predicted clear margin408 consecutive patients with MRI before total mesorectal excision | primaryMERCURY 2006 | 94% (91–96%) | — | MRI predicted a clear margin in 349 patients; 327 of them had a clear margin at surgery. |
| OSFollow-up cohort of 374 patients with rectal cancer; analysis population otherwise not stated in the abstractFive-year overall survival by MRI-assessed circumferential resection margin (clear vs involved) | long term-updatemedian follow-up 62 monthsTaylor 2014 | 42.2% at 5 y | 62.2% at 5 y | HR 1.97 (95% CI 1.27–3.04)P<.01Surviving patients were followed for a median of 62 months. The abstract does not label the direction of the hazard ratio; because 5-year survival is lower with an MRI-involved margin and the ratio is above 1, it can only describe an MRI-involved relative to an MRI-clear margin. Observational association, not a treatment effect. |
| DFSFollow-up cohort of 374 patients with rectal cancer; analysis population otherwise not stated in the abstractFive-year disease-free survival by MRI-assessed circumferential resection margin (clear vs involved) | long term-updatemedian follow-up 62 monthsTaylor 2014 | 47.3% at 5 y (33.7–60.9%) | 67.2% at 5 y (61.4–73.0%) | HR 1.65 (95% CI 1.01–2.69)P<.05The abstract does not label the direction of the hazard ratio; because 5-year DFS is lower with an MRI-involved margin and the ratio is above 1, it can only describe an MRI-involved relative to an MRI-clear margin. MRI-involved margin was the only preoperative staging parameter that remained significant for OS, DFS and local recurrence on multivariate analysis (abstract). |
| Local recurrence after rectal surgeryFollow-up cohort of 374 patients with rectal cancer; analysis population otherwise not stated in the abstractTime to local recurrence by MRI-assessed circumferential resection margin; the abstract gives a hazard ratio only, no per-group rates | long term-updateTaylor 2014 | — | — | HR 3.50 (95% CI 1.53–8.00)P<.05Hazard ratio for an MRI-involved versus an MRI-clear margin. |
Data table
| Endpoint | HR | 95% CI |
|---|---|---|
| OS | 1.97 | 1.27–3.04 |
| DFS | 1.65 | 1.01–2.69 |
| Local recurrence after rectal surgery | 3.5 | 1.53–8 |
Appears in
Sources
MERCURY Study Group. Diagnostic accuracy of preoperative magnetic resonance imaging in predicting curative resection of rectal cancer: prospective observational study. BMJ. 2006;333(7572):779.
Taylor FG, Quirke P, Heald RJ, Moran BJ, Blomqvist L, Swift IR, et al. Preoperative magnetic resonance imaging assessment of circumferential resection margin predicts disease-free survival and local recurrence: 5-year follow-up results of the MERCURY study. J Clin Oncol. 2014;32(1):34-43.