Abstract P2-02-01: Did established clinical practice regarding MRI bias the COMICE trial?

Abstract

Abstract Introduction: The negative COMICE study and reports of inappropriate mastectomies worldwide have served to discredit the use of preoperative MRI for the purpose of aiding conservative breast surgery. We postulate that established clinical practice regarding MRI at the time of the COMICE trial lead to bias in case selection, with more complicated cases being preselected out prior to randomisation. We reviewed the local practice at the time of COMICE. Methods and materials: Retrospective analysis of all cases of Breast MRI performed to assess disease extent pre-operatively during recruitment to COMICE (December 2001 - January 2007) was undertaken. Size on mammogram/ultrasound, MRI and histology was documented. As pre-PAC's era, the information was obtained from imaging and pathology reports. Where reports did not include distance between lesions in multifocal/multicentric disease, the sum of the lesions was used (taken as immediately adjacent) so as not to overestimate the size. All cases were reviewed blinded to COMICE status. Cases with mammogram/ultrasound size >/ = 40mm were excluded as these were deemed unsuitable for conservative surgery. Results: A total of 318 breast MRI examinations were performed in this interval to assess disease extent pre-operatively of which, 81 were excluded appropriately, for inadequate information (n = 47), size on conventional imaging >/ = 40mm (n = 18), receiving neo-adjuvant treatment (n = 6), non cancer diagnosis (n = 7) and non invasive disease (n = 3). 242 cancers from 237 patients were included comprising 77COMICE and 160NON-COMICE patients. Statistical difference was noted in the types of surgery between the groups, p<0.001. Re-excision rates were similar: COMICE = 11.7% and NON-COMICE = 8%. Mastectomy rates were however different, 15.6% of the COMICE group and 42% of the NON-COMICE group. There was a significant difference in histological size between the 2 groups, mean size in NON-COMICE cases 32mm versus 26mm in the COMICE group (p = 0.009). There was a significant difference in the tumour types between the 2 groups (p<0.001). Notably 37% of cases in the NON-COMICE group were of lobular type compared to 13% of the COMICE group. Conclusion: Data from this well established MRI unit has demonstrated clinical bias in the COMICE trial with more complicated cases, which benefited from MRI, being pre-selected out prior to randomisation which understated the value of MRI. MRI in appropriately selected patients in conjunction with modern oncoplastic surgical techniques will increase the opportunities for conservative surgery rather than increase mastectomy rates. An aggressive biopsy policy for MRI detected lesions is required to avoid inappropriate mastectomies. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P2-02-01.