Pre-OP Breast MRI Is Not a Mastectomy-Driver

It is well known that breast MRI provides more accurate information about a cancer’s extent than conventional mammography or ultrasound. Is MRI therefore of advantage for preoperative assessment or does its broader use only lead to overtreatment, as many fear? The multicenter MIPA trial gives surprising results.

Presenter Francesco Sardanelli from Milan, Italy, and colleagues already demonstrated in 2004, that breast MRI provides a significant higher sensitivity for assessing breast cancer extent than digital mammography (DM; 81% vs. 66%, P<0.001) with a comparable positive predictive value (69% vs. 72%, AJR 2004). Nowadays sensitivity of breast MRI even achieves 94-96% with a specificity of 90-97%, as Sardanelli further demonstrated in a multicenter trial (Invest Radiol 2016).

Interestingly, these results did not lead to a triumph of breast MRI in the preoperative setting, but to a certain fear: Many radiologists believe that breast MRI could be too accurate and consequently lead to an increase of unnecessary mastectomies.

The Studies’ Weaknesses

As an advocate for preoperative breast MRI, Sardanelli has a different opinion: “What we find is clinically relevant.” He believes that MRI helps surgeons to better estimate the tumors extent for breast conserving surgery (BCS). It could support surgeons to completely eradicate the tumor and therefore reduce the rate of positive margins, which is still quite high with BCS.

Unfortunately, most study results do not support his opinion: Sardanelli reviewed 19 studies dealing with approximately 86,000 patients, which clearly show that preoperative breast MRI is associated with an increase of the mastectomy rate without influencing the positive margin rate and the number of re-excisions. Sardanelli, however, analyzed the study designs and detected a certain selection bias: The studies did not consider the different reasons, why surgeons and radiologists ordered a breast MRI, such as dense breasts, patient age and aggressiveness of the tumor or tumor size.

To overcome this general weakness, Sardanelli and colleagues started the MIPA trial (Multicenter International Prospective Meta-Analysis of Individual Woman Data). A total of 39 medical centers participated in this large study and provided data of 5,365 consecutive breast cancer patients. 52% of the patients underwent MRI, 48% did not. Most patients of the MRI group underwent this assessment in a preoperative setting (79%), while the rest had already received it for screening (7%) or diagnostic purposes (14%). Sardanelli highlighted that 30% of the MRIs were ordered by surgeons only: “They ask for it!”


With a median age of 56 years, patients in the MRI group were significantly younger than the patients of the non-MRI group (62 years, p< 0.001). In the MRI group, there were significantly more patients with heterogeneously dense breasts and completely dense breasts as in the non-MRI group (50% vs. 33%, P<0.001). In 22% of the cases of the MRI group mastectomy was already planned due to DM/US results. In the non-MRI group, only 14% should undergo mastectomy. The Odds ratio (OR) of the MRI group was significantly higher than OR of the non-MRI group (1.7 vs. 1.0; P<0.001).

In the non-MRI group, the rate of finally performed mastectomies did not change (14%), while it slightly got reduced in the MRI-group (from 22% to 21%). OR was 1.6 for the MRI group and 1.0 for the non-MRI group. But the OR of the MRI group dramatically changed after overcoming the selection bias by adjusting OR for age, breast density and planned mastectomy at DM/US. It finally was 1.0, so totally comparable to the non-MRI group. “There is no difference,” Sardanelli repeated twice with a loud and excited voice.

In the MRI group, the MRI assessment finally led to a non-significant reduction of mastectomies. This refers to a total of ten out of the 598 planned mastectomies. Significant changes were also seen regarding the reoperation rate due to positive margins: In the MRI group, reoperation rate was 2.6% lower than in the non-MRI group (P=0.005). Above that, the MRI assessment often led to changes of the excision extents of a breast conserving surgery. In 7%, MRI resulted in a less extensive excision, while it caused a more extensive surgery in 8.2%. “This is personalized medicine,” Sardanelli said.


For Sardanelli, preoperative breast MRI is an important innovation for breast cancer assessment. He believes that radiologists and surgeons will soon provide it the rightful place.

Presentation Title: Preoperative staging with MRI: results of preoperative MRI (MIPA) trial
Speaker: Francesco Sardanelli, Milan, Italy
Date: Thursday, February 28th, 2019
Session Code: RC 502 – A0233