Preoperative MRI to Predict Upstaging of DCIS to Invasive Cancer at Surgery.

Background: Ductal carcinoma in situ (DCIS) is overtreated, in part because of inability to predict which DCIS cases diagnosed at core needle biopsy (CNB) will be upstaged at excision. This study aimed to determine whether quantitative magnetic resonance imaging (MRI) features can identify DCIS at risk of upstaging to invasive cancer.

Methods: This prospective observational clinical trial analyzed women with a diagnosis of DCIS on CNB. All the participants underwent preoperative 3T MRI. Quantitative MRI features from routine dynamic contrast-enhanced (DCE) MR images (e.g., peak percent enhancement [PE]) and from advanced high temporal-resolution DCE MR images (e.g., K) were measured. Clinical, pathologic, and mammographic features were reviewed. Associations with upstaging were summarized using the area under the receiver operating characteristic curve (AUC).

Results: Of 58 DCIS lesions at CNB, 15 (26%) were upstaged to invasive cancer at surgery. Of the 58 lesions, 46 (79%) enhanced on MRI, although enhancement alone was not significantly associated with upstaging (p = 0.71). Among the DCIS lesions that enhanced, higher PE was most strongly associated with upstaging (AUC, 0.81; adjusted p = 0.009) and outperformed MRI features acquired via high temporal resolution DCE-MRI (AUC, 0.50-0.73). Lesion span on MRI was not significantly associated with upstaging risk (AUC, 0.55; adjusted p = 0.61), nor were any clinical, pathologic, or mammographic features (p > 0.24).

Conclusions: Quantitative features acquired from routine clinical breast MRI and advanced DCE-MRI demonstrated good performance in identifying which DCIS lesions were upstaged to invasive cancer at excision. These features may prove valuable for appropriate selection of active surveillance in future DCIS de-escalation trials.