Abstract 150P
Background
Systemic neoadjuvant therapy (NAC) aims to achieve complete pathological remission (pCR), which is associated with more favourable outcomes. The basic assumption is to determine the extent of residual disease after NAC more precisely. MRI imaging has shown superiority to other modalities regarding pCR prediction; however, its accuracy differs between different breast cancer subtypes. This study is focused on the limitations of MRI in preoperative assessment of residual disease extent for individual tumour subtypes.
Methods
232 patients with invasive breast cancer received NAC in 2016-2022 and were imaged with 3T breast MRI preoperatively. Tumour characteristics, preoperative MRI results and definitive pathology outcomes were recorded. Comparisons of molecular subtypes' influence on MRI sensitivity, specificity, positive/negative predictive value (PPV, NPV), and accuracy in assessing pCR were performed.
Results
The dataset exhibits an overall sensitivity of 94.5%, specificity of 71%, PPV of 88.5%, NPV of 84.5%, and accuracy of 87.5% of MRI for predicting pCR. Among the subtypes (HER2+, TN, Lum B HER2+, Lum A/B) MRI is very reliable for ruling out residual disease in HER2+ breast cancer post-NAC (sensitivity 94.1%, NPV 92.86%), but less reliable for confirming the presence of residual disease (specificity 62%, PPV 66.7%), its accuracy is 76.3%. For TN MRI correctly identified residual tumour with a sensitivity of 93.1%, PPV of 85.1% and pCR with NPV of 84.6%, moderate specificity of 71% contributes positively to the overall accuracy of 85.4%. The reliability in confirming pCR in Lum B/ HER2+ is high (NPV 88.8%), in Lum B much lower (NPV 66.6%.).
Conclusions
MRI achieves overall accuracy in predicting pCR and is a valuable tool in assessing the response to NAC. However, it has limitations, which are particularly evident when considering the different breast cancer subtypes, as each can respond differently to treatment and thus present differently on MRI. This variability affects the confidence with which clinicians can recommend forgoing surgical treatment based solely on MRI results. Further investigation and identification of imaging biomarkers can improve MRI accuracy.
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.