We analyzed mammographically undetected lesions in dense breast in this study; therefore, the tumors would have been missed in the absence of supplementary examinations. With increasing scholarly research on AP and UE-MRI, there is new hope that breast MRI can be used as a rapid and economical supplementary screening scheme for dense breasts9,16.
Different research groups have reported consistent results 17–19, suggesting that the sensitivity of AP ranged from 89–96%, and was equivalent to that of FDP. Previous reports 9,20 indicated that the specificity of common abbreviated protocols did not differ from that of FDP. However, Grimm et al. 10 and Chen et al. 11 reported that the specificity of AP (52%, 86.5%), was lower than that of FDP. These results may indicate that although AP had high sensitivity similar to FDP for breast cancer, it cannot completely replace FDP in clinical practice. In our study, we further confirmed this by showing that AP had a lower AUC than FDP. In previous studies, the protocols for AP-MRI have not been consistent. Heacock et al. 21 and Grimm et al. 10 concluded that it was no significantly applicable to the sensitivity or specificity of breast lesions when adding T2-weighted imaging or the second post-contrast series to the protocol of abbreviated. Chen et al. 19 found that the combination of common AP and DWI sequence improved the specificity from 86.5–95.0%. Oldrini et al. 22 combined ultrafast MRI with the AP protocol, and the specificity of diagnosis was significantly improved, but the sensitivity did not change significantly. Therefore, further research is needed to achieve consistent standardized AP protocols.
Unlike DCE-MRI and AP-MRI, which rely on injection of gadolinium-based contrast agent to show tissue hemodynamics, diffusion-weighted imaging measures endogenous water movement in tissues. Zhang et al. 23 found that DW-MRI alone (75.6%) in the specificity of diagnosed breast cancer was equivalent to DCE MRI alone (71.1%), but the sensitivity was significantly reduced (86.0% vs 93.2%). The previous studies found that comparisons of DWI sequences alone, integrated DWI with T2WI, improved the detection of the location of lesions, and also prevented missing lesions on DWI 24–26. So, our study used the combination sequence as UE-MRI. Amornsiripanitch et al. 13 concluded in a study that the average diagnostic sensitivity of UE was 72%, which was lower than the specificity (90%). The study reported that the pathological type of DCIS, namely, mucinous carcinoma or triple-negative carcinoma with extensive necrosis, may lead to the decrease of sensitivity 13. In addition, Pinker et al. 27 and Rotili et al. 28 suggested that another factor resulting in the decrease of sensitivity is the smaller tumors, especially lesions ≤10 mm, which are not well detected in UE-MRI blind reading studies. This was also reflected in our study, as we found that smaller tumors were not only easy to miss on DWI and difficult to assess regarding the ADC value, but also affected the observation of tumor morphology in the T2WI sequence. However, for lesions >10 mm, it showed extremely high sensitivity at 93.88% for both readers, similar to previous studies 28.
The diagnostic efficacies of AP and UE-MRI were compared for occult lesions of mammography. The common advantage of the two was that they significantly reduced the scanning time and interpretation time compared with FDP-MRI. AP is superior to UE in its sensitivity for the diagnosis of breast cancer of different sizes. Telegrafo et al. 25 and Kang et al. 29 used advanced read-out segmented echo planar imaging (EPI), DW, MRI, and background suppression techniques in their research, resulting in high sensitivity (93%, 94%). If the problem of low sensitivity in small size lesions could be solved, UE-MRI may be a potential supplementary screening tool for breast cancer because it does not require gadolinium-based contrast agent, and the public is increasingly concerned about the unknown health problems of gadolinium deposition in the brain and other tissues by injecting gadolinium contrast agent repeatedly 30. Therefore, it is difficult to balance the benefits of AP-MRI with the issue of the safety of gadolinium deposition 31.
The limitations of this study were as follows: First, because this study lacked a sufficient number of non-mass enhanced (NME) lesions, a subgroup analysis of mass and non-mass enhanced lesions was not formed. Avendano et al. 32 reported that 31% of the lesions with NME on DCE-MRI could not be assessed by DWI. So far, we have not found the study of AP on non-mass lesions. Second, in the past few years, there was no unified standard for slice thickness in DWI scanning; the slice thickness used in different studies was 3–5 mm 15,25,28,33,34. Before 2019, the “European Society of Breast Radiology” established a minimum slice thickness of 4 mm 35. However, patients from 2015 to 2018 were included in our study, and a slice thickness of 5 mm was selected, which was based on previous studies. Therefore, this may affect the sensitivity of UE-MRI, especially for small-size lesions. In the future, we will reduce the slice thickness of DWI. Third, there are no unified guides for UE-MRI and AP-MRI interpretation. In UE-MRI, the optimal cutoff value of ADC has not been established. The b value and cutoff value of ADC based on each scholar's research are different, which may lead to differences between our study and other research results. The sequence of the AP-MRI protocol also needs to be optimized in further prospective research.
In conclusion, most occult lesions in mammography can be found by UE-MRI or AP-MRI; however, both of them cannot be an alternative protocol of FDP according to our study. Regarding the sensitivity of diagnosis, AP-MRI may be superior to UE-MRI, especially in sizes ≤10 mm. However, the safety based on gadolinium use cannot be ignored in future research.