Impact of the Kaiser score on clinical decision-making in BI-RADS 4 mammographic calcifications examined with breast MRI

Calcifications detected on mammography are found in almost one-third of patients screened for breast cancer [1]. While mammographic calcifications are considered an early indicator of breast cancer, in particular DCIS (ductal carcinoma in situ), not all biopsies yield malignant results. Therefore, mammographic calcifications regularly require further diagnostic workup with stereotactic-guided biopsy, a significant proportion of which yield benign findings, and could thus potentially be avoided [2‐4]. A systematic review and meta-analysis reported varying malignancy rates in mammographic calcifications ranging between 6 and 82% depending on their mammographic appearance. Interestingly, the paper reveals that malignancy rates in all BI-RADS feature combinations exceed BI-RADS 3 benchmarks [5]. Therefore, invasive workup rather than short-term follow-up is formally required. Second, due to the preferred use of one-way vacuum-biopsy needles and marker placement for follow-up or surgery, stereotactic biopsy is costlier than ultrasound-guided biopsies. One potential way to exclude malignancy in these cases, and thereby reduce the number of unnecessary biopsies, would be to use contrast-enhanced magnetic resonance imaging (MRI). However, while several reports have demonstrated a diagnostic role for breast MRI with regard to calcifications, its application in this setting has yielded variable results, and it is not yet recommended for this purpose [6‐8]. One reason is the lack of definite diagnostic criteria that would enable objective and reliable exclusion of malignancy in case of lesions that present as mammographic calcifications. An interpretation of breast MRI using a clinical decision rule could alleviate this issue. The Kaiser score is such a decision-making tool and has demonstrated its usefulness in downgrading MRI-detected suspicious breast lesions through a formalized combination of the BI-RADS lexicon descriptors [9‐11].

Consequently, we investigated the diagnostic performance and inter-reader agreement of the Kaiser score in the diagnostic workup of mammographic calcifications by breast MRI to potentially avoid unnecessary breast biopsies.

The results of our study highlight the diagnostic potential of applying the Kaiser score to breast MRI in the diagnostic workup of suspicious mammographic calcifications. This clinical decision tool, which guides the reader through an intuitive, stepwise decision tree, allows a distinction of the underlying pathology by means of objective criteria. According to our results, the Kaiser score largely allows the exclusion of clinically significant breast cancer and, thus, obviates the need for unnecessary invasive procedures, with an effective reduction of up to 65.3% of unnecessary stereotactic breast biopsies.

The Kaiser score, named after Werner Alois Kaiser, a German radiologist and pioneer in clinical breast MRI, has shown substantial usefulness in the completion of lesion descriptors of the BI-RADS lexicon to reliably differentiate malignant from benign lesions on dynamic contrast-enhanced breast MRI [9‐11]. According to the Kaiser score, the probability of a lesion being malignant can range from 1 (minimal) to 11 (maximal). Our findings revealed that Kaiser scores ≤ 4 largely exclude malignancy and yielded a low number of false-negative findings, i.e., all but one non-invasive and hormonal receptor–positive tumor. Our findings were applicable to mass and non-mass lesions alike, thereby stressing the applicability of the Kaiser score in the investigated population. The Kaiser score may thus be used to avoid further invasive diagnostic approaches, such as percutaneous breast biopsies or open surgical procedures. In addition to the potential to avoid biopsies, MRI might also be helpful after inconclusive negative biopsy results. As higher Kaiser scores associate with high malignancy rates, a positive MRI might indicate the need to repeat a negative biopsy. This provides the opportunity to stratify women with suspicious mammographic calcifications into those who require or do not require further diagnostic workup, and has, therefore, the economic potential to reduce stereotactic biopsy or surgery time slots, and, ultimately, healthcare costs. Although, due to inconsistent healthcare policies in Europe, definitive numbers are difficult to provide, a diagnostic MRI scan is usually cheaper and may be less time-consuming compared with stereotactic breast biopsies, though procedure times vary depending on experience and can be substantially reduced using more modern equipment such as tomosynthesis guidance [14]. The relation of these costs currently is 2:1 in our country, as costs vary between countries and settings (academic, public, private); thus, a dedicated analysis regarding the cost effectiveness of using MRI before stereotactic biopsy of mammographic calcifications is not possible at this point. In addition, while MRI is not an invasive procedure such as stereotactic biopsy, the potential risks of contrast medium application need to be considered [15, 16]. The main potential disadvantage of MRI, however, is waiting time in case of limited availability, leading to stress and anxiety in patients. In this respect, contrast-enhanced mammography is receiving increased attention as it promises similar diagnostic information as MRI without the need to re-schedule the patient, thus avoiding a diagnostic delay [17]. Contrast-enhanced mammography, however, increases the patient’s exposure to ionizing radiation and does also require the application of contrast media associated with higher rates of adverse effects compared with MRI contrast media. It should also be stressed that this rather novel method has not been extensively tested regarding the sensitivity for non-invasive cancer and non-mass enhancements.

Since its introduction, the Kaiser score has been applied in a limited number of clinical settings [10]. Marino et al demonstrated the applicability of the Kaiser score in a problem-solving setting [11]. They found a higher inter-reader agreement when using the Kaiser score compared with BI-RADS. Further, the authors showed an improved diagnostic performance for inexperienced readers. Another study by Woitek et al demonstrated the potential to avoid unnecessary MRI-guided biopsies in MRI-only lesions using the Kaiser score. Using a cutoff value ≤ 2, more than 25% of the false-positive biopsies could have been correctly classified as benign without yielding false-negative cases [18]. The authors further demonstrated a consistently high diagnostic accuracy and inter-reader agreement for the Kaiser score, even though their diagnostic MRI scans came from multiple units and field strengths. This may be one explanation why their cutoff for malignancy was lower compared with ours, and to the results of Marino et al. Our study adds to the body of evidence and also suggests a role for the Kaiser score in the emerging indication of resolving mammographic calcifications. A recent meta-analysis demonstrated the potential of contrast-enhanced breast MRI to downgrade BI-RADS 4 calcifications while pointing out that there are no established diagnostic criteria in this setting [19]. According to our results, the Kaiser score fills this gap.

Some limitations of the present study should be mentioned. First, our study was retrospective; thus, though we could demonstrate the potential of the Kaiser Score to diagnose breast MRI lesions found in patients with suspicious mammographic calcifications, we did not prospectively demonstrate the potential to avoid unnecessary biopsies. The rather high prevalence of malignancy that was higher than the general prevalence of malignancy in mammographic calcifications suggests a certain selection bias toward more suspicious cases [5]. The effect of this potential selection bias due to time slot availability cannot be confirmed or excluded based on the retrospective nature of our study. Potential effect is overestimation of sensitivity while underestimating specificity. Therefore, prospective studies are needed to both confirm the potential of breast MRI and the Kaiser score in this setting. In addition, the analysis of the inter-reader agreement to differentiate benign from malignant breast lesions revealed a moderate result, indicating that diagnosis of lesions presenting as mammographic calcifications is challenging on MRI, even when using a standardized diagnostic algorithm. It should be noted that morphological interpretation of MRI images is always a subjective task—this is why conventional BI-RADS feature interpretation has a limited inter-reader agreement which has been reported as being lower compared with the Kaiser score [11]. However, considering the variations in sensitivity and specificity observed within this study, the results were fairly robust and thus encouraging in the setting of a breast MRI workup for suspicious mammographic calcifications. In addition, the analyzed images came from several different MRI units and protocols. This should, however, not be considered a limitation, but rather, a strength of this study, as it reflects clinical practice in a tertiary care assessment center, and thus, confirms the applicability of our findings to clinical practice. Due to the heterogeneous protocols, we did not evaluate whether background parenchymal enhancement does influence the diagnostic certainty of the Kaiser score in the investigated setting: this remains an interesting research gap as the effect of background parenchymal enhancement on the diagnostic performance of breast MRI remains controversial [20, 21].

In conclusion, we demonstrated that the application of the Kaiser score in breast MRI can downgrade up to 65.3% of lesions that present as suspicious mammographic calcifications. This suggests the potential to avoid unnecessary stereotactic biopsies in a substantial proportion of cases.