Pathologic Diagnostic Correlation With Breast Imaging Findings: A College of American Pathologists Q-Probes Study of 48 Institutions Open Access

The study was offered and conducted as a College of American Pathologists voluntary Q-Probes program, the basic mechanism of which has been previously described.8 Participants in this study, conducted in the fall of 2010, retrospectively reviewed 30 consecutive, initial, diagnostic needle core biopsy cases performed for abnormal radiologic findings. If 12 months of accessioned cases were reviewed without identifying 30 qualifying cases, participants stopped the retrospective review and included all cases identified. For each case or specimen, the participants provided detailed information about the radiologic and pathologic findings. Specimens with no prebiopsy breast imaging, with tissue specimens processed at another institution, with surgical excision specimens (incisional biopsy, excisional biopsy/lumpectomy, and mastectomy), with previous breast carcinoma, or with fine-needle aspiration biopsies were excluded from the study. Correlation, in this study, was defined as pathologic findings providing an acceptable and reasonable explanation for the radiologic findings, for example, calcifications identified in pathology for biopsies performed for radiologic calcifications. The participants were asked to determine, based on their judgment, whether the pathologic finding of each case provided an acceptable explanation for the radiologic finding. We also independently evaluated correlation based on specific reported radiologic and pathologic findings. In that evaluation, we expanded our definition of correlation to include cases with in situ or invasive ductal lesions without calcifications for biopsies performed either for radiologic calcifications or for masses. The rate of radiologic-pathologic correlation of breast needle core biopsies was determined based on (1) participating pathologists' judgments, and (2) reported radiologic and pathologic findings, judged independently. We also tested these rates for associations with institutional demographic practice variables and aggregate biopsy case characteristics.

Individual associations between the 2 correlation rates and the demographic and practice variables were investigated with Kruskal-Wallis tests for discrete-valued, independent variables and with regression analysis for the continuous independent variables.

Variables with associations (P < .10) were then included in a forward-selection, multivariate-regression model. A P value of < .05 was considered significant. In addition, the case-level radiologic-pathologic correlation was tested for association with case-specific characteristics by the Pearson χ² test.

All analyses were performed with SAS 9.2 statistical software (SAS Institute Inc, Cary, North Carolina).

Participants from 48 institutions submitted data for this study. In aggregate, a radiologic-pathologic correlation was found in 94.9% (1328 of 1399) of the cases reviewed based on the participants' judgment. There were no significant differences in the correlation rates of the following: whether surgeons or radiologists performed the biopsy, whether cores with calcifications were identified in any way, and whether radiologic reports and images were reviewed before verification of pathology reports. However, there were significant differences in the correlation rates when cases were discussed at interdepartmental, multidisciplinary conferences (P < .001) and by biopsy types (P  =  .03; see Table 1).

Twenty-one percent (292 of 1391) of the breast needle core biopsies were performed by the surgeons. The BIRADS score was not provided in 59% of the cases; however, when provided, 79.5% (449 of 565) and 12.9% (73 of 565) of the biopsies were performed for BIRADS 4 and 5, respectively. Cores with calcifications were identified in some way in 66% (386 of 585) of the cases by the radiologists (Table 2). Table 3 highlights the demographics of participating institutions.

Of the 47 participants answering, 28 (59.6%) reported having the same turnaround time (TAT) for breast and nonbreast needle biopsy specimens; 54.3% (25 of 46) of the participants have an expected TAT of 2 working days for needle core breast biopsy, whereas 43.5% (20 of 46) of all institutions had an expected TAT of 1 working day. In these institutions, 74.5% (35 of 47) did not have a designated breast pathologist, 89.4% (42 of 47 ) had an interdepartmental or multidisciplinary breast conference, and 34.8% (16 of 46) indicated they had a formal mechanism or system in place to ensure radiologic-pathologic correlation in the pathology department. For cases performed for radiologic calcifications with no calcifications identified on histology, 36.2% (17 of 47) of the participants x-ray the paraffin blocks and only cut additional sections if calcifications are present on the specimen x-ray images. Table 4 highlights other laboratory practices related to needle core breast biopsy cases in the participating institution.

The correlation rates are not significantly different based on whether the institutions use digital mammography, film-based mammography, or both. Similarly, there was no significant difference in correlation rates by whether the institution had one or more designated breast pathologists. Having a formal system in place in the pathology department to ensure correlation does not appear to be associated with increased correlation; about two-thirds of the participants did not have such a system in place (see Table 5). For cases determined to have no correlation, the pathology reports indicated the lack of correlation in 53.6% (37 of 69) of the instances and indicated additional steps taken to evaluate the specimen in 53.7% (36 of 67) of the cases (Table 2). The radiologic-pathologic correlation rates based on participating pathologists' judgments and the correlation rates judged independently based on reported radiologic and pathologic findings had a similar level of significance (Tables 1 and 5). Table 6 highlights distribution of surgical pathology cases and staffing volumes.

The study evaluated the rate of radiologic-pathologic correlation of specific radiologic lesions (calcifications, masses, or distortions) and observed a high correlation rate between radiologic and pathologic findings overall. Some practices that might be expected to increase correlation rates, such as separating cores with calcifications, reviewing radiologic reports or images before signing out cases, and having a designated breast pathologist(s), were not significantly associated with greater correlation rates in this study. One practice that was significantly associated with a greater correlation rate was having a breast biopsy multidisciplinary conference. Although not all breast biopsy cases can be discussed at multidisciplinary or interdepartmental conferences, the presence of such a conference may alert the pathologist to the need for correlation because of the possibility that the case might be discussed.

Pathologists generally do not need extensive knowledge of breast radiologic imaging to accurately interpret breast biopsies; some knowledge of radiologic imaging may help in correlating the pathologic findings with the radiologic findings. In formal interdepartmental or multidisciplinary conferences, the pathologist is exposed to the radiologic findings, the management options, and the sampling difficulties encountered by the radiologists. These pathologists gain experience and knowledge that helps ensure pathologic correlation with radiographic findings.

Being reasonably proficient in the interpretation of breast lesions may not be synonymous with having the designation of breast pathologist. Only a few institutions (25.5%; 12 of 47) in our study reported having a designated breast pathologist. Although that may be due to the participants' practice characteristics, we did not observe a significant difference in the rate of correlation between institutions with or without a designated breast pathologist. Some investigators believe that expert breast pathology assessment is necessary for adequate management of breast carcinoma,9,10 but others observed no statistical difference in the discrimination scores or the degree of inconsistency between general pathologists and pathologists with this special interest.11 Many hospitals have a policy to routinely review outside pathology materials before treatment of patients in their institutions12,13 to ensure diagnostic accuracy and complete pathology reports. We must, however, emphasize that our study was not designed to evaluate pathologic diagnostic accuracy or differences in the diagnostic opinions between designated breast pathologists and nondesignated breast pathologists, nor was it designed to evaluate intradepartmental or extradepartmental consultation practices for borderline or difficult breast needle core biopsy cases. However, possible reasons for the lack of association between having a designated breast pathologist and the correlation rates may include (1) nondesignated breast pathologists being reasonably proficient in interpreting breast lesions, (2) pathologists with interests in breast pathology not having the designation of a breast pathologist, (3) internal consultations among pathologists taking place before a case is verified (as an internal quality assurance measure), and (4) extradepartmental consultations with an expert being sought for difficult cases. Azam and Nakhleh14 observed that breast specimens were among the specimens most commonly sent for extradepartmental expert consultation. Nakhleh and colleagues15 also demonstrated that breast biopsies were reviewed internally before sign-out at a greater frequency than other specimens.

The lack of significant difference in the correlation rate observed for needle core biopsies performed by surgeons or by radiologists, although unexpected, is not particularly surprising. The American College of Radiology requires certain minimum qualifications be met by radiologists and nonradiologists (eg, surgeons) who perform stereotactic breast biopsies or ultrasound-guided breast biopsies.16,17 Therefore, any qualified physician performing image-guided breast biopsies should be certified and reasonably proficient.

Although some believe radiologic-pathologic correlation rates for needle core biopsies performed for calcifications are greatly enhanced by separating cores with calcifications from those without calcifications,18 others believe the practice is not particularly useful to pathologists.19 Separating cores with calcifications appears to be a common practice because 66% (385 of 585) of the cases in this study were reported to have cores with calcifications identified in some way by the radiologist. However, the correlation rates showed no association with whether cores with calcifications were separated. This study was not designed to assess whether this practice improves diagnostic accuracy or reduces time spent examining specimens. Therefore, we cannot comment specifically on the utility of this practice. Nevertheless, whether cores with calcifications are separated from those without calcifications, the pathologist is expected to carefully examine all the cores and specimen. This may explain the observed lack of difference in correlation with this practice.

The practice of reviewing radiologic reports and images is not common among pathologists. When available, the participants indicated that the radiologic images and reports were actually reviewed in 22.3% (313 of 1401) and 47.9% (669 of 1398) of the cases respectively before verification of pathology reports. Although reviewing the radiologic images or reports is a good practice, there does not appear to be an indication in this study that the practice significantly improves the correlation rates. This may further highlight that a vast knowledge of breast radiology is not necessary to accurately interpret breast core biopsies.

Similarly, no significant difference was found in the correlation rate for digital versus film-based mammography. One may expect digital imaging to be better, given that digital mammography is reportedly easier to manipulate for further evaluation, thereby reducing the follow-up procedures and repeat images needed.20 It is easier to share digital images for consultation.20 However, there are divergent views on which method has better cancer detection rates.21–23 Furthermore, some investigators noted that digital mammography may lead to overtreatment because of greater cancer detection rates24 and that digital mammograms may actually take longer to read25 than film-based mammograms. Our study did not evaluate the accuracy rates of cancer detection between digital and film-based mammography. Nevertheless, an accurate interpretation of the breast images is largely dependent on the expertise and experience of the radiologists or surgeons interpreting the images, regardless of the type of mammography used.

Most of the laboratories (65.2%; 30 of 46) have no formal system in place to ensure radiologic-pathologic correlation in the pathology department, but when such a mechanism is in place (34.8%; 16 of 46 institutions), the most frequent mechanism used is to discuss the incongruent case with the radiologist (56.3%; 9 of 16) responsible for the biopsy (Table 4). Surprisingly, there appears to be no significant difference in correlation rates depending on whether such a system is in place. It is not clear why a lack of a formal system for correlation does not affect the correlation rate; perhaps, it is because radiologists perform the correlation because they are required by the Mammography Quality Standards Act26 to have some formal system in place. Currently, there is no regulatory requirement in place for pathologists to have a system in place, and the awareness of the existence of such a system in radiology may be influencing pathologist behavior, regarding active participation in radiologic-pathologic correlation. However, not all needle core breast biopsies are being performed by the radiologists, not all breast biopsy pathology reports get to the radiologist(s) for correlation with findings on imaging, and perhaps not all radiologists perform radiologic-pathologic correlation. Therefore, it is a good practice for pathologists to evaluate breast biopsy specimens for correlation with the stated breast-imaging findings. Of note, for cases determined to have no correlation by the pathologists, the pathologist indicated that lack of correlation in 53.6% (37 of 69) of the reports, and indicated additional steps taken to evaluate the specimen in 53.7% (36 of 67) of the cases (Table 2). We consider documenting a lack of correlation and additional steps taken to ensure correlation in the pathology report a good practice. Such documentation may also be useful for risk management if there is a significant delay in diagnosing a malignancy because of inadequate radiologic sampling of the lesion.

This study has some inherent limitation because the data are self-reported. We also did not ask for follow-up information on the cases to evaluate whether follow-up breast imaging or surgical excision confirmed the participants' judgment of correlation for all cases. As such, we are unable to determine the false-negative and false-positive correlation. We attempted to mitigate this by evaluating the correlation rates based on specific pathologic findings that may correlate with specific radiologic findings (Table 7); however, lack of follow-up information limits the conclusion that may be drawn from the data. The sample size may be another limitation of this study. Lastly, most of the data (76.1%; 35 of 46) came from laboratories in institutions with 300 or fewer beds, which may skew that data somewhat toward smaller institutions. In spite of these apparent limitations, we believe that the data provide a glimpse into the practice patterns of pathologists from multiple institutions regarding radiologic-pathologic correlations in breast biopsies.