After appropriate inclusion and exclusion criteria were applied, our study cohort included 935 lumps in 861 patients. The mean patient age at initial imaging was 49 ± 12 (SD) years. In all cases, mammography preceded diagnostic ultrasound, and 88% of the patients underwent same-day evaluation with both modalities. In 12% of cases, only a targeted ultrasound examination was performed, because the radiologist interpreted the mammographic images from the preceding screening mammogram as part of the imaging evaluation.

The most common associated symptom with a palpable lump was tenderness or pain (81/935 [8.7%]). Painful breast lumps had both benign (55/81 [67.9%]) and malignant (4/81 [4.9%]) imaging findings. Findings of diagnostic workup with mammography and ultrasound were negative for 22 of 81 (27.2%) painful lumps.

Of the 935 palpable lumps, 77 (8.2%) had a malignant outcome and 858 (91.8%) had a benign outcome (Fig. 1). According to clinical follow-up findings alone, 172 patients were determined to have a benign outcome.

Seventy-five palpable malignancies had a corresponding lesion identified at diagnostic imaging. However, two of the 935 (0.2%) palpable lumps had negative imaging findings but were subsequently found to be malignant. One patient was 49 years old, and the other 66 years old. One underwent MRI within 1 month and the other at 6 months because of clinical concern about a persistent lump. Both had suspicious MRI findings and an ultimate diagnosis of invasive lobular carcinoma.

Overall, 568 of 935 (60.7%) palpable breast masses had a corresponding benign or malignant finding at imaging, and 367 of 935 (39.3%) had negative imaging findings, categorized as BI-RADS 1. There was no difference in the presence or absence of an imaging correlate based on whether the lump was initially palpated by a primary care provider, breast specialist, obstetrician-gynecologist, or patient. When analyzing all palpable imaging findings in the setting of breast density, we found no difference in ultrasound detection of the finding in the four tissue densities. However, mammography had diminishing sensitivity from 64.7% in fatty breasts to 50.0% in dense breasts.

When we subdivided the malignant lumps by age, we found increasing prevalence of malignancy corresponding to increasing age (30–39 years, 2.5%; 40–49 years, 5.9%; ≥ 50 years, 13.6%), as would be expected (Table 2).

Figure 1 outlines the imaging modalities with which the palpable malignancy was seen, and Table 3 provides this breakdown by age group. Our study showed that 75 of 77 (97.4%) malignant palpable lumps were visible with mammography combined with ultrasound or with ultrasound alone. Sixty-six (85.7%) malignancies were visible with both mammography and ultrasound. The additional nine (11.7%) malignancies were detected with ultrasound only. No palpable malignancies were detected with mammography alone. Two (2.6%) malignant lumps had no mammographic or sonographic correlate. Figure 2 shows the visibility of palpable malignancies based on breast density. Most (6/9 [66.7%]) of the mammographically occult malignant lumps subsequently found with ultrasound were in dense breasts. The two imaging-occult malignancies occurred in women with scattered fibroglandular density.

Of the 66 palpable malignancies visible with both modalities, mammography contributed in 27 of 66 (40.9%) cases by better delineating extent of disease for surgical planning (Fig. 3). This contribution was seen in all age groups, that is, one of six (16.7%) 30- to 39-year-old women, 8 of 13 (61.5%) 40- to 49-year-old women, and 18 of 47 (38.3%) women 50 years old or older (Table 4). Overall, mammography contributed to the workup in 27 of the total 77 malignancies (35.0%). Thirty-two of 66 palpable malignancies contained calcifications within the associated mass. In 16 of these, mammography showed that calcifications extended well beyond the target palpable mass, sometimes with an associated satellite mass. These accounted for 16 of 27 (59.2%) cases in which mammography was contributory, other cases having the other features shown in Table 1.

Analysis of the 858 benign palpable lumps showed that 493 (57.5%) had a corresponding lesion seen at imaging and 365 (42.5%) had negative imaging findings or were BI-RADS category 1 (Fig. 1). Of the 858 benign lumps, 259 (30.2%) were visible with both mammography and ultrasound (Fig. 1 and Table 3). Ultrasound depicted another 224 (26.1%) lumps, and only 10 (1.2%) benign lumps were detected with mammography alone. Of the 259 benign palpable cases visible on both modalities, mammography contributed to the workup of 26 of 259 (10.0%) benign lumps by confirming a benign diagnosis (Table 4 and Fig. 4). Overall, mammography contributed to the workup in 26 of the total 858 benign lumps (3.0%). A benign appearance on mammograms was most often related to the presence of fat or characteristically benign calcifications in a palpable mass (Table 1).

Statistical analysis of the contribution of mammography to the diagnostic workup of palpable lumps visible with both modalities revealed that mammography contributed significantly more to the evaluation of malignant lumps than to evaluation of benign lumps (chi-square, p < 0.001) (Table 4). Therefore, the prime contribution of mammography was to help identify malignancy rather than to confirm benignity. The results of logistic regression analysis also suggested a significant increase in the proportion of cases in which mammography contributed to lump workup corresponding to increase in age overall (p = 0.02). Breast density did not confound these results.

We separately analyzed the incidental (nonpalpable) findings at both mammography and ultrasound (Fig. 5). Fifty-two (6.0%) of the 861 patients had incidental findings for which biopsy was recommended (BI-RADS category 4 or 5). Suspicious incidental findings were initially identified with mammography in 29 cases (55.8%) and with ultrasound in 23 cases (44.2%). Of the biopsied incidental lesions initially seen with mammography, 22 of 29 (75.9%) were benign and seven of 29 (24.1%) were malignant. Incidental cancers were detected with mammography in all age groups (30- to 39-year-old women, one; 40- to 49-year-old women, three; women 50 years old or older, three). These malignancies presented as a contralateral suspicious mass (n = 2) or a mass with pleomorphic calcifications (n = 1), ipsilateral calcifications (n = 1), or a suspicious mass in a separate quadrant from the palpable concern (n = 3). The number of incidental malignancies increased the cancer detection rate of mammography by approximately 8 in 1000 patients, from 76.7 to 84.8 (Table 5). Of the biopsied incidental lesions initially seen with ultrasound, 22 of 23 (95.7%) were benign, and one (4.3%) was malignant, presenting as a suspicious mass near but separate from a palpable lump in a patient in the 50 years old and older group.

For this cohort of women with palpable findings, the overall performance statistics of mammography and ultrasound are summarized in Table 5. Mammography did not depict 11 of 77 malignancies, for sensitivity of 85.7% and negative predictive value of 98.6%. With mammography, 74 cases were falsely identified as suspicious, for specificity of 91.4%. The positive predictive value 2 was 47.1% for mammography and 30.4% for ultrasound.

Compared with mammography, ultrasound depicted a larger number of palpable malignancies but at the expense of a higher false-positive rate (20.0% for ultrasound vs 4.0% for mammography). With ultrasound, 172 cases were falsely identified as suspicious, for specificity of 80.0%. Ultrasound depicted 75 of 77 malignant lumps alone or combined with mammography, for sensitivity of 97.4%.

Combined diagnostic mammography and targeted breast ultrasound are commonly performed for patients presenting with palpable breast masses [2–8]. Our breast care center follows this approach for women 30 years old and older.

Our study showed that the vast majority of malignant palpable lumps were detected either with mammography combined with ultrasound or with ultrasound alone. Among palpable breast lumps interpreted as benign, 0.2% were further evaluated with MRI owing to high clinical suspicion and were found to be malignant. Our results are concordant with literature reports of false-negative rates of combined mammography and ultrasound of palpable breast lumps [4–8]. Although the false-negative rate is exceptionally low, the presence of malignancy in patients with normal imaging findings at mammography and ultrasound highlights the importance of the clinical breast examination for identifying cases that may warrant further imaging and histologic sampling. Mammography alone did not depict any malignancies presenting as palpable abnormalities; this finding was similar to previously published findings [9, 10].

However, unlike Lehman et al. [9] and Leung et al. [10], who minimized the role of mammography in the workup of palpable lumps, we found that that mammography did add clinical value for women 30 years old and older with palpable breast lumps. This is largely because we defined the value of mammography as more than showing whether a lump could be identified as malignant; we included its role in identifying disease extent, incidental disease, and benign disease. Our study showed the greatest utility of mammography to be in the workup of palpable malignancies: mammography improved delineation of disease extent in 27 of 66 (40.9%) of cancers seen with both modalities, which was statistically significant (p < 0.001). In such cases, mammographic results better informed or altered surgical management. For some palpable lumps, mammographic findings helped establish a benign diagnosis, obviating the need for follow-up or biopsy in 26 cases. These results have implications for decreasing potential anxiety associated with biopsy and costs associated with continued diagnostic follow-up. Mammography also depicted seven incidental cancers. This finding validates the adjunct screening function of mammography in evaluating for ipsilateral or contralateral malignancy separate from the primary palpable area of concern [12, 13].

Because of the small sample size, we are unable to draw statistically significant conclusions about the 30- to 39-year-old group. However, there was a trend toward the contribution of mammography both in identification of malignant disease and confirmation of benign disease. A larger study focusing on this age group is a worthwhile goal for better understanding the value of mammography to these patients.

There were several other limitations. The findings represent a single institution and therefore may not be generalizable. However, ours is a tertiary care center with electronic medical records that serves a diverse patient population. Our analysis was subject to the inherent limitations of a retrospective analysis. Findings relied primarily on the report of the original interpreter rather than a review of the images. This included a subjective assessment by the reviewer based on reports as to whether mammography contributed to the workup. In addition, most patients had combined reporting of mammography and ultrasound, which limited differentiation of BI-RADS categories between the two modalities. Moreover, the imaging study that was performed first may have biased the interpretation of the subsequent study. Thus, a suspicious finding at ultrasound performed first may have biased the interpretation at subsequent mammography. To better understand the difference between these two modalities, future prospective study is warranted in which the radiologist readers are blinded to the other modality. In a few cases, mammography preceded diagnostic ultrasound by up to 3 months. This affected the diagnostic accuracy of mammography in at least two cases in which a new palpable breast lump was traced back to an interval trauma that occurred between screening mammography and diagnostic ultrasound. Last, although we had pathologic proof of malignancy based on core biopsy and surgical excision, we did not assess final tumor size or reexcision rates.

Although our study shows the clinical value of mammography, the results do not indicate the time interval within which mammography should be performed. Leung et al. [10] investigated whether interval change is detected on a mammogram when a patient presents with a palpable breast lump 6–12 months after a normal mammogram. They found that 13% of mammograms showed an interval change, 38 presenting as a more prominent finding and 42 as a new finding. Unlike our study, however, their study evaluated only whether the cancer was identified with mammography. Given the new results of our study revealing the added value of mammography in establishing benignity or extent of malignancy and incidental cancer, a worthwhile future aim would be to examine the time intervals necessary to actualize such a benefit.

Moving forward, evaluating the role of tomosynthesis for palpable breast masses may be useful because of its improved cancer detection rate compared with that of conventional mammography [14, 15]. Similarly, whole-breast screening ultrasound has been associated with an increased cancer detection rate compared with conventional mammography alone [16]. It is possible that performing a bilateral whole-breast ultrasound examination at presentation for a palpable lump may have increased detection of the incidental contralateral or ipsilateral cancers seen with mammography alone in our study. However, performing bilateral whole-breast ultrasound for all patients with palpable lumps would likely result in many unnecessary biopsies due to the lower positive predictive value of that modality [16].

Our investigation substantiated mammography as a beneficial adjunct in the diagnostic evaluation of palpable breast lumps in women 30 years old and older by elucidating the extent of disease, screening for nonpalpable malignancy, and aiding in the diagnosis of benign entities. With the incorporation of 3D tomosynthesis, mammography may add still more value to the diagnostic workup of palpable lumps. Future studies should investigate the added value of advanced mammographic techniques in the workup of palpable masses—such as tomosynthesis and contrast-enhanced spectral mammography—and the optimal timeline for repeat mammography for a patient presenting with a newly palpable breast lump.

We thank A. Brook for assistance with statistical analysis and checking tables for accuracy.