Ductal carcinoma in situ (DCIS) is a common pre-invasive malignancy of the breast, representing approximately 20% of all breast cancer diagnoses.12 It is widely believed that DCIS is a precursor lesion to invasive ductal carcinoma, but the exact biologic nature is not completely understood and debated by some.3-5 DCIS is unarguably a heterogeneous disease with variable malignant potential. Evidence shows that high-grade DCIS is an aggressive subtype with an overall poorer prognosis than non-high-grade disease. There have been many studies evaluating the role of the radiologist in the diagnosis of high-grade DCIS with emphasis on radiologic-pathologic correlation using standard mammography and magnetic resonance imaging. Our current understanding of the clinical importance of high-grade DCIS from the perspective of a radiologist and characteristic imaging features are discussed in detail.
The diagnosis of DCIS has increased dramatically over the last several decades from an incidence of less than 2 per 100,000 in the early 1970s to 32.5 in 2004.1 Much of this increase has been attributed to the advent of screening mammography. Some advocates of screening see this as a victory, achieving one of the goals of a screening program: the prevention of life threatening invasive cancer by detection and treatment at the in situ stage.6 Detractors, however, believe the detection of DCIS leads to a substantial number of patients being over-diagnosed and over treated for a non-life threatening condition.7 While large scale trials have shown survival benefit of screening mammography in the range of 30 percent, the screening debate goes on and is beyond the scope of this discussion.8-11
Much of the controversy regarding the increase in diagnosis of DCIS over the years lies in our limited knowledge of the natural history of the disease. There is little argument that DCIS is likely a precursor to invasive ductal carcinoma. However, it is very clear that some — but not all — of DCIS will progress over the lifetime of a patient. The evidence well summarized by Erbas et al. showed that 14-53% of DCIS misdiagnosed as benign will progress to invasive carcinoma over a 10-15 year interval.4 In a study by Sanders et al., low-grade DCIS progressed in 11 of 28 patients with most occurring within 10 years; 3 were diagnosed between 23 and 42 years after the initial biopsy; and 5 of 11 died of breast cancer.3
Autopsy studies suggest that a substantial number of DCIS cases may remain subclinical, although the interpretation and significance of these findings is debated.12,13 Papers such as these underscore the concept that some DCIS is effectively benign, but it remains evident that there is no way to prospectively determine if and when DCIS will progress to invasive disease. Moreover, the available data are not representative of the full spectrum of DCIS and largely exclude high-grade lesions. High-grade DCIS is rarely misdiagnosed pathologically and is routinely surgically excised, owing to the perceived malignant potential. This has allowed for very limited long-term observations.
In spite of the unknown, the overall prognosis for DCIS is excellent with appropriate surgical and oncologic management (approximately 98% long-term survival).14 DCIS is typically treated with wide surgical resection with or without radiation therapy; there is an evolving role for hormonal therapy.15,16 With a breast cancer specific mortality of less than 2%, it has proven difficult to demonstrate significant survival benefit with more advanced treatment options, such as radiation or hormonal therapy. This has led to considerable effort to stratify patients with a diagnosis of DCIS based on the risk of local recurrences, as invasive recurrence (either local or systemic) appears to be the primary source of breast cancer specific mortality in these patients. Approximately half of patients with recurrence after breast conservation surgery are diagnosed with invasive disease, and 12- 15% of these patients ultimately die of breast cancer.17,18 While the overall prognosis for conservation therapy is good, the risk of recurrence or death is relatively negligible when DCIS is treated with mastectomy. 18,19
The pathologic evaluation of DCIS is one of the primary considerations in stratifying patients and has shifted from a purely architectural classification, which offered little prognostic information, to a focus on the nuclear grade and degree of cellular necrosis. This is reflected in the Van Nuys system which simply divides DCIS lesions into high-grade and non-high-grade; the latter group is further divided into those with or without necrosis.20 Moreover, the Consensus Conference on Classification of Ductal Carcinoma In Situ (1997) recommends stratifying DCIS first by nuclear grade (high, intermediate, and low) and then determining the presence or absence of necrosis due to the potential treatment implications.16 The consensus reflects the current understanding of high-grade DCIS as an aggressive subtype of DCIS with an overall poorer prognosis than non-high-grade disease. Analysis of the data has shown that a high nuclear grade may increase the risk of local recurrence after breast conservation therapy, shorten the time to recurrence, increase the rate of distant metastases, increase the rate of recurrence with invasion, and increase mortality with recurrent invasion.17,19,25 High grade DCIS at core needle biopsy also appears to be a significant risk factor for underestimation of invasive breast cancer, a phenomena which occurs in approximately 25% of all DCIS diagnoses.26
High-grade DCIS represents the majority of screening detected in situ lesions in multiple series, further emphasizing the importance of this diagnosis.27-29 Diagnosing high-grade DCIS represents a relatively frequent opportunity for radiologists to impact patient care. Thorough knowledge of the characteristic imaging features of high-grade DCIS, as well as the limitations of imaging, is imperative.
Microcalcifications are found in an estimated 50-75% of all DCIS diagnosed on mammography and in approximately 90% of clinically occult DCIS.30-34 Radiologic-pathologic correlation has shown that these calcifications develop as a consequence of calcified intraluminal cellular debris secondary to a high concentration of calcium in adjacent necrotic cells and from ductal secretions, such as mucin or other calcific product.31,35
Many studies have demonstrated a correlation between the type of mammographic calcification and the pathologic diagnosis, suggesting that certain calcification types are more likely to be associated with high-grade lesions at histopathology. Specifically, linear branching calcifications are generally predictive of high-grade DCIS (Figure 1)21,36-38 These result from extensive intraluminal necrosis and calcifications which form “casts” of the ducts, yielding the characteristic linear branching pattern. A variety of mammographic descriptors have been used that with nuanced differences appear to be synonymous to linear branching calcifications, including “fine linear branching,” “casting,” and “comedo” calcification. The range of non-standardized microcalcification descriptors in the literature may indicate a measure of subjectivity in the analysis, although generally the differences can be reconciled.
While linear branching calcifications are characteristic of high-grade DCIS, the significance of this finding as a histologic predictor of disease is debated. High-grade DCIS appears to present with this finding in a majority of cases. In one study by Lee et al., 15 of 16 high-grade DCIS lesions presented with linear calcifications and showed excellent correlation.36 In another study by Dinkel et al., 14 of 18 high-grade DCIS lesions showed linear calcifications. This pattern represented high-grade DCIS 56% of the time. Though this is the majority, compared with intermediate and low-grade DCIS, this was not a statistically significant result. The remaining 32% and 12% of linear calcifications represented intermediate and low-grade DCIS, respectively.39 First evaluated by Tabar et al., multiple studies have provided evidence that casting calcifications import a poor prognosis when associated with small invasive cancers.3,40,41 Analysis shows that these casting calcifications consistently represent the presence of extensive high-grade DCIS (Figure 2).
Unfortunately, as the Dinkel study illustrates, there remains considerable overlap in the imaging appearance of not only different grades of DCIS but also between DCIS and benign processes, as high grade DCIS is not confined to the linear branching pattern of calcifications. While this morphology has shown good positive predictive value for high-grade DCIS, other calcification patterns have not produced significant negative predictive values or correlation with low-grade disease that might be used to confidently reduce biopsy rates. Indeed, high-grade DCIS is seen with varying degree in all of the ACR Bl RADS suspicious calcification types, including amorphous or indistinct, coarse heterogeneous, and fine pleomorphic forms, though less frequently(>Figure 3).30,36,37 The challenge for radiologists is most evident in early stages when high-grade DCIS lesions are small and more confined; the appearance of associated calcifications is often non-specific. Appropriate biopsy technique and sampling provide an opportunity to limit under-diagnosis in these situations.26
When not calcified, DCIS presents in numerous ways on mammograms, including masses, asymmetries, architectural distortion, and even as a negative exam.32 DCIS may present as a mass, either palpable or screen detected in up to 10% of cases of DCIS; this presentation is seen more commonly in low-grade lesions, rather than high-grade DCIS.42,43 Presentation as a focal asymmetry may be especially challenging to radiologists when the finding is questionable, not seen with ultrasound, or difficult to localize stereotactically. Moreover, a negative mammogram may harbor DCIS, as demonstrated by occult cases identified only with the use of MRI. While such presentations are less common, it remains important to be aware that high grade DCIS may present as a non-calcified mammographic abnormality.
The overalI sensitivity of breast MRI for the detection of all grades of DCIS was previously considered to be relatively low with authors reporting various sensitivity data for DC IS as low as 77%.44 However, with improved MRI techniques and high spatial resolution, as many as 98% of DCIS cases are now detectable by MRI with an additional 6-23% of mammographically occult DCIS lesions detectable only by MRl.44,45 This is largely attributed to the enhancement of non-calcified DCIS which cannot be identified with a mammogram.45
The most common MR imaging finding in DCIS falls under the category of “nonmasslike enhancement” (NMLE) and is demonstrated in 60-80% of cases.47,48 Though there is some variability in the literature regarding the exact descriptors, the NMLE seen with DCIS is typically in a segmental or linear distribution. Morakkabati-Spitz et al. demonstrated a positive predictive value of 34% and specificity of 96% for segmental and linear enhancement patterns.49 The most commonly seen internal enhancement pattern among NMLE lesions associated with DCIS is clumped or heterogeneous enhancement (Figure 4)48 Less commonly, purely DCIS lesions manifest as a mass (14-34%) or focal enhancement (1-12%).44
The kinetic characteristics of pure DCIS lesions are more heterogenous and less predictive than those of invasive cancers, which are more likely to demonstrate early enhancement followed by rapid washout kinetics. The majority of pure DCIS lesions have rapid initial phase of enhancement in up to 77% of cases.46,48,50 The type of delayed enhancement is variably reported, but most commonly described as plateau or washout. Less often, DCIS lesions may demonstrate slow, progressive delayed enhancement (Figure 5). As with enhancing masses, a suspicious morphology, such as unilateral segmental or linear enhancement, will trump an associated benign appearing dynamic enhancement curve.
While one might expect that enhancement morphology and kinetics would reflect biologic behavior and by extension nuclear grade, there is no definitive evidence to suggest that either can be used to predict the presence of high-grade DCIS.46,48 When morphologic features of high-grade ver>us non-high grade DCIS are compared, there is simply no statistical difference that would separate these categories.51,52 However, some potentially significant observations have been made regarding features of high-grade DCIS on MRI. High-grade DCIS appears to be more easily detected than low-grade, suggesting MRI may have a significant benefit in excluding high-grade disease with a negative exam.49 Additionally, high-grade DCIS is significantly more likely to be detected with MRI than conventional mammography, with as many as 48% of high-grade cases detected with MRI alone.45
Enhancement features of pure high-grade DCIS, including a focal branching pattern or irregular contour, may also be helpful in prospectively differentiating from pure invasive disease.52
Overall, pure DCIS has an excellent prognosis; however, high-grade DCIS is an aggressive subtype with significantly greater morbidity and risk of mortality with recurrent invasive disease. Appropriate use of mammography and MRI affords radiologists an opportunity to identify this population and guide the most appropriate surgical and oncologic management based upon our current understanding of the disease. Research has extensively evaluated mammographic and MRI features of high-grade DCIS, and its appearances are well documented. Unfortunately, imaging currently shows little prospective value in cases of pure high-grade DCIS beyond the ability to make the initial diagnosis. Future research is necessary to determine the full impact of imaging patients with high-grade disease and to further define the best clinical treatment strategies.
The views expressed in this material are those ofthe author, and do not reflect the official policy or position of the U.S. Government, the Department of Defense, or the Department of the Air Force.
Ballard LJ, Ballard GR. High‐Grade Ductal Carcinoma In Situ: An Overview for the Radiologist. J Am Osteopath Coll Radiol. 2013;2(1):18-25.