MRI for diagnosis of pure ductal carcinoma in situ: a prospective observational study

Authors: Kuhl CK, Schrading S, Bieling HB, Wardelmann E, Leutner CC, Koenig R, Kuhn W, Schild HH
Source: Lancet 2007; 370:485-92
Affiliation: University of Bonn, Bonn, Germany

No conflicts of interest to report

Background

While ductal carcinoma in situ (DCIS) is a heterogeneous disease, there is general agreement that DCIS should be treated at least by local excision to avoid recurrence or progression to invasive carcinoma. The advent of mammographic screening has increased early detection of DCIS and the percentage of breast cancers diagnosed at a pre-invasive stage has increased from 2% in 1980 to 20% now. The general belief is that mammography is more reliable for the detection DCIS than MR. This is based on studies that use MRI to validate mammographically detectable DCIS, which has not been shown to be reliable since MRI does not visualize calcium well and does not have sufficient resolution. However, more recently, it has been proposed that improvement in technology and refinement of diagnostic criteria allows sensitive and reliable detection of DCIS. This study presents a large prospective comparison of MRI and mammography for the detection of pure DCIS.

Methods

• Patients: 7319 women between 2002-2006 who presented to the Breast Center at the University of Bonn Hospital who had one of the following (Table 1):

• a non-normal screening mammogram
• normal conventional imaging studies but clinical symptoms of breast cancer
• normal conventional imaging studies but at increased risk for primary or recurrent breast cancer
• normal conventional imaging studies and at average risk but wished to undergo additional screening with MRI

• Design: Prospective observational analysis of mammography compared to MRI for the detection of pure DCIS.
  • Imaging studies were completed and interpreted before any biopsy procedure was done. They were read and scored independently by different radiologists trained in breast imaging and with 10-15 years experience.
  • Interpretation of imaging were prospectively collected and recorded.
  • MRI scans were first read without mammographic films to avoid biased readings.
  • Mammograms were scored according to the American College of Radiology Breast Imaging and Data System (BI-RADS) 1-6. MRI was scored according to an MRI BI-RADS lexicon including lesion morphology, contrast enhancement kinetics, and enhancement pattern. A "negative" test is defined as categories 1-3. A "positive" test is defined as categories 4-5.
  • If at least one of the imaging studies was positive, or if the patient exhibited clinical findings (bloody discharge, palpable lumps, nipple retraction, Paget’s disease) the patient underwent biopsy.
  • The Breast Center’s pathology database was used to identify all women who received the final surgical pathology diagnosis of DCIS only, independent of the imaging study results.
  • Wilcoxon matched pairs signed ranks and the Student t test for paired samples were used to check for statistical significance.

Results

• 7319 patients underwent imaging. 1208 (15%) of these women had positive imaging diagnosis. Of these, 574 had benign pathology, 469 had invasive carcinoma, and 165 had pure DCIS. 2 additional cases of pure DCIS were diagnosed based on clinical findings and not by imaging result.
• MRI was more sensitive than mammography at detecting DCIS (92% vs. 56%). This is largely due to the increased sensitivity to detect intermediate and high grade DCIS by MRI. There was no significant difference in the ability to detect low grade DCIS.
• In only 49% of the cases was DCIS seen using both MRI and mammography
• In 7% of the cases, DCIS was seen only by mammography
• In 43% of the cases, DCIS was seen only by MRI
• No statistical difference between the demographics of patients with DCIS that is detected by mammography and patients with DCIS that is detected by MRI alone.
• DCIS detectable by mammography only had a high percentage of non high-grade DCIS, 83%. Compared to DCIS detectable by MRI only that had a high percentage of high grade DCIS, 60%. But no statistically significant differences.
• Overall imaging positive predictive value (PPV) = 47%
• Mammography PPV = 55%
• MRI PPV = 59%

Conclusions

• MRI is overall more sensitive at diagnosis of DCIS than mammography
• Each imaging modality is able to diagnose distinct subsets of DCIS that overlap
• However, other than DCIS grade, it is unclear what characteristics contribute to the differential ability of mammography or MRI to detect these distinct subsets.
• These data suggest that MRI might be used as an independent rather than an adjuvant study to mammography.
• However, the cost-benefit ratio of using MRI to detect DCIS is unknown
  • There is increased cost of MRI vs. mammography, as well as of over-diagnosis and likely increased rate of biopsy (with associated monetary as well as physical and emotional costs to the patient.)
  • It is unknown if detecting more cases of DCIS will improve patient outcome.