Habib Rahbar MD, Presenter: Nothing to Disclose

Sana Parsian MD, Abstract Co-Author: Nothing to Disclose

Savannah Corrina Partridge PhD, Abstract Co-Author: Research grant, Koninklijke Philips Electronics NV

Wendy Burton Demartini MD, Abstract Co-Author: Nothing to Disclose

Brenda Kurland PhD, Abstract Co-Author: Nothing to Disclose

Constance D. Lehman MD,PhD, Abstract Co-Author: Research Consultant, Bayer AG Research Consultant, General Electric Company Research Consultant, Koninklijke Philips Electronics NV

Ductal carcinoma in situ (DCIS) is a variably aggressive malignancy that has increased in incidence with improved imaging techniques. Individualized risk-based treatment strategies remain few due to insufficient risk stratification parameters, raising concerns of overtreatment. The pathologic Van Nuys (VN) classification is known to correlate with risk of progression to invasive disease and local recurrence. In prior studies, we identified promising MR biomarkers that correlated with DCIS nuclear grade (NG). We sought to investigate the potential of these MR biomarkers to further assess DCIS risk by evaluating their ability to predict VN DCIS lesion class.

After IRB approval, we retrospectively identified consecutive biopsy proven DCIS lesions in patients who underwent 3 Tesla (T) dynamic contrast-enhanced (DCE) and diffusion weighted (DW) (b=0, 800 s/mm2) breast MR from 1/2010 to 6/2011. Maximum lesion size, morphology, kinetics, apparent diffusion coefficient (ADC), and DW lesion to normal contrast-to-noise ratio (CNR) were recorded. VN classification was determined for each lesion (VN1=non-high NG without necrosis, VN2= non-high NG with necrosis, VN3=high NG with or without necrosis). Ordinal regression with a random patient effect to account for multiple lesions described associations between VN classification and MR parameters. The predictive performance of imaging features was assessed by area under the receiver operating characteristic curve (AUC).

We assessed 36 DCIS lesions (VN1: n=8, VN2: n=8, VN3: n=20) in 32 women (mean age=56±10.7 years). DW CNR was lower for higher VN class (p=0.03) and significantly discriminated VN1 lesions from VN2 and VN3 (AUC=0.70, p=0.02). The C-statistic for discrimination of individual VN classes with DW CNR was 0.61. Maximum lesion size, lesion morphology, kinetics, and ADC were not predictive of VN groups (p>0.05). 

VN1 DCIS lesions, which are known to have a lower risk of recurrence and progression to invasive disease, exhibited higher DW CNR than VN2 and VN3 lesions. Further study is warranted to investigate the role of this MR biomarker to improve risk stratification of DCIS lesions.

DW CNR shows promise as a distinct imaging biomarker that correlates with DCIS risk and may aid in the development of individualized therapies.  

Rahbar, H, Parsian, S, Partridge, S, Demartini, W, Kurland, B, Lehman, C, MR Biomarkers at 3 Tesla for Prediction of Van Nuys Pathological Classification of Ductal Carcinoma in Situ.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12025473.html