ParticipantsGaiane M. Rauch, MD, PhD, Houston, TX (Presenter) Nothing to Disclose
Hongtu Zhu, PhD, Houston, TX (Abstract Co-Author) Nothing to Disclose
Heng Li, Houston, TX (Abstract Co-Author) Research funded, Varian Medical Systems, Inc
Beatriz E. Adrada, MD, Houston, TX (Abstract Co-Author) Nothing to Disclose
Lumarie Santiago, MD, Houston, TX (Abstract Co-Author) Nothing to Disclose
Wei T. Yang, MD, Houston, TX (Abstract Co-Author) Consultant, General Electric Company; Medical Advisory Board, Seno Medical Instruments, Inc
Rosalind P. Candelaria, MD, Houston, TX (Abstract Co-Author) Nothing to Disclose
Jessica W. Leung, MD, Houston, TX (Abstract Co-Author) Scientific Advisory Board, Hologic, Inc; Speakers Bureau, Hologic, Inc; Speakers Bureau, FUJIFILM Holdings Corporation
Mohamed Elbanan, MD, Houston, TX (Abstract Co-Author) Nothing to Disclose
Alper H. Duran, MD, Houston, TX (Abstract Co-Author) Nothing to Disclose
Ebru Unlu, Houston, TX (Abstract Co-Author) Nothing to Disclose
Minhua Wang, MD,PhD, Houston, TX (Abstract Co-Author) Nothing to Disclose
Stacy Moulder, MD, Houston, TX (Abstract Co-Author) Research funded, AstraZeneca PLC; Research funded, F. Hoffmann-La Roche Ltd; Research funded, Oncothyreon; Research funded, Novartis AG; Research funded, Merck KGaA
Yun Wu, Houston, TX (Abstract Co-Author) Nothing to Disclose
Elizabeth Mittendorf, MD, Houston, TX (Abstract Co-Author) Nothing to Disclose
Alastair Thompson, Houston, TX (Abstract Co-Author) Nothing to Disclose
gmrauch@mdanderson.org
PURPOSETo evaluate associations of quantitative MRI texture features and tumor infiltrating lymphocytes (TIL) levels in HER2+ and triple negative (TN) subtypes of breast cancer (BC) receiving neoadjuvant chemotherapy (NAC), as potential prognostic non-invasive imaging markers for pathologic complete response prediction (pCR).
METHOD AND MATERIALSRetrospective review of BC patients who had MRI at staging, neoadjuvant chemotherapy and surgery from January 1, 2008 through December 31, 2015 was performed. Demographic, imaging, and pathologic data including TIL levels were documented. Quantitative MRI texture analysis was performed using 3 types of textural features (TF): local binary patterns (LBP), gray-level co-occurrence matrix (GLCM), and threshold adjacency statistics (TAS). Associations between MRI quantitative TF, TIL levels, and pCR were evaluated by Pearson correlation and logistic regression.
RESULTSThere were 50 HER2+ and 38 TN patients (median age 51 years, range 29-59) with pretreatment MRI and TIL status for analysis; 27 HER2+ patients and 15 TN patients had pCR at surgery. For HER 2+ patients 9 TF significantly correlated with pCR (p<0.05): f1 (angular 2nd moment), l3 (75 percentile), l4 (standard deviation), t1-t6 (adjacency 0-5). Four TF were significantly associated with high TIL levels (p<0.05): texture l4 (standard deviation), t2 and t3 (adjacency 1 and 2). Additional 4 TF had weak association with TIL (p<0.1): feature f8 (sum entropy), t1, t3 and t4 (adjacency 0, 3 and 4). Three TF were significantly associated with both, pCR and TIL (p<0.05): texture l3 (75 percentile), l4 (standard deviation), t9 (adjacency 8). For TN patients 4 TF f2 (contrast), t1,t3 and t4 (adjacency 0 ,2,3) were significantly associated with pCR (p < 0.005). No TF were significantly associated with TIL levels for TNBC, only t3 and t4 (adjacency 4 and 5) showed weak association with TIL levels (p<0.1).
CONCLUSIONQuantitative tumor MRI texture analysis in HER2+ BC showed 9 TF associated with pCR, 8 TF with TIL and 3 TF with both pCR and TIL; for TNBC 4 TF were associated with pCR, and 2 TF weakly associated with TIL.
CLINICAL RELEVANCE/APPLICATIONAnalysis of associations of MRI quantitative TF with pCR and TIL in HER2+ and TNBC may help to develop prognostic non-invasive imaging markers for treatment response prediction.