A Nomogram To Predict Oncotype DX Recurrence Scores for ER Positive Breast Cancer Based on Routine Histopathologic Characteristics.
Geza Acs, Kiran Turaga, Nicole N Esposito, Christine Laronga. Ruffolo Hooper & Associates, Tampa, FL; Moffitt Cancer Center, Tampa, FL; University of South Florida College of Medicine, Tampa; Medical College of Wisconsin, Milwaukee
Background: Oncotype DX is an RT-PCR based 21-gene molecular assay validated to provide prognostic and predictive information in patients with ER positive, node negative breast cancers. The Oncotype DX Recurrence Score (RS) is divided into three risk categories as low (<18), intermediate (18-30) and high (>30) risk of distant tumor recurrence at 10 years. Although it has been recently shown that the RS risk category can also be estimated using traditional histopathologic variables in approximately 2/3rd of cases using the Magee study equation, RS has not been adequately compared to prediction based on traditional histologic features. Our objective was to generate a nomogram to predict the Oncotype DX RS for ER positive breast cancers based on routine histopathologic characteristics.
Design: The study included 348 patients with ER positive primary breast cancer who underwent Oncotype DX testing between 2006-2010. The histopathologic features of the cases were prospectively determined by 2 pathologists (GA and NNE) without knowledge of the RS results. Multivariate linear regression modeling and transformation of data were used to develop the model which was then tested on an independent validation dataset. Receiver operating characteristics (ROC) curves and concordance statistics were calculated.
Results: There were 153 patients in the training dataset and 195 patients in the validation dataset. Variables included in the final nomogram were histologic type, tubule formation, nuclear grade, number of mitosis per 10 high power fields, presence of lymphatic invasion, percentage of ER and PR positivity, HER2/neu status (all with p<0.05) and total number of lymph nodes examined (p=0.15). The area under the ROC curve was 0.76 (0.68-0.82) for prediction of RS≥18 (increased risk of recurrence) and 0.64 (0.54-0.74) for prediction of score >30 (benefit from adjuvant chemotherapy). In the validation dataset, the ROC curve area was 0.66 (0.60-0.72) for prediction of score ≥18 and 0.82 (0.67-0.95) for prediction of score >30.
Conclusions: Our results suggest that a nomogram based on routine histopathologic parameters can be used to predict the Oncotype DX RS. Although the nomogram may be refined based on more cases and needs to be further validated, it may serve as a surrogate marker for RS and may be judiciously utilized in making treatment decisions.
Tuesday, March 1, 2011 9:30 AM
Poster Session III # 52, Tuesday Morning