Background: Prognostic factors have been used for years to determine the benefit of adjuvant chemotherapy in breast cancer. However, reporting of the size and grade of the tumor are affected by interobserver variability in reporting. This can result in a change in the results of adjuvant online and an impact on decision making of chemotherapy. On the contrary, genomic testing such as oncotype Dx is reproducible. The purpose of our study was to assess the effect of pathological discordance on the adjuvant online results on a cohort of patients who also underwent oncotype Dx testing.
Materials and Methods: A total of 143 patients' histologies were included in this study. The results of the Phase III WSG-Plan B trial concerning central vs. local grade discrepancy rates were utilized to randomly change the grade of the tumors. 61 percent of grade 1 cancers were upgraded to grade 2 and 2% upgraded to grade 3. 4 percent of grade 2 cancers were downgraded to grade 1 and 26% were upgraded to grade 3. 1 percent of grade 3 cancers were downgraded to grade 1 and 25% were downgraded to grade 2. Likewise, change was made in the size of the tumor in 20 percent of patients. 8 to 10mm, 18-20, 28-30, and 48-50mm changed to 11, 21, 31 and 51mm respectively. 11-13mm, 21-23, 31-33, and 51-53mm was changed to 10, 20, 30, and 50mm. Ten percent of patients had the ER and Her 2 status changed.
Results: The simulation results showed that when the grade was only altered, the spearman correlation of the predicted 10 year mortality on adjuvant online with the original data was significantly changed from 1 to a result of 0.788. When the changed size was additionally added, the coefficient was 0.836. With the altered ER status, the result was 0.749 and with the Her 2 change, the spearman correlation was minimally changed to 0.742. The scattergrams showed a large number of outliers when the alteration in size was added to the altered grade.
Conclusion: Our simulation study confirms that with minimal changes in the clinical parameters because of the lack of perfect correlation between pathologist's results, there is a significant difference in the 10 year predicted mortality on adjuvant online. This is one step further in understanding the lack of correlation between adjuvant online and oncotype Dx, and the inconsistency of chemotherapy decision making with the sole use of adjuvant online.
Citation Format: Khawaja S, Thomas D, Udayasankar S, Munir A, Huws A, Sharaiha Y, Holt S. A simulation study depicting the inconsistency of adjuvant online compared to genomic testing when determining the benefit of chemotherapy [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-03-12.
