Real-Life Implementation of Treatment Guidelines for Adjuvant Chemotherapy in Invasive Breast Cancer: Retrospective Analysis of a Large Single Centre 10-Year Database.

Abstract

Abstract Background and objective: Decision to administer adjuvant chemotherapy to N-/M- invasive breast cancer (IBC) patients remains a clinical challenge. Several patient-related factors, such as age, tumor size and grade, or hormone receptor (HR) and HER2 status are incorporated in treatment guidelines. Recommendations have become more complex with time and availability of additional tests. Using the breast cancer database developed at Institut Paoli Calmettes (IPC) from 1999 onwards, our objective was to analyze the importance of key patient-related factors in chemotherapy decision and their evolution over time.Methods: Using our database containing 5854 N-/M- IBC patients, descriptive statistics and classification tree analysis (CART) examined the following factors: age (cut-off 40y), tumor size (cut-off 20 mm), SBR histologic grade, HR positivity, perivascular infiltration (PVI). At IPC, HER2 became clinical routine from 2000 onwards, and cut-off for tumor size was reduced to 15 mm after 2005.Consequently, a second cohort comprised of patients diagnosed between 2000 and 2005 with homogeneous diagnosis and management practices was analysed.Results: SBR-1 and SBR-2 represented 32% and 46% of all cases (n=5854) resp., and SBR-3 22%. Distribution of patients for the other factors of interest was: tumor size > 20 mm for 27%, HR+ for 80%, age >= 40 yr for 91%, PVI+ for 24%. Adjuvant chemotherapy was administered to 1597 patients (27,3%). The optimal tree model based on pruning and 10-fold cross-validation correctly predicted 83.5% of decision, with 93% sensibility and 61% specificity. SBR was the most important t decision factor, allowing to improve the prediction rate from 72.7% to 76.4%. The next factors were date of diagnosis, then size for SBR 1-2, and age for SBR3. Adjuvant chemotherapy was more frequent after 2000 than before: 34.2% vs 8.9% resp. for SBR1-2, and 83.2% vs 39.3% resp.for SBR3. In the 2000-2005 cohort (n=1658, SBR1=37%, SBR2=40%, SBR3=23%), chemotherapy was chosen for 754 patients (45.5%). The optimal tree model based on pruning and 10-fold cross-validation correctly predicted 79.3% of cases with 82.8% sensibility and 76.4% specificity. SBR was again the primary decision factor, improving the prediction rate from 55.5% to 75.0%. For SBR1-2 chemotherapy was used in 35% of patients, and tumor size, PVI and age were the next decision factors. For the SBR-3, grade was the only relevant decision factor in the algorithm, leading to chemotherapy in 82% of patients.Conclusion: Hierarchy of factors driving chemotherapy decision is remarkably stable over time. Tumor grade, a major indicator of proliferation, remains the most contributive factor for adjuvant chemotherapy in N-/M- IBC, despite the progressive availability of new pathologic indicators over time. Considering the variability of tumor grading by histology across centers, and the limited informative power of intermediate SBR-2 grade, improvement in tumor grading may provide a simple approach to refine currently validated treatment algorithms and allow for a more individualized treatment strategy. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 2089.