Abstract P2-11-15: A novel scoring system based on the clinical and genomic variables to predict the early relapse in estrogen receptor-positive/HER2-negative (ER+/HER2-) breast cancer

Abstract

Abstract Background: Despite low invasiveness in tumor biology and high sensitivity to endocrine therapy (ET), about 5% to 10% patients with ER+ breast cancer will relapse early in the first two years (yrs) after initiation of ET. Resistance to ET remains one of the leading causes of treatment failure. Methods: Patients with ER+/HER2- breast cancer were retrospectively identified from Shanghai Ruijin Hospital. Expression of 16 cancer-related genes were measured using RT-PCR based on 21-gene assay. Cox proportional hazard model was developed to identify the clinical and genomic variables associated with pre-2-yr relapse with P value of less than .10. Both the landmark analyses and tests for the interaction between genes expression and time were performed to identify the inconsistent effects of individual genes on relapse in two time periods. Finally, a risk score was established based on the clinical and genomic variables with the shrinkage correction and validated using a bootstrap method. Relationship between the risk score and log-hazard ratio of early relapse was presented by the cubic smoothing spline method. The primary endpoint was invasive disease-free survival (IDFS) and secondary endpoints were distant relapse-free survival (DRFS) and overall survival (OS). Results: A total of 1,227 patients were identified. In univariate Cox regression, both the age over 50 yrs (HR 0.55, 95% CI 0.27 - 1.10) and tumor size greater than 2cm (HR 3.26, 95% CI 1.61 - 6.60) were independent prognostic factors for early relapse. As for genomic variables, higher expression of ER (HR 0.79, 95% CI 0.64 - 0.97), PGR (HR 0.83, 95% CI 0.71 - 0.96), BCL2 (HR 0.75, 95% CI 0.55 - 1.03), CD68 (HR 0.66, 95% CI 0.47 - 0.91), GSTM1 (HR 0.78, 95% CI 0.58 - 1.04), and BAG1 (HR 0.69, 95% CI 0.49 - 0.97) were associated with increased early relapse. Of the genomic variables, the HR of PGR, CD68, and BAG1 were in the opposite direction for 2 time periods (interaction P .026, .003, and .011, respectively; Table 1). A scoring system was established based on 2 clinical variables and 6 genomic variables with the shrinkage adjustment (C-index, 0.68; bootstrap validated C-index, 0.69; Table 2). The risk score tended to be associated with early relapse linearly using cubic spline method. When used as a categorical variable with the cutoff point decided by X-tile, risk score higher than 2.5 was also associated with increased early relapse (HR 3.23, 95% CI 1.56 - 6.69, P< .001). Likewise, the score remained an independent predictor for 2-yr DRFS (HR 6.18, 95% CI 1.79 - 21.34, P = .001) and OS (HR 4.57, 95% CI 1.02 - 20.43, P = .029). Conclusion: The scoring system reported herein, taking into account both the clinical and genomic variables, may inform prognosis and endocrine responsiveness. For patients with high risk of early relapse, treatment escalation may be considered. Table 1. Univariate analysis for individual genes by time periods.0-5 yrs0-2 yrs2-5 yrsInteraction PHR (95% CI)PHR (95% CI)PHR (95% CI)PEstrogen moduleER0.92 (0.79-1.07).2730.79 (0.64-0.97).0241.05 (0.86-1.28).651.053PGR0.93 (0.84-1.04).2030.83 (0.71-0.96).0131.05 (0.91-1.21).545.026BCL20.84 (0.68-1.04).1050.75 (0.55-1.03).0740.91 (0.68-1.22).540.374SCUBE20.92 (0.81-1.04).1990.93 (0.77-1.12).4570.91 (0.77-1.08).290.878Proliferation moduleKi671.25 (1.01-1.55).0391.31 (0.94-1.83).1081.21 (0.92-1.60).178.710STK151.12 (0.94-1.33).2221.03 (0.77-1.37).8421.18 (0.94-1.47).152.467Survivin1.09 (0.91-1.30).3691.07 (0.80-1.43).6551.10 (0.87-1.38).428.888CCNB11.07 (0.87-1.32).5260.88 (0.64-1.21).4231.23 (0.94-1.61).126.110MYBL21.16 (0.97-1.40).1071.13 (0.84-1.52).4321.19 (0.94-1.50).151.783Invasion moduleMMP111.11 (0.95-1.29).2051.03 (0.82-1.31).7901.16 (0.95-1.43).154.460CTSL21.07 (0.89-1.29).4491.16 (0.87-1.54).3221.02 (0.80-1.30).863.516HER2 moduleGRB71.11 (0.88-1.41).3700.89 (0.62-1.26).5041.30 (0.96-1.75).085.105HER20.91 (0.75-1.11).3510.89 (0.66-1.21).4620.92 (0.71-1.20).550.865GSTM10.81 (0.67-0.99).0360.78 (0.58-1.04).0880.84 (0.65-1.09).192.673CD680.96 (0.76-1.21).7040.66 (0.47-0.91).0111.28 (0.95-1.73).106.003BAG10.97 (0.76-1.22).7690.69 (0.49-0.97).0311.24 (0.92-1.68).158.011 Table 2. Univariate β coefficients and shrinkage factor for clinical and genomic variables.BetaSEHR95% CIPShrinkage factorClinical variables0.857Age, vs ≤ 50 yrs-0.6000.3560.550.27-1.10.093Tumor size, vs ≤ 2cm1.1810.3603.261.61-6.60.001Genomic variables0.313ER-0.2410.1070.790.64-0.97.024PGR-0.1900.0760.830.71-0.96.013BCL2-0.2860.1600.750.55-1.03.074CD68-0.4210.1650.660.47-0.91.011GSTM1-0.2540.1490.780.58-1.04.088BAG1-0.3780.1750.690.49-0.97.031Abbreviation: SE, standard error; HR, hazard ratio; CI, confidence interval. Citation Format: Caijin Lin, Jiayi Wu, Shuning Ding, Lisa Andriani, Weilin Chen, Deyue Liu, Li Zhu. A novel scoring system based on the clinical and genomic variables to predict the early relapse in estrogen receptor-positive/HER2-negative (ER+/HER2-) breast cancer [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P2-11-15.