1558P - Independent validation of the Breast Cancer Risk Assessment Tool (Gail model) for predicting breast cancer risk in Egyptian population

Background

The Gail model (GM) has proven to be a valuable tool for breast cancer risk assessment in the Western population. There is limited evidence regarding its accuracy and validity in MENA populations.

Methods

The study included women who attended a breast cancer screening program at Cairo University Hospitals from November 2019 till May 2021. GM calculates the risk compared to matched women of the same age from the USA population, a further independent screening cohort of 8400 British women aged > 50 years was used for cross-comparison. The USA-validated risk score cutoff of 1.67 was used to define high-risk women. GM clinical validity was evaluated using calibration measured by 5-year observed to expected cancer events ratio (O/E) and discrimination measured by the AUC of the ROC curve analysis.

Results

A total of 10604 women attended for screening and 175 (1.7%) were finally diagnosed with breast cancer. The median age of cases was significantly older compared to controls (53 vs 46 years, p<0.001). The mean 5-year risk and lifetime risk scores were 0.85 and 8.49, respectively. The mean 5-year-risk in cases vs controls was 1.18 vs. 0.84. The O/E ratio was 0.47 with a relative difference of -4.64%. The cutoff for high-risk showed a sensitivity of 18.3%, specificity of 91.4%, and AUC of 0.55 (p= 0.029) in our cohort. The cases had a significantly higher rate of high-risk women compared to controls (18.9% vs 8.6%, p< 0.0001). Compared to the USA women of matched age, our cohort had a significantly lower rate of high-risk women (9% vs 13.5%, p<0.0001). Limiting the analysis to women aged > 50 years, Egyptian women had less rate of high risk compared to the USA (21% vs 37%, p< 0.001) and British women (21% vs 47%, p<0.001). An age-adjusted logistic regression analysis identified other significant non-GM risk factors including body mass index (odds ratio (OR): 1.02, 95% CI: 1.00-1.03, p= 0.025) and mammographic density (OR: 1.50, 95%CI: 1.08- 2.10, p= 0.017).

Conclusions

GM showed a modest discriminatory power in identifying high-risk women in our cohort. In addition, GM overpredicted the risk of breast cancer occurrences. Breast cancer risk models validated in Western populations should be used cautiously in other populations.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Cairo University.

Disclosure

K.S. Shohdy: Financial Interests, Personal, Invited Speaker: Pfizer; Financial Interests, Institutional, Research Grant: AstraZeneca, Novartis; Financial Interests, Other, Educational Grant: Adaptimmune. All other authors have declared no conflicts of interest.