Recalibration of the Gail model for predicting invasive breast cancer risk in Spanish women: a population-based cohort study

Abstract

The Gail model for predicting the absolute risk of invasive breast cancer has been validated extensively in US populations, but its performance in the international setting remains uncertain. We evaluated the predictive accuracy of the Gail model in 54,649 Spanish women aged 45–68 years who were free of breast cancer at the 1996–1998 baseline mammographic examination in the population-based Navarre Breast Cancer Screening Program. Incident cases of invasive breast cancer and competing deaths were ascertained until the end of 2005 (average follow-up of 7.7 years) through linkage with population-based cancer and mortality registries. The Gail model was tested for calibration and discrimination in its original form and after recalibration to the lower breast cancer incidence and risk factor prevalence in the study cohort, and compared through cross-validation with a Navarre model fully developed from this cohort. The original Gail model overpredicted significantly the 835 cases of invasive breast cancer observed in the cohort (ratio of expected to observed cases 1.46, 95 % CI 1.36–1.56). The recalibrated Gail model was well calibrated overall (expected-to-observed ratio 1.00, 95 % CI 0.94–1.07), but it tended to underestimate risk for women in low-risk quintiles and to overestimate risk in high-risk quintiles (P = 0.01). The Navarre model showed good cross-validated calibration overall (expected-to-observed ratio 0.98, 95 % CI 0.92–1.05) and in different cohort subsets. The Navarre and Gail models had modest cross-validated discrimination indexes of 0.542 (95 % CI 0.521–0.564) and 0.544 (95 % CI 0.523–0.565), respectively. Although the original Gail model cannot be applied directly to populations with different underlying rates of invasive breast cancer, it can readily be recalibrated to provide unbiased estimates of absolute risk in such populations. Nevertheless, its limited discrimination ability at the individual level highlights the need to develop extended models with additional strong risk factors.Electronic supplementary materialThe online version of this article (doi:10.1007/s10549-013-2428-y) contains supplementary material, which is available to authorized users.

Assessment of performance of the Gail model for predicting breast cancer risk: a systematic review and meta-analysis with trial sequential analysis

“…However, Gail model 2 overpredicted the risk for breast cancer (pooled E/O = 1.20; 95% CI 1.07–1.35) (see Additional file 3 ). When further stratified by different versions of Gail model 2, the pooled E/O ratios of Caucasian-American Gail model 2 in American [ 11 , 12 , 15 – 20 , 31 ], European [ 28 – 30 , 32 – 35 ] and Asian [ 39 , 41 ] women were 0.98 (95% CI 0.91–1.06), 1.07 (95% CI 0.66–1.74) and 2.29 (95% CI 1.95–2.68), respectively. The pooled E/O ratio for Asian women was significantly higher than that in American and European females ( p < 0.001).…”

“…Twenty-six articles with 29 datasets describing the C-statistic/AUC of the Gail model were combined to evaluate its pooled discrimination [ 11 , 15 , 18 – 24 , 27 , 29 – 32 , 34 – 36 , 39 – 46 , 51 ] (Table 2 ). The pooled AUC was 0.60 (95% CI 0.58–0.62) with substantial heterogeneity ( I 2 = 97.0%; p < 0.01) (Fig.…”

“…Results also showed the pooled AUC for Gail model 1 was 0.70 (95% CI 0.57–0.77), and when stratified by the geographic region the pooled AUCs for Gail model 1 in American and European women [ 22 , 36 ] and Asian females [ 42 , 44 , 46 ] were 0.55 (95% CI 0.53–0.56) and 0.75 (95% CI 0.63–0.88), respectively (see Additional files 10 and 13 ). Additionally, the pooled AUC for Gail model 2 was 0.59 (95% CI 0.57–0.61), and when stratified by the geographic region and different versions of Gail model 2 the pooled AUCs for Caucasian-American Gail model 2 in American [ 15 , 17 – 21 , 23 , 24 , 27 ], Asian [ 39 , 41 , 43 , 45 ] and European [ 29 , 30 , 32 , 34 , 35 , 51 ] females were 0.61 (95% CI 0.59–0.63), 0.55 (95% CI 0.52–0.58) and 0.58 (95% CI 0.55–0.62), respectively (see Additional file 13 ). However, only one study clearly stated that they used Asian-American Gail model 2, and the AUC was reported to be 0.54 (95% CI 0.50–0.59) [ 41 ].…”

“…However, these studies showed variability in their calibration (expected-to-observed (E/O) ratio) and discrimination (Concordance-statistic (C-statistic) or area under the curve (AUC)). Although three systematic review and meta-analyses validated the Gail model previously [ 53 – 55 ], 19 studies [ 13 , 14 , 17 – 20 , 22 – 24 , 32 – 36 , 38 , 40 , 41 , 51 , 52 ] with inconsistent results have been published subsequently or were not included in the previous meta-analyses. However, the evaluation studies launched in China [ 39 , 42 – 50 ] have not been incorporated before and the diagnostic accuracy of the Gail model has not been fully evaluated.…”

Preventive treatments for breast cancer: recent developments

“…In other words, they are effective at the population level, with predictive capacities over 90 %, but perform poorly in relation to concordance, or the ability to determine individual risk (they barely obtain areas under the curve higher than 0.6). Two recent Spanish studies have produced similar results, also confirming the importance of mammographic density as a decisive risk factor [22, 23]. It is interesting that although the results of the MAP.3 study [24] are cited, exemestane is not expressly included among the agents to be considered (this also occurs with the NICE guidelines), perhaps because of not being approved yet for the preventive indication in the US.…”