Article Figures & Data
Figures
- Figure 1.
Flowchart of selection process to include studies in systematic review and meta-analysis. The flowchart depicts the flow of information through the different phases of the systematic review and meta-analysis. It maps out the number of records identified, included and excluded, and the reasons for exclusions.
- Figure 2.
Meta-analysis of the improvement in AUC, from the addition of SNPs into breast cancer risk prediction models by increasing number of SNPs (38 models). A forest plot for the improvement in AUC and increasing number of SNPs is shown. The 95% CIs are denoted by the black tails originating from the black square, which is the point estimate for improvement in AUC in each model. The combined improvement estimate across all models is represented by the black diamond, where the diamond width corresponds to the 95% CI bounds. Heterogeneity was assessed using I2-statistic to determine the extent of variation between study population estimates. As the I2-statistic is more than 50%, the random-effects model is applied.
- Figure 3.
Meta-analysis of the improvement in AUC, from the addition of SNPs into breast cancer risk prediction models by increasing number of SNPs and baseline model group. A–F, BCRAT (5–6 risk factors of BCRAT; A); partial BCRAT (2–4 risk factors of BCRAT; B); partial BCRAT + additional risk factors (C), BCSC (D), IBIS (E), and other models (F). A forest plot for the improvement in AUC and increasing number of SNPs by baseline model group is shown. The 95% CIs are denoted by the black tails originating from the black square, which is the point estimate for improvement in AUC in each model. The combined improvement estimate within each baseline model group is represented by the black diamond, where the diamond width corresponds to the 95% CI bounds. Heterogeneity was assessed using I2-statistic to determine the extent of variation between study population estimates. As the I2-statistic is more than 50% for all model groups, the random-effects model is applied.
Tables
- Table 2.
Reported NRI values from included studies
Model No. of risk factors No. of SNPs NRI Value (95% CI)a Classification categories Author, year BCRAT (5–6 risk factors of BCRAT) 6 7 Overall: 0.085 <1.5%, Mealiffe, 2010 5 7 Overall: 0.028 1.5%–2.0%, Dite, 2013 35–39 years: 0.021 >2.0% 40–49 years: 0.074 50–59 years: −0.029 5 77 0.066 (0.019–0.110) Dite, 2016 6 75 0.033 (0.025–0.089) <1.5%, Allman, 2015 6 71 0.082 (0.003–0.162) 1.5%–<2.0%, ≥2.0% Partial BCRAT (2–4 risk factors of BCRAT) 2 + 4 51 0.062 <1.0%, Lee, 2014 1.0%–<1.5%, 1.5%–<2.0%, 2.0%–<2.5% ≥2.5% BOADICEA 8 77 0.040 (0.007–0.073) <1.5%, Dite, 2016 BRCAPRO 8 77 0.063 (0.030–0.094) 1.5%–2.0%, IBIS 12 77 0.052 (0.015–0.088) >2.0% 6 75 0.060 (0.005–0.113) <1.5%, Allman, 2015 6 71 0.181 (0.085–0.273) 1.5%–<2.0%, ≥2.0% BCSC 5 76 Case: 0.110 (0.070–0.150), Control 0.020 (−0.010–0.050) <3%, Vachon, 2015 ≥3% Other models 10 18 0.083 <1%, Hüsing, 2012 1%–<1.66%, 1.66%–<3.5% >3.5% Abbreviations: BCRAT, Breast Cancer Risk Assessment Tool (6); BCSC, Breast Cancer Surveillance Consortium (11, 12); BOADICEA, Breast and Ovarian Analysis of Disease Incidence and Carrier Estimation Algorithm (9, 10); BRCAPRO (8); CI, confidence interval; IBIS, International Breast Intervention Study (7); NRI, net reclassification improvement.
↵aPositive NRI values indicate an overall correct reclassification.
Supplementary Data
- Online Supplementary Material - Table S1. Search terms applied to each database Table S2a. 25-item "Strengthening the Reporting of Genetic RIsk Prediction Studies (GRIPS)" checklist* Table S2b. Evaluation of studies against the 25-item "Strengthening the Reporting of Genetic RIsk Prediction Studies Prediction Studies" (GRIPS) checklist Table S3a. Newcastle-Ottawa Quality Assessment Scale (NOS) coding manual for case-control studies* Table S3b. Newcastle-Ottawa Quality Assessment Scale (NOS) of Studies Table S4. Detailed characteristics of studies that evaluated the predictive ability of SNP-enhanced breast cancer risk prediction models Table S5. Improvement in AUC in SNP-enhanced breast cancer risk prediction model of, grouped by baseline model Table S6. Risk factors included in the various breast cancer risk prediction models Figure S7. Meta-analysis of the overall area under the curve (AUC) for SNP-enhanced breast cancer risk prediction models by increasing baseline AUC and baseline model group (a) BCRAT (5-6 risk factors of BCRAT) (b) Partial BCRAT (2-4 risk factors of BCRAT) (c) Partial BCRAT + additional risk factors (d) BCSC (e) IBIS (f) Other models Figure S8. Summary of pooled estimates obtained from meta-analysis, grouped by similar baseline model
Volume 28, Issue 3
