-
Views
-
Cite
Cite
Alice S. Whittemore, Gail Gong, Re: On the Use of Familial Aggregation in Population-Based Case Probands for Calculating Penetrance, JNCI: Journal of the National Cancer Institute, Volume 95, Issue 1, 1 January 2003, Pages 76–77, https://doi.org/10.1093/jnci/95.1.76
Close - Share Icon Share
Extract
Breast cancer risks in BRCA1 and BRCA2 gene mutation carriers may vary with other modifying genes or personal attributes. Begg (1) noted that such heterogeneity causes upward bias in risk estimates based on cancer occurrence in families of population-based samples of case patients with breast cancer. We argue that this bias is small compared with the standard errors of the estimates. Thus, the large variability in risk estimates across studies does not appear to be caused by their biases but rather by the large standard errors of their estimates. This assertion is supported both by the data reviewed by Begg and by our computer simulations, as we discuss below. Our simulations also show that estimates from multiple-case families are more precise than those from families of population-based case patients.
Table 1 shows results from the four studies reviewed by Begg that give estimates and 95% confidence intervals (CIs) for breast cancer risk among carriers of mutations in BRCA1 or BRCA2. The CIs are wide, reflecting the large standard errors of the risk estimates. We used the CIs in Table 1 to calculate a variance for each risk estimate in the table and then used the estimates and their variances to test for differences in risk across the three studies with data for each gene. We found no evidence that risk differs across studies, which does not support Begg's hypothesis that some are more biased than others.
