Evaluation of genotype data in clinical risk assessment: methods and application to BRCA1, BRCA2, and N-acetyl transferase-2 genotypes in breast cancer

Associations of numerous susceptibility genes with disease risk have been reported. However, objective methods have not been developed to evaluate the conditions under which translation of knowledge about susceptibility genotypes may be clinically informative. We describe and apply a statistical approach to evaluate when genotype information may be clinically informative in disease risk assessment. We estimate an interval of cumulative cancer incidences where it may be appropriate to use these genes in disease risk assessment. We also estimate the magnitude of a log odds ratio (H) that measures genotype-disease association. We illustrate this method with three breast cancer susceptibility genotypes: population screening data evaluating the 185delAG mutation at BRCA1 and the 6174delT mutation at BRCA2 in a Ashkenazi Jewish population, and case control data for the slow acetylation genotype at the N-acetyl transferase 2 (NAT2) gene in combination with smoking. Knowledge of the 185delAG mutation in BRCA1 (HdelAG = 3.42; 95% CI: 3.04, 3.79) or the 6174delT mutation in BRCA2 (HdelT = 1.98; 95% CI: 1.16, 2.30) can be clinically informative in distinguishing individuals who are and are not at breast cancer risk in populations with cumulative breast cancer incidences of > or = 4% and > or = 13%, respectively. NAT2 genotypes alone are much less clinically informative in predicting breast cancer risk (HNAT2 = 0.10). However, knowledge of both heavy smoking 20 years ago and NAT2 genotype is a more clinically informative predictor of postmenopausal breast cancer risk with HNAT2 = 2.19, when the cumulative breast cancer incidence in the target population is at least 31%. These results indicate that knowledge of the 185delG mutation-status may be clinically informative even in populations with low cumulative breast cancer incidences, whereas the 6174delT mutation and NAT2 genotypes may only be clinically informative in a population with higher cumulative breast cancer incidence. The proposed approach can be used to objectively evaluate the conditions under which susceptibility genotypes may be applied for risk assessment or genetic screening.

Multivariate assessment of computer-analyzed corneal topographers

Methodological aspects of multivariate statistical models to describe and to assess data from computer-analyzed corneal topographers (CACT) are considered. The data generated by repeated curvature mappings of calibrated steel balls are discussed in detail with the objective of formulating the basic questions and suggesting directions to assess data from clinical applications of CACT. The interpretation of seemingly straightforward concepts such as accuracy and precision is revisited with the objective of understanding its meaning in future clinical and experimental applications of CACT. Some of the statistical problems related to the analysis of corneal astigmatism based on CACT data are also discussed.