Statistical Methods for Analysis of NF Clinical Data

Abstract

Abstract : This project describes research in statistical methods that would be useful for statistical modeling and analysis of clinical data from neurofibromatosis 1 (NF1) and neurofibromatosis 2 (NF2) subjects. The statistical methods are classified into the following areas: (a) estimation of familial correlation for different types of data, and (b) assessment of multi-hit mutation models for incidence of tumors. Some of the statistical methods to be developed are either new or partly new and require further research for computer software implementation. Clinical data exist in many formats, including binary, categorical, count, and continuous information. Furthermore, a common real life problem is censored data (where the beginning or end point is not known for all cases but some intermediate data exist). One goal of the project is to produce a software package for familial data analysis for different types of data, such as binary, count, and censored survival data. To date, seven manuscripts have been accepted for publication, and one other has been submitted. Presentations were made at the 2000-2002 meetings of the American Society of Human Genetics and are planned for the 2003 meeting. In addition, this year five presentations were made at the NNFF International Consortium for the Molecular Biology of NF1 and NF2, at the Fourth International Conference on Vestibular Schwannoma and Other CPA Lesions, Cambridge UK, and at the 10th European Neurofibromatosis Meeting, Turku, Finland. The report contains the following preprints: Genotype-phenotype correlation for cataracts in NF2; Familial analysis of binary traits; and Analysis of neurofibromatosis lesions by body segment. Also included are the outline for Yinshan Zhao's thesis, excerpt from a summary section on estimation methods from Zhao's thesis, and 6 abstracts published or accepted in 2003.