Genetic Analysis: Biomolecular Engineering 14 (1999) 229–233 Sources of bias in the detection and reporting of p53 mutations in human cancer: analysis of the IARC p53 mutation database Tina Hernandez-Boussard, Ruggero Montesano, Pierre Hainaut * International Agency for Research on Cancer Unit of Mechanisms of Carcinogenesis 150 cours Albert Thomas, 69372 Lyon, France Abstract p53 gene encodes a transcription factor with tumor suppressive properties and to date, somatic mutation of this gene is the most common genetic event in human cancer. A relational database has been developed to facilitate the retrieval and analysis of these mutations at the International Agency for Research on Cancer (IARC) and it currently contains information on over 8000 individual tumors and cell lines. Many factors may influence the detection and reporting of mutations, including selection of tumor samples, study design, choice of methods, and quality control. There is also concern that several biases may affect the way data appear in the literature. Minimizing these biases is an essential methodological issue in the development of mutation databases. In this paper, we review and discuss these main sources of bias and make recommendations to authors in order to minimize bias in mutation detection and reporting. © 1999 Elsevier Science B.V. All rights reserved. Keywords: p53; Mutations; Database; Quality control; Reporting bias; Human cancer 1. Introduction p53 gene encodes a transcription factor with tumor suppressive properties. To date, somatic mutation of this gene is the most common genetic event in human cancer, and over 8000 p53 mutations have been re- ported in tumor samples and cell lines. The majority of these mutations are missense and affect codons that encode residues within the DNA-binding domain (ex- ons 5 – 8). Less frequent are insertions and deletions, which represent approximately 12.5% of all reported mutations. In all, about 1150 different point mutations have been described spreading over 310 codons, with more than 95% falling in exons 5–8. A database has been developed to facilitate the retrieval and analysis of this mutation data. This database is available to the scientific community in several electronic formats [1] 1 and provides a unique source of information on muta- tions associated with cancer in humans. Analysis of p53 mutations is informative in molecular epidemiology (they give clues on how carcinogens can specifically damage DNA) and molecular pathology (they may help to predict the evolution and outcome of a tumor) (for detailed reviews see Refs. [2 – 4]). Of particular interest is the identification of specific mutation spectra in skin cancers linked with UV exposure, in hepatocellular carcinoma associated with dietary intoxication with Aflatoxin B1 and in lung cancers of smokers [2,5]. The p53 mutation database, maintained and devel- oped at IARC, started in 1991 as a list of mutations by M. Hollstein and C.C. Harris [6]. The database exclu- sively contains mutations identified by sequencing (ei- ther direct or from cloned DNA fragments) and published in peer-reviewed journals. It contains a de- scription of each mutation, of the biological sample in which the mutation was detected, and, when available, limited information on individual risk factors. The database does not contain information on tumors or Abbreiations: DGGE, denaturing gradient gel electrophoresis; PCR, polymerase chain reaction; rt-PCR, reverse-transcriptase poly- merase chain reaction; SSCP, single-strand-conformation-polymor- phims; TGGE, temperature gradient gel electrophoresis. * Corresponding author. Tel.: +33-472728485; fax: +33- 472738322; e-mail: hainaut@iarc.fr. 1 http://www.iarc.fr/p53/Homepage.htm or ftp://ftp.ebi.ac.uk/pub/ database/p53 1050-3862/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved. PII:S1050-3862(98)00030-8