Multivariate analysis of karyotypic abnormality in leukemia facilitated by numerical encoding of cytogenetic data

A system is proposed whereby human karyotype data is expressed in quantitative terms rather than in the ISCN (1985) terminology used at present. This recoding facilitates application of multivariate analysis using standard statistical packages. As an example, karyotypes of 714 cases from 11 leukemias (Mitelman, 1983) are here recorded and subjected to discriminant analysis (SPSSx, 1983). Significant karyotypic specificity is apparent in six of the 11 FAB leukemia types. Four others show insignificant levels of specificity, while the last is equivocal. These results merely confirm present views. However, their generation by means of computerized multivariate analysis is novel, and confirms the feasibility of the approach. In this quantitative form, karyotypic data may be combined with any other data of diagnostic or prognostic value. Given such a consolidated data set, desired information concerning any aspect of neoplasia could be extracted via a single procedure.