Lead articleCoping with complexity: multivariate analysis of tumor karyotypes
Introduction
Most human cancers are characterized by chromosomal aberrations and cytogenetic analysis has proved a fruitful approach to detect pathogenetically important genetic changes. In particular, the cytogenetic investigations of hematological malignancies have been rewarding in that an increasing number of disease-specific balanced rearrangements, most often translocations, have been found, which all exert their action at the molecular level through one of two alternative mechanisms: deregulation, usually overexpression, of a seemingly normal gene is one of the breakpoints, or the creation of a hybrid gene through fusion of parts of two genes, one in each breakpoint 1, 2. However, many solid tumors exhibit a much more complex pattern of aberrations. Although these chromosome changes invariably show a nonrandom distribution over the chromosome complement, tumor-specific aberrations are uncommon. Solid tumors also tend to contain a higher number of aberrations than leukemias and often exhibit extensive variability in the pattern of aberrations, even within the same histopathological entity [1]. It has been suggested that this variability reflects a multistep pathogenetic process, resulting in both loss of tumor suppressor gene function and oncogene activation [3]. The high level of karyotypic complexity in these tumor types has made a systematic characterization of the cytogenetic evolution patterns difficult and dependent upon a large body of cytogenetic information. Obviously, novel strategies are needed to understand the biological relevance of these highly abnormal karyotypes. This article presents a set of statistical methods for describing and analyzing cytogenetic data from tumor cell populations, taking both complexity and heterogeneity into account.
Section snippets
Preparation of the data
A large body of karyotypic data according to ISCN 1995 [4] on chromosomal changes in human neoplasia has been collected and is now available through the Mitelman Database of Chromosome Aberrations in Cancer [5]. Karyotypes may be retrieved from this database and used to identify frequent chromosome aberrations and imbalances [6]. Each tumor can then be assessed for the presence (1) or absence (0) of selected aberrations/imbalances and the results tabulated in a binary form. These data matrices,
Distribution analysis
In most tumor types, the number of chromosome rearrangements is roughly proportional to the grade of malignancy [1], indicating that chromosome aberrations accumulate continuously during tumor progression. The number of cytogenetic imbalances per tumor (NIPT) will thus, to some extent, reflect the biological age of the tumor. Furthermore, NIPT distributions in a population of tumor samples may give clues to the mode of karyotypic evolution. For instance, a fairly stable tumor type would show a
Temporal analysis
Intuitively, one expects that imbalances appearing early in tumor progression would be seen in both simple and complex karyotypes, whereas imbalances that appear late would predominantly be seen in highly complex tumors. Hence, by plotting the NIPT distribution for tumors containing a given imbalance it would be possible to determine if this imbalance occurs early or late. The majority of such distributions, of which examples may be seen in Fig. 2, are broad and thus no strict classification of
Principal component analysis
To identify possible karyotypic pathways, one may argue that aberrations acting in a synergistic or complementary fashion in the carcinogenic process should frequently be seen in the same tumor cases, whereas biologically incompatible aberrations would rarely be present in the same case. Thus, when calculating the correlation between the presence of different imbalances in a given tumor type, a positive correlation would indicate membership of the same karyotypic pathway, whereas a negative
Conclusions
The large amount of karyotypic data available makes it possible to perform elaborate statistical analyses of the cytogenetic variation in cancer. Particularly, general patterns of chromosome changes are now open for investigation. This necessitates the use of methods organizing the data into comprehensive biologically relevant patterns instead of focusing on specific changes. Such patterns may not be obvious from the mere inspection of a limited number of cases since the data is likely to be
Acknowledgments
Supported by the Swedish Society and the Swedish Child Cancer Fund.
References (13)
- et al.
Abnormal nuclear shape in solid tumours reflects mitotic instability
Am J Pathol
(2001) - et al.
Cancer Cytogenetics
(1995) - et al.
A breakpoint map of recurrent chromosomal rearrangements in human neoplasia
Nat Genet
(1997) Clonal chromosome abnormalities in neoplastic cellsevidence of genetic instability?
Cancer Surv
(1996)- ISCN 1995. An international system for human cytogenetic nomenclature. Mitelman F, editor. Basel: S Karger,...
- Mitelman Database of Chromosome Aberrations in Cancer (2001). Mitelman F, Johansson B and Mertens F, editors. Available...