Chromosomal aberrations such as amplifications, deletions and complex translocations are pervasive in human epithelial cancers. These are the main cancers affecting the aged, which has motivated efforts to elucidate the mechanisms leading to chromosomal instability1. Several mechanisms of genetic instability have been proposed, including mutations in mitotic checkpoint genes that control chromosome segregation, and loss of telomere capping function resulting in dysfunctional telomeres2. The latter model is supported by mouse knockout studies that have established a link between telomere dysfunction, increased epithelial cancers and radically altered cytogenetic profiles typical of those found in human epithelial cancers3. Studies of human primary tumors and epithelial cultures have also supported the idea that telomere dysfunction and its associated bridge-fusion-breakage (BFB) cycles are important in shaping the cancer genome4,5,6. But it is not yet known at which stage of tumorigenesis telomere-induced chromosomal instability unfolds.

Burst of instability

On page 984 of this issue, Chin et al.7 report that telomere-based BFB events coincide with a burst of chromosomal instability associated with the transition from benign to malignant growth in human breast cancers (Fig. 1). Human breast cancer evolves through a well-defined series of histological stages from normal luminal epithelium to usual ductal hyperplasia (UDH), atypical ductal hyperplasia (ADH), ductal carcinoma in situ (DCIS) and, ultimately, invasive and metastatic cancer. Using fluorescence in situ hybridization, Joe Gray and colleagues observed genome integrity during these early stages of breast cancer development. They identified 'episodic instability' during the transition from ductal hyperplasia to DCIS and, notably, only a limited increase in chromosomal alterations in more advanced disease tissues (invasive cancers). The documentation of telomere erosion and anaphase bridging at this transition, coupled with the known activation of telomerase activity in DCIS and invasive cancers, bolsters the idea that telomere-based crisis is a crucial event in breast tumorigenesis that drives genomic instability. Although only a limited number of tissue samples were analyzed, the data are convincing and reinforced by recent studies describing substantial differences between normal mammary epithelial cells and DCIS, but not between DCIS and invasive tumors8,9.

Figure 1: Schematic model of breast tumor progression.
figure 1

Clonal selection drives the accumulation of genetic changes that lead to the development of progressively more aggressive tumor phenotypes. The most notable transition is from ductal hyperplasia (UDH or ADH) to in situ carcinoma (DCIS), with a marked increase in genomic instability, genetic changes and cell death due to shortened telomere–induced crisis (adapted from ref. 7).

Full size image

To obtain additional proof that telomere-based crisis fuels genomic instability in breast cancer, Chin et al. analyzed at different passages human mammary epithelial cells immortalized with ZNF217, a putative oncogene on 20q13. As expected, early-passage human mammary epithelial cells were devoid of genetic changes and had normal telomere lengths. But telomeres progressively shortened with passaging and reached a critical length, leading to crisis and telomerase reactivation, around passage 22. Coincidentally, at this same passage, the frequency of cells with anaphase bridging increased significantly, and numerous genetic changes, similar to the ones detected in primary tumors, became detectable by comparative genomic hybridization, a platform designed to audit regional amplifications and deletions in cancer genomes. Although the cells might have acquired mutations in other genes leading to this change in phenotype, the results suggest that telomere function and genomic instability are linked.

Consistent with the results of Chin et al., previous studies analyzing UDH, DCIS and invasive ductal carcinomas using comparative genomic hybridization detected extensive recurrent chromosomal changes in both DCIS and invasive lesions, but no such changes in UDH10. Based on these results, the authors of these previous reports speculated that UDH and DCIS are not clonally related, but rather UDH and similar benign tumors might represent a pathologic 'dead end' of tumor evolution. This view is at odds with the model proposed by Chin et al. Although epidemiologic data from individuals with breast cancer support the hypothesis that UDH is not the direct precursor of DCIS, definitive resolution of this point will require further molecular studies in human tumors and model systems.

Paths to aneuploidy

The model of telomere-linked genomic instability is supported by data from both human tumors and animal models. But crisis is not the sole force driving aneuploidy in epithelial cancers2. For example, genes known to have a role in aneuploidy include mitotic spindle checkpoint genes (encoding Bub1, BubR1 and MAD2) and genes involved in recombination and repair (encoding MRE11) and cell cycle control (encoding CDC4 and cyclin E). As each of these genes is mutated in only a small fraction of cases, it seems unlikely that mutations in these genes could provide a rational and unifying explanation for the age-associated increase in epithelial cancers. In this regard, the role of telomere erosion in driving the age-associated increase in breast cancer incidence becomes important, given the correlation between the number of life-time menstrual cycles (i.e., amount of epithelial turnover and, hence, telomere attrition) and breast cancer risk.

Even though chromosomal instability and aneuploidy are common features of most cancer types, it is unclear whether genomic instability must be continuously present as tumors progress or whether, after a transient period of crisis resulting in aberrant karyotypes, the cancer genome remains stable unless cells are placed under selective pressure. This latter hypothesis is supported by the finding that DCIS tumors and their invasive local recurrences, or primary tumors and their metachronous metastases, are molecularly similar. Transitory erosion of the telomeres due to sudden clonal expansion may explain the varying degree of genomic instability during tumor progression. At the time of crisis, unprotected chromosome ends can engage in illegitimate recombination, resulting in end-to-end fusion cycles and unstable dicentric chromosomes. In premalignant cells, this would initiate a checkpoint control and lead to elimination by apoptosis. But rare cells could emerge from crisis through activation of telomere maintenance mechanisms or mutations in genes encoding proteins involved in checkpoint control or apoptosis (such as p53; ref. 11). Correspondingly, during breast tumor progression, a sharp increase in apoptosis is observed in DCIS tumors, which declines as they progress to invasive lesions. Thus, telomere dysfunction may have multiple roles in tumorigenesis, contributing in several different ways to the development of the increasingly aggressive tumor phenotype.

The importance of understanding the molecular mechanism of genomic instability goes far beyond solving a crucial and interesting problem in tumor biology, as instability is one of the main reasons for current therapeutic failures and acquired resistance in cancer therapy. Thus, molecular targeting of pathways responsible for genomic instability or selective killing of cells carrying chromosomal imbalances would have a tremendous impact in the clinical management of individuals with cancer.