Mirroring similar observations in most other cancers, initial studies measuring bulk telomere lengths by TRF analysis [40–42] or the slot blot assay  demonstrated that the majority of invasive mammary carcinomas had shorter telomeres than adjacent, benign breast tissues. Telomere lengths in cancer cells were shorter in high-grade tumours  and short telomeres correlated with aneuploidy and the development of lymph node metastases . Subsequently, high resolution in situ telomere length assessment combined with immunostaining to differentiate specific cell types , confirmed that significant telomere shortening is prevalent in ∼70% of invasive mammary carcinomas . Interestingly, ∼25% of invasive breast carcinomas contain telomeres that are either similar or longer than the adjacent stromal fibroblasts . Additionally, two studies have identified breast tumours displaying the ALT phenotype, a telomerase-independent telomere length maintenance mechanism characterized by remarkable telomere length heterogeneity, ranging from ultra-short to ultra-long telomeres [45, 46]. The ALT phenotype  has been primarily observed in sarcomas, but is relatively rare in most carcinomas . Similar telomere length distributions seen in cancer cells have been observed in the preneoplastic lesions, ductal carcinoma in situ and lobular carcinoma in situ, although the ALT phenotype has not been reported.
Surprisingly, not only does telomere shortening occur in the majority of in situ and invasive breast carcinomas , but telomere length alterations also occur in seemingly histologically normal breast tissues. These alterations have been observed in normal terminal ductal lobular units adjacent to a tumour and even in disease-free breast tissues obtained from reduction mammoplasties [44, 49, 50]. Using Q-FISH, telomere lengths were assessed in normal lobules and normal lactiferous ducts. Strikingly, telomere shortening was observed in the majority of normal lobules, but not in normal lactiferous ducts. Notably, short telomeres were only seen in the luminal cells and not in the myoepithelial cells. This finding was confirmed in a recent study that observed telomere shortening in normal luminal and tumour cells, but not shortening in the myoepithelial or fibroblast cell populations . Although interesting biologically that telomere shortening may contribute to breast cancer promoting genomic alterations, these observations most likely preclude the use of telomere length measurements, particularly bulk measurements, as a diagnostic marker in breast tissues (e.g. needle core biopsy specimens). Telomere length measurements in cell preparations, for example fine-needle aspirates (FNAs), may be difficult because the histological information is lost. However, telomere length measurements in breast tissues adjacent to tumours may still have clinical utility. In particular, telomere DNA content measured by the slot blot assay was observed to be decreased in benign tissues 1 cm away from the visible tumour margin, but not to the same extent as in tissue 5 cm from the tumour margin, suggesting a possible cancer field effect . The concept of cancer field effect, or field cancerization, refers to the occurrence of molecular alterations in histologically normal tissues surrounding tumours . Although still preliminary, the identification of molecular alterations, such as telomere shortening, in histologically normal cells may have clinical implications for breast-sparing surgery by defining appropriate molecular tumour margins and assessing risk factors for the development of recurrent disease .
A newer area of investigation has focused on measuring telomere lengths in peripheral blood lymphocytes, either by TRF analysis or Q-PCR, to assess potential links between constitutive telomere length and risk of breast cancer development. However, to date, the results have been largely conflicting. When comparing telomere lengths to healthy controls, three studies have observed shorter telomeres in cancer patients [53–55], one study did not observe any significant difference , and two studies observed longer telomeres in cancer patients [57, 58]. Recently, a prospective cohort of postmenopausal women did not show a significant association between increased risk of developing breast cancer and telomere length in peripheral blood lymphocytes . Pooley et al. showed that decreased mean telomere length in peripheral blood lymphocytes was associated with a significant, yet modest increased risk of developing breast cancer in a retrospective study; however, the association was not replicated in a prospectively collected cohort . The authors postulated that the observed telomere shortening predominantly occurs after diagnosis, thus diminishing its potential value as a predictive risk marker for breast cancer. More research is warranted to clarify these issues.
Whereas, the above studies have investigated mean telomere length, Zheng and colleagues have focused on assessing telomere alterations on individual chromosome arms [61, 62]. Suggesting that critically short telomeres on specific chromosome arms may be an underlying mechanism for chromosome specific instability, chromosome arm-specific telomere lengths were measured by Q-FISH in short-term cultured blood lymphocytes. In the first case–control study, short telomere lengths on chromosome 9p were associated with an increased risk of breast cancer . Interestingly, the CDKN2A gene which encodes for p16INK4 and p14ARF, tumour suppressor proteins that regulate the Rb and p53 pathways, is located on chromosome 9p. In premenopausal women, these findings where confirmed and extended; in addition to 9p, short telomere lengths on 15p, 15q and Xp were also associated with an increased risk of the development of breast cancer . Chromosomal arm-specific telomere length analysis could be incorporated in a panel of biomarkers used for risk assessment of breast cancer.
A major characteristic of prostate cancer is prominent chromosomal instability. Prostate cancer is thought to develop from benign epithelium through high-grade prostatic intraepithelial neoplasia (PIN), the earliest precursor lesion, to invasive adenocarcinoma. Because telomere dysfunction causes chromosomal instability, Sommerfeld and colleagues studied telomere dynamics in the prostate by measuring telomere lengths in matched samples of benign prostatic hyperplasia (BPH), benign nodules composed of stromal components and epithelial cells and invasive prostatic adenocarcinomas obtained by radical prostatectomy . Prostate cancer tissue telomere lengths were significantly shorter than the telomeres from cells in BPH tissues and from adjacent normal tissues. These results were confirmed [64, 65], and extended to demonstrate an association between reduced telomere lengths in prostate tumours and disease recurrence .
Using high-resolution in situ methods, telomere shortening was observed in tumour epithelial cells compared to normal prostatic epithelial cells in the vast majority of prostate tumours , as well as in high-grade PIN lesions [67, 68]. Within these PIN lesions, telomere shortening only occurred in the luminal epithelial cells and not in the basal epithelial cells or the surrounding stromal cells. The high prevalence of short telomeres in high-grade PIN lesions, of which only a small fraction progress to invasive carcinomas, would preclude the use of this molecular marker for prostate cancer detection. However, telomere length measurements in tumour specimens obtained at time of surgery or taken at time of biopsy do seem promising as a potential prognostic marker in prostate cancer [66, 69, 70].
To date, only one study has assessed the relationship between constitutive telomere length in peripheral blood lymphocytes and prostate cancer. Using Q-PCR, this nested case–control study showed no association between mean telomere length and risk of aggressive prostate cancer development . More interestingly, recent studies have reported telomere shortening in histologically normal prostate tissues from diseased prostates. In one study, telomere lengths were assessed by Q-FISH in biopsies from a cohort of men diagnosed with high-grade PIN, but without evidence of prostate cancer . The degree of telomere shortening in the surrounding stromal cells and within cells from the high-grade PIN lesions were associated with the eventual diagnosis of prostate cancer . In another report, mapping of the spatial distributions of telomere DNA content, measured by the slot blot assay, revealed telomere length variations in fields of histologically normal tissues surrounding tumours in a small set of radical prostatectomies specimens . Expanding on their previous work, Joshua et al. assessed telomere lengths topographically by Q-FISH in normal epithelium, adjacent stroma, BPH, high-grade PIN and cancer in whole mount tissue sections . Here, the presence of short telomeres in different prostatic histologies correlated with telomere lengths within adjacent stromal cells, suggesting microenvironmental effects within the prostate gland, such as increased oxidative stress .
The lack of association between constitutive telomere length, as measured in peripheral blood lymphocytes, and increased prostate cancer risk and the existence of telomere shortening in seemingly histologically normal prostate tissues suggests measurement of telomere lengths will not be useful as a direct diagnostic marker of prostate cancer. However, it is possible that telomere length analysis could nonetheless be potentially useful. For example, in settings where there is suspicion of cancer, i.e. in men with persistently elevated serum prostate specific antigen (PSA) but negative biopsy results. If abnormal telomeres in the stromal or epithelial cell populations are present on the biopsy, then it may suggest an underlying defect, thus triggering a repeat biopsy.
Other cancer types
As illustrated with breast and prostate cancers, telomere lengths have been extensively studied in most cancer types . In general, the majority of studies have compared tumour samples to either histologically normal, tumour-adjacent tissues or truly disease-free tissues. Although telomere alterations are found in the majority of cases, the direction of the alteration, either shortening or lengthening, appears tissue dependent and may vary within a particular tumour type. For example, a fraction of colorectal carcinomas have cancer cells with telomere lengths longer than adjacent normal cells and these cases tend to have a poor survival [75, 76]. Additionally, telomere alterations frequently occur in precursor lesions of most human epithelial cancers [77, 78]; therefore, telomere length analysis alone cannot differentiate between the presence of a precursor lesion or invasive cancer. Because most precursor lesions are not treated and will subsequently never progress to invasive carcinomas, the direct measurement of telomere length in tissue, or even in cytological preparations, is not a suitable molecular marker for the diagnosis of cancer. However, an area of investigation that may be promising is in situ telomere length analysis to identify patients that are good, or poor, candidates for a particular therapy. For example, recently developed telomerase inhibitors work most effectively on cells with short telomeres; therefore, assessment of the telomere lengths, along with telomerase activity, prior to treatment would be necessary to differentiate the patients most likely to benefit from the treatment .
Another avenue of investigation has concentrated on telomere length measurement in peripheral blood lymphocytes which can be easily obtained. Taking advantage of the high-throughput Q-PCR assay, numerous groups have assessed mean telomere length in peripheral blood as a possible marker for the risk of development of different cancer types, including lung, bladder, oesophagus, skin, head and neck and kidney; the field has been recently reviewed by Svenson and Roos . Since that comprehensive review, additional population-based investigations correlated shorter mean leucocyte telomere length with an increased risk of gastric cancer  and serous ovarian adenocarcinoma . Conversely, longer leucocyte telomere length was associated with an increased risk of developing non-Hodgkin lymphoma , whereas no association between telomere length and risk of incident colorectal carcinoma was found in two recent prospective studies [83, 84]. One particularly intriguing study performed by Willeit and colleagues, analysed mean leucocyte telomere length in 787 participants free of cancer at baseline and prospectively followed for 10 years . In this cohort, short telomeres were associated with subsequent cancer development independent of other cancer risk factors. Although associations could not be assessed between telomere length and each specific cancer type due to the relatively small number of cases, this inverse association is of particular interest and warrants further investigation. As with the tissue-based findings, the correlations with mean leucocyte telomere length and the risk of the development of cancer depends on the cancer type. Although, larger, prospective studies are needed, these preliminary findings suggest that telomere length analysis may provide some screening diagnostic benefit, most likely in conjunction with other molecular markers, to identify a subset of patients at risk for development of a particular cancer.