Based on the available data, the present study has provided insights into the relationship between PBL telomere length and the CRC risk. Due to small number of studies the results from the current analysis is insufficient to provide evidence on the role of PBL telomere length and CRC risk.
Telomere length in CRC
Findings in this review showed there is no significant relationship between telomere length measured in PBL DNA and the CRC risk in both retrospective and prospective studies. There are many possible reasons for such relationship. The risk of CRC might be elevated by shorter or longer length of telomere, indicating a U-shape association [
25]. It is possible that loss of telomeric DNA in relation to degradation or incomplete replication is apparently balanced by telomere elongation [
1,
42,
43]. This also implied that telomere length in an ‘appropriate range’ may be necessary to maintain chromosomal stability and normal programmed cell death - functions which are protective against tumour development [
25,
44]. It has been hypothesized that cells with long telomeres may favour a delayed cell senescence and apoptosis, leading to an increased chance of various genetic and environmental insults and subsequent accumulations of genetic abnormalities attributed to a higher risk of carcinogenic transformation [
25].
On the other hand, the relationship between short telomere length and the CRC risk is biologically plausible [
28]. A large proportion of human cancers are made up of cells with very short telomeres (5 kb), which is attributable to telomere dysfunction [
45]. Although the exact mechanism is not fully understood, it is possible that, in some situations, cells with critically short telomere length may reactivate the telomerase enzyme, and this further promotes malignant transformation [
28]. The putative pathway for telomere-associated neoplasia is that shortening of telomeres in the colorectal mucosa increases the chromosomal susceptibility to instability [
40] as well as the microsatellite instability [
47].
Reasons for no significant associations in our findings also rest on discrepancy of the study participants such as gender, age, life styles and the study design related factors (e.g. sample size, prospective or retrospective study). For instance, an association of telomere length and the CRC development in age-depending manner was shown in an individual study included in this analysis [
28]. Older patients with shorter telomeres may be prone to an increased risk for various types of cancers due to telomere crisis [
28] and subsequent chromosomal aberrations [
25] and the diseases that result [
25]. Extremely long telomere in younger individuals may be indicative of dysregulation in telomere regulation process [
28].
We found a change in the effect estimates in a subgroup analysis solely with females. This highlighted that hormonal influence (estrogen) on the length of telomere needed to be given due attention in interpreting the results. Studies documented that the rate of terminal restriction fragment length shortening per year in men was significantly greater than that in women [
45]. An estrogen-responsive element is present in telomerase reverse transcriptase (hTERT), so the hormone can stimulate telomerase [
46]. This might have reduced the shortening of telomere. Another possible mechanism is in the light of reactive oxygen species (ROS), which are number of reactive molecules and free radicals derived from molecular oxygen. As such, hydrogen peroxide and other ROS accelerate telomere erosion at least in cultured cells. Women are known to produce fewer reactive oxygen species [
42] and this might contribute to less telomere attrition.
Most of the primary studies in this review have estimated an individual’s telomere length from a single blood collection. In a study, multiple measurements showed a good reliability (intraclass correlation coefficient 0.64) indicating a one-time measurement is a good representation of an individual’s telomere length within a short period of time [
48]. Hence, a one-time determination in the primary studies would not affect the confidence in the effect estimate of the current review.
Study limitations and strengths
There are some limitations that need to be acknowledged. Although we had done an exhaustive literature search, it was possible that some publications might have been overlooked, especially when abstracts in English were not available. Variation in sample collection time in relation to study design among included studies is a concern. Samples in the prospective study were collected (many) years before diagnosis and therefore changes in participant’s diet and lifestyle related factors and drug history (e.g. treatment with estrogen in women) overtime could have influenced telomere length, or conversely telomere length could be influenced by diagnosis and treatment in the retrospectively designed study [
24,
27].
Variation in measurement methods of telomere lengths in the primary studies could have affected the effect estimates. This was, however, not the case in the present review as all studies applied qPCR for the measurement of telomere length. Stage-specific relationship could not be analysed as staging of disease with Dukes’ or TNM systems had been provided inconsistently in a few studies. This might lead to a misclassification bias in interpreting the results.
Since relevant data were limited, we could not adjust for confounding factors. For instance, age of patients was likely to be a confounding factor for the association between PBL telomere length and the CRC risk. It is difficult to render the pooled analysis stratified by age groups, which were not consistently classified among these studies. Individual studies [
45,
49] as well as a systematic review had suggested an interaction between aspirin or non-steroidal anti-inflammatory drug use [
49,
50] and genes encoding transcription factor [
49] or a combined effect of life style with gene encoding [
27] in oncogenesis, inflammatory and drug metabolic pathways in relation to risk of CRC [
50]. Hence, our results might be influenced by these confounding factors.
Due to a small number of studies with limited sample sizes, there is a limited power to detect a true difference of telomere length between cases and controls. It calls for future well designed, large prospective studies in this field.