Main

Colorectal cancer (CRC) is the third most commonly diagnosed cancer in males and the second in females, with over 1.2 million new cases and 608 700 deaths estimated to have occurred worldwide in 2008 (Jemal et al, 2011). Despite improved treatments, increased awareness and early detection which have all contributed to prolonged survival, CRC is still an important cause of cancer-related deaths.

CRC is a heterogeneous complex of diseases with different molecular pathways and biological characteristics, arising through a multistep process of which several genetic and epigenetic events have been characterised (Markowitz and Bertagnolli 2009). Although surgical resection and adjuvant chemotherapy are effective curative treatments, the risk of recurrence cannot be foreseen, even in patients with tumours at the same stage. Although 5-fluorouracil-based adjuvant chemotherapy is the standard of care for patients with stage III CRCs, the role of adjuvant therapy for stage II CRCs is still controversial (Andrè et al, 2009). It is therefore critical to identify the subgroup of patients with stage II tumours at high risk of recurrence; however, this is still difficult owing to the molecular characteristics of these tumours as well as their intrinsic heterogeneity (Engstrom et al, 2009; O’Connor et al, 2011).

Many efforts have been made to identify molecular markers that predict the outcome of CRC patients, and several genetic and epigenetic alterations involved in the development of CRC have been proposed as prognostic markers of disease progression. Genetic instability has a critical role in the carcinogenetic process. Most CRCs have chromosomal instability (Pino and Chung, 2010), while other CRCs have a high-grade microsatellite instability (MSI) phenotype, generated by a deficient DNA mismatch repair pathway, frequently associated with the GpG island methylator phenotype (Boland and Goel, 2010). KRAS and TP53 gene mutations, involved in chromosome instability, and the MSI phenotype have been extensively studied for their potential prognostic and predictive roles. Although a predictive value of KRAS mutation status for response to EGFR-targeted therapy has been defined (Van Cutsem et al, 2009; Cunningham et al, 2010), its prognostic value is still unclear (Menin et al, 2006; Kim et al, 2007; Cunningham et al, 2010).

Telomere/telomerase interplay is an important mechanism involved in the maintenance of genetic chromosome stability and its dysfunction has emerged as having a causative role in carcinogenesis. Telomeres provide genomic stability by protecting chromosome ends. When the shortening of telomeres, which occurs at each cell division cycle because of end-replication problems of DNA polymerase, reaches a critical length, cells cease to proliferate and undergo senescence (Blackburn et al, 2006). Further erosion of telomeres may impair their function, resulting in genetic instability (Calado and Young, 2009; Artandi and DePinho 2010). Several studies demonstrated that telomeres were shorter in CRCs than in adjacent mucosa, suggesting that telomere shortening is a key initial event in colorectal carcinogenesis (Hastie et al, 1990; Takagi et al, 1999; Gertler et al, 2004; Garcia-Aranda et al, 2006; Rampazzo et al, 2010). Nevertheless, the maintenance of telomere length by telomerase is critical to preserving the replicative potential of cancer cells. Telomerase is a ribonucleoprotein complex containing a catalytic protein with telomere-specific reverse transcriptase (hTERT), which synthesizes telomeric sequences de novo utilising an internal RNA template (hTR). hTERT is the rate-limiting component of the telomerase complex (Nakamura et al, 1997) and its expression, usually absent from normal somatic cells, is essential for unlimited cell growth (Kelland, 2007). Notably, recent studies have suggested that, besides maintenance of telomere length, hTERT is involved in several other cellular functions. Expression of hTERT increases replicative potential (Rufer et al, 2001; Terrin et al, 2007), promotes cell growth in adverse conditions and may act as an anti-apoptotic agent (Del Bufalo et al, 2005; Rahaman et al, 2005).

Previous studies have shown that hTERT expression and/or telomerase activity is higher in CRCs than in adjacent non-cancerous mucosa (Gertler et al, 2002; Nowak et al, 2003; Garcia-Aranda et al, 2006; Terrin et al, 2008) and increases along with cancer progression (Maláska et al, 2004; Terrin et al, 2008; Rampazzo et al, 2010). However, the prognostic value of telomerase is controversial. Different results may be partly owing to the assessment of different end-points, disease-free survival (DFS) or overall survival (OS) and/or variations in the methodology employed to measure this marker, that is, telomerase activity or hTERT genetic expression (Tatsumoto et al, 2000; Kawanishi-Tabata et al, 2002; Gertler et al, 2004; Garcia-Aranda et al, 2006; Sanz-Casla et al, 2005; Vidaurreta et al, 2007; Safont et al, 2011). Of note, the only study, to the best of our knowledge, that addressed the prognostic role of telomerase in patients with stage II CRCs found that positivity for telomerase activity was associated with a better clinical outcome (Kawanishi-Tabata et al, 2002). In previous studies we developed a real-time PCR assay to quantitate hTERT mRNA, and have found that levels of hTERT expression correlated with telomerase activity and increased with tumour stage and histological grade (Terrin et al, 2008; Rampazzo et al, 2010).

The aim of this study was to elucidate the role of hTERT expression, quantified in tumour tissues by real-time PCR, as a prognostic marker of survival in CRC patients. In addition, we focused our study on verifying whether hTERT levels may be a prognostic marker of disease recurrence in stage II patients.

Materials and Methods

Patients

The study population consisted of 137 patients with a diagnosis of CRC who underwent surgery at a single Institution from 1995 to 2005. The surgical specimens, assessed in a standardised way, were reviewed by one pathologist (CM), following the American Joint Committee on Cancer TNM stage system, 2010. The initial inclusion criteria of patients were the availability of tumour tissues in which the neoplastic component was at least 80% of cells, the availability of adjacent non-cancerous mucosa and an adequate clinical follow-up (116 patients). Because of interest in defining the prognostic value of hTERT levels in stage II CRC patients, an additional group of 21 patients with stage II CRCs, for whom adequate tumour samples and clinical follow-up were available, was included in the study. Although the study was retrospective, the data were prospectively recorded in an institutional database. At surgery, each patient signed the informed consent approving that biological samples could be stored at our biological bank and used for research purposes.

The MSI status, performed by analysing five microsatellites of the Bethesda recommended panel (BAT-25, BAT-26, D2S123, D5S346, and D17S250) (Pucciarelli et al, 2003), was available for all tumours (Table 1).

Table 1 Clinical–pathological characteristics of 137 colorectal cancer patients
Full size table

Quantification of hTERT transcripts

RNA samples were extracted from cryostat sections of primary tumours using Trizol reagent (Invitrogen, Life Technologies, Carlsbad, CA, USA) and were reverse transcribed into cDNA using the SuperScript TM III RNase reverse transcriptase assay (Invitrogen) (Terrin et al, 2008). The expression of hTERT transcripts was quantified by real-time PCR using an ABI prism 7900 HT Sequence Detection System (Applied Biosystems, Life Technologies). Absolute quantification was carried out using five-fold dilutions of hTERT amplicon as a reference curve for hTERT copies, and five-fold dilutions of housekeeping hypoxanthine–guanine phosphoribosyl transferase 1 (HPRT1) amplicon for HPRT1 copies. Values of hTERT were then normalised for 103copies of HPRT1, as previously reported (Terrin et al, 2008).

Statistical analyses

OS was considered the end-point of the study and was defined as the interval between the date of surgery and the date of death or the last follow-up. DFS was defined as the interval between the date of surgery and the date of disease recurrence. Univariate and multivariate survival analysis was performed by means of the Cox regression model. The proportional hazard assumption was tested using the Grambsch-Therneau method (proportional hazard-test), which is based on Schoenfeld residuals (Therneau and Grambsch, 2000). The Breslow method for handling ties was used while running the Cox model. For univariate analysis, the following prognostic factors were considered: age (years), site of tumours (colon vs rectum), MSI vs microsatellite stability (MSS), pTNM stage (I, II, III, IV) and hTERT levels. hTERT levels were categorised as high (above median value) and low (below median value). The fractional polynomial method was employed to study the functional form of continuous covariates and to relax the linearity assumption of the Cox regression model. The selection of the covariates that significantly and independently contribute to the prognostic model was performed using the stepwise mode. As a measure of overall prognostic accuracy of the final model we calculated the concordance index, which estimates the probability of concordance between predicted and observed events and ranges between 0.5 (no prognostic accuracy) and 1 (perfect prognostic accuracy) (Harrell et al, 1996). OS and DFS empirical curves were estimated with the Kaplan–Meier method and the survival functions were compared using the log-rank test. The median follow-up was calculated according to the reverse Kaplan–Meier method. All tests are two-sided, the α-level of significance being set to 5%. All statistical analyses were performed using Stata 11 SE (StataCorp, College Station, TX, USA).

Results

Characteristics of patients and relationship with hTERT expression

Clinical–pathological characteristics of the 137 CRC patients included in the study are reported in Table 1. Seventy-six patients were males and 61 were females; the median age was 67 years (range, 30–86 years). Tumours were located in the colon in 105 cases (77%) and in the rectum in 32 cases (23%). According to the American Joint Committee on Cancer staging, 33 tumours were classified as stage I, 50 as stage II, 27 as stage III, and 27 as stage IV. Of the 122 patients for whom data were available, 66 received adjuvant therapy. One hundred and twelve CRCs (82%) had stable microsatellites, whereas 25 CRCs (18%) had MSI. Quantification of hTERT transcripts was obtained for all 137 CRC samples. The median hTERT level was 93.8 (interquartile range 48.0–254.4) copies. In accordance with previous studies (Terrin et al, 2008; Rampazzo et al, 2010), hTERT levels increased with severity of disease. Indeed, median hTERT levels were 52.4 (32.4–121.1), 85.4 (52.8–140.0), 114.52 (65.6–268.1) and 469.2 (210.6–2099.8) copies in tumours stage I, II, III, and IV, respectively (overall, P<0.0001). Furthermore, in agreement with a previous study (Rampazzo et al, 2010), no correlation was found between hTERT levels and MSI or tumour location. Median hTERT levels were 92.3 (62.8–140.0) and 94.3 (44.3–275.0) copies in MSI and MSS CRC, respectively (P=0.68), and 106.1 (54.3–275.1) and 80.3 (36.2–140.1) copies in colon and rectal CRC, respectively (P=0.144).

Relationship of hTERT level with OS of patients

With a median follow-up of 70 months (interquartile range 57–122), the 5-year survival rate was 55.4% (95% confidence interval (CI) 46–64). With univariate analysis, TNM stage and hTERT levels were significantly associated with survival (Table 2). The survival curves according to TNM stage are shown in Figure 1A. With regard to hTERT, the study of its functional form suggested that dichotomization by the median value (93.8 copies) would yield a better fit of the survival model; considering this cutoff, patients with high hTERT levels (above the median) showed a significantly worse OS (median 37 months) than those with lower hTERT levels (below the median; (median OS not reached; log-rank test P<0.0001) (Figure 1B). This difference corresponded to a hazard ratio (HR)=3.30 (95% CI 1.98–5.52), P<0.0001.

Table 2 Univariate and multivariate survival analysis (Cox regression model).
Full size table
Figure 1
figure 1

(A) Kaplan–Meier OS curves of patients according to the pTNM stages. (B) Kaplan–Meier OS curves of patients according to high (above median value) or low (below median value) hTERT levels.

Full size image

With multivariate analysis, age, TNM stage, and hTERT were retained in the final regression model (Table 2). The prognostic value of hTERT was maintained even after adjustment for the other covariates (age, site, MSI, pTNM stage), supporting the independent role of this biomarker in predicting the OS of these patients. In particular, the risk of death for patients with high hTERT levels was approximately double that of patients with low levels (HR=2.09 (95% CI 1.20–3.64), P=0.009). The overall prognostic accuracy of the final model was remarkable, as suggested by the high concordance index value (0.78).

Considering the subgroup of 122 patients for whom the chemotherapy data were available, multivariate analysis showed that hTERT level, but not chemotherapy, was significantly associated with OS (HR=2.40 (95%1.14–3.76), P=0.017). The distribution of patients who received chemotherapy did not significantly differ between patients with high or low hTERT levels (Fisher's exact test P=0.102).

Relationship of hTERT level with disease-free and OS of stage II patients

In order to evaluate the role of hTERT as a prognostic factor in patients with stage II CRC, we tested the hypothesis that hTERT might identify patients at high risk in this subgroup. As shown in Figure 2A, stage II patients with high hTERT levels showed a significantly worse median OS (55 months) than those with low hTERT levels (median OS not reached; log-rank test P=0.0048; HR=3.24 (95% CI 1.37–7.71), P=0.008). The median OS of patients with stage II disease and high hTERT levels was very similar to that of patients with stage III disease (55 and 59 months, respectively). Furthermore, as shown in Figure 2B, high hTERT levels might identify patients with higher risk of disease recurrence; indeed, patients with high hTERT levels showed a significantly worse median DFS (46 months) than those with low hTERT levels (median DFS not reached; log-rank test P=0.03; HR=3.06 (95% CI 1.03–9.04), P=0.043).

Figure 2
figure 2

(A) Kaplan–Meier OS curves of stage II patients according to high or low hTERT levels. (B) Kaplan–Meier DFS curves of stage II patients according to high or low hTERT level.

Full size image

Relationship of MSI with disease-free and OS of patients

Of note, MSI status did not correlate with OS in our cohort of patients (MSI vs MSS): HR=1.08 (95% 0.59–1.98), P=0.802). It has been suggested that MSI defines a subset of patients with stage II tumours with a good prognosis who do not require adjuvant 5-fluorouracil-based chemotherapy (Sargent et al, 2010). In the stage II patients we studied, 11 had tumours with MSI and 39 with MSS, 15 of whom (4 MSI and 11 MSS) received adjuvant chemotherapy. In this subset of stage II patients neither MSI nor chemotherapy had any significant impact on prognosis in terms of OS (log-rank test P=0.231 and P=0.126, respectively) and DFS (log-rank test P=0.938 and P=0.502, respectively).

Discussion

Pathological tumour staging remains a key determinant of CRC prognosis and treatment. Although radical resection is the primary treatment for patients with loco–regional CRC, adjuvant chemotherapy provides additional survival benefits for patients with stage III tumours. The same benefit has not been conclusively demonstrated in patients with stage II tumours (O’Connor et al, (2011). The controversial results obtained from different studies (Ribic et al, 2003; Benson et al, 2004; Gill et al, 2004; Quasar Collaborative Group, 2007; Sargent et al, 2010) may reflect the molecular heterogeneity of CRCs and highlight the need for strong prognostic markers able to stratify patients.

In the present study, we found that a high level of hTERT is a prognostic marker of shorter OS and its negative prognostic value is independent of pathological stage. In addition, in stage II patients, a high hTERT level identified patients at higher risk of disease recurrence and death.

Several studies have addressed the prognostic value of telomerase expression in CRC, but results are quite controversial. This might be due to the different end-points assessed in these studies, DFS or OS, and the methodology employed to measure the level of hTERT expression or telomerase activity. Most of the studies utilised a semi-quantitative Telomere Repeat Amplification Protocol assay to evaluate telomerase activity (Tatsumoto et al, 2000; Kawanishi et al, 2002; Gertler et al, 2004; Garcia-Aranda et al, 2006; Sanz-Casla et al, 2005; Vidaurreta et al, 2007). Only a few studies, using different reference curves and housekeeping genes, employed real-time PCR to evaluate hTERT mRNA values (Gertler et al, 2002; Saleh and Lam 2008; Safont et al, 2011). Furthermore, only one study addressed the prognostic role of telomerase activity in patients with stage II CRCs. Positivity for telomerase activity, estimated by the Telomere Repeat Amplification Protocol assay, was associated with a better prognosis and patients with telomerase-positive CRCs had longer DFS than patients with telomerase-negative tumours (Kawanishi-Tabata et al, 2002).

In the present study, the values of hTERT were normalised for expression of the HPRT1 gene, which did not vary according to tumour stage (de Kok et al, 2005), thus allowing a more reliable estimate of hTERT levels in CRCs. Notably, hTERT levels estimated by this approach have been found to be related to telomerase activity in CRCs (Terrin et al, 2008).

Of interest, in our study, we found that hTERT levels significantly stratified stage II patients, regardless of their MSI status. MSI has been reported to be associated with improved DFS compared with MSS; however, no difference in OS was observed between MSI and MSS tumours, and MSI was not predictive of chemotherapy benefit (Kim et al, 2007). Moreover, it has been recently reported that defective DNA mismatch repair, measured by MSI or by loss of MLH1 and MSH2 proteins, is a prognostic marker of DFS; however, its prognostic effect was not maintained in multivariate models (Sargent et al, 2010). In agreement with these studies, our findings do not support the prognostic value of MSI. Indeed, OS did not significantly differ in patients with MSI and MSS tumours. It is of note that in a previous study we found that MSI tumours had shorter telomeres than MSS CRCs, but hTERT expression did not differ (Rampazzo et al, 2010). The present study, in agreement with these findings, did not find a relationship between MSI status and hTERT levels. Notably, the TP53 gene is known to be frequently altered in MSS CRCs. As the p53 is a well-known negative regulator of the hTERT promoter (reviewed in Dolcetti and De Rossi, 2012), mutated p53 may result in an earlier activation of hTERT, with an earlier stabilisation of telomeres, rather than higher levels of hTERT.

Previous studies, indicating that telomerase activity and/or hTERT expression increased along with tumour progression (Maláska et al, 2004; Terrin et al, 2008; Rampazzo et al, 2010; Kojima et al, 2011), have suggested that higher levels of hTERT may compensate for the greater shortening of telomeres due to higher proliferative activity. However, several studies, including ours, did not find a significant relationship between telomere length and telomerase levels and/or activity (Frías and Morán 2009; Rampazzo et al, 2010). Several studies suggested that hTERT might increase the malignant potential of tumours beyond just preservation of telomere length. hTERT might act as a growth-promoting and anti-apoptotic factor, independent of its telomere-elongating activity (Cao et al, 2002; Del Bufalo et al, 2005; Rahman et al, 2005; Massard et al, 2006; Folini et al, 2007; Jin et al, 2010). Our findings that hTERT levels have a prognostic value, regardless of chemotherapy, support the previous in vitro observation that hTERT expression inhibited p53-dependent apoptosis in response to 5-fluorouracil in HCT116 colon carcinoma cells (Rahman et al, 2005). Specific studies are required to investigate the predictive value of hTERT levels in response to specific chemotherapies.

In conclusion, our study strongly indicates that hTERT level is an independent prognostic indicator of survival in patients with CRC. In addition, measurement of hTERT could improve the stratification of stage II patients by the risk of disease recurrence. These effects are likely associated with several functions of hTERT, in addition to its ability to maintain telomere length. Understanding the functions of hTERT is important to deepening the knowledge of CRC pathogenesis and, ultimately, to designing new therapeutic strategies, including hTERT inhibitors.