One hundred and ninety-eight newly diagnosed and histologically confirmed patients diagnosed with CRC (median age 66 years; range 32–88 years; 127 men; 71 women) were enrolled in the study. Peripheral blood samples were collected from all patients at Thomayer University Hospital in Prague, the Czech Republic, from June 2008 to May 2018. In total, 478 samples were collected from 198 patients. The characteristics of the study group are shown in Table 1 . Blood specimens were drawn from patients at the baseline before any treatment (n = 137), followed by collection at the sixth (n = 156), twelfth (n = 135), and eighteenth (n = 50) month after diagnosis. The 6-month sampling encompassed, for the most part, patients (56%) for whom the anticancer therapy had been completed. An age-matched control group was not included, based on our previous results confirming that the lymphocyte telomere length of cancer patients is age-independent, in contrast to healthy controls (14).

In our study, we analyzed the effect of both neoadjuvant and adjuvant therapies on LTL. In general, for CRC, the type of treatment varies according to the tumor localization. The treatment strategy for colon cancer patients typically involves surgical removal, followed by adjuvant chemotherapy based on 5-fluorouracil (often in combination with oxaliplatin, irinotecan, or either of the above drugs with added folic acid (15). In contrast, rectal cancer treatment is based on neoadjuvant (5-fluorouracil-based chemo)radiotherapy to downstage the tumor before surgical removal. Patients with locally advanced tumors may sometimes receive adjuvant chemotherapy (based on drugs described above for colon cancer treatment). In the study, we also stratified patients based on their response to the therapy. Good therapy responders were defined as patients who did not require a switch of chemotherapy regimen, achieved remission after a prescribed number of cycles, and did not experience any chemotherapy side effects.

The cohort analyzed for the impact of 5-fluorouracil-based chemotherapy as standard care for CRC in the adjuvant setting comprised 42 individuals with a complete record of a given treatment. To minimize the influence of other treatment strategies, only patients without previous neoadjuvant treatment were enrolled. The effect of neoadjuvant therapy was investigated in 23 patients with rectal cancer with complete treatment records. To eliminate the possible influence of other administered treatments, the cohort included only patients who did not receive postoperative therapy. Neoadjuvant therapy for the patients involved (5-fluorouracil-based chemo)radiotherapy or chemotherapy with 5-fluorouracil used alone. The therapies prescribed for each patient are presented in Supplementary Table 1 .

All participants provided written informed consent to participate in this study. Peripheral blood sampling was performed in accordance with the ethical standards outlined in the Declaration of Helsinki. The clinicopathological characteristics of all patients were collected from their medical records. Personal data were acquired using lifestyle questionnaires. Only collaborating clinicians could disclose participants’ identities since they entered the anonymized study.

Blood was collected into Vacuette K3EDTA blood collection tubes by venipuncture. After sampling, fresh specimens were processed to produce mononuclear cells. These peripheral mononuclear cells were isolated from 2 ml of whole blood by Ficoll-Paque PLUS (GE Healthcare Life Sciences) flotation, as recommended in the manufacturer’s handbook, and frozen in TRI Reagent (Sigma-Aldrich). The remaining whole blood was stored at -80°C.

Genomic DNA from whole blood (i.e., leukocytes) and total RNA from mononuclear cells were used as starting materials for qPCR-based analyses. DNA was extracted from 200 µL of blood according to the DNeasy Blood and Tissue Kit protocol (Qiagen, Valencia, CA). RNA was extracted from Ficoll-isolated mononuclear cells using the standard TRI Reagent protocol.

DNA and RNA concentrations were determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). RNA quality (RIN ≥ 8) was assessed using a Bioanalyzer 2100 (Agilent, Santa Clara, CA, USA) and the Agilent RNA 6000 Nano kit.

Relative LTL was estimated using singleplex qPCR based on Cawthon (16) and later optimized by Gil and Coetzer (17). The method was adopted in collaboration with the Department of Biology and Medical Genetics, First Faculty of Medicine, Charles University, Prague, which has published articles on this topic (18, 19).

Briefly, LTL measurements were performed using the ViiA 7 Real-Time PCR System (Applied Biosystems, Foster City, CA, USA). qPCR reactions were run on a MicroAmp Fast Optical 96-Well Reaction Plate (Life Technologies, Carlsbad, CA, USA) in triplicate. The 24 µL reactions consisted of PowerUp SYBR Green Master Mix (Life Technologies, Carlsbad, CA, USA), primers (300 nM for 36B4, 900 nM for telomeres), and 5 ng of DNA and were prepared according to the manufacturer’s instructions. To detect the reference 36B4 gene, which encodes the acidic ribosomal phosphoprotein P0, and telomere sequences, primers (Sigma-Aldrich, St. Louis, MO, USA) with the same annealing temperature were used. Primer sequences are listed in Supplementary Table 2 . The reaction conditions were identical for both targets. The holding stage was performed at 50°C for 2 min and 95°C for 2 min, followed by 40 cycles at 95 °C for 30 s and 54°C for 1 min (ramp rate: 1.6°C/s). Primers’ specificity was assessed by melting curve analysis ( Supplementary Figure 1 ) at the end of the amplification procedure. A calibration curve and interplate calibration were established using a 17 ng/µL reference DNA sample pooled from 30 healthy middle-aged hospital volunteers: 15 females and 15 males ranging from 41 to 55 years of age with no previous experience with cancer (20). Participants were recruited at the Faculty Hospital Kralovské Vinohrady in Prague, Czech Republic. Amplification efficiency was verified using a 5-point calibration curve created through a 2-fold serial dilution of the reference sample with three replicates at each point. The DNA amount ranged between 17 ng to 1.0625 ng per well. The reaction efficiency was calculated from the slope of the standard curves (see Supplementary Figure 2 ) using the equation: E = -1 + 10^(-1/slope) and evaluated using Viia7 software. The efficiency was 97% for telomeres and 101% for 36B4. The interplate and intraplate coefficients of variation were 1.03% and 0.35% for 36B4 and 1.92% and 0.24% for telomeres, respectively. Samples assayed in triplicate with the standard deviation (SD) > 0.3 were omitted from the analysis, and their measurement was performed again. When two replicates were nearly identical, and one was distant with a suspected dilution error, the outlier was omitted from the calculations.

LTL was estimated as the ratio of telomere repeat (T) copy number to single gene (S) copy number (T/S ratio) and calculated by the ΔΔCt method (21). The mean cycle threshold (Ct) of 36B4 was subtracted from the mean Ct of telomeres. The calculated ΔCt of each sample and ΔCt of the standard (derived from the ΔCts of standard replicates within the calibration curve) were substituted into the formula 2^{-(ΔCt(sample)-ΔCt(standard)}. The obtained T/S ratio is proportional to the average TL in the cell. A ratio above 1 indicated that the patient had longer telomeres than the reference sample.

TERT mRNA expression levels were measured by two-step singleplex quantitative reverse-transcription PCR (RT-qPCR) using the TaqMan chemistry.

Complementary DNA (cDNA) was synthesized using 300 ng of total RNA with a High Capacity cDNA Reverse Transcription Kit (Life Technologies, Carlsbad, CA, USA) in 15 μL reactions. All components were combined and incubated in an MJ Research PTC-200 Thermal Cycler (Bio-Rad), following the manufacturer’s instructions. Subsequent qPCR reactions were carried out in a 384-well format using a MicroAmp Optical 384-Well Reaction plate (Life Technologies, Carlsbad, CA, USA) on the Viia7 Real-Time PCR System (Applied Biosystems). All reactions were performed in triplicates. To normalize TERT expression levels, endogenous control genes glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and beta-actin (ACTB), which are known to be stably expressed in mononuclear cells (22), were used. All 20 μl qPCR reactions included TaqMan Universal Master Mix II, with UNG (Applied Biosystems, Carlsbad, CA, USA), particular FAM/MGB TaqMan Gene Expression Assay (Hs00972650_m1 (TERT), Hs01060665_g1 (ACTB), or Hs02786624_g1 (GAPDH)), RNase-free water, and 2 μl of 3:1 diluted cDNA. Thermocycling conditions were 50°C for 2 min and 95°C for 10 min, followed by 40 cycles at 95°C for 15 s and 60°C for 1 min. The fluorescence signal was captured at the end of the 60 °C segments. Using the peripheral blood mononuclear cell RNA from a healthy donor, the interplate and intraplate coefficients of variation were 2.21% and 0.45% for TERT, 0.70% and 0.42% for GAPDH, and 0.40% and 0.49% for ACTB. Technical replicates with the SD of the average Ct > 0.3 were omitted from the study, and samples were measured in a new run.

The relative expression of TERT was calculated analogically to the previous description (see Section 2.3) using the mean Ct of the two housekeeping genes as a reference.

Statistical analysis was performed using STATISTICA (version 11Cz; TIBCO Software Inc., Palo Alto, CA, USA) and MATLAB (version 2019b; The MathWorks, Inc., Natick, MA, USA). All reported p-values were two-tailed and the level of statistical significance was set at α = 0.05.

Owing to their mostly normal distribution, LTL and log₂-transformed TERT expression values (i.e., -ΔΔCt values) were compared between relevant group pairs (according to sex, smoking habit, microsatellite instability, Union for International Cancer Control (UICC) tumor-node-metastasis (TNM) classification, and treatment response) using a two-sample t-test. The Mann-Whitney U test was used to compare the LTL of rectal cancer patients measured before and after neoadjuvant therapy (in an independent arrangement because of a small overlap between the patient groups). In the samples collected at the time of the diagnosis, an ordinal association between LTL and TERT expression was analyzed using the Kendall rank correlation coefficient. Kendall’s τ rank test was also performed to investigate the relationship between age and LTL or TERT expression and the correlation between LTL and time from adjuvant therapy termination. One-way analysis of variance (ANOVA) was used to examine the differences in LTL and TERT expression in patients diagnosed with proximal colon, distal colon, and rectal cancer. To examine LTL changes over time, these diagnosis-related groups were further investigated using a repeated-measures ANOVA. Determination of whether LTL values of all patients changed over time was also performed using repeated-measures ANOVA. Lastly, repeated-measures ANOVA was used to compare LTL in patients with different TNM stages and treatment responses over time. Kruskal-Wallis ANOVA was used to compare the mean LTL values in time intervals from adjuvant therapy. Subsequently, to verify the time-related LTL changes in the context of other variables and account for possible confounding, while also allowing for patients with incomplete data to remain in the analysis (as opposed to repeated-measures ANOVA), a linear mixed-effects model was fit in the data, including the sampling time as a continuous predictor, TNM category, primary tumor location (colon/rectum), radiotherapy (yes/no), and application of 5-fluorouracil (yes/no) as fixed categorical factors, and the patient ID as a random factor.

Overall survival (defined from diagnosis to death or censored at the last follow-up) and its association with LTL were analyzed using the Kaplan-Meier method, Cox proportional-hazards model, and after stratification by LTL median, the Gehan-Wilcoxon test. The median follow-up time was 26.1 months (determined using the inverse Kaplan-Meier method).

The first three samplings (collected at diagnosis and 6 and 12 months after diagnosis) were available for 49 patients. Specifically, we monitored the overall LTL dynamics and dynamics across the patient subgroups. Monitoring of these 49 patients identified LTL shortening during the treatment and follow-up examinations. In the whole group, there was a significant downward trend across the first three 6-month observation periods independent of age, cancer stage, tumor localization, or treatment (repeated measures ANOVA, F(2, 96) = 16.93, n = 49, p < 0.0001). The Bonferroni post hoc test revealed that the most pronounced decrease in LTL was between 6 and 12 months after the diagnosis (Bonferroni corrected p < 0.0001, Figure 1 ). Furthermore, patients analyzed in the subgroups categorized according to their tumor location showed different trajectories of LTL decreases. (repeated measures ANOVA F(4, 90) = 2.83, p = 0.03, proximal colon n = 11, distal colon n = 24, rectum n = 14, Figure 2 ). Subsequent analysis with a linear mixed-effects model confirmed the downward time trend of LTL also in the whole patient cohort (average decrease of 1.05% of the initial value per month with a 95% CI of 0.64–1.47%, p < 0.0001), independent of TNM stage, primary tumor location, radiotherapy, or 5-fluorouracil application. The dynamics of LTL in time can also be seen in the differences in LTL values between the measured time points calculated for each patient and summarized in Table 2 . Concerning therapy response and cancer staging, we did not observe a difference in the rate of LTL decrease between good (n = 19) and poor responders (n = 4) (repeated-measures ANOVA, F(2, 42) = 1.30, p = 0.28), or between TNM I + II (n = 33) and III + IV (n = 13) (repeated-measures ANOVA, F(2, 88) = 0.52, p = 0.60). We also performed sampling 18 months after the diagnosis, but the four complete samples were drawn from 6 patients only.

We observed no correlation between LTL and age at diagnosis (Kendall’s τ = 0.03, p = 0.63, n = 137), gender (females: 0.86 ± 0.35, n = 51; males: 0.86 ± 0.35, n = 86; two-sample t-test t(135) = -0.01, p = 0.99), or smoking status (smokers: 0.88 ± 0.34, n = 32; non-smokers: 0.87 ± 0.35, n = 98; two-sample t-test t(128) = -0.13, p = 0.90). There was no difference in LTL between patients with microsatellite stability (0.86 ± 0.34, n = 94) and instability (0.82 ± 0.29, n = 22; two-sample t-test t(114) = 0.48, p = 0.63). Patients diagnosed with TNM stage I + II (0.84 ± 0.35, n = 82) and TNM stage III + IV (0.89 ± 0.34, n = 47) did not show statistically significant differences in LTL (two-sample t-test t(127) = -0.693, p = 0.49). Likewise, there was no difference in LTL in patients with tumors in the proximal colon, distal colon, and rectal locations (one-way ANOVA (F(2, 132) = 0.28, p = 0.76). The complete LTL results are listed in Table 3 .

In incident CRC cases, comparison of LTL between good (0.94 ± 0.34, n = 41) and poor (0.91 ± 0.30, n = 20) therapy responders revealed no difference (two-sample t-test t(59) = 0.34, p = 0.74). With regard to adjuvant therapy, LTL was inversely correlated with the time elapsed from the therapy completion (Kendall’s τ = -0.22, p = 0.03), but the correlation was computed only by using n = 49 data points. Further, LTL was significantly non-uniform across the time intervals defined with respect to the treatment (Kruskal-Wallis ANOVA H(5, 96) = 12.74, p = 0.03, Figure 3 ), although the post hoc test did not confirm the differences between any of the time point pairs. Additionally, within the period < 90 days after the end of adjuvant therapy, we found that increase in LTL showed a borderline association with an increased overall survival (Cox proportional-hazards model, hazard ratio (HR) = 0.01, 95% confidence interval (CI) 0.00–0.01, p = 0.05, n = 16). However, stratification based on median LTL (cut-off value 0.94), in contrast, did not show that LTL positively affected the overall survival of the patients (Gehan-Wilcoxon test, p = 0.06). When focusing on neoadjuvant therapy in rectal cancer cases only, LTL of the patients measured before (0.73 ± 0.28, n = 7) and after the therapy administration (0.82 ± 0.36, n = 32) did not differ (Mann-Whitney U = 94, p = 0.53). We also did not find any correlation with time after neoadjuvant therapy (Kendall’s τ = -0.06, p = 0.65). The results are summarized in Table 4 .

The 3-year overall survival rate of CRC patients was 83.3% (95% CI 76.2–90.4%). Investigating overall survival according to telomere length measured in newly diagnosed patients suggested that LTL had no relation to the survival rate (Cox proportional-hazards model, HR = 1.36, 95% CI 0.33–5.59, p = 0.67, n = 137).

We also measured TERT expression in mononuclear cells at the time of diagnosis. TERT expression did not correlate with LTL (Kendall’s τ = 0.00, p = 1, n = 104). TERT expression was higher in females (n = 43) than males (n = 66) (two-sample t-test t(107) = -3.01, 1.35-fold, 95% CI 1.11–1.65, p = 0.003, Supplementary Figure 3 ) and smokers (n = 29) than non-smokers (n = 74) (two-sample t-test t(101) = -2.11, 1.27-fold, 95% CI 1.01–1.61, p = 0.04, Supplementary Figure 4 ). No correlation was observed between TERT expression and patient age (Kendall’s τ = -0.08, p = 0.23, n = 109). Microsatellite unstable patients (n = 15) showed only negligibly lower TERT expression than microsatellite stable patients (n = 75) (0.91-fold, 95% CI 0.67–1.23, two-sample t-test t(88) = 0.66, p = 0.51), as did TNM III + IV (n = 37) in comparison with TNM I + II (n = 66) (0.96-fold, 95% CI 0.79–1.17, two-sample t-test t(101) = 0.37, p = 0.72). When compared to patients with the tumors in the distal colon (n = 53), decrease in TERT expression was observed for the localization in proximal colon (n = 24, 1.12-fold, 95% CI 0.85–1.47) as well as in rectum (n = 31, 1.11-fold, 95% CI 0.86–1.44), however without statistically significant (one-way ANOVA F(2, 105) = 0.55, p = 0.58). Similarly, TERT was insignificantly decreased in poor therapy responders (n = 18) compared to good responders (0.87-fold, 95% CI 0.65–1.15, two-sample t-test t(45) = 0.96, p = 0.34).