Association of relative telomere length with progression of chronic kidney disease in two cohorts: effect modification by smoking and diabetes

Chronic kidney disease (CKD) is a highly progressive disease. We studied the association between relative telomere length (RTL) and CKD progression and tested whether this association is modified by smoking and diabetes mellitus. RTL was measured by qPCR in two prospective cohort studies, the MMKD-Study (n = 166) and the CRISIS-Study (n = 889) with a median follow-up of 4.5 and 2.8 years, respectively. Progression was defined as doubling of baseline serum creatinine (MMKD-Study) and/or end stage renal disease (both studies). 59 and 105 of the patients from MMKD and CRISIS experienced a progression of CKD. Mean standardized pooled RTL was 0.74 ± 0.29. In the meta-analysis shorter RTL at baseline showed a borderline association with CKD progression (HR = 1.07 [95%CI 1.00–1.15]; p = 0.06). We observed an effect modification of RTL and CKD progression by smoking and diabetes (p-values of interaction p = 0.02 and p = 0.09, respectively). Each 0.1 unit shorter RTL was significantly associated with an increased hazard for CKD progression in active-smokers by 44% (HR = 1.44 [1.16–1.81]; p = 0.001) and in patients with diabetes mellitus by 16% (HR = 1.16 [1.01–1.34]; p = 0.03). Estimates were adjusted for baseline age, sex, proteinuria and GFR. This study in two independent cohorts reinforces that RTL is a marker and potentially a pathogenetic factor for CKD progression.


Mild to Moderate Kidney Disease Study (MMKD Study)
Exclusion criteria were defined as: serum creatinine >6 mg/dL (531 µmol/L), diabetes mellitus of any type, malignancy, liver, thyroid, or infectious diseases, nephrotic syndrome (defined as daily proteinuria >3.5g/1.73m 2 ), organ transplantation, immunosuppressive treatment, allergy to ionic contrast media, treatment with fish oil or erythropoietin and pregnancy.
All patients were recruited by a single investigator who visited all participating centers to avoid interobserver differences. Information on age, gender, smoking habits, comorbidities and antihypertensive treatment at baseline was recorded by patient interview and confirmed by checking patient records. A clinical examination completed the procedure. Hypertension was defined as BP >140/90 mmHg and/or the use of antihypertensive medication.
Patients received regular reviews in the outpatient ward. Endpoints were assessed by medical record abstraction and reported to the coordinating center.
Unless they had reached the primary endpoint, patients were followed until the end of the whole follow-up time period of 7 years. A total of 177 (78%) patients from the baseline cohort of 227 patients could be followed prospectively. Due to inappropriate or missing laboratory samples (n=11) or missing of follow-up data (n=50), the values of RTL were available in 166 patients.
A total of 50 patients were lost to follow-up. They moved away or were not referred by their physicians for follow-up visits in the renal units. Significantly better renal function at baseline was observed when compared to patients who were not lost to follow-up (i.e., a higher mean GFR [91±44 versus 64±39 ml/min per 1.73 m 2 ; P <0.01]). Both groups, however, did not differ significantly with respect to age and gender.

Chronic Renal Insufficiency Standards Implementation (CRISIS Study)
Adult patients referred for management of renal disease at Salford Royal NHS Foundation Samples were normalized in 96-well microtiter plates. Samples were ascertained in a singleplex, quadruplicate approach to measure the T/S-ratios. These T/S-ratios are proportional to individual relative telomere length (RTL). RTL was measured with some modifications as described below by using a quantitative real-time polymerase chain reaction (qPCR) assay, which was first described by Each qPCR was carried out in 384-well format which was vertically segmented in two parts: one for the telomeres (T) and one for the housekeeping gene 36B4 (S). Each 384-well plate contained the standard DNA, a quality control (commercially available DNA-Human Genomic DNA, Roche) and a non template control (NTC) in quadruplicate. All sample transfers and dilution steps were performed with a Tecan robotic workstation. Relative qPCR was carried out on an Applied Biosystems Taqman Fast Real-Time PCR 7900HT System.
The thermal cycling began with the initial polymerase activation step (10 min at 95°C) and was followed by 40 cycles of 95°C for 15 s, 60°C for 1 min. A melting curve analysis to verify the specificity and identity of the products was performed.
The relative quantities were determined by the efficiency correction method 3 To test the reproducibility of RTL measurement, the inter-assay CV of T/S-ratios was calculated according to the following formula: About 5% of all samples in both studies were analyzed in duplicate. Duplicate samples were never positioned on the same plate or at the same plate position. They were taken from different original DNA plates. Inter-assay CV of T/S-ratios of duplicates was 7.8% in CRISIS.
As second RTL measurement quality control, we analyzed the relative telomere length of a commercially available DNA, which was positioned on each 384-well plate (a total of 24 plates). Inter-assay CV of T/S-ratios of this 24 times analyzed sample was 3.4%. In case of MMKD, inter-assay CV of T/S-ratios of duplicates was 12% and of quality control 6.9% (6 plates).
T/S-ratio values of CRISIS and MMKD are not directly comparable as different DNA extraction methods were used but can be interpreted in the same manner within both studies.
Several methods of TL measurement have been published so far. Compared to the Southern blot method, the quantitative polymerase chain reaction (qPCR) requires far less amount of DNA and is a high-throughput method. It is therefore widely used for large epidemiological studies. The qPCR assay provides the ratio of the telomere (T) and the housekeeping gene 36B4 (S) that should be proportional to the RTL of an individual. The comparison between these two methods has been published previously 5 .

Line plot
Marginal mean values and 95% CI of relative telomere length (RTL) (age-and sex-adjusted) were derived by general linear regression models. In a next step, differences in marginal mean values were compared over chronic kidney disease (CKD) stages defined by the Kidney Disease Outcomes Quality Initiative (KDOQI) guidelines.

Proportional hazards assumption
All calculated Cox regression analyses in the MMKD Study did not depart from the proportional hazards assumption. In the CRISIS Study, in the total group and in non-smokers and non-diabetics, the fully adjusted Cox regression Model 3 did depart from the proportional hazards (PH) assumption. As the deviation from the PH assumption was mainly due to GFR, we included a covariate in the respective models that accounts for the time-dependency of GFR. Thus, the HRs shown for relative telomere length are adjusted for this time effect (see Table 2).

Sensitivity analysis
To exclude that the association of RTL with progression of CKD in active smokers and patients with diabetes is confounded by angiotensin-converting-enzyme inhibitors (ACEi) (

Investigation of the yearly GFR change in the CRISIS Study
The first and the last eGFR calculation of the CRISIS study were used to calculate an average yearly GFR decline. A non-linear P-spline was used to check for linearity of yearly GFR change in a general linear regression analysis on RTL in active smokers. Yearly GFR change was inverse-normal transformed due to its skewed distribution. This analysis was adjusted for age, sex, proteinuria and GFR at baseline. As can be seen from Supplementary   Figure 1 linearity was not present and we therefore performed the analysis using tertiles.