Introduction

Phosphate is an essential mineral in the body1 and plays a crucial role in many physiological processes, such as energy generation, signal transduction, acid-base balance and bone mineralization2. Phosphate homeostasis is modulated by dietary phosphate intake, intestinal absorption, renal reabsorption and excretion3. There are various regulators of phosphate metabolism, including parathyroid hormone (PTH), calcitriol, vitamin D and fibroblast growth factor 23 (FGF23)2. And imbalance of phosphate homeostasis may induce hypo- and hyperphosphatemia4. Hyperphosphatemia is correlated with a higher risk of cardiovascular diseases5 and all-cause mortality6. In contrast, hypophosphatemia is associated with hypertension7 and reduced insulin resistance8. Kalaitzidis, R. et al. reported that individuals with metabolic syndrome had remarkably lower serum phosphate levels than did healthy participants9. Hence, higher and lower serum phosphate levels have been associated with, respectively, increased cardiovascular risks and more parameters of metabolic syndrome.

Telomeres capping the end of eukaryotic chromosomes protect chromosomes from loss and end-to-end fusion10. Telomeres shorten with repeated cell division and DNA replication11. Telomere length has been positively correlated with healthy life years in elderly individuals12. Furthermore, shorter telomere length is correlated with higher all-cause mortality13. Numerous factors have been found to be associated with telomere length14. Accelerated telomere attrition was correlated with lower socioeconomic status and poor diet15. A review in 2011 demonstrated that various nutrients, such as vitamin A, vitamin D, folate, vitamin B12, magnesium, zinc and iron, may affect telomere length16. In a cross-sectional analysis, Xu, Q., et al. found a positive relationship between multivitamin use and telomere length in women17. Higher dietary intake of magnesium was also associated with longer telomere length17. In contrast, shorter telomere length was found in iron supplement users18.

Despite a large amount of evidence on factors affecting telomere length, little research has clarified the relationship between serum phosphate and telomere length. The aim of our study was to investigate the correlation between serum phosphate and mean telomere length in the US general population.

Results

Characteristics of the study population

The characteristics of the participants divided by serum phosphate quartiles are listed in Table 1. In the study, the mean age of all participants was 49.42 ± 18.82 years, and 48.2% of all participants were men. Mean telomere length, creatinine and total calcium levels were significantly positively associated with serum phosphate levels, whereas age, body mass index (BMI), SBP, serum FG and serum albumin were negatively correlated with serum phosphate levels.

Table 1 Characteristics of Study Participants of Quartiles of serum phosphate.
Full size table

Association between serum phosphate levels and mean telomere length after stratification by sex and ethnicity

Tables 2, 3 presents the results of the correlation between serum phosphate levels and mean telomere length stratified by sex and ethnicity. A significant correlation was found between higher serum phosphate levels and longer mean telomere lengths in the three models (p < 0.05). Additionally, male participants demonstrated an association between higher serum phosphate levels and longer mean telomere lengths in model 2 and in the fully adjusted model (model 2: β = 0.043, p = 0.011; model 3: β = 0.045, p = 0.009). In the non-Hispanic white group and other racial groups, a positive association between serum phosphate levels and mean telomere length was found in 3 adjusted models (p < 0.05).

Table 2 Gender-specific association between the mean Telomere Length and the serum phosphate level.
Full size table
Table 3 Ethnicity specific association between the mean Telomere Length and the serum phosphate level.
Full size table

After serum phosphate levels were stratified into four quartiles, gender-specific associations between serum phosphate levels and mean telomere length were determined and are listed in Table 4. Table 4 shows a significantly positive association between the highest serum phosphate level (Q4) and mean telomere length for all designed models (p < 0.05). The mean telomere length was greater in the higher quartiles than in the lowest quartile of serum phosphate, and a dose-dependent association was observed. Male participants with the highest serum phosphate levels (Q4) had a longer mean telomere length than did those with serum phosphate levels in Q1 in the 3 adjusted models.

Table 4 Gender-specific association between the mean Telomere Length and the serum phosphate level.
Full size table

Table 5 presents the ethnicity-specific association between serum phosphate levels and mean telomere length after multivariable logistic regression analysis. Only in the non-Hispanic white group, participants with higher quartiles of serum phosphate levels (Q2, Q3 and Q4) had a longer mean telomere length than did participants with serum phosphate levels in Q1 in the 3 regression models (p < 0.05).

Table 5 Ethnicity specific association between the mean Telomere Length and the serum phosphate level.
Full size table

Discussion

In the present cohort study of the US general population, a positive relationship between serum phosphate levels and mean telomere length was observed. These findings are not consistent with those of previous studies5,19,20,21,22,23. Higher serum phosphate levels were noted to be linked with increased cardiovascular risk and mortality risk5,19,20,21,23,24. Furthermore, McClelland, R., et al. demonstrated that hyperphosphatemia was associated with accelerated aging, which was evaluated by markers of biological age (i.e., telomere length and DNA methylation content)22. Recently, emerging evidence on klotho expression and aging has been published25,26,27,28. Klotho was associated with phosphate homeostasis via the fibroblast growth factor (FGF) receptor, and decreased klotho expression may induce hyperphosphatemia29. Thus, a plausible association between phosphate toxicity and accelerated aging in klotho-deficient mice was also reported30,31.

Telomere length was viewed as a biomarker of age32. Accelerated telomere attrition was associated with various diseases, such as coronary heart disease33, diabetes mellitus34, hypertension35 and cancer36. Telomere length was influenced by multiple factors, including psychosocial, environmental, and behavioral factors14. Different lifestyles and diets also played prominent roles in telomere length37,38. Mirabello, L., et al. demonstrated that a healthy lifestyle with more exercise, cigarette abstinence and a diet high in fruit and vegetables were significantly correlated with longer telomere length38. Cassidy, A., et al. reported that intake of dietary fiber was positively linked with telomere length, while waist circumference, body mass index (BMI) and the intake of polyunsaturated fatty acids were negatively linked37. The inverse correlation between obesity and telomere length was observed in females39. Increased adiposity and increasing BMI were correlated with shorter telomere length40. A previous study by Song et al. revealed that BMI was significantly positively associated with biomarkers of DNA damage41. Furukawa et al. proposed that elevated chronic oxidative stress was observed in accumulated fat42. Kurz, D.J., et al. demonstrated that oxidative stress may accelerate telomere erosion43. Hence, we speculated that a higher BMI may increase oxidative stress, which could cause DNA damage and might induce telomere attrition.

Few studies have demonstrated a negative association between serum phosphate and BMI44,45. Lower BMI might increase the risk of hyperphosphatemia46. Haglin et al. observed that hypophosphatemia was correlated with higher BMI in females47. Our results were in agreement with these articles that showed an inverse relationship between serum phosphate and BMI. Several possible mechanisms have been proposed for this phenomenon. Haglin et al. speculated that it might be due to a high caloric diet with low nutrient density and low protein intake, which could cause phosphate depletion47. Obeid, O.A. hypothesized that lower serum phosphate may reduce ATP production, which is important for energy expenditure45. Depending on the inverse relationship, we hypothesized that a higher serum phosphate level might be associated with a lower BMI and that a lower BMI may be associated with a greater telomere length. Therefore, these hypothesized relationships might serve as a possible explanation for our results. The hypotheses that participants with higher serum phosphate levels had longer telomere lengths are biologically plausible.

Our study demonstrated that statistically significant relationships between higher serum phosphate levels and longer mean telomere length, particularly in male participants. Previous published studies have shown gender differences in serum phosphate levels48,49. Few studies have examined the effects of gonadal steroids on phosphorus homeostasis. Sex hormone deprivation was found to be correlated with elevated serum phosphate levels in men50. Moreover, testosterone and estrogen were associated with mean telomere length51,52. A possible mechanism may account for this phenomenon53.

There are several limitations in the present study. First, because our study was a cross-sectional observational analysis of a database, causal inferences were not clear. Further longitudinal studies are warranted. Second, adjustments were not made in regression models for unmeasured confounding variables, such as lifestyle factors and dietary patterns that might interfere with both mean telomere length and serum phosphate levels. Third, there might be a possible selection bias in our study. Our participants were mostly Caucasian; therefore, the generalizability of the results might be limited to specific racial populations.

In conclusion, a significant positive correlation between serum phosphate levels and mean telomere length was observed in the study. The findings of the paucity of currently published studies were consistent with our results, and our study provided epidemiologic evidence for further studies on the relationship between mean telomere length and serum phosphate levels.

Materials and methods

Study populations

All data were obtained from the National Health and Nutrition Examination Survey (NHANES) from 1999 to 2002. NHANES was a cross-sectional study of noninstitutionalized US citizens and was conducted by the Centers for Disease Control and Prevention (CDC) and the National Center for Health Statistics (NCHS). NHANES database included personal data including demographic information, past medical history and laboratory data. The participants’ information was collected by a household interview and a subsequent physical examination. All of the study protocol, consent documents and relevant information were detailed on the NHANES website. The NHANES study protocol was conducted according to the NCHS Institutional Review Board (IRB). Before data collection and the health examinations, all informed consents had been obtained. All the experimental protocols were approved by NCHS IRB. Initially, a total of 7,817 participants who aged 20 years old or older were included in our study.

Covariates

For causal relationship from cross-sectional study, it was essential to control confounding, but it was hard to recognize a potential confounder. A confounding variable was associated with the exposure and with the outcome or the occurrence of a disease. A confounder obscured the real causal path between the exposure and outcome. Based on the variables correlation and previous studies, we drew our Directed Acyclic Graph (DAG) analysis of the study, showed in Fig. 1. Demographic confounders were collected with a computer-assisted personal interviewing system, including age, gender, race, personal history, and past medical status. Other medical histories were recorded, including congestive heart failure, coronary artery disease, angina, stroke and cancer/malignancy, which were diagnosed or revealed by a doctor.

Figure 1
figure 1

Directed Acyclic Graph of our study.

Full size image

The level of serum C-reactive protein (CRP) was measured by the Dade Behring Nephelometer II Analyzer System using latex-enhance nephelometry (Dade Behring Diagnostics Inc., Somerville, NJ). The level of serum fasting glucose (FG) was measured by the Instrumentation Cobas Mira Chemistry System (Roche Diagnostic Systems, Inc., Montclair, New Jersey). The resting biochemical profiles, including creatinine, alanine aminotransferase (ALT), total calcium, serum total cholesterol (TC) and serum albumin, were measured with the Beckman Synchron LX20. All protocols followed the standardized guidelines and record accuracy based on CDC reference methods.

Serum phosphate measurement

The level of serum phosphate was measured by a Hitachi model 737 multichannel analyzer (Boehringer Mannheim Diagnostics, Indianapolis, IN). Inorganic phosphorus reacted with ammonium molybdate in an acidic solution to make a colored phosphomolybdate complex. The quantification of serum phosphate was performed using the change in absorbance at 365 nm of phosphomolybdate.

Mean telomere length measurement

The telomere length assay for the measurement of telomere length relative to standard reference DNA (T/S ratio) was performed using the quantitative polymerase chain reaction (PCR) method in the laboratory of Dr. Elizabeth Blackburn at the University of California, San Francisco. More detailed information can be found on the NHANES web site in the laboratory section.

Statistical analysis

We used SPSS version 18 (SPSS Inc., Chicago, IL, USA) to perform all statistical analyses. Continuous variables are indicated as the means and standard deviations (SDs); categorical variables are indicated as numbers and percentages. The chi-square test and one-way ANOVA were used for categorical data and continuous data, respectively. To examine the variables of interest, we divided 7,817 participants into quartiles based on serum phosphate levels. Two-sided p-values <0.05 were considered statistically significant.

We investigated the relationship between serum phosphate levels and mean telomere length using multivariable logistic regression analysis. Covariate adjustments were conducted using 3 extended-model methods: model 1 was adjusted for age, gender and race; model 2 was further adjusted for associated clinical laboratory data; and model 3 was further adjusted for smoking history and past medical history. We tested for effect modification by serum phosphate levels and race and sex by including interaction terms in the models for the mean telomere length. Based on the statistically significant findings of the interaction effect, we used stratified test to perform further analyses.