Accelerated ageing and renal dysfunction links lower socioeconomic status and dietary phosphate intake

Introduction

In the Scottish city of Glasgow, there is a large difference in life expectancy between affluent and deprived communities. This remarkable difference is one of the largest reported in the developed world, despite common risk factors for age-associated morbidities [1–4], despite the most and least deprived living within a 15 km geographical radius. This exceptional gradient of socio-economic difference is reflected in the associated variation in mortality and morbidity in this city.

The reasons for this remain unclear, but these have been investigated as a part of the psychological, social, and biological determinants of ill health (pSoBid) study cohort [1] the characteristics of which have been described in depth elsewhere [2].

In the pSoBid cohort, we have demonstrated previously that lower socio-economic status is associated with the expression level of the pro-inflammatory cytokine Interleukin 6 (IL-6), a marker of cardiovascular disease and with features of insulin resistance, such as elevated adipokine levels [3]. Additionally, we have demonstrated an association between accelerated ageing amongst the most deprived in association with poor diet, telomere shortening and genomic DNA hypomethylation [4, 5].

A common direct link between ageing and nutrition has also been demonstrated recently within mammalian genera, based around the uptake of dietary phosphate (Pi), with a strong correlation between Pi levels and longevity [6]. This is supported by data from Ohnishi et al [7, 8] who have demonstrated that Pi toxicity accelerates mammalian ageing. Recent evidence has shown that inorganic Pi shortens life span in klotho-deficient mice via activation of the AKT/mammalian target of rapamycin complex 1 (mTORC1) [9].

Phosphate is naturally present in basic foodstuffs, including meats, fish, eggs, dairy products and vegetables. Intestinal absorption of naturally occurring Pi is minimally regulated, as absorption is efficient, hence high Pi supplementation results in markedly elevated levels of serum phosphate. In the human body, Pi levels are regulated via vitamin D, parathyroid hormone (PTH) and FGF-23/klotho. High Pi levels, as a consequence of dietary intake, have been linked to higher all-cause and cardiovascular mortality risk in the NHANES III study [10] and to premature vascular ageing in the general population [11–15].

Extracellular Pi exerts cytotoxic effects by forming insoluble nanoparticles, termed calciprotein particles. Calciprotein particles are highly bioactive ligands that can induce various cellular responses, including the osteogenic transformation of vascular smooth muscle cells and the cell death of vascular endothelial and renal tubular epithelial cells [10]. Indeed, high Pi levels are a common feature of various metabolic diseases [16–18] including the rare Hutchinson-Gilford's Progeria syndrome [19].

Notably, the effects of calciprotein particles can be mitigated via the action of Fetuin A, a proven a mediator of biological ageing in a number of pathologies, including end stage renal disease and cancer [20, 21], where telomere length, a marker of biological age in circulating blood cells, is dependent on circulating Fetuin A levels.

Collectively, these observations in humans and other mammals are consistent with the hypothesis that increased Pi intake drives biological ageing processes [16, 20–24] and suggest that Pi intake may contribute to human health span at a fundamental level.

This is particularly intriguing in the context of Glasgow, as an association between serum Pi and SES has also been described previously [23], raising the possibility that the accelerated ageing observed among Glasgow's most deprived is contributed to by increased intake of dietary Pi.

We explored if variation in serum Pi levels were associated with the socioeconomic differences, accelerated ageing and epigenetic status among the participants in the pSoBid cohort. In turn, we have evaluated whether there were any associations with intake of specific dietary factors, including processed foods and their frequency of consumption.

Results

Subject characteristics relative to socioeconomic status (SES) and pertinent biochemical, biophysical lifestyle and nutritional parameters are listed in Table 1.

Determination of associations between serum phosphate levels, SES factors, biomarkers of ageing and biochemical parameters associated with chronic inflammation and metabolic syndrome.

In unadjusted analyses Pi levels showed significant associations with age (r= 0.09, 95% CI 0.008 to 0.167; p=0.032), log relative telomere length (95%CI −0.224 to −0.02; p=0.040), global DNA methylation content (95%CI −0.284 to −0.03; p=0.028), IL-6 (95%CI 0.017 to 0.179; p=0.043), eGFR (95%CI −0.186 to −0.026; p=0.025), adiponectin (95%CI 0.016 to 0.176; p=0.043) and D-dimer (95%CI 0.027 to 0.188; p=0.025)(Table 2), in keeping with previous data showing a relationship between features of biological ageing and deprivation in this cohort.

These parameters remained significant after adjustment for age alone, in combination with gender and/or deprivation status (Table 2) again in keeping with previous data showing a relationship between features of biological ageing and deprivation in this cohort. Other factors including smoking status, alcohol consumption, high blood pressure, incidence of cancer and diabetes were additionally analyzed in the context of aging markers and biochemical parameters (Supplementary Table 1). Negative correlations were observed between Pi levels and telomere length, global DNA methylation content and eGFR, while positive correlations were observed with age in males, adiponectin, cholesterol, IL-6 and D-dimer.

No overt associations were observed with diet or other SES measures in unadjusted analyses, nor were any observed between Pi and vitamin D in unadjusted analyses. When adjusted for age and gender, however, the relationship between Pi and vitamin D was significant (95%CI −0.253 to −0.093; p=0.010). No overt relationship was observed between vitamin D levels and deprivation status, however a significant association with Pi level was noted for the least deprived males (95%CI −0.392 to −0.089; p=0.010).

The relationship between serum Pi and relative telomere length was explained by age and gender (95%CI −0.283 to −0.083; p=0.036) as expected on the basis of previous analyses [4]. Women displayed higher Pi levels than men (1.02 mmol/L vs. 0.97 mmol/L; p=0.004), though only males showed a significant association between Pi levels and chronological age, in keeping with the most deprived males having the shortest telomeres and highest serum Pi (95%CI 0.431 to −0.063; p=0.036) (Figure 1).

Similarly, differences observed in global DNA methylation content were explained by age, gender and deprivation (Table 2). Notably, the least deprived females had higher DNA methylation content (95%CI −0.593 to −0.138; p=0.012) associated with lower Pi levels, than with their most deprived counterparts and with males, again in keeping with previous observations [5]. A negative association was observed between serum Pi levels and relative telomere length and DNA methylation status in the group of regular smokers (p<0.05, Supplementary Table 1). Similarly, subjects with cancer also displayed a significant association. However, due to the small number of such individuals, this association requires further investigation and may be a type 1 statistical error.

A weak correlation was observed between Pi and adiponectin levels (Table 2). This relationship was strengthened when adjusted for age (95%CI 0.045 to 0.205; p=0.026) and gender (95%CI 0.052 to 0.211), p=0.025). The gender effect was principally explained by deprivation status (95%CI 0.035 to 0.195; p=0.043) and most prevalent among the least deprived (95%CI Additionally, the association between Pi and cholesterol was principally explained by age and gender (95%CI 0.089 to 0.246; p=0.006), with female gender being concomitant with higher cholesterol (95%CI 0.042 to 0.265; p=0.027). This relationship was not weakened when additionally adjusted for deprivation status. Indeed, in the most deprived females high cholesterol levels correlated with high Pi (95%CI 0.034 to 0.342; p=0.042).

Phosphate levels also showed an association with IL-6 levels, which was strengthened when adjusted for gender (95%CI 0.087 to 0.247; p=0.002, Table 2).

A similar circumstance was also observed for the relationship between Pi and D-dimer (p=0.025); this association was strengthened when adjusted for age, gender and deprivation (95%CI 0.089 to 0.247; p=0.011).

No association between serum Pi levels and biochemical and ageing markers were observed in relation to alcohol consumption, high blood pressure, or diabetes. In contrast ageing markers (telomeres, global DNA methylation content) were negatively correlated with the serum Pi levels, while other biochemical para-meters displayed positive correlations (Sup Table 1).

Discussion

These data provide novel insight into a possible mechanism for the accelerated ageing and early onset disease observed amongst the most deprived in Glasgow.

Our data indicate that high Pi levels are associated with features of biological ageing, namely decreasing telomere length, increasing inflammatory status and genomic DNA hypomethylation [4, 5]. Similarly, Pi levels correlated with renal function, as determined by eGFR measurement, and parameters associated with insulin resistance, including adiponectin and cholesterol levels.

Notably, the correlation between Pi levels and accelerated biological ageing is prominent in males and most specifically in the most deprived males, where it is principally associated with increased red meat consumption. This effect appears to be independent of telomere attrition, as the association between Pi levels and telomere length is not strengthened as a result of adjustment for age and deprivation and gender. Deprived males also show other features consistent with accelerated biological ageing, including reduced genomic methylation content and elevated IL-6 levels.

No other foodstuff appears to adversely affect Pi levels in males. In females only fizzy drink consumption appeared to influence Pi levels. The association with fizzy drink consumption is striking; as those drinking less than one fizzy drink a month have significantly lower Pi levels than those consuming fizzy drinks more than once per month. This effect was strongest among the least deprived and no such association was found among the most deprived females or among males.

A gender difference in Pi levels was observed in this cohort. Females had higher Pi levels than males, a difference that increased with higher chronological age. This observation is non intuitive in the context of hyperphosphataemia being associated with decreased lifespan and health span in mammals [6], as human females in general, live longer than males, despite having higher Pi levels. One possible explanation for this is oestrogen status, and bone loss of Pi in post-menopausal females. Burnett-Bowie et al [25] have reported that the experimental suppression of testosterone levels in healthy volunteers was associated with increasing Pi concentration. Oestrogens are thought to facilitate the incorporation of Pi in remodeling bone; thus with advancing age Pi levels in women might increase.

The relationship between Pi levels and adiponectin is also striking. Adiponectin is typically a regulator of glucose metabolism and inflammatory signaling [26, 27] and may also influence bone turnover [28]. It is highest among the least deprived females, indicative of a better relative health status among this group and in keeping with a better health and longer lifespan. Notably, this group also possesses the greatest genomic methylation content and better eGFR than males, consistent with superior biological ageing relative to the other groups.

Our observations are consistent with a mechanism whereby elevated Pi is associated with increased CPP formation and cellular dysfunction and death. Moreover, among the most deprived elevated Pi is associated with high IL-6 levels, a measure of inflammatory status and increased comorbidity (Table 2) consistent with accelerated biological ageing.

That a Pi associated acceleration in biological ageing correlates with nutrition accords with the hypothesis that differences in biological ageing contribute to Glasgow's health disparity. Our finding that red meat consumption amongst the most deprived males correlated strongly with the advanced biological age in this group, relative to least deprived males and females of both socioeconomic groups, is novel. Notably, this association was not observed for other meat products (non-red meat including pates, sausages, haggis or ham). Red meat consumption has previously been related to increased cancer risk and overall mortality in a number of studies [29, 30]. Recent studies have reported that red meat consumption is associated with higher risk of cancer [31] and CVD [32]. In the present study, the observations indicate that elevated red meat consumption has adverse effects amongst a subject group that has a poor diet and sub optimal fruit and vegetable intake [4], where the effects of high Pi intake may be exacerbated. Notably, these effects are not apparent among less deprived males, or in females, especially in the context of a more balanced diet.

The prevalence of Pi containing additives is highest in low cost processed foods [33], most frequently consumed by the most deprived. Our study does not address this directly, but indicates that further study is merited for this aspect of diet. As Pi in beverages is readily absorbed these could also be postulated as having a significant impact on serum Pi levels. Fizzy drink consumption has recently been subject to much debate regarding the adverse impact on health from their elevated sugar content [34]. In this study, the effect of Pi levels from fizzy drink consumption was striking. This study indicates that the Pi content of these drinks actually had a limited impact on age related health. It should be noted, however that considerable amounts of Pi are also found in beer and wine [35] and this will need to be evaluated in more detail in pSoBid.

Strikingly, many of the subjects had eGFR values indicative of incipient or early onset CKD. Our observations are consistent with accelerated ageing [15, 16] being associated not only with premature vascular disease and inflammation but mild-moderate CKD [1–3]. These findings are directly relevant to the growing incidence of CKD globally [36, 37]. As the general Glaswegian population has a high prevalence of early stage CKD, especially so among the most deprived males. This indicates hyperphosphataemia associated with deprivation as a potential contributory factor to CKD. Indeed, it has been speculated that a high dietary Pi intake increases FGF-23 synthesis, which increases the Pi excretion per nephron. The Pi overload precipitates CaPi in renal tubular cells and leads to tubular injury and interstitial fibrosis, as extracellular Pi is toxic to the cell at high concentrations [38]. Indeed, data on 2269 healthy subjects from the Framingham Heart Study showed the odds ratio of developing CKD is much higher in healthy subjects with a serum Pi >4 mg/dL [39].

Our data suggest that Pi levels merit further study in larger and longitudinal cohorts, as well as in more controlled mechanistic studies. As high Pi intake has also been associated with increased mortality in an ostensibly healthy population [12]. Our findings may have important public health implications, both locally in Glasgow and in a wider public health context.

Weaknesses of the present study include its cross sectional nature and its limitation in size and limited data on type of red meat consumed (i.e. non processed beef, pork etc. versus processed meat). A number of potential confounders that were not taken into consideration also bear mentioning. While the genetics of the pSoBid cohort is relatively homogeneous, differences in epigenetics and circadian effects on ageing remain to be fully elucidated. These data are, however, biological plausible based on consistency with previous findings in this and other cohorts [17, 23, 38, 40]. In this respect, it is worth bearing in mind that analyses of large general population cohorts have inherent logistical issues with respect to the need for high throughput telomere analyses, hence the inclusion of techniques enabling individual telomere measurements, such as STELA, are not easily applied [41, 42], though the inclusion of an alternative technique in a sub set of samples, should enhance the robustness of any findings.

In summary, our data suggest that moderate hyper-phosphataemia promotes a gap between chronological and biological age. As high Pi intake is associated with increased mortality in an ostensibly healthy population [12] our findings have public health implications. In the light of both the recent non-domestic livestock red meat scandal and the WHO report on red meat consumption and cancer [33], it has not escaped our attention that red meat product quality and preservation may have an impact upon the diets of the most deprived and their associated health.

Materials and Methods

Supplementary Materials

Supplementary Table