Our study demonstrated that dietary phosphorus should be taken into account as a potential cardiovascular risk factor not only in renal patients, but also in the general population. We found positive linear trends in associations among energy-adjusted dietary P intake and FAP intake, and IMT. Moreover, IMT of females was significantly higher in the highest TP intake quintile compared to the lowest intake quintile. The results strengthen the hypothesis that not only elevated serum phosphate concentrations, but also higher dietary P intake may be associated with higher IMT in a normal, healthy population. Data on health effects of high P intake on the normal population are scarce; most of the studies carried out on P intake have focused on bone health [21–24].
We wanted to address our research question separately for TP intake and FAP intake because FAP is more absorbable in the intestine than natural P [7, 8]. However, the proportion of FAP in TP is unknown and FAP amounts in foodstuffs between manufacturers vary greatly. We also paid attention to TP intake and, when adjusted for energy intake, describes the P density of the diet, excluding energy intake as a potential confounding factor. While marked variation in TP intake can be seen between the smallest intake quintile (ca. 1100 mg/d) and the highest one (ca. 2300 mg/d), on average all subjects exceeded recommended intakes [4, 5]. However, it should be noted that persons in Finland usually have a high dietary Ca intake , as also our study subjects had (mean intake about 1200 mg/d).
IMT, the main outcome of the study, is a marker of pre-clinical atherosclerosis and is associated with such traditional CVD risk factors as smoking, diastolic blood pressure, age, serum glucose, and LDL-C concentration [25, 26]. Additionally, male sex and black race are associated with higher IMT . Already in younger populations changes in carotid IMT can be seen . FGF-23, a bone-derived hormone-like phosphatonin, has been speculated to participate in the effects of P in CVD, but the mechanisms are unknown . High FGF-23 concentrations have been associated with CVD risk factors  and CVD itself  but results of many studies are discordant , and to our knowledge, no studies have focused on FGF-23 and IMT in the general population. Dietary P load has been shown to increase FGF-23 concentration in serum acutely , however, in our study P-FGF-23 concentration was not associated with dietary P intake which may be due to the cross-sectional design and not adjusting for potential covariates.
The regulation of Pi plays a role in vascular calcification. Smooth vascular muscle cells differentiate to osteoblast/chondrocyte-like cells under high Pi concentrations which induce calcification [33, 34]. Ruan et al.  reported that high S-Pi was associated with higher IMT, and Park et al.  found that lower S-Pi was associated with less coronary calcification. However, in our study FAP intake was associated with IMT, but IMT was inversely associated with S-Pi concentrations. It should be noted that analyses on S-Pi were not adjusted for potential covariates, and, in addition, our blood samples were collected when the S-Pi is at its lowest concentration, which may explain the inverse association between FAP intake and S-Pi. Notable is that in the Health Professionals Follow-Up Study, plasma phosphate concentration was not associated with the development of coronary heart disease among males . However, we have no information about the sample collection time of the other studies, and the contradictories may be explained by this. In the studies of Mataix et al.  and de Boer et al. , dietary P intake correlated only weakly with S-Pi and Mataix et al.  concluded that because S-Pi is tightly controlled, dietary P does not affect S-Pi very efficiently. The cross-sectionality of these studies and our study may affect the results because clinical studies carried out at our unit indicate an acute dose–response effect of dietary P on S-Pi concentration [8, 21, 22].
One limitation of the study is its cross-sectional design, which precludes an evaluation of causality. However, it is also difficult to know if the dietary intake methods in the current study are indicative of usual diet. Selection bias may also exist; our study subjects may be healthier than the normal population because they were interested in participating in this kind of study. There may be some differences between the sexes in adaptation to P; here the association was affected by sex. Women being more vulnerable to P, especially through Ca and PTH metabolism , may explain not finding differences among males. Nevertheless, the distribution of sexes in our study was not equal; the number of females was twice the number of males, and the sizes of FAP groups differed widely, with the smallest group sizes having the highest FAP intake scores. In the case of males, the size was only ten persons, which may also cause the lack of power in the analyses.
Strengths of our study are that many biomarkers were analyzed and extensive background data were available. In large studies where the main focus was not nutritional, food intake data may have been collected less accurately than in our study. The way that FAP intake is calculated may be criticized, but one should realize that the information on the exact amounts of FAP in foodstuffs is not available.