Impact of Phosphorus-Based Food Additives on Bone and Mineral Metabolism

Abstract

Context:

Phosphorus-based food additives can substantially increase total phosphorus intake per day, but the effect of these additives on endocrine factors regulating bone and mineral metabolism is unclear.

Objective:

This study aimed to examine the effect of phosphorus additives on markers of bone and mineral metabolism.

Design and Setting, and Participants:

This was a feeding study of 10 healthy individuals fed a diet providing ∼1000 mg of phosphorus/d using foods known to be free of phosphorus additives for 1 week (low-additive diet), immediately followed by a diet containing identical food items; however, the foods contained phosphorus additives (additive-enhanced diet). Parallel studies were conducted in animals fed low- (0.2%) and high- (1.8%) phosphorus diets for 5 or 15 weeks.

Main Outcome Measures:

The changes in markers of mineral metabolism after each diet period were measured.

Results:

Participants were 32 ± 8 years old, 30% male, and 70% black. The measured phosphorus content of the additive-enhanced diet was 606 ± 125 mg higher than the low-additive diet (P < .001). After 1 week of the low-additive diet, consuming the additive-enhanced diet for 1 week significantly increased circulating fibroblast growth factor 23 (FGF23), osteopontin, and osteocalcin concentrations by 23, 10, and 11%, respectively, and decreased mean sclerostin concentrations (P < .05 for all). Similarly, high-phosphorus diets in mice significantly increased blood FGF23, osteopontin and osteocalcin, lowered sclerostin, and decreased bone mineral density (P < .05 for all).

Conclusions:

The enhanced phosphorus content of processed foods can disturb bone and mineral metabolism in humans. The results of the animal studies suggest that this may compromise bone health.

Disorders of phosphorus metabolism are associated with bone disease and excess cardiovascular disease independently of established risk factors (1). Although the mechanisms for these associations remain incompletely understood, a considerable body of data implicate local and systemic disturbances in phosphorus metabolism in the pathogenesis of bone and cardiovascular disease (2–7). These results have fueled interest in identifying potentially modifiable risk factors for disordered phosphorus metabolism in the general population. Dietary factors have garnered the most attention given that excess phosphorus intake plays a central role in the pathogenesis of disordered phosphorus metabolism.

Dietary phosphorus consumption in the United States far exceeds current recommendations for daily intake (8). This is partly due to the nearly ubiquitous distribution of phosphorus-based food additives in the food supply (9). Phosphorus-based additives serve a number of critical functions for food manufacturing, including pH stabilization, leavening, and bactericidal actions. Because of this wide diversity of applications, phosphorus-based additives are heavily used by the food manufacturing industry and are estimated to account for 10–50% of total phosphorus intake per day in the typical Westernized diet. Although these levels of intake are considered generally safe for public consumption by current US Food and Drug Administration guidelines (10), the vast majority of studies used to derive these guidelines were conducted more than four decades ago (9). As a result, these studies were unable to examine the effect of phosphorus-based additives on more recently discovered endocrine factors such as fibroblast growth factor 23 (FGF23), osteopontin, and sclerostin, which are critically involved in regulating bone and mineral metabolism and associated with bone and cardiovascular disease (11–14). Understanding the effect of phosphorus additives on these more recently discovered markers is needed to fully assess the public health effect of high amounts of phosphorus-based food additives in the US food supply. The focus of the current study was to examine the effect of phosphorus-based food additives on markers of bone and mineral metabolism in healthy individuals, and to compare and contrast physiological changes in these markers in humans to those of animals fed progressively increased amounts of phosphorus over longer periods of intervention.

Materials and Methods

Human studies

Healthy volunteers 19–45 years of age from the Birmingham, AL, metropolitan area were recruited to participate in the study. Exclusion criteria included the following: known history or laboratory evidence of kidney disease as indicated by an estimated glomerular filtration rate less than 60 mL/min per 1.73 m² or an abnormal urinalysis; pregnancy, breast-feeding, or a history of irregular menses for females; medical conditions known to effect phosphorus metabolism (eg, thyroid disease); use of medications known to affect phosphorus metabolism (eg, high-dose vitamin D supplements); body mass index at least 30 kg/m²; or screening laboratory evidence of abnormal serum phosphorus (> 4.6 or < 2.5 mg/dL), serum calcium (total calcium > 10.6 or < 8.5 mg/dL), or severe anemia (hemoglobin < 8 g/dL for women and < 9 g/dL for men). The University of Alabama at Birmingham Institutional Review Board for Human Use approved the study and all participants provided written informed consent.

Study diets

Separate menus for a low-additive diet and an additive-enhanced diet were developed by the Bionutrition Core of the Clinical Research Unit (CRU) at University of Alabama at Birmingham using the Nutrition Data System for Research (2011) as previously detailed (15). Briefly, each menu consisted of 4 separate days of food (breakfast, lunch, and dinner; full menus are presented in Supplemental Table 1). The low-additive menu was designed to provide 1800–2000 kcal/d (15% from protein, 55–60% from carbohydrates, and 25–30% from fat) and 900 mg of phosphorus per day using minimally processed foods. These targets were meant to approximate current U.S. Department of Agriculture recommendations (16). The additive-enhanced menu was designed to provide identical energy and nutrient content per day as the low-additive menu, but substituted highly processed for minimally processed foods (15). To confirm that the study diets adequately differed in the main nutrient of interest (phosphorus), the nutrient contents of each menu were measured by Covance Laboratories Inc. as previously described (15).

Study protocol

After completing a screening visit, eligible participants entered a 2week run-in period during which two 24-hour urine collections were obtained for measurement of creatinine clearance and urinary excretion of calcium and phosphorus while consuming ad libitum diets (Supplemental Figure 1). In addition, a 4-day food record was obtained to assess baseline diet intake. At the end of the run-in period, participants started the intervention period during which they were provided study meals prepared by the metabolic kitchen to consume at home during the following 2 weeks. During the first week, participants consumed the low-additive diet to standardize phosphorus intake given wide variability in ad libitum phosphorus consumption. Immediately following the end of the first week, participants consumed the additive-enhanced diet for 1 week. Participants returned to the CRU every 3–4 days to pick up packaged meals and also to have follow-up weight and blood pressure measurements and blood and urine samples collected. In addition, 24-hour urine collections were obtained at the end of each week of the study diets. Compliance with interventions was monitored by self-reported intake and measurements of 24-hour urinary phosphorus excretion.

Data collection

At each visit during which participants picked up study meals, blood and urine were collected to measure phosphorus, calcium, and creatinine concentrations using standard assays. In addition, serum and plasma samples were collected and frozen at −80°C for batched analysis of FGF23, PTH, amino-terminal propeptide of type 1 collagen (P1NP), carboxy-terminal collagen cross-links (CTX), osteopontin, osteocalcin, and sclerostin. Details on assay characteristics are listed in the Supplemental Materials and Methods.

Animal studies

Animal care and experimental procedures were conducted with the approval of the Emory Institutional Animal Care and Use Committee. C57BL/6 mice were obtained at 8 weeks of age from The Jackson Laboratories (5-wk study) or bred at Emory University (15-wk study). Mice were housed in a facility with controlled conditions (temperature, 21–24°C; humidity, 40–70%; light/dark cycle, 12/12 h light/dark). At 10 weeks of age, mice were randomly assigned (ten female mice per group) to either a low-phosphorus diet (LPD; 0.2% phosphorus [Cat No. TD.110360]) or a high-phosphorus diet (HPD; 1.8% phosphorus [TD.110362]), all manufactured by Harlan Laboratories, Inc. The study diets were designed to keep calcium constant (0.6%) and represent estimated intakes of ∼500 mg/day (LPD) and ∼2100 mg/day (HPD). Protein, fat, and other mineral elements were matched between the diets providing essentially identical energy and caloric values (3.7–3.9 kcal/g). Vitamin D was 2.2 IU for each diet. Mice were fed ad libitum. A second study was performed under similar conditions in which mice (seven per group; 5F/2M) were fed the varying phosphorus diets for 15 weeks. After 5 or 15 weeks on the diets, mice were killed for analysis of bone mineral density (BMD), cortical and trabecular indices, and serum factors; and bone and bone marrow were collected for RNA (see Supplemental Methods for full details). Weight was recorded at the completion of the study. In pilot studies food consumption was measured weekly for greater than 15 weeks with no significant difference between diets in agreement with our previous results (17).

Statistical analysis

For the human studies, linear mixed-effects models were used to examine changes in blood and urine analytes over time. In these models, time represented the repeated measures factor and individuals were treated as random effects terms. When we detected a significant effect of time, we localized individually significant changes in postbaseline time points by comparing them with the baseline values using ordinary least-squares linear regression. The primary outcome measure was the percent change in FGF23 at the end of the additive-enhanced dietary period compared with the end of the low-additive period. Secondary outcome measures included changes in mean blood and urine analyte values over time. Plasma FGF23 concentrations were not normally distributed, so log-transformed values were analyzed and back-transformed into geometric means (95% confidence intervals [CIs]) for clarity of interpretation. For the animal studies, differences in BMD, and blood analytes between diet groups were assessed using Student t test. Two-sided P < .05 was considered statistically significant.

Results

Human studies

A total of twelve volunteers were enrolled onto the 2-week feeding study. Two participants withdrew shortly after enrolling: one due to inability to tolerate the study diet and the other due to diagnosis of severe hypothyroidism on screening labs, leaving ten participants who completed the entire study protocol.

Table 1 depicts demographic characteristics, estimated daily energy, and nutrient intake, and laboratory measurements of study participants at baseline. The mean age of the study sample was 31.9 ± 7.9 years, 30% were male, and 70% were African American. Measured energy and nutrient contents of the study diets, averaged over the 4 menu days, are depicted in Table 2 (measured energy and nutrient contents for each menu day are presented in Supplemental Table 2). There were no statistically significant differences in total energy, calcium, or potassium contents between the two diets. The 4-day average phosphorus contents of the low-additive and additive-enhanced diets were 1070 ± 58 and 1677 ± 167 mg, respectively (mean difference of 606 ± 125 mg/d; P < .001). The 4-day average sodium contents of the low-additive diet and additive-enhanced diets were 89 ± 28 and 148 ± 28 mmol, respectively (mean difference of 58 ± 27 mmol/d; P = .02).

Table 1.

Participant Characteristics at Baseline Presented as Arithmetic Mean ± SD, Geometric Mean (95% CI), or Number (frequency)

Variable	n = 10
Age	31.9 ± 7.9
Male, No. (%)	3 (30)
Black, No. (%)	7 (70)
Body mass index, kg/m2	22.8 ± 3.1
Systolic blood pressure, mm Hg	121 ± 15
Diastolic blood pressure, mm Hg	73 ± 10
Baseline diet, 4-d food record
Energy, kcal/d	1800 ± 360
Calories from fat, %	35.4 ± 3.6
Calories from carbohydrate, %	48.1 ± 5.9
Calories from protein, %	16.3 ± 3.0
Phosphorus, mg/d	968 ± 227
Calcium, mg/d	578 ± 168
Sodium, mmol/d	142 ± 34
Baseline Laboratory Data
Serum phosphorus, mg/dL	3.66 ± 0.9
Serum calcium, mg/dL	9.22 ± 0.2
Serum 25-hydroxyvitamin D, ng/mL	24.9 ± 6.2
Intact PTH, pg/mL	41.6 ± 12.4
FGF23 (RU/mL)	77.2 (49.3, 120.9)
24-h urine phosphorus excretion, mg/d	625 ± 137
24-h urine calcium excretion, mg/d	88.3 ± 42.6
24-h urine sodium excretion, mmol/d	129.1 ± 49.7
Creatinine clearance, mL/min	127 ± 35

Abbreviation: RU, reference units.

Table 2.

Characteristics of the Study Diets Presented as Means ± SD

Characteristic	Low-Additive Diet	Additive-Enhanced Diet
Calories, kJ/d	2294 ± 258	2278 ± 323
Fat, %	24 ± 2	28 ± 5a
Protein, %	17 ± 3	16 ± 4
Carbohydrates, %	59 ± 2	56 ± 9
Calcium, mg/d	732 ± 232	677 ± 178
Phosphorus, mg/d	1070 ± 58	1677 ± 167b
Sodium, mmol/d	89 ± 28	148 ± 28a
Potassium, mg/d	3252 ± 758	3152 ± 580

^aP < .05 compared with low-additive diets.

^bP < .001 compared with low-additive diet.

Table 3 depicts the changes in 24-hour urinary phosphorus, calcium, sodium, and potassium during the 2-week dietary intervention. Mean 24-hour urinary phosphorus and sodium excretion significantly declined after 1 week of the low-additive diet (visit 5) and then increased back to baseline values after 1 week of the additive-enhanced diet (visit 7). Mean 24-hour urinary potassium excretion decreased from baseline during the 2 weeks of study diets, with no differences between study diet periods. There were no statistically significant changes in mean 24-hour urinary calcium excretion throughout the study.

Table 3.

Change in Laboratory Data Over Time

Laboratory Value	Baseline	Low-Additive Diet		Additive-Enhanced Diet		P-Time
	Day 0	+3 Days	+7 Days	+10 Days	+14 Days
Urine analyte excretion
Urine phosphorus, mg/d	625 ± 137	—	477.2 ± 217.3	—	601 ± 228.9	.006
Urine calcium, mg/d	88.3 ± 42.6	—	121.6 ± 62.5	—	110.3 ± 63.3	.57
Urine sodium, mmol/d	129.1 ± 49.7		69.8 ± 25.9	—	110.1 ± 40.4	.01
Urine potassium, mmol/d	47.1 ± 8.5	—	38.3 ± 12.1	—	38.0 ± 13.7	.03
Blood markers of mineral metabolism
Phosphorus, mg/dL	3.66 ± 0.95	3.71 ± 0.52	3.95 ± 0.46	3.83 ± 0.46	3.85 ± 0.60	.84
Calcium, mg/dL	9.22 ± 0.22	9.27 ± 0.37	9.24 ± 0.31	9.28 ± 0.24	9.30 ± 0.27	.58
FGF23, RU/mL	77.2 (49.3, 120.9)	70.4 (44.9, 110.3)	71.8 (45.8, 112.6)	74.5 (47.5, 116.7)	85.7 (54.7, 134.3)	.19
PTH, pg/mL	41.6 ± 12.3	—	39.2 ± 14.6	—	39.2 ± 15.1	.91
Blood markers of bone metabolism
P1NP, μg/L	46.1 ± 22.4	—	43.2 ± 22.1	—	41.1 ± 19.6	.06
CTX, ng/mL	0.39 ± 0.18	—	0.41 ± 0.19	—	0.46 ± 0.26	.66
Osteopontin, pg/mL	73.4 ± 12.7	—	72.3 ± 10.2	—	79.3 ± 12.9	.02
Osteocalcin, pg/mL	13.3 ± 6.3	—	12.7 ± 6.6	—	13.9 ± 6.9	.02
Sclerostin, pmol/mL	42.6 ± 12.9	—	42.0 ± 15.7	—	37.3 ± 13.9	.02

Abbreviations: RU, reference units.

Results presented as Arithmetic Means ± SD or Geometric Means (95% CI).

Mean changes in measured blood analytes throughout the 2-week dietary period are summarized in Table 3. There were no statistically significant changes in arithmetic or geometric mean values of circulating phosphorus, calcium, FGF23, PTH, P1NP, or CTX over the course of the 2-week intervention period. Whereas there were no changes in mean serum osteopontin, osteocalcin, or sclerostin concentrations from baseline to the end of the low-additive diet period, osteopontin and osteocalcin concentrations significantly increased and mean sclerostin concentrations significantly decreased after 1 week of consuming the additive-enhanced diet (P for main effect of time = .02 for all).

Relative to baseline values, FGF23 did not significantly change after 1 week of the low-additive diet period (Figure 1). After 1 week of consuming the additive-enhanced diet, plasma FGF23 concentrations increased by 23% relative to values at the end of the low-additive diet period (P = .03). Similarly, osteopontin and osteocalcin concentrations did not significantly change after 1 week of the low-additive diet but significantly increased by 10% and 11%, respectively, after 1 week of consuming the additive-enhanced diet (P = .01 and .04, respectively). There were no significant changes in sclerostin concentrations relative to values from the immediate preceding dietary period (P = .10).

Figure 1.

Percent change in circulating FGF23, osteopontin, sclerostin and osteocalcin relative to the immediate preceding dietary period. Results are depicted as means ± SDs.

Animal studies

Mouse models were used to complement the clinical studies because of the ability to perform longer studies and to harvest tissues at completion. C57BL/6 mice were randomly assigned to LPDs (0.2%) or HPDs (1.8%). These diets were designed to keep calcium constant at 0.6% and therefore, resulted in progressively higher calcium-to-phosphorus ratios of 3:1 (LPD) and 1:3 (HPD). The mice were fed the diets ad libitum for 5 or 15 weeks. At the completion of the studies the average weight of the animals was not statistically different between groups (5-wk: 21.3 ± 3.2 g [LPD] vs 19.5 ± 1.3g [HPD]; 15-wk: 31.3 ± 6.6 g [LPD] vs 30.8 ± 4.2 g [HPD]). Analysis of serum for circulating endocrine factors identified that similar to the human studies, compared with the LPD, the HPD resulted in significantly higher levels of osteopontin, FGF23, osteocalcin, and a trend toward lower sclerostin levels after 5 weeks of diet (Table 4). After 15 weeks on the diet, serum osteopontin, FGF23, and osteocalcin were significantly higher and serum sclerostin concentrations were significantly lower in HPD compared with LPD mice (Table 4). The changes in endocrine factors and serum markers of bone metabolism between the diets were more pronounced after 15 weeks of diet, other than CTX.

Table 4.

Changes in Serum Factors in Mice.

Factor	LPD	HPD	% Difference	P
5 weeks of diet (mouse, n = 7–10)
FGF23, pg/mL	240 ± 67	1233 ± 715	+414	.001
Osteopontin, ng/mL	174 ± 56	240 ± 80	+38	.044
Sclerostin, pg/mL	213 ± 35	207 ± 17	−3	ns
CTX, ng/mL	21 ± 16	49 ± 21	+133	.005
Osteocalcin, ng/mL	5.4 ± 0.8	6.7 ± 1.0	+24	.006
15 weeks of diet (mouse, n = 7)
FGF23, pg/mL	334 ± 145	1607 ± 512	+381	<.001
Osteopontin, ng/mL	207 ± 49	334 ± 63	+61	.001
Sclerostin, pg/mL	182 ± 47	122 ± 27	−33	.012
CTX, ng/mL	20 ± 9	24 ± 13	+20	ns
Osteocalcin, ng/mL	2.6 ± 0.7	4.9 ± 1.1	+88	<.001

Abbreviation: ns, not significant.

Results are depicted as Mean ± SD.

To provide some insight into the mechanism, we examined bone and bone marrow for changes in RNA expression in response to the varying phosphorus diets. FGF23 and sclerostin (Sost) are both expressed predominantly by osteocytes, terminally differentiated osteoblasts that are embedded in the hydroxyapatite of bone (18, 19). Osteocytes are estimated to comprise ∼90% of living cells in bone (20). To determine whether the changes in serum concentrations of FGF23 or sclerostin correlate with changes in gene expression in osteocytes, bone (femur) was collected for gene expression analysis at the completion of the diet studies. Bone marrow was flushed from the femurs by centrifugation prior to RNA isolation. Results identified no significant change in FGF23 expression in bone from mice fed HPD relative to LPD at either 5 or 15 weeks (Supplemental Figure 2). Although not statistically significant, Sost expression decreased at both time points. Osteopontin (spp1) is more ubiquitously expressed, including by cells of the hematopoietic lineage and bone marrow stromal cells; and therefore, we analyzed the bone marrow flushed from the femurs for RNA levels. Gene expression analysis identified a significant increase in osteopontin expression (Supplemental Figure 2).

Many of these phosphorus-responsive serum factors are known to either affect bone metabolism or represent markers of changes in bone homeostasis. We therefore performed an analysis of the bones of the mice. Dual-energy x-ray absorptiometry (DXA) was used to determine BMD. The mice fed the HPD had significantly reduced spine and femur BMD relative to the LPD (Figure 2A). The reduction in BMD was significantly more pronounced after 15 weeks relative to 5 weeks on the diets (Figure 2B). To determine whether the change in BMD was reflected as changes in structural indices, microcomputed tomography (μCT) was used to analyze the bones from the 5-week-study mice. In agreement with the changes identified by DXA, μCT quantification of structural indices of lumbar spine also found significantly reduced bone volume/trabecular volume (BV/TV; %), decreased trabecular number (Tb.N; 1/mm) and thickness (Tb.Th; 1/mm³) and increased trabecular spacing (Tb.Sp; mm) from mice on the high relative to low phosphorus diet (Figure 2, C and D). Cortical indices from the mid-diaphysis of the femur were also quantified by μCT and in agreement with the trabecular indices, showed a significant decrease in cortical thickness (Ct.Th), total cross-sectional area (Tt.Ar), cortical area (Ct.Ar), as well as an increase in porosity (Ct.Po) from the mice fed the HPDs compared with LPDs (Figure 2, E and F).

Figure 2.

A, Femurs and vertebrae from the C57BL/6 mice fed HPD (1.8%) and LPD (0.2%) diets for 5 weeks were analyzed by DXA (n = 10). B, Percent change in femur BMD relative to LPD was calculated for the 5-week (n = 10) and 15-week (n = 7) mice fed HPD. Vertebrae from the C57BL/6 mice fed HPD and LPD for 5 weeks were analyzed by μCT. C, Images of representative vertebrae. D, Quantitative indices generated by μCT analysis; BV/TV (%);Tb.N, trabecular number (1/mm); Tb.Th, trabecular thickness (1/mm³); Tb.SP, trabecular spacing (mm) (N = 5). E, Images of representative femurs from the same mice as in panel C at the mid-diaphysis. F, Femurs were analyzed for cortical quantitative indices; cortical thickness (Ct.Th), total cross-sectional area (Tt.Ar), cortical area (Ct.Ar), and percent porosity (Ct.Po) (N = 9). Results are average ± SEM. *, P < .05 relative to LPD; #, P < .05 relative to 5-week HPD; Student t test.

Discussion

In this study, consumption of a diet rich in phosphorus-based food additives but stable for calcium for 1 week increased circulating FGF23, osteopontin, and osteocalcin concentrations relative to baseline values and decreased mean sclerostin concentrations in healthy individuals. Similar results were observed in animals fed diets with increased phosphorus content; findings that were accompanied by substantial decreases in BMD and structural indices. This is the first study to our knowledge to demonstrate that the enhanced phosphorus content of highly processed foods is sufficient to stimulate FGF23 secretion in healthy individuals. Further, this is the first study to demonstrate that high dietary phosphorus intake can stimulate osteopontin and decrease sclerostin concentrations in humans and animals.

Prior studies have shown that high phosphorus intake can stimulate FGF23 secretion in healthy individuals. However, these studies primarily involved loading healthy volunteers with oral phosphorus supplements (21–23), which does not take into account the effects of food processing or cooking on phosphorus-based additives found in highly processed foods. Moreover, these studies used phosphorus supplement doses meant to maximize stimulation of FGF23 secretion (≥ 2200 mg/d), and not to mimic levels of phosphorus additive intake more typical of Westernized diets. As such, it was unclear how faithfully these studies captured the effects of commercial food additives on FGF23 secretion in humans. Therefore, our finding that these additives can significantly increase FGF23 concentrations is a novel and important finding. Given the growing body of literature showing that elevated FGF23 concentrations associate with poor cardiovascular outcomes in community-dwelling adults (13)—perhaps via direct toxic effects on cardiac tissue (24)—these results provide new insights into potential adverse effects of high levels of phosphorus additive intake in the food supply.

Another novel finding of the current study was that 1 week of consuming foods with high phosphorus additive content was sufficient to stimulate osteopontin secretion. Osteopontin is a protein with pleiotropic actions including regulation of bone mineralization, insulin sensitivity, and immune function (12, 14). Studies have shown that bone lineage cells secrete osteopontin in response to stress-induced bone remodeling (25, 26). Given that high phosphorus intake has well-known adverse effects on bone metabolism, it is conceivable that the changes in osteopontin we observed were an early response to phosphorus-induced disturbances in bone mineralization or remodeling. In fact, the osteopontin null mouse is resistant to high phosphorus-induced bone loss identifying this circulating factor as a key intermediate in the systemic response to changes in dietary phosphorus intake (27). It is also possible that osteopontin secretion was directly stimulated by high phosphorus intake. We have previously shown that inorganic phosphorus can directly induce osteopontin gene expression in osteoblast-like cells as well as bone marrow stromal cells (28). Although the lack of changes in serum phosphorus concentrations in individuals consuming the additive-rich diet for 1 week would argue against direct induction of osteopontin expression by high phosphorus intake, we measured phosphorus concentrations at one time of day and, thus, were unable to detect potential increases in time-averaged serum phosphorus during the day. We have also recently demonstrated a synergistic effect of elevated phosphorus and FGF23 for the induction of osteopontin in cell culture (29) and therefore, short-term increases in serum phosphorus in the presence of elevated FGF23 may have a more pronounced effect. Given that high osteopontin is associated with a variety of nonskeletal adverse effects such as insulin resistance, tumorigenesis, and vascular calcification (14, 30), future studies should determine longer-term adverse systemic effects of additive-induced increases in osteopontin.

We also identified significant changes in serum osteocalcin and sclerostin. Sclerostin is a hormone that modulates bone mass by inhibiting the Wnt/β-catenin pathway, thereby inhibiting bone apposition (11). Decreases in sclerostin promote bone formation. As such, it is possible that sclerostin concentrations decreased in individuals after consuming the high-additive diet as an adaptive response to preserve bone mass in the setting of phosphorus-induced resorption. Our finding that high dietary phosphorus intake stimulates osteocalcin secretion is in agreement with at least two studies that detected an increase in serum osteocalcin in response to phosphorus loading in humans (31, 32). Future studies will need to determine whether the additive-induced changes in osteocalcin observed in the current study have a measurable effect on bone or metabolic health in humans.

We took advantage of our mouse model to investigate the potential source(s) of changes in circulating FGF23, osteopontin, and sclerostin. FGF23 and sclerostin are highly expressed by osteocytes that reside in the hydroxyapatite of bone and we therefore examined RNA expression in the femur. Surprisingly, we did not detect an increase in FGF23 expression. The results suggest that either a different tissue is mainly responsible for the serum changes in response to a high phosphorus diet or the increase is due to post-translational stability. In fact, FGF23 protein levels are known to be regulated by protein stability. Post-translational O-glycosylation inhibits cleavage of the protein and subsequent inactivation (33, 34). With regard to sclerostin, the general pattern of decreased serum levels and gene expression levels in the bone are consistent. The bone marrow is a complex mixture of cells containing cells types ranging from B and T cells to macrophages to pluripotent mesenchymal stem cells, many of which are known to express osteopontin. Here we identified the bone marrow as a potential source of changes in serum osteopontin levels based on increased gene expression, although this does not rule out contribution by other tissues such as the kidney. These results suggest that changes in dietary phosphorus consumption can influence gene expression in different tissues even with normal renal function. However, whether these changes are due to a direct effect or due to changes in systemic factors still must be determined.

Our study had limitations. The small sample size may have limited our ability to detect modest changes in a number of the markers of bone and mineral metabolism in humans that we measured. However, this is also a potential strength in that we were able to show statistically significant changes in FGF23, osteopontin, osteocalcin, and sclerostin despite this potential limitation. The length of intervention was relatively short, precluding us from being able to determine whether the observed changes were sustained over a longer period of exposure. It is possible that physiological adaptations to increased phosphorus intake might change over longer periods of exposure, a possibility that will need to be examined in future studies with longer periods of intervention. We only had a single phosphorus measurement during follow-up visits and so were unable to examine potential effects of the diets on diurnal variation of phosphorus. We had imbalances in race and sex, with most participants being African-American women.

In conclusion, phosphorus-based food additives significantly increased circulating FGF23, osteopontin, and osteocalcin concentrations relative to baseline values and significantly decreased circulating sclerostin concentrations. These results suggest that high phosphorus-additive intake may have adverse effects even in individuals with normal kidney function. In addition, the diet-induced endocrine changes were correlated in a mouse model of dietary phosphorus-modulated bone loss providing a means to investigate the underlying mechanisms.