Association of Calcium‐Phosphorus Product With Blood Pressure in Dialysis

Abstract

J Clin Hypertens (Greenwich). 2010;12:96–103. ^©2009 Wiley Periodicals, Inc.

Hypertension is very common in dialysis patients. Disorders of mineral metabolism have been linked to vascular calcification and hypertension in dialysis. Fifty‐four hemodialysis patients were included in a cross‐sectional study in a dialysis unit during a 6‐month period. Linear regression analysis was done between averages of calcium and phosphorus (ca × ph) product and blood pressures (BPs). Ca × ph was significantly associated with systolic BP predialysis (P=.03, R=0.28), diastolic BP predialysis (P=.001, R=0.44), predialysis mean arterial pressure (MAP) (P=.002, R=0.4), and diastolic BP postdialysis (P=.03, R=0.26). No relationship was found with pulse pressures. Multilinear regression analysis was then done between ca × ph product and BPs adjusting for age, sex, hemoglobin, diabetes, albumin, parathyroid hormone, ultrafiltration volume, and average BP medications per patient. There was a strong positive association with predialysis systolic BP (P=.003, R ²=0.49), predialysis MAP (P=.001, R ²=0.51), and postdialysis MAP (P=.02, R ²=0.65). No associations with pulse pressures were detected. The study findings suggest that ca × ph product is significantly associated with dialysis MAP and not pulse pressure. This is likely secondary to the stronger relationship with diastolic BP than with systolic BP. Prospective studies looking into the associated hemodynamic parameters related to arterial stiffness and endothelial dysfunction along with measures for calcifications would be very beneficial.

Hypertension is a major public health problem and is considered a major risk factor for cardiovascular disease. ¹ Hypertension is also the second most common etiology for end‐stage renal disease in the United States. ² Unfortunately the number of patients with hypertension is likely to grow with the aging population and it is expected to occur in more than half of the population older than 65 years. ³

Hypertension is very common among hemodialysis patients, with a reported prevalence from 60% to 86%. ⁴ , ⁵ Etiology of hypertension in dialysis is multifactorial. Volume overload, increased sympathetic activity, activation of the renin‐angiotensin system, and altered endothelial function play major roles. ⁶ Administration of erythropoietin for anemia has also been linked to hypertension. ⁷ Parathyroid hormone (PTH) and plasma calcium (ca) levels have also been associated with blood pressure (BP) in dialysis. ⁸ Phosphorus (ph) load has been shown to have a role on endothelial dysfunction by increasing production of reactive oxygen species and decreasing nitric oxide production via inhibitory phosphorylation of endothelial nitric oxide synthase. ⁹ Iseki and colleagues ¹⁰ showed that the prevalence of systolic or diastolic hypertension was significantly associated with volume excess and serum levels of albumin, ca, and phosphorous in chronic hemodialysis patients. In a cross‐sectional study, Chow and colleagues ¹¹ showed no correlation between pulse pressures (PPs), serum PTH level, and ca and ph (ca × ph) product in dialysis patients. PP, however, was directly associated with age and the presence of diabetes. In another cross‐sectional analysis of patients in the Hemodialysis (HEMO) study, Rocco and colleagues showed that diabetes, older age, increased BP medications, and lower hematocrit level were positively associated with systolic hypertension in dialysis patients. Diastolic hypertension, however, was associated with younger age and increased number of BP medications. ¹² In 1999, Marchais and colleagues ¹³ showed an association between serum phosphate and higher mean and diastolic BP (DBP) in dialysis. In a recent study by Huang and colleagues, elevated serum phosphate was associated with higher predialysis systolic BP (SBP) and PP. Each 1 mg/dL higher serum phosphate was associated with 1.77 mm Hg higher SBP. ¹⁴

Cardiovascular disease is still the leading cause of mortality in dialysis patients. ² Damage of large arteries is a major contributor to the high cardiovascular morbidity and mortality in end‐stage renal disease. Arterial stiffness is significantly affected by the amount of arterial calcifications. ¹⁵ , ¹⁶ Increase in stiffness raises SBP and decreases DBP, thus increasing PP. ¹⁵ Dialysis patients have stiffer arteries than age‐ and hypertensive‐matched nonuremic patients. ¹⁷ Arterial stiffness in dialysis has also been associated with mean arterial pressure (MAP). In a study of patients with chronic kidney disease, Covic and colleagues ¹⁸ showed a significant linear relationship between atherosclerosis burden and arterial stiffness measured through pulse wave velocity and augmentation index. Independent predictors of arterial stiffness, however, were age and mean arterial pressure. Vascular and coronary calcifications are also more common and progressive in end‐stage renal dialysis patients. ¹⁹

Arterial medial calcification was found to be a strong prognostic marker of all‐cause and cardiovascular mortality in dialysis, with the principal effect of calcification causing an increase in arterial stiffness. ¹⁵

Higher ca‐ph product and serum ph have been linked to risk of vascular calcification. ²⁰ , ²¹ No clear association was detected with PTH levels, but lower serum albumin was associated with higher odds ratios for calcification. ²⁰ Other studies have reported association of calcification with PTH levels, however. ²² Severity of calcification was also associated with patients’ age, SBP, and plasma phosphate concentrations. ²² Other studies linking calcification with BPs, however, have shown variable results. Goodman and colleagues ¹⁹ showed no significant differences in systolic or diastolic BPs among patients with or without calcifications. Adragao and colleagues ²³ have shown that higher vascular calcification scores in dialysis patients were significantly associated with mean arterial BPs. In a recent paper by Guillaume and colleagues in dialysis patients with vascular calcification, no relationship was found with BP. ²⁴ Guerin and colleagues ¹⁵ detected an association between vascular calcification and PP mainly due to a decrease in diastolic pressure, not an increase in systolic pressure.

We elected to study the relationship of hypertension with elements related to vascular stiffness and calcification, mainly associated with ca‐ph metabolism in dialysis patients. Due to scarce data related to ca‐ph product and hypertension, and due to varying reports linking SBP, DBP, PP, and MAPs to disorders of mineral metabolism in dialysis, we decided to include various ways of pressure measures and study the association. Our hypothesis is that there is a significant relationship between ca‐ph product and BP in dialysis patients.

Patients and Methods

Fifty‐four hemodialysis patients were included in a cross‐sectional study in a dialysis unit during a 6‐month period from May 2008 to October 2008. A sample size of 54 was determined based on the available data to the investigator at the time of the study. Patients were chosen from one dialysis unit and the investigator included only his patients in this study. The study was adequately powered (0.85) with a significance level of α=0.05 based on a correlation coefficient of 0.4, derived from previous regression‐based studies linking serum ca‐ph with BP, left ventricular mass index, and valvular calcification. ¹³ , ²⁵ , ²⁶ , ²⁷

Predialysis and postdialysis BPs and amount of fluid removal are available on records for each dialysis treatment. Trained technicians digitally measured BPs with standard cuffs incorporated into the dialysis machines for each individual patient. Serum ca and ph were routinely measured per dialysis protocol once or twice a month. Serum albumin and PTH levels were usually measured once a month; Hemoglobin was measured on average twice a month. All data were recorded and available in the patient’s individual charts. Data used in the analysis was completely de‐identified/de‐linked from identifiers.

BP measures were averaged in each of these patients during the 6‐month period. Measures included predialysis SBP and postdialysis SBP, predialysis DBP, postdialysis DBP, predialysis MAP, postdialysis MAP, predialysis PP, and postdialysis PP.

Age, sex, and diabetic status were also included in the analysis. Six‐month averages of serum ca, ph, ca‐ph product, PTH, albumin, and hemoglobin were calculated. Average amount of fluid removed in dialysis per patient was also determined. Average number of BP medications per patient was also incorporated into the analysis.

Statistical Analysis

Linear regression analysis was done to measure the association of predialysis and postdialysis PP, MAP, and systolic and diastolic pressures with ca‐ph product. Multilinear regression analysis was then performed between predialysis and postdialysis systolic, diastolic, mean, and pulse pressures with ca‐ph products adjusting for albumin, PTH, ultrafiltration volume, diabetic status, hemoglobin, age, sex, and average number of BP medications used per patient. Two‐sample Student t test was performed to compare averages of MAPs and PPs. Statistical analyses were performed using STATA software version 10.1 (Stata Corporation, College Station, TX). Power analysis was performed using SigmaStat software version 3.5 (Systat Software, San Jose, CA).

Results

Baseline characteristics of patients are listed in Table I. The average age of the study population was 58.6 years, 59% were women, 81.5% were African American, and diabetes was present in 57% of patients. Average duration of dialysis was 3.8 years. Each patient used an average of 2.6 BP medications.

Table I.

 Baseline Characteristics of Study Patients

Total No. of Patients	54
Average age, y	58.62 (15.53)a
Sex, %
Female	59
Male	41
Race, %
Black	81.5
White	18.5
Diabetes, %
Present	57
Absent	43
Average duration of dialysis at time of study, y	3.83 (2.28)a
Average blood pressure medications per patient	2.62 (1.41)a

^aData presented as mean (SD).

In the study patients, the average PP before dialysis was 76 mm Hg, and the average PP after dialysis was 69 mm Hg. The average MAP before dialysis was 103 mm Hg, whereas the average MAP after dialysis was 94 mm Hg. The average fluid removed per dialysis session was 2.86 L (range, 0.7–4.7 L), and the average ca‐ph product was 51.4 (range, 26.5–92.3).

Using linear regression analysis, there was no association between PP predialysis or postdialysis and ca‐ph product. When MAP was analyzed, there was a statistically significant positive relationship between MAP predialysis and ca‐ph product (P=.002, R=0.4) (Figure 1). No association was detected with MAP postdialysis, however.

Figure 1.

 Linear regression between predialysis pressures and calcium‐phosphorus (ca × ph) for mean arterial pressure (MAP) and pulse pressure (PP).

Studying for individual BP parameters, a positive association was found between SBP predialysis and ca‐ph product (P=.03, R=0.28) (Figure 2). No relationship was detected with SBP postdialysis. A strong positive association was found between DBP predialysis and ca‐ph product (P=.001, R=0.44), and a weaker association was detected with DBP postdialysis (P=.03, R=.26) (Table II).

Figure 2.

 Linear regression between predialysis pressures and calcium‐phosphorus (ca × ph) for systolic blood pressure predialysis (SBPpre) and diastolic blood pressure predialysis (DBPpre).

Table II.

 Linear Regression Between Ca × Ph Product and Blood Pressures

	Ca × Ph
	P Value	R	Coefficient
MAPpre	.002a	0.40a	+0.36
MAPpost	.157	0.19	+0.16
Pulsepre	.93	0.01	+0.011
Pulsepost	.29	0.14	−0.13
SBPpre	.03a	0.28a	+0.37
SBPpost	.49	0.095	+0.12
DBPpre	.001a	0.43a	+0.35
DBPpost	.03a	0.29a	+0.2

Abbreviations: Ca × Ph, calcium‐phosphorous product; DBPpost, diastolic blood pressure postdialysis; DBPpre; diastolic blood pressure predialysis; MAPpost, mean arterial pressure postdialysis; MAPpre, mean arterial pressure predialysis; Pulsepost, pulse pressure postdialysis; Pulsepre, pulse pressure predialysis; SBPpost, systolic blood pressure postdialysis; SBPpre, systolic blood pressure predialysis. ^aStatistical significance.

Average MAPs predialysis were then compared among patients with ca‐ph products <50 with those with a product >50. Among the patients with ca × ph >50, the average predialysis MAP was 107. MAP dropped to 100 in patients with ca × ph <50. This difference was statistically significant using 2‐sample t test analysis (P=.02) (Table III).

Table III.

 Two‐Sample t test Comparing Averages in Patients With Ca × Ph > or <50

	Ca × Ph >50	Ca × Ph <50	Difference	P Value
MAPpre	106.88	99.98	6.90	.02a
MAPpost	93.67	94.4	−0.72	.51
Pulsepre	77.88	74.24	3.64	.17
Pulsepost	67.37	71.08	−3.7	.84

Abbreviations: Ca × Ph, calcium‐phosphorous product; MAPpost, mean arterial pressure postdialysis; MAPpre, mean arterial pressure predialysis; Pulsepost, pulse pressure postdialysis; Pulsepre, pulse pressure predialysis. ^aSignificant difference detected in MAPpre among patients with ca × ph >50 compared with patients with ca × ph <50.

When PPs where compared (ca × ph >50, PP=77; ca × ph <50, PP=74) the difference was not statistically significant (P=.17). No significant differences were detected among mean pressures and PPs in the two subgroups postdialysis.

When patients were divided into two other subgroups based on ca × ph ≥45, there were no significant differences noted between their average MAPs or PPs.

Multilinear regression analysis was then performed to compare predialysis and postdialysis MAPs and PPs with ca‐ph product, adjusting for age, sex, hemoglobin, albumin, PTH, ultrafiltration volume (reflecting fluid gains in between dialysis), diabetic status, and average number of BP medications used per patient (Table IV).

Table IV.

 Multilinear Regression Analysis Between Blood Pressures and Ca × Ph, Adjusting for Age, Sex, Diabetes, Hemoglobin, Parathyroid Hormone, Albumin, Ultrafiltration Volume, and Average Blood Pressure Medications per Patient

	Ca × Ph
	P Valuea	R  2
MAPpre	.001b (0.42)	0.51b
MAPpost	.02b (0.22)	0.65b
Pulsepre	.143	0.50
Pulsepost	.939	0.55
SBPpre	.003b (0.55)	0.49b
SBPpost	.08	0.6
DBPpre	.002b (0.36)	0.54b
DBPpost	.009b (0.22)	0.63b

Abbreviations: Ca × Ph, calcium‐phosphorous product; DBPpost, diastolic blood pressure postdialysis; DBPpre; diastolic blood pressure predialysis; MAPpost, mean arterial pressure postdialysis; MAPpre, mean arterial pressure predialysis; Pulsepost, pulse pressure postdialysis; Pulsepre, pulse pressure predialysis; SBPpost, systolic blood pressure postdialysis; SBPpre, systolic blood pressure predialysis. ^aNumbers in parentheses represent coefficient. ^bRepresent statistical significance.

MAP predialysis was very strongly positively associated with ca‐ph product adjusting for the other independent variables (P=.001, R ²=0.51). Of note was the strong association of hemoglobin (P=.001) and average number of BP medications (P=.017) with predialysis MAP. No association was detected with albumin, PTH, age, and sex.

When MAP postdialysis was analyzed, there was a significant positive association with ca × ph product (P=.02, R ²=0.65). Ultrafiltration volume, BP medications, and hemoglobin were also significantly associated with MAP postdialysis (P=<.0001, <.0001, and .003, respectively). No relationship was detected with albumin, PTH, age, sex, or diabetic status.

In the multilinear regression analysis there was also no relationship between PP predialysis and postdialysis with ca‐ph product (Table IV).

PP predialysis was only positively associated with age, diabetes, and BP medications (P=.002, .001, and .03, respectively.) Postdialysis PP was also related to age, diabetes, and BP medications (P=.007, .022, and .001, respectively).

When individual BP parameters were evaluated in the multilinear regression, there was a strong positive relationship between predialysis SBP and ca × ph. (P=.003, R ²=0.49). No association was detected with postdialysis SBP. Diastolic pressures were strongly associated with ca × ph in the multilinear model (DBP predialysis, P=.002, R ²=0.54 and DBP postdialysis, P=.009, R ²=0.63).

To try to differentiate between the effects of the ca × ph vs serum ph on BP, we ran a multiple linear analysis between various BPs and serum ca and serum ph (excluding ca × ph), adjusting for age, sex, ultrafiltration volume, BP medications, hemoglobin, albumin, PTH, and diabetic status. There were significant positive relationships between serum ph and SBP predialysis (P=.004, R ²=0.5), DBP predialysis (P=.001, R ²=0.55), DBP postdialysis (P=.007, R ²=0.64), MAP predialysis (P=.001, R ²=0.52), and MAP postdialysis (P=.019, R ²=0.66). No relationships were detected between SBP postdialysis, PPs, and serum ph levels.

Discussion

In this 6‐month cross‐sectional study, we demonstrated a significant association between BP and ca‐ph product in dialysis patients. In a linear regression analysis, ca‐ph product was positively associated with predialysis SBP, DBP, and MAP. Postdialysis, only DBP was related to ca‐ph product. No relationship was detected with PP.

In the multiple linear analysis, ca‐ph product was positively related to predialysis SBP, DBP, and MAP. Postdialysis, the product was only associated with MAP and DBP. No relationship was detected with PPs. The strong positive association with SBP predialysis, MAP predialysis, DBP predialysis, DBP postdialysis, and MAP postdialysis persists only when serum ph was used in the regression equation (excluding ca × ph), pointing to the independent role of ph in the physiologic process.

Significant differences in MAPs where detected among patients whose ca‐ph products were >50 compared with patients whose products were <50.

Serum ph has been associated with vascular calcification in chronic kidney disease. ²⁸ This effect is mediated via phenotypic changes in vascular smooth muscle cells in part via osteogenic factors osteocalcin and Cbfa‐1. ²⁹ Serum ph and ca‐ph product were also found to be higher in dialysis patients with calcifications. There was also an increase in relative risk of arterial medial calcifications with increasing ph levels. ²⁰ Studies have shown a correlation between elevated ph levels in dialysis and mortality. This association has been attributed to the roles of ph and elevated ca‐ph products in vascular calcification. ³⁰

Arterial calcifications contribute to arterial stiffness. The increase in stiffness causes an elevation in SBP and a decrease in DBP, thus an increase in PP. ¹⁵ , ¹⁶ Other studies, however, have shown a significant association between arterial stiffness and MAP. ¹⁸ , ³¹

There are only a few studies evaluating the relationship between serum ph and BP in dialysis. No studies have detected an association between ca‐ph product and BP in dialysis. Chow and colleagues ¹¹ showed no correlation between PPs, serum PTH level, and ca‐ph product in dialysis patients. Volume excess and serum levels of albumin, ca, and ph in chronic hemodialysis patients were associated with systolic or diastolic hypertension in one study. ¹⁰ In 1999, Marchais and colleagues ¹² showed an association between serum ph and higher mean and diastolic BP in dialysis. In a recent study by Huang and colleagues, ¹⁴ elevated serum ph was associated with higher SBP, DBP, and PP. Klassen and colleagues ³² reported that ph concentration was directly associated with PP, which was a strong predictor of mortality.

PP is an index of the pulsatile component of the cardiac cycle. It is governed by the relationship between ventricular ejection and arterial stiffness that is determined by the viscoelastic effects of the large arteries and the effect of arterial wave reflection from the periphery back to the central arteries. ³²

Patients with end‐stage renal disease exhibit many vascular abnormalities contributing to elevated PP, including increased arterial stiffness, pulse wave velocity, and early wave reflection. ³³ , ³⁴ As previously discussed, arterial calcifications associated with elevated ca‐ph product ²⁰ contribute to arterial stiffness, causing an increase in PP. ¹⁵ , ¹⁶ Other studies, however, have shown that higher vascular calcification scores in dialysis patients were significantly associated with mean arterial BPs. ²³ Some studies even showed no significant differences in SBPs or DBPs among patients with or without calcifications. ¹⁹ , ²⁴ Major determinants of MAP are ventricular ejection and peripheral vascular resistance. In the presence of a normal cardiac output, MAP reflects static resistance to blood flow by arterioles during diastole. ³⁵ Marchais and colleagues have shown that hyperphosphatemic dialysis patients were characterized by higher diastolic and mean BPs. These patients had a “hyperkinetic” circulation associated with high cardiac output due to increased stroke volume and heart rate. There was also elevation in tensile strength of arteries (measured in the common carotid artery lower wall to lumen ratio), ¹³ a mechanism possibly related to arterial wall hypertrophy by metastatic calcifications of the arterial media or by activation of fibroblasts and development of smooth muscle vascular fibrosis. ³⁶

In this study, we demonstrated a significant positive association between MAPs in dialysis patients with the ca‐ph product. No association was detected with PPs. Absence of this association was also demonstrated by Chow and colleagues. ¹¹ The lack of relationship of ca‐ph product with PP in our study is likely related to the stronger association with diastolic pressure and the weaker association with systolic pressure (predialysis). The stronger association with MAP and DBP could be related to the effect of ca × ph or serum ph on increasing peripheral vascular resistance. Ph has been shown to increase production of reactive oxygen species and decrease nitric oxide production via inhibitory phosphorylation of endothelial nitric oxide synthase, thus inhibiting endothelium‐dependent vasodilation. ⁹

In our study, the subgroup with ca‐ph product >50 had higher average predialysis MAP than the patients whose ca × ph was <50. This difference was not detected if the groups were separated with ca × ph > or <45. According to the National Kidney Foundation Kidney Disease Outcome Quality Initiative, the ca × ph product should be <55. This is based on evidence linking higher product with increased risk of calcifications. ²¹ The findings in our study thus suggest that in the subgroup with ca × ph >50, there is a significant link between increased risk of calcification and hypertension.

The positive relationship between age and PP demonstrated in this study has been proven in other analyses. ³² , ³⁵ In addition, the negative association between age and diastolic pressure has also been shown in other studies. ¹² Age and BP data essentially reflect the effect of age on vascular stiffness and elastic recoil as previously discussed. ³⁷

The strong association between diabetes and PP in our study has also been demonstrated in other data, ¹¹ , ³² , ³⁵ again highlighting the importance of diabetes on cardiovascular risk and physiology. Hemoglobin in our study was positively associated with predialysis MAP and predialysis diastolic pressure. Since no adjustment to the erythropoietin dose was made in our study, BP increase could have reflected the effect of increasing erythropoietin doses since higher levels correlate with more hypertension. ³⁵ , ³⁸ In other studies, hemoglobin concentration was found to be negatively associated with BP in dialysis, reflecting hyperdynamic circulation associated with increased stroke volume. ¹² , ³² , ³⁵

An important confounding factor associated with elevated ca‐ph product and hypertension in this study is noncompliance to dialysis treatments, diet, and medications (ph binders and antihypertensives). It can be argued that noncompliant dialysis patients are the ones who are the most fluid gainers. These patients are also not adherent to ph restriction and medication intake. We accounted for nonadherence and fluid gains by adjusting for BP medications and ultrafiltration volume in the multilinear regression analysis.

The relationship between MAP, SBP predialysis, and ca‐ph product persisted after adjusting for these confounding variables. Other confounding factors that could have been adjusted for include vitamin D analog use and type of ph binder use (especially with the association of ca‐based binders and increased calcifications in some studies). ³⁹

Limitations

This study has several limitations. It is an observational cross‐sectional study. No direct causality could be established although several features suggest a causal association, including strength of association and biological plausibility. Sample size limitation is related to the available data for the study by the investigator. Most of the patients in this study were African American; therefore, different pathophysiologic features could occur in this population. In addition, no interdialytic BP readings were included in the data analysis, although interdialytic readings may be more accurate hemodynamic determinants in dialysis. ⁴⁰

Conclusions

In this study we demonstrated a strong association between ca × ph, and serum ph with MAP before and after dialysis. There was also an independent association of ca × ph and ph with predialysis systolic and diastolic BPs and postdialysis diastolic pressure. No relationship was demonstrated with predialysis or postdialysis PPs. The importance of the relationship between ca × ph and hypertension seem to occur when the product is >50, reflecting the role of vascular calcification in the physiologic process.

The role of ca‐ph product and ph with hypertension, arterial stiffness, and vascular compliance requires more investigation. More is needed with regard to the possible connection of the product to cardiac output and systemic vascular resistance, especially in connection to endothelial dysfunction. Prospective studies looking into the relationships of ca‐ph product and ph with BP in dialysis, including hemodynamic parameters dealing with arterial stiffness, compliance, systemic vascular resistance, and endothelial function, along with measures for calcifications, would be very beneficial.