Association of serum phosphorus and calcium x phosphate product with mortality risk in chronic hemodialysis patients: A national study

Association of Serum Phosphorus and Calcium ⴛ Phosphate Product With Mortality Risk in Chronic Hemodialysis Patients: A National Study Geoffrey A. Block, MD, Tempie E. Hulbert-Shearon, MS, Nathan W. Levin, MD, and Friedrich K. Port, MD, MS ● Elevated serum phosphorus is a predictable accompaniment of end-stage renal disease (ESRD) in the absence of dietary phosphate restriction or supplemental phosphate binders. The consequences of hyperphosphatemia include the development and progression of secondary hyperparathyroidism and a predisposition to metastatic calcification when the product of serum calcium and phosphorus (Ca ⴛ PO4) is elevated. Both of these conditions may contribute to the substantial morbidity and mortality seen in patients with ESRD. We have analyzed the distribution of serum phosphorus in two large national, random, cross-sectional samples of hemodialysis patients who have been receiving dialysis for at least 1 year. Data were obtained from two special studies of the United States Renal Data System, the Case Mix Adequacy Study (1990) and the Dialysis Morbidity and Mortality Study Wave 1 (1993). The relative risk of death by serum phosphorus quintiles is described after adjusting for age at onset of ESRD, race, sex, smoking status, and the presence of diabetes, the acquired immunodeficiency syndrome, and/or neoplasm. Logistic regression analysis is then used to describe the demographic, comorbid, and laboratory parameters associated with high serum phosphorus. Serum phosphorus was similar in these two study populations and averaged 6.2 mg/dL. Ten percent of patients had levels greater than 9 mg/dL and at least 30% of each group had serum phosphorus levels greater than 7 mg/dL. The adjusted relative risk of death by serum phosphorus level was not uniform across all quintiles, being constant below a level of 6.5 mg/dL and increasing significantly above this level. The relative risk of death for those with a serum phosphorus greater than 6.5 mg/dL was 1.27 relative to those with a serum phosphorus of 2.4 to 6.5 mg/dL. This increased risk was not diminished by statistical adjustment for coexisting medical conditions, delivered dose of dialysis, nutritional parameters, or markers of noncompliance. Evaluation of predictors of serum phosphorus greater than 6.5 mg/dL revealed in multivariate analysis that younger age at onset of ESRD, female sex, white race, diabetes, active smoking, and higher serum creatinine levels were all significant predictors. Analysis of serum calcium revealed no correlation with relative risk of death. The Ca ⴛ PO4 product, however, showed a mortality risk trend similar to that seen with serum phosphorus alone. Those in the highest quintile of the Ca ⴛ PO4 product (G72 mg2/dL2) had a relative mortality risk of 1.34 relative to those with products of 42 to 52 mg2/dL2. The relative mortality risk by log parathyroid hormone (PTH) level was elevated for patients with higher levels, but the mortality risk associated with hyperphosphatemia was independent of PTH. For hemodialysis patients who have been receiving dialysis for at least 1 year, we conclude that a large percentage have a serum phosphorus level above 6.5 mg/dL and that this places them at increased risk of death. This increased risk is independent of PTH. The mechanism(s) responsible for death is unknown, but may be related to an abnormally high Ca ⴛ PO4 product. Although mechanisms are not clearly established, this study supports the need for vigorous control of hyperphosphatemia to improve patient survival. r 1998 by the National Kidney Foundation, Inc. INDEX WORDS: Phosphorus; PTH; calcium ⴛ phosphate product; hemodialysis; mortality.

D

UE TO THE diminished ability of the kidneys to excrete a phosphorus load as glomerular filtration decreases, most patients with end-stage renal failure have a predisposition toward elevated levels of serum phosphorus. Even with high-efficiency, high-flux membranes, hemodialysis falls far short of completely removing the dietary load of phosphorus.1,2 As a result, supplemental measures including dietary restriction and/or the use of phosphate binders must be prescribed in an effort to normalize serum phosphorus. Each of these treatments, however, has potential side effects. Aggressive restrictions on dietary intake must be balanced against the importance of maintaining adequate nutrition, particularly since malnutrition is a common conse-

quence of advanced renal failure. If nutritional intake is adequate, there is almost always a requirement for phosphate binders if serum phosphorus is to be controlled appropriately.2 Phosphate binders currently in use are limited to

From the US Renal Data System and the Departments of Medicine and Epidemiology, University of Michigan, Ann Arbor, MI; and the Department of Medicine, Beth Israel Medical Center, New York, NY. Received March 25, 1997; accepted in revised form October 10, 1997. Address reprint requests to Friedrich K. Port, MD, MS, Kidney Epidemiology and Cost Center, 315 W Huron Ave, Suite 240, Ann Arbor, MI 48103. E-mail: [email protected]

r 1998 by the National Kidney Foundation, Inc. 0272-6386/98/3104-0006$3.00/0

American Journal of Kidney Diseases, Vol 31, No 4 (April), 1998: pp 607-617

607

608

either calcium- or aluminum-containing compounds. Calcium salts may contribute to the development of hypercalcemia, while aluminum hydroxide is known to be potentially toxic in even small doses.3 Newer agents, such as a cationic polymer, may avoid these risks.4 The recent discovery that secondary hyperparathyroidism can be effectively treated with supplemental vitamin D3 has led to an increased use of this therapy. This may potentially aggravate the tendency toward hyperphosphatemia by stimulating increased intestinal absorption of phosphorus. Efforts to develop synthetic analogues of vitamin D3 that suppress parathyroid hormone (PTH) and that do not stimulate intestinal vitamin D receptors remain experimental at this time, but offer a promising alternative for the future. The result of this complex interplay of factors is that a substantial number of patients with end-stage renal disease (ESRD) continue to have elevated levels of serum phosphorus. The development and progression of secondary hyperparathyroidism is the primary consequence of this elevated serum phosphorus.5 An additional consequence is a predisposition to metastatic calcification when the Ca ⫻ PO4 product is elevated.6 Both of these may contribute to the high morbidity and mortality of patients with end-stage renal failure. The former results in the development of renal osteodystrophy and exposes the patient to excessively high serum levels of PTH, which may itself be a uremic toxin.7 The latter often results in calcification of soft tissues, joints, blood vessels, and internal viscera, such as myocardium, lung, liver, and kidney.6,8-10 It is certainly conceivable that vascular and cardiac calcification in particular lead to complications and increased mortality. The purpose of this study was to assess the level to which serum phosphorus is maintained in two large, national, random samples of patients who have been receiving hemodialysis for at least 1 year. Data from the US Renal Data System (USRDS) Case Mix Adequacy Study (CMAS) and the Dialysis Morbidity and Mortality Study Wave 1 (DMMS) were used to describe the distribution of serum phosphorus and also to identify the demographic, clinical, and laboratory variables associated with poor control of serum phosphorus. In addition, the relationship of serum phosphorus and mortality risk was

BLOCK ET AL

assessed while adjusting for comorbid and other factors. The potential contributions of serum calcium, the Ca ⫻ PO4 product, and PTH to the relationship between serum phosphorus and mortality were also examined. MATERIALS AND METHODS The data used for this analysis were from two USRDS special studies: the CMAS and the DMMS. Both studies represent a random national sample of prevalent chronic hemodialysis patients in the United States.

Data Sources The CMAS is a USRDS special study in which data were collected to study the relationship of dose of dialysis and dialyzer characteristics with patient outcomes. A systematic sampling methodology (based on dialysis unit size) was used to identify a sample of dialysis units, from which patients were subsequently randomly selected (proportional to unit size) based on the last two digits of their Social Security number. This process was undertaken to ensure patient and provider samples that were nationally representative. A random sample of 7,096 hemodialysis patients who were alive on December 31, 1990, was drawn from these 523 hemodialysis facilities. Data were abstracted from the patients’ dialysis facility medical records by the 18 ESRD Networks using a data form11 that was developed and tested by the USRDS Coordinating Center in cooperation with the National Institute of Diabetes and Digestive and Kidney Diseases and the Health Care Financing Administration. The study start date was December 31, 1990, for patients who developed ESRD before January 1, 1990. To yield a patient sample in which residual renal function was likely minimal or absent, we excluded patients who started dialysis within 12 months of the study start date (N ⫽ 1,746). Data were abstracted between April 1, 1992, and March 31, 1993, and included patient characteristics, information on patient history, the presence or absence of a variety of comorbid health conditions occurring within 10 years before the study start date, and laboratory data obtained just before the study began. For the present analyses, the follow-up period for each patient was up to 2 years from the start of the study. The DMMS is also a historic prospective study in which data were collected to study dialysis prescription, dialyzer reuse, anemia, nutrition, and vascular access. A random sample of 550 dialysis units was selected. A random sample of hemodialysis patients treated in these units on December 31, 1993, was selected. Data were abstracted on 5,670 in-center hemodialysis patients from the patients’ medical records by the dialysis facility personnel using data forms11 developed by the USRDS Coordinating Center and tested by the ESRD Networks. The study start date was December 31, 1993. Again, to yield a patient sample in which residual renal function was likely minimal or absent, we excluded patients who started dialysis within 12 months of the study start date (N ⫽ 2,134). In addition, because of some data collection problems, patients from one Network were excluded from the DMMS dataset (N ⫽ 580). Data were abstracted between March 1995 and July 31, 1995, for

PHOSPHORUS AND MORTALITY IN HEMODIALYSIS

essentially the same items as in the CMAS study. For these analyses, each patient was also monitored for up to 2 years from the start of the study. The USRDS database was linked to the CMAS and DMMS datasets to abstract supplemental information on race, primary cause of ESRD, and date of death. Since the USRDS database includes only Medicare patients, some patients (approximately 7%) in the DMMS could not be studied past the date of data abstraction (1- to 2-year follow-up period). Patients were excluded if they did not undergo dialysis treatments three times per week (NCMAS ⫽ 658, NDMMS ⫽ 314) or if bicarbonate was not used as the dialysate bath (NCMAS ⫽ 939, NDMMS ⫽ 175). In addition, patients were excluded if more than half of the variables on comorbid conditions were missing (NCMAS ⫽ 86, NDMMS ⫽ 43), if information was missing for phosphorus (NCMAS ⫽ 230, NDMMS ⫽ 218), serum albumin (NCMAS ⫽ 292, NDMMS ⫽ 371), or age (NCMAS ⫽ 4, NDMMS ⫽ 8); and if the number of days at risk could not be calculated (NCMAS ⫽ 24, NDMMS ⫽ 0). Calcium and PTH information was collected in the DMMS but not in the CMAS. Patients in the DMMS with missing calcium information were also excluded from the analyses (N ⫽ 210). Some patients were excluded for more than one reason. After these exclusions there were 6,407 patients included in these analyses (NCMAS ⫽ 3,738, NDMMS ⫽ 2,669).

Analytical Methods For variables indicating the presence of comorbid conditions, missing values were coded as not present and suspected conditions were coded as present. Missing values for creatinine, hematocrit, dose of dialysis (Kt/V), and body mass index (BMI) were set to the mean for patients in the corresponding study. Cox proportional hazards regression techniques were used to estimate the relationship of serum phosphorus and number of days to death. Patients were censored (that is, removed from analysis) at the time of transplantation, 60 days after a switch to peritoneal dialysis, or study completion date (usually 2 years after study start), whichever was earliest. Main analyses were adjusted for age at onset of ESRD, race, sex, active smoking, and the presence of diabetes, neoplasm, or the acquired immunodeficiency syndrome (AIDS). In one model, serum phosphorus was entered as a continuous variable (0.1 mg/dL increments). Another model used categorical variables with patients grouped into quintiles by serum phosphorus. Having determined an elevated risk for serum phosphorus above 6.5 mg/dL, it was entered as a categorical variable, with patients divided into a high phosphorus group (⬎6.5 mg/dL) and a reference phosphorus group (2.4 to 6.5 mg/dL). For this last model, patients with serum phosphorus in the lowest first percentile (serum phosphorus ⱕ2.3 mg/dL) were excluded (N ⫽ 67). These models were fit for patients in each study separately as well as for patients in the two studies combined. Similar models were fit predicting mortality with calcium and the Ca ⫻ PO4 product (in mg2/dL2 ) as well as corrected calcium and the corrected Ca ⫻ PO4 product with patients grouped by quintiles. Corrected calcium (in mg/dL) was

609

defined as corrected calcium ⫽ calcium if serum albumin ⱖ 4 g/dL and corrected calcium ⫽ calcium ⫹ 0.8 ⴱ (4 ⫺ serum albumin) if serum albumin ⬍ 4 g/dL. The 5corrected6 Ca ⫻ PO4 product (in mg2/dL2 ) used the 5corrected6 calcium level multiplied by the serum phosphorus level (each in mg/dL). The models involving calcium included only patients from DMMS because calcium information was not available in the CMAS. For the models with corrected calcium, additional analyses were conducted, adding serum albumin to the covariate list already mentioned. An interaction between serum phosphorus and the presence of diabetes was evaluated. Analyses to determine the predictors of an elevated serum phosphorus were completed using logistic regression while adjusting for age at onset of ESRD, race, sex, active smoking, BMI, inability to ambulate, dialysis sessions skipped (ⱖ1/month), serum albumin, serum creatinine, delivered dose of dialysis, and the presence of diabetes, neoplasm, or AIDS. For these analyses, patients with serum phosphorus in the lowest first percentile (serum phosphorus ⱕ2.3) were again excluded, and the odds of a patient’s having a phosphorus level greater than 6.5 mg/dL (compared with 2.4 to 6.5 mg/dL) were estimated. Sensitivity analyses were performed using Cox proportional hazards techniques looking for modifiers of the relative risk of death and serum phosphorus. In this series of models, covariates were added to adjust for the presence of coronary heart disease (including a past history of abnormal coronary angiography, angioplasty, or coronary artery bypass grafting), nutritional parameters (including serum albumin, serum creatinine, and BMI), the presence of peripheral vascular disease (including the presence of claudication, absent foot pulses, or limb amputation), the presence of cerebrovascular disease, a marker of noncompliance (ie, having missed one or more dialysis session during the first month of the study), and delivered dose of dialysis. These added covariates are important mortality risk factors according to prior analyses.11 Delivered dose of dialysis was calculated as Kt/V using the following formula12: Delivered Kt/V ⫽ ⫺ln (R ⫺ 0.008 ⴱ t) ⫹ (4 ⫺ 3.5R) ⴱ UF/W where R ⫽ postdialysis/predialysis blood urea nitrogen, t ⫽ dialysis hours, UF ⫽ predialysis-postdialysis weight change, and W ⫽ postdialysis weight. The average value of several Kt/V readings (average, three per patient) over a 6-month period near the start date of the study was used in these analyses. To better judge the appropriateness of dividing patients into two groups, a Cox model was fit similar to the main model, with patients grouped by deciles of serum phosphorus rather than quintiles. This finer specification was intended to allow for a better determination of the cut-point for the two-group analyses and to check for possible increased mortality risk in the lowest range of serum phosphorus. To

610

BLOCK ET AL

test the sensitivity of the two-group models further, which categorized patients as having high or average phosphorus, to the exclusion of the patients with serum phosphorus levels ⱖ2.3 mg/dL, the models were fit without the exclusion. To test the sensitivity of these models to the phosphorus cutpoint chosen, the models were fit using 5 and 6 mg/dL as the cut-point. PTH data were available for 2,087 of the DMMS patients. Several sensitivity analyses were conducted on this subgroup of patients to determine the role of PTH in the relationship between hyperphosphatemia and increased mortality risk. A Cox model was fit, adjusting for age at onset of ESRD, race, sex, active smoking, and presence of diabetes, neoplasm, or AIDS, and including PTH as a categorical variable with patients grouped by quintile of PTH. Two similar models were fit with either phosphorus or Ca ⫻ PO4 product added to the covariates in the first model. These three models were also fit with a continuous specification for log PTH.

RESULTS

Demographic, comorbidity, and laboratory characteristics of the study population are shown in Table 1. The mean age at onset of ESRD in the 6,407 study patients was 53 years. Fifty-three percent of the patients were white, 50% were male, 30% reported diabetes as the cause of ESRD, and an additional 9% had diabetes as a comorbid condition. (These percentages differ from reported data on prevalent patients11 because patients starting ESRD replacement therapy within 1 year of the study start date were excluded here.) The mean serum phosphorus was 6.2 mg/dL. The distribution of serum phosphorus was similar in the two study populations, as shown in Fig 1. At least 30% of each study population had serum phosphorus levels greater than 7 mg/dL and roughly 10% had serum phosphorus greater than 9 mg/dL. Approximately 12% of patients had serum phosphorus less than 4 mg/dL, with 1% of patients having a level less than 2.4 mg/dL in each study group. The overall relative mortality risk associated with serum phosphorus as a continuous variable adjusting for age at onset of ESRD, race, sex, and the presence of diabetes, smoking, AIDS, and neoplasm was 1.06 per 1 mg/dL higher serum phosphorus. This suggests a 6% higher mortality risk for each 1 mg/dL higher serum phosphorus, eg, for patients with a phosphorus level of 8 mg/dL versus similar patients with a phosphorus level of 7 mg/dL, or 6 mg/dL versus 5 mg/dL. A more detailed examination of this

Table 1. Baseline Variables for the Study Population in DMMS Wave I and CMAS Combined (N ⴝ 6,407 [N ⴝ 2,669 for Calcium Information])

Variable

Demographics Mean age at onset of ESRD (yr) Race (% white) Gender (% male) Cause of ESRD (% diabetes) Mean duration of ESRD at study start date (yr) Comorbid conditions (% yes or suspected) Diabetes (history and/or nephropathy) Coronary heart disease (history)* Left ventricular hypertrophy (history) Congestive heart failure (history) Peripheral vascular disease (history)† Cerebrovascular disease (history) Chronic obstructive pulmonary disease (history) Neoplasm (history)‡ Smoking (active) AIDS Mean BMI (kg/m2) Laboratory values Albumin (g/dL) Creatinine (mg/dL) Hematocrit (%) Phosphorus (mg/dL) Calcium (mg/dL) Dialysis dose and compliance Delivered dose (Kt/V) Skipping dialysis ⱖ1/mo (%)

Mean ⫾ SD or %

53 ⫾ 16.6 53 50 30 4.5 ⫾ 3.7 39 45 40 43 26 13 12 9.5 19 0.3 24.5 ⫾ 5.3 3.8 ⫾ 0.40 11.6 ⫾ 3.6 29.8 ⫾ 4.6 6.2 ⫾ 2.1 9.4 ⫾ 1.0 1.16 ⫾ 0.22 8.8

NOTE. Variables shown in bold are adjusted for in the main model; italics indicate additional variables adjusted for in other models. *Includes history of coronary heart disease or coronary artery disease, coronary artery bypass surgery, angioplasty, or abnormal angiography. †Includes history of peripheral vascular disease, amputation, absent pulses, or claudication. ‡Excludes basal and squamous cell carcinoma of the skin.

relationship in which patients were grouped by serum phosphorus quintile indicates that this increased risk was not uniform across the spectrum of serum phosphorus levels. The results of this analysis are shown in Fig 2, in which each category represents approximately 20% of the study population. The relative mortality risk of serum phosphorus by quintiles was flat below 6.6 mg/dL and was markedly elevated above this level. Patients with serum phosphorus in the 6.6 to 7.8 mg/dL range had a 13% higher mortality

PHOSPHORUS AND MORTALITY IN HEMODIALYSIS

Fig 1. Distribution of serum phosphorus in CMAS (N ⴝ 3,738) and DMMS (N ⴝ 2,669).

risk than patients in the reference range of 4.6 to 5.5 mg/dL. Patients in the highest phosphorus quintile (7.9 to 16.9 mg/dL) had a relative mortality risk 34% higher than patients with serum phosphorus in the reference range. The shape of the relative mortality risk curve was similar when patients were categorized by deciles of serum phosphorus (results not shown), with elevated risk in patient groups with serum phosphorus levels above 6.5 mg/dL compared with those in the reference group of 4.4 mg/dL to 4.9 mg/dL. The sensitivity of the results discussed above to the use of 6.5 mg/dL as the cut-off level for ‘‘poor control’’ of phosphorus was examined by fitting the same models with different cut-off values. Analyses using cut-off values of 5.0 and 6.0 mg/dL resulted in relative risks that decreased in absolute levels. There was a relative mortality risk of 1.27 for phosphorus greater than

611

6.5 mg/dL (in the main analyses), a relative risk of 1.19 for phosphorus greater than 6.0 mg/dL, and a relative risk of 1.11 for phosphorus greater than 5.0 mg/dL. This is the expected result if one progressively moves increasing numbers of patients with actually ‘‘low risk’’ to the ‘‘high risk’’ category. Thirty-nine percent of patients had phosphorus levels greater than 6.5 mg/dL. The main models were also fit, including all patients (with 6.5 mg/dL as the phosphorus cut-off) to examine the sensitivity of the results when including patients with phosphorus less than 2.4 mg/dL. The results were essentially the same. Since mild hyperphosphatemia of 5.0 to 6.5 mg/dL was not associated with an elevated mortality risk, we divided patients into two groups based on the presence or absence of marked hyperphosphatemia, defined as greater than 6.5 mg/dL. For this analysis, to limit the reference group to those patients most commonly encountered in clinical practice, patients were excluded if the serum phosphorus level was less than 2.4 mg/dL (representing approximately 1% of the study population). As shown in Table 2, the main model indicates that the adjusted relative risk of mortality for patients with serum phosphorus greater than 6.5 mg/dL was 1.27, relative to those with levels between 2.4 and 6.5 mg/dL (P ⬍ 0.001). To evaluate how much of this observed relative risk of mortality seen with markedly eleTable 2. Relative Mortality Risk for Patients With a Serum Phosphorus Level Greater Than 6.5 mg/dL (Versus 2.4 to 6.5 mg/dL): Sensitivity Analyses for DMMS and CMAS Combined (N ⴝ 6,340)

Covariate Adjustments

Fig 2. Relative mortality risk by serum phosphorus quintiles (N ⴝ 6,407). The vertical bars indicate the 5% to 95% confidence intervals.

Main model (age at ESRD onset, race, sex, diabetes, AIDS, neoplasm, active smoking) Main model ⫹ skipping ⱖ1 session/mo Main model ⫹ dose of dialysis (Kt/V) Main model ⫹ nutritional parameters (BMI, creatinine, albumin) Main model ⫹ artherosclerotic disease (coronary, cerebral, peripheral) All of the above combined *PO4 ⬎ 6.5 v 2.4 to 6.5 mg/dL.

Relative Risk (95% CI)*

P Value

1.27

⬍0.001

1.26

⬍0.001

1.26

⬍0.001

1.45

⬍0.001

1.26 1.41

⬍0.001 ⬍0.001

612

BLOCK ET AL

vated serum phosphorus could be explained by other factors, sensitivity analyses were performed adjusting for pre-existing medical conditions, estimates of noncompliance, estimates of nutritional status, and delivered dose of dialysis. These results are also shown in Table 2. Adjusting for any of the additional covariates did not reduce the risk of mortality for patients with serum phosphorus greater than 6.5 mg/dL. Specifically, the increased mortality risk was not simply because hyperphosphatemic patients missed dialysis more often, received inadequate dialysis, or had a higher incidence of atherosclerotic vascular disease. Interestingly, when adding adjustment for estimates of nutritional status, the relative risk associated with high serum phosphorus increased markedly (RR ⫽ 1.45). Using the same cut-point of 6.5 mg/dL to indicate ‘‘poor control’’ of serum phosphorus, we then sought to identify risk factors for poor phosphorus control. Table 3 shows adjusted odds of having a serum phosphorus above 6.5 mg/dL by logistic regression analysis, adjusting for the variables shown. In this analysis, significant predictors of high serum phosphorus included younger age at onset of ESRD, female sex, white race, presence of diabetes, active smoking, and high serum creatinine. Serum albumin and BMI were predictors of high phosphorus levels in univariTable 3. Predictors of Serum Phosphorus Greater Than 6.5 mg/dL (Versus 2.4 to 6.5 mg/dL) by Multivariate Analysis* in DMMS Wave I and CMAS Combined (N ⴝ 6,340)

Variable

Odds Ratio*

P Value

Age at onset of ESRD (per 1 yr older) Sex (male v female) Race (black v white) Diabetes (presence of) Smoking (active) Neoplasm (presence of) AIDS (presence of) Independent ambulation BMI (per 1 kg/m2) Serum albumin (per 1 g/dL) Serum creatinine (per 1 mg/dL) Skipping dialysis (ⱖ1/mo) Delivered dose (per 0.1 Kt/V)

0.985 0.774 0.620 1.293 1.453 1.078 1.114 1.022 1.007 1.005 1.135 1.406 0.985

0.0001 0.0001 0.0001 0.0001 0.0001 0.43 0.83 0.82 0.22 0.95 0.0001 0.0003 0.25

*Odds by logistic regression model, adjusting for all variables shown. Bold indicates significant results with Bonferroni correction (P ⬍ 0.0038).

Fig 3. Relative mortality risk by serum calcium quintiles (N ⴝ 2,669). The vertical bars indicate the 5% to 95% confidence intervals.

ate analysis (data not shown), but not when adjusting for the other variables, including serum creatinine. The latter may serve as a nutritional marker and is also a strong predictor of hyperphosphatemia. Having skipped at least one dialysis session during 1 month at the study start (a potential marker of noncompliance) was strongly associated with having ‘‘high’’ serum phosphorus. The delivered dose of dialysis was not a statistically significant predictor of phosphorus greater than 6.5 mg/dL. Black patients were 38% less likely to have a high serum phosphorus (⬎6.5 mg/dL) than whites (RR ⫽ 0.62; P ⬍ 0.0001). Because one hypothesis would explain the association of elevated serum phosphorus levels with mortality risk through an elevated Ca ⫻ PO4 product, further analyses were performed using only patients from the DMMS (calcium data were not collected in the CMAS). The relative risk of mortality by quintiles of serum calcium levels controlling for age at onset of ESRD, race, sex, and the presence of diabetes, smoking, AIDS, and neoplasm is shown in Fig 3. No statistically significant difference in mortality risk was seen across the spectrum of calcium levels. Figure 4 describes the relative mortality risk by quintiles of Ca ⫻ PO4 product, again controlling for age at onset of ESRD, race, sex, and the presence of diabetes, smoking, AIDS, and neoplasm. This curve reveals that higher Ca ⫻ PO4 product was associated with higher risk. This effect reached statistical significance when the product was greater than 72 mg2/dL2. The group of patients with a Ca ⫻ PO4 product

PHOSPHORUS AND MORTALITY IN HEMODIALYSIS

613

(⬎975 pg/mL) where the mortality risk was 1.34 (P ⫽ 0.09). The results appeared to be similar when phosphorus or the Ca ⫻ PO4 product was added to the model, although results for the continuous PTH model became statistically insignificant (P ⫽ 0.10 to 0.12). The relative mortality risk associated with higher levels of serum phosphorus remained the same when PTH was controlled for in the model (RR ⫽ 1.06 per 1 mg/dL higher serum phosphorus; P ⬍ 0.01). DISCUSSION Fig 4. Relative mortality risk by Ca ⴛ PO4 product quintiles (N ⴝ 2,669). The vertical bars indicate the 5% to 95% confidence intervals.

greater than 72 mg2/dL2 had a 34% higher risk of death compared with those with a Ca ⫻ PO4 product in the reference range between 42 and 52 mg2/dL2 (RR ⫽ 1.34; P ⬍ 0.01). This reference range was chosen because it has been cited in the literature as a desirable range for this product in the ESRD population.3 Both calcium and the Ca ⫻ PO4 product were also examined using calcium levels corrected for serum albumin. The results were similar to those shown in Figs 3 and 4, except that in the model with corrected calcium, patients with the lowest corrected calcium levels (3.7 to 8.9 mg/dL) had significantly lower mortality risk (RR ⫽ 0.78; P ⫽ 0.03) than those in the reference range of 9.4 to 9.9 mg/dL. This decreased risk among patients with the lowest corrected calcium levels became smaller and statistically insignificant when serum albumin was controlled for in the model (RR ⫽ 0.84, P ⫽ 0.12). To determine the effect PTH has on the increased risk associated with elevated serum phosphorus levels, models were fit for the subgroup of patients from the DMMS with PTH available (78% of patients). There was a significant difference in relative risk of mortality by log PTH as a continuous variable, controlling for age at onset of ESRD, race, sex, and the presence of diabetes, smoking, AIDS, and neoplasm (P ⫽ 0.03). Using PTH in quintiles with the same covariates, patients with PTH in the highest quintile (PTH ⬎511 pg/mL) appeared to have a higher mortality risk (RR ⫽ 1.18; P ⫽ 0.19) than those with PTH between 34 and 91 pg/mL (Fig 5). This trend was greater for the highest decile of PTH

This study describes the level of serum phosphorus in two national, random, cross-sectional samples of hemodialysis patients who have been receiving hemodialysis for more than 1 year (mean, 4.5 years). Fifty percent of patients have a serum phosphorus greater than 6.0 mg/dL and one quarter of patients have a level greater than 7.4 mg/dL. These results are similar to those published by Lowrie and Lew13 for over 17,000 patients receiving hemodialysis in 1988. Their analysis also showed that 25% of patients had a phosphorus level greater than 7.2 mg/dL, suggesting that the ability to control serum phosphorus has changed very little between 1988 and 1993. The frequent finding of hyperphosphatemia is due partly to the fact that phosphorus clearance by dialysis itself is not the major factor involved in the control of serum phosphorus.2,14,15 Phosphate kinetic studies in hemodialysis patients show that there is an initially high phosphate clearance from the extracellular compartment followed by a slow efflux from an intracellular compartment.15-18 In fact, serum phosphorus be-

Fig 5. Relative mortality risk by serum PTH level quintiles (N ⴝ 2,087). The vertical bars indicate the 5% to 95% confidence intervals.

614

gins to increase before the termination of the hemodialysis session.18 These data also show that despite a major shift from the use of aluminum-containing to calcium-containing phosphate binders in the 1980s, no major change can be seen in the distribution of serum phosphorus levels. Interestingly, a recent study reported that slow nocturnal home hemodialysis (six nights weekly, 8 hours per night) was successful in lowering the serum phosphorus, despite a 50% increase in phosphorus intake and the cessation of the use of phosphate binders.19 This successful therapy of hyperphosphatemia supports the idea of slow release of phosphate from an intracellular source. This new dialytic technique, however, appears to be practical for only a small fraction of patients. Since the overwhelming majority of dialysis patients receive standard thrice-weekly hemodialysis, dietary restriction and prescription of phosphorus binders play a much more important role. Unfortunately, poor compliance with both diet and medication use is quite common among patients with ESRD.20,21 Beyond this description of the actual level of serum phosphorus control in hemodialysis patients in the United States, this study demonstrates that the presence of high serum phosphorus confers a substantial mortality risk, even after adjusting for the presence of comorbid illness. Statistical adjustment for dose of dialysis did not modify these results, suggesting that Kt/V for urea does not substantially affect the phosphorus level. Thus, the present study provides further evidence for the poor dialytic clearance of phosphorus. Further analysis by phosphorus levels shows that those patients with serum phosphorus of greater than 6.5 mg/dL had a 27% higher mortality risk ((RR ⫽ 1.27) than patients with phosphorus of 2.4 to 6.5 mg/dL. As an example, if there were 22 deaths per 100 patients with phosphorus between 2.4 to 6.5 mg/dL, there would be 28 (22 ⫻ 1.27) deaths per 100 patients with phosphorus above 6.5 mg/dL, all else being equal. These results are consistent with the report by Lowrie and Lew,13 who found a twofold higher relative risk of death for their chronic hemodialysis patients with a serum phosphorus between 7 and 11 mg/dL relative to those patients with a level between 5 and 7 mg/dL. Their analysis, however, did not adjust for comorbid conditions.

BLOCK ET AL

The exact mechanism whereby hyperphosphatemia contributes to excess patient mortality is unknown, but may be related to its association with secondary hyperparathyroidism. Recent evidence suggests that phosphorus has a direct stimulatory effect on PTH secretion and that hypophosphatemia decreases PTH mRNA.22-26 Denda et al27 have demonstrated in uremic rats that phosphorus restriction prevents parathyroid gland growth and secondary hyperparathyroidism independent of changes in ionized calcium or 1,25 (OH) 2 vitamin D3. Further contributing to the problem is the fact that poor control of serum phosphorus has been shown to lessen the ability of vitamin D analogues to suppress PTH secretion.5,28,29 Hyperphosphatemia therefore directly contributes to secondary hyperparathyroidism and results in prolonged exposure to excessively high levels of PTH. Massry and Smogorzewski7,30 proposed that hyperparathyroidism results in many of the manifestations of the uremic syndrome and accounts for substantial morbidity and perhaps mortality in patients with ESRD. The analyses of patients with PTH data available indicate that there was a weak association between mortality and PTH levels, possibly driven by the 10% of patients with the highest levels of PTH (⬎975 pg/mL). This association is not explained by phosphorus or corrected calcium levels, and does not affect the relationship between hyperphosphatemia and increased mortality risk. In contrast to the results of a previous study,31 these analyses showed no increased risk for patients with low values of PTH. This may be related to patient selection because the present study excluded patients who had ESRD for less than 1 year, whereas the previous study considered only patients who were recently diagnosed with ESRD. An alternative hypothesis to explain the increased mortality risk of hyperphosphatemia may be its contribution to an elevated Ca ⫻ PO4 product. In the current study, analysis of the large subset of patients with available data for both calcium and phosphorus revealed that a Ca ⫻ PO4 product greater than 72 mg2/dL2 was associated with a significantly higher relative risk of death (RR ⫽ 1.34). Since isolated hypercalcemia was not associated with any increased relative mortality risk, it appears that the increased risk in this group is driven by the hyperphospha-

PHOSPHORUS AND MORTALITY IN HEMODIALYSIS

temia. There was significantly decreased risk among patients with the lowest corrected calcium, but this effect was partially explained by serum albumin. An increased risk of mortality with a Ca ⫻ PO4 product greater than 72 mg2/ dL2 correlates well with previously published data suggesting an increased risk of metastatic calcification at levels above 60 to 75 mg2/ dL2.32,33 Hyperphosphatemia has been a strong risk factor for extraskeletal calcification,33,34 and it is tempting to relate the altered mineral metabolism to the excess cardiovascular risk seen in this population. Previous work in laboratory animals showed that the visceral calcium content in the heart, lung, aorta, and kidney increased dramatically immediately following the development of acute or chronic renal failure and that this effect could be completely abolished by phosphorus restriction.4 Arterial calcification has been linked to an elevated Ca ⫻ PO4 product, and dialysis patients have been shown using electron beam computed tomography to have 2.5 to five times the coronary artery calcium score compared with nondialysis patients.35 The use of this technique to detect coronary artery calcification has been shown to have a high sensitivity (84%) for predicting coronary disease, and the level of vascular calcification has been found to correlate well with luminal stenoses.36 Early reports on calcification in dialysis patients include an autopsy study by Kuzela et al8 that showed severe visceral calcification in 36% of dialysis patients, including lesions located in the myocardium, which were felt to have contributed to death. Ten years later, increased myocardial calcium content and vascular calcifications were shown to have a strong positive correlation with an elevated Ca ⫻ PO4 product and an inverse correlation with left ventricular function.37 Furthermore, dialysis patients have elevated levels of myocardial and pulmonary calcium content as well as a markedly higher incidence of valvular calcifications.37-41 Specific causes of death for patients with either elevated phosphorus levels or elevated Ca ⫻ PO4 products were not addressed in this study; however, given that cardiovascular disease (cardiac arrest, acute myocardial infarction, and other cardiac causes) is the cause of death in more than 45% of all dialysis patients,42 it seems that an elevated Ca ⫻ PO4

615

product may act through a cardiovascular mechanism. The search for risk factors predictive of poorly controlled serum phosphorus revealed that markers traditionally associated with improved nutritional status were significantly associated with higher serum phosphorus levels. This explains the observation that the relative risk of death from high serum phosphorus was particularly large when the model was adjusted for nutritional covariates. There appears to be a competing risk/benefit from having higher serum phosphorus levels and improved nutrition. When patients with higher phosphorus and good nutrition are compared only with patients with ‘‘equally’’ good nutrition by adjustment for albumin, creatinine, and BMI, the benefit of nutrition is removed (by the adjustment) and only the risk is retained. This is documented by the relatively large mortality risk with adjustment for nutrition (RR ⫽ 1.45) compared with no adjustment for nutrition (RR ⫽ 1.27). The finding that black race was associated with lower odds of having a high phosphorus than white race was unexpected. When the adjustment for serum creatinine was removed from the model, this association was small and no longer statistically significant. In other words, with a serum creatinine of 12 mg/dL and all else being equal, a black patient would be much less likely than a white patient to have a serum phosphorus greater than 6.5 mg/dL. Interpretation of this result is most likely related to race-related differences in creatinine metabolism or body composition (ie, relatively more muscle and less fat). Black hemodialysis patients in this study had significantly higher mean serum creatinine levels (blacks, 13.0 mg/dL; whites, 10.7 mg/dL). It appears, therefore, that the reliability of serum creatinine as a marker of nutritional intake might be different for black and white ESRD patients. It is possible, although perhaps less likely, that dietary differences could account for the discrepancy between black and white hemodialysis patients’ tendency toward hyperphosphatemia. If confirmed in subsequent studies, this would represent a potentially modifiable risk factor particularly for white ESRD patients. It is unclear how active smoking contributes to high serum phosphorus levels, although it could be speculated that smokers might be less compliant to other

616

BLOCK ET AL

medical recommendations, such as dietary restrictions and/or phosphate binders. Smokers have been recently found to also be less compliant with respect to shortening or skipping dialysis treatment sessions or having excessive interdialytic weight gain.43 Limitations This study was not prospective in design, and observed relationships do not necessarily imply causality. Although these analyses tried to control for a significant number of comorbid conditions, there may be other differences that were not measured and therefore not adjusted. Data on acid-base status were not evaluated. This study used surrogate markers of nutritional status, such as serum albumin, serum creatinine, and BMI. The use of the normalized protein catabolic rate may provide a better estimate of recent protein nutrition. Although this can be estimated from predialysis and postdialysis blood urea nitrogen and knowledge of the time interval between dialysis, there has been concern expressed of a mathematical linkage between normalized protein catabolic rate and Kt/V. Since adjustment for surrogate markers of nutrition increased the correlation of hyperphosphatemia with mortality, we suspect that a better nutritional measure such as normalized protein catabolic rate would increase, not diminish, the magnitude of the mortality risk found in this study. It is important to recall that the findings of the present study were based on patients who had been receiving hemodialysis for at least 1 year. Therefore, the results should not be extrapolated to patients new to dialysis or to those treated by peritoneal dialysis. Any error that may have occurred as a result of using a single laboratory value to classify patients into ‘‘high’’ versus ‘‘low’’ phosphorus groups would have tended to yield smaller and less significant results, thus suggesting that true correlations with mortality may be even stronger than described here. Similarly, the correlations of PTH with mortality may have been underestimated because the assay for PTH was not uniformly the same. CONCLUSION

This national study of two large, random samples of patients receiving hemodialysis for at least 1 year shows that 39% of patients had a serum phosphorus level greater than 6.5 mg/dL.

Patients with levels above this value had an excess risk of death (RR ⫽ 1.27) compared with those with serum phosphorus between 2.4 and 6.5 mg/dL. Risk factors that might identify a ‘‘poorly controlled’’ serum phosphorus include diabetes, active smoking, female sex, white race, high serum creatinine levels and having skipped dialysis in the preceding month. The mechanism by which high serum phosphorus results in mortality may be related to an elevated Ca ⫻ PO4 product but are not explained by PTH levels. Ca ⫻ PO4 product levels above 72 mg2/dL2 were observed in 20% of patients and were associated with a significant increase in the relative risk of death (RR ⫽ 1.34) compared with those with a Ca ⫻ PO4 product between 42 and 52 mg2/dL2. The excess risk seen with an elevated phosphorus level does not appear to be related to having received a lower dose of dialysis. Beyond the well-recognized importance in the treatment and prevention of metabolic bone disease, the results of the present study suggest that vigorous phosphorus control to levels below 6.5 mg/dL is of vital importance to chronic hemodialysis patients. This can be accomplished under the current thrice-weekly dialysis prescription, primarily by manipulating phosphorus absorption from the gastrointestinal tract. REFERENCES 1. Winchester JF, Rotellar C, Goggins M, Robino D, Rakowski TA, Argy WP: Calcium and phosphate balance in dialysis patients. Kidney Int 43:174-178, 1993 2. Hsu CH: Are we mismanaging calcium and phosphate metabolism in renal failure? Am J Kidney Dis 29:641-649, 1997 3. Kates DM, Andress DL: Control of hyperphosphatemia in renal failure: Role of aluminum. Semin Dial 9:310315, 1996 4. Chertow GM, Burke SK, Lazarus JM, Stenzel KH, Wombolt D, Goldberg D, Bonventre JV, Slatopolsky E: Poly[allylamine hydrochloride] (RenalGel): A noncalcemic phosphate binder for the treatment of hyperphosphatemia in chronic renal failure. Am J Kidney Dis 29:66-71, 1997 5. Slatopolsky E, Delmez JA: Pathogenesis of secondary hyperparathyroidism. Nephrol Dial Transplant 11:130-135, 1996 6. Alfrey AC, Ibels LS: Role of phosphate and pyrophosphate in soft tissue calcification. Adv Exp Med Biol 103:187193, 1978 7. Massry SG: Pathogenesis of uremic toxicity: Parathyroid hormone as a uremic toxin, in Masry SG, Glassock RJ (eds): Textbook of Nephrology (ed 3). Baltimore, MD, Williams & Wilkins, 1995, pp 1270-1303 8. Kuzela DC, Huffer WE, Conger JD, Winter SD, Hammond WS: Soft tissue calcification in chronic dialysis patients. Am J Pathol 86:403-424, 1977

PHOSPHORUS AND MORTALITY IN HEMODIALYSIS

9. Parfitt AM: Soft tissue calcification in uremia. Arch Intern Med 124:544-556, 1969 10. Milliner DS, Zinsmeister AR, Lieberman E, Landing B: Soft tissue calcification in pediatric patients with endstage renal disease. Kidney Int 38:931-936, 1990 11. US Renal Data System: USRDS 1996 Annual Data Report. Bethesda, MD, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 1996 12. Daugirdas JT: Second generation logarithmic estimates of single pool variable volume of Kt/V: An analysis of error. J Am Soc Nephrol 4:1205-1212, 1993 13. Lowrie EG, Lew NL: Death risk in hemodialysis patients: The predictive value of commonly measured variables and an evaluation of death rate differences between facilities. Am J Kidney Dis 15:458-482, 1990 14. Delmez JA, Slatopolsky E: Hyperphosphatemia: Its consequences and treatment in patients with chronic renal disease. Am J Kidney Dis 19:303-317, 1992 15. Zuccheli P, Santoro A: Inorganic phosphate removal during different dialytic procedures. Int J Artif Organs 10:173178, 1987 16. Haas T, Hillion D, Dongradi G: Phosphate kinetics in dialysis patients. Nephrol Dial Transplant 6:108-113, 1991 (suppl 2) 17. Chauveau P, Poignet JL, Kuno T, Bonete R, Kerebrun A, Naret C, Delons S, Man NK, Rist E: Phosphate removal rate: A comparative study of five high flux dialyzers. Nephrol Dial Transplant 6:114-115, 1991 (suppl 2) 18. Desoi CA, Umans JG: Phosphate kinetics during high flux hemodialysis. J Am Soc Nephrol 4:1214-1218, 1993 19. Mucsi I, Hercz G, Ouwendyk L, Wallace L, Francoeur B, Uldall R: Phosphate removal during conventional hemodialysis and slow nocturnal home hemodialysis. J Am Soc Nephrol 7:498, 1996 20. Wolcott DL, Maida CA, Diamond R: Treatment compliance in end-stage renal disease patients on dialysis. Am J Nephrol 6:329-338, 1986 21. Bailie GR, Eisele G, Low CL: Hemodialysis patients’ knowledge of phosphate binders. J Am Soc Nephrol 7:1791, 1996 22. Finch JL, Ross FP, Brown AJ, Denda M, Slatopolsky E: Phosphorus increases PTH secretion and synthesis in cultured rat parathyroid glands. J Am Soc Nephrol 7:1813, 1996 23. Fine A, Cox D, Fontaine B: Elevation of serum phosphate affects parathyroid hormone levels in only 50% of hemodialysis patients, which is unrelated to changes in serum calcium. J Am Soc Nephrol 3:1947-1953, 1993 24. Silver J, Moallem E, Kilav R, Epstein E, Sela A, Naveh-Many T: New insights into the regulation of parathyroid hormone synthesis and secretion in chronic renal failure. Nephrol Dial Transplant 11:2-5, 1996 25. Ho LT, Coward M, Negrea L: The effect of serum phosphorus on PTH secretion in hemodialysis patients. J Am Soc Nephrol 7:1791, 1996 26. Slatopolsky E, Finch J, Denda M: Phosphorus restriction prevents parathyroid gland growth. High phosphorus directly stimulates PTH secretion in vitro. J Clin Invest 97:2534-2540, 1996 27. Denda M, Finch J, Slatopolsky E: Phosphorus accel-

617

erates the development of parathyroid hyperplasia and secondary hyperparathyroidism in rats with renal failure. Am J Kidney Dis 28:596-602, 1996 28. Quarles LD, Yohai DA, Carroll BA, Spritzer CE, Minda SA, Bartholomay D, Lobaugh BA: Prospective trial of pulse oral versus intravenous calcitriol treatment of hyperparathyroidism in ESRD. Kidney Int 44:1259-1265, 1994 29. Llach F, Hervas J, Yudd M, Goldblat MV: Hyperphosphatemia worsened parathyroid function in dialysis (HD) patients with severe hyperphosphatemia receiving intravenous calcitriol. J Am Soc Nephrol 7:1791, 1996 30. Massry SG, Smogorzewski M: Mechanisms through which parathyroid hormone mediates its deleterious effects on organ function in uremia. Semin Nephrol 14:219-231, 1994 31. Avram MM, Sreedhara R, Avram DK, Muchnick RA, Fein P: Enrollment parathyroid hormone level is a new marker of survival in hemodialysis and peritoneal dialysis therapy for uremia. Am J Kidney Dis 28:924-930, 1996 32. Velentzas C, Meindok H, Oreopoulos DG, Meema HE, Rabinovich S, Sutton D, Ogilvie R: Visceral calcification and the Ca ⫻ P product. Adv Exp Med Biol 103:195201, 1978 33. Drueke TB: A clinical approach to the uraemic patient with extraskeletal calcifications. Nephrol Dial Transplant 11:37-42, 1996 34. Southwood RL, Mueller BA, Copley JB: Soft tissue calcification in renal failure. DICP 24:855-859, 1990 35. Braun J, Oldendorf M, Moshage W, Heidler R, Zeitler E, Luft FC: Electron beam computed tomography in the evaluation of cardiac calcifications in chronic dialysis patients. Am J Kidney Dis 27:394-401, 1996 36. Rumberger JA, Simons DB, Fitzpatrick LA: Coronary artery calcium area by electron beam computed tomography and coronary atherosclerotic plaque area: A histopathologic correlative study. Circulation 92:2157-2162, 1995 37. Rostand SG, Sanders C, Kirk K, Rutsky EA, Fraser RG: Myocardial calcification and cardiac dysfunction in chronic renal failure. Am J Med 85:651-657, 1988 38. Maher ER, Young G, Smyth WB, Pugh S, Curtis JR: Aortic and mitral valve calcification in patients with endstage renal disease. Lancet 2:875-877, 1987 39. Maher ER, Pazianas M, Curtis JR: Calcific aortic stenosis: A complication of chronic uremia. Nephron 47:119122, 1987 40. Mako J, Lengyel M, Szucs J: Intracardiac calcification in patients under chronic hemodialysis. Int Urol Nephrol 19:441-446, 1987 41. Akmal M, Barndt RR, Ansari A, Mohler JG, Massry SG: Excess PTH in CRF induces pulmonary calcification, pulmonary hypertension and right ventricular hypertrophy. Kidney Int 47:158-163, 1995 42. US Renal Data System: USRDS 1997 Annual Data Report. Bethesda, MD, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 1997 43. Leggat JE Jr, Orzol S, Hulbert-Shearon TE, Golper TA, Held PJ, Port FK: Noncompliance with hemodialysis: Predictors and survival analysis. J Am Soc Nephrol 8:200A, 1997 and Am J Kidney Dis 1998 (in press)