KDIGO Clinical Practice Guideline for the Diagnosis, Evaluation, Prevention, and Treatment of Chronic Kidney Disease-Mineral and Bone Disorder (CKD-MBD)

Chapter 3.1: Diagnosis of CKD-MBD: biochemical abnormalities

Kidney International (2009) 76 (Suppl 113), S22-S49. doi:10.1038/ki.2009.191

INTRODUCTION

Biochemical abnormalities are common in chronic kidney disease (CKD) and are the primary indicators by which the diagnosis and management of CKD-mineral and bone disorder (CKD-MBD) is made. The two other components of CKD-MBD (bone abnormalities and vascular calcification) are discussed in Chapters 3.2 and 3.3.

RECOMMENDATIONS

3.1.1 We recommend monitoring serum levels of calcium, phosphorus, PTH, and alkaline phosphatase activity beginning in CKD stage 3 (1C). In children, we suggest such monitoring beginning in CKD stage 2 (2D).

3.1.2 In patients with CKD stages 3-5D, it is reasonable to base the frequency of monitoring serum calcium, phosphorus, and PTH on the presence and magnitude of abnormalities, and the rate of progression of CKD (not graded).Reasonable monitoring intervals would be:

• In CKD stage 3: for serum calcium and phosphorus, every 6-12 months; and for PTH, based on baseline level and CKD progression.
• In CKD stage 4: for serum calcium and phosphorus, every 3-6 months; and for PTH, every 6-12 months.
• In CKD stages 5, including 5D: for serum calcium and phosphorus, every 1-3 months; and for PTH, every 3-6 months.
• In CKD stages 4-5D: for alkaline phosphatase activity, every 12 months, or more frequently in the presence of elevated PTH (see Chapter 3.2).

In CKD patients receiving treatments for CKD-MBD, or in whom biochemical abnormalities are identified, it is reasonable to increase the frequency of measurements to monitor for trends and treatment efficacy and side-effects (not graded).

3.1.3 In patients with CKD stages 3-5D, we suggest that 25(OH)D (calcidiol) levels might be measured, and repeated testing determined by baseline values and therapeutic interventions (2C). We suggest that vitamin D deficiency and insufficiency be corrected using treatment strategies recommended for the general population (2C).

3.1.4 In patients with CKD stages 3-5D, we recommend that therapeutic decisions be based on trends rather than on a single laboratory value, taking into account all available CKD-MBD assessments (1C).

3.1.5 In patients with CKD stages 3-5D, we suggest that individual values of serum calcium and phosphorus, evaluated together, be used to guide clinical practice rather than the mathematical construct of calcium-phosphorus product (Ca X P) (2D).

3.1.6 In reports of laboratory tests for patients with CKD stages 3-5D, we recommend that clinical laboratories inform clinicians of the actual assay method in use and report any change in methods, sample source (plasma or serum), and handling specifications to facilitate the appropriate interpretation of biochemistry data (1B).

Summary of rationale for recommendations

• As the diagnosis of CKD-MBD depends on the measurement of laboratory and other variables, it is important to provide a guide to clinicians regarding when to commence measurement of those variables. Although changes in the biochemical abnormalities of CKD-MBD may begin in CKD stage 3, the rate of change and severity of abnormalities are highly variable among patients.
• Thus, the recommendations and suggestions above indicate that assessment of CKD-MBD should begin at stage 3, but the frequency of assessment needs to take into account the identified abnormalities, the severity and duration of the abnormalities in the context of the degree and rate of change of glomerular filtration rate (GFR), and the use of concomitant medications. Further testing and shorter time intervals would be dependent on the presence and severity of biochemical abnormalities.
• Furthermore, the interpretation of these biochemical and hormonal values requires an understanding of assay type and precision, interassay variability, blood sample handling, and normal postprandial, diurnal, and seasonal variations in individual parameters.
• The serum phosphorus fluctuates more than the serum calcium. As the mathematical construct of the calcium X phosphorus product (Ca X P) is largely driven by serum phosphorus and generally does not provide any additional information beyond that which is provided by individual measures, it is of limited use in clinical practice.

BACKGROUND

The laboratory diagnosis of CKD-MBD includes the use of laboratory testing of serum PTH, calcium (ideally ionized calcium but most frequently total calcium, possibly corrected for albumin), and phosphorus. In some situations, measuring serum ALPs (total or bone specific) and bicarbonate may be helpful. It is important to acknowledge that the biochemical and hormonal assays used to diagnose, treat, and monitor CKD-MBD have limitations and, therefore, the interpretation of these laboratory values requires an understanding of assay type and precision, interassay variability, blood sample handling, and normal postprandial, diurnal, and seasonal variations. Derivations of these assays compound the problems with precision and accuracy. It is important for the practicing clinician to appreciate the potential variations in laboratory test results to avoid overemphasizing small or inconsistent changes in clinical decision making. Educating patients and primary-care physicians as to these subtleties is also important to ensure the appropriate interpretation by non-nephrologists who may also receive the results of the tests.

This chapter is the result of a comprehensive literature review of selected topics by the Work Group with assistance from the evidence review team to formulate the rationale for clinical recommendations. Thus, it should not be considered as a systematic review.

RATIONALE

3.1.1 We recommend monitoring serum levels of calcium, phosphorus, PTH, and alkaline phosphatase activity beginning in CKD stage 3 (1C). In children, we suggest such monitoring beginning in CKD stage 2 (2D).

Abnormalities in calcium, phosphorus, PTH, and vitamin D metabolism (collectively referred to as disordered mineral metabolism) are common in patients with CKD. Changes in the laboratory parameters of CKD-MBD may begin in CKD stage 3, but the presence of abnormal values, the rate of change, and the severity of abnormalities are highly variable among patients. To make the diagnosis of CKD-MBD, one or more of these laboratory abnormalities must be present. Thus, measuring them once is essential for diagnosis. Although the initial assessment should begin at this stage, the frequency of assessment is based on the presence and persistence of identified abnormalities, the severity of abnormalities, all in the context of the degree and rate of change of GFR and the use of concomitant medications.

The interpretation of the biochemical and hormonal values also requires an understanding of normal postprandial, diurnal, and seasonal variations, with differences from one parameter to the other. For example, serum phosphorus fluctuates more than serum calcium within an individual, and is affected by diurnal variation more than is serum calcium. Given the complexity of changes within any one parameter, it is important to take into account the trends of changes rather than single values to evaluate changes in the degree of severity of laboratory abnormalities of CKD-MBD.

The best available data to guide diagnostic monitoring consist of that which is obtained from population-based or cohort-based prevalence studies. Although subject to specific biases, these studies do guide the clinician with respect to expected proportions of abnormal test results at specific levels of CKD. However, even this is problematic, given the inconsistent definitions of 'abnormal' (be it insufficient, deficient, or in excess). Moreover, there are additional issues with specific assays, especially for PTH and 25(OH)D, which further complicate and limit our ability to characterize specific levels as pathological.

Limitations of current data sources

Most of the studies describing observational data and relationships between individual parameters and clinical outcomes have been conducted in hemodialysis (HD) populations. Furthermore, those HD population studies are generally from cohorts who did not always receive predialysis care or early identification. In addition, the analysis of the observational data uses cohort-specific cut points or KDOQI recommendations from 2003.

Limited data exist regarding the prevalence of biochemical and hormonal abnormalities in CKD stages 3-5, because of the general absence of registry data, population-based studies, or large cohort studies. There are increasingly recognized differences in referred vs nonreferred populations, and in those with kidney transplants. Data are limited in all of these non-dialysis groups. Even in national dialysis databases, a routine collection of data on MBD is uncommon, and in those databases that do have the information, they are generally available only for a single time point, such as dialysis initiation, or confounded by treatment.

Thus, establishing diagnostic and management criteria on the basis of data obtained from the sources described above, and in the context of individual person and assay variability, is problematic. Nevertheless, utilizing trends, consistency of data direction, and biological plausibility, the Work Group has made recommendations and suggestions for the diagnosis and management of laboratory parameters.

Examples of studies that describe the prevalence of abnormalities

CKD stages 3-5. Levin et al.²⁸ have described the prevalence of abnormalities in serum calcium, phosphorus, and PTH in a cross-sectional analysis of 1800 patients with CKD stages 3-5 in North America (Study To Evaluate Early Kidney Disease). Calcium and phosphorus values did not become abnormal until GFR fell below 40 ml/min per 1.73 m², and were relatively stable until GFR fell below 20 ml/min per 1.73 m² (Figure 4). However, 12% of patients with GFR>80 ml/min per 1.73 m² had a high PTH (defined as >65 pg/ml, the upper limit of normal of the assay used) and nearly 60% of patients with GFR<60 ml/min per 1.73 m² had elevated PTH levels. Similar findings have been recently reported from a community-based screening program sponsored by the National Kidney Foundation, the Kidney Education and Evaluation Program.²⁹ It is to be noted that both cohorts were primarily nonreferred populations, with a diagnosis of CKD made on the estimated GFR.

Figure 4 | Prevalence of abnormal mineral metabolism in CKD. (a) The prevalence of hyperparathyroidism, hypocalcemia, and hyperphosphatemia by eGFR levels at 10-ml/min per 1.73 m² intervals. (b) Median values of serum Ca, P, and iPTH by eGFR levels. (c) Median values of 1,25 (OH)2D₃, 25(OH)D₃, and iPTH by GFR levels. CKD, chronic kidney disease; eGFR, estimated glomerular filtration rate; GFR, glomerular filtration rate; iPTH, intact parathyroid hormone. Reprinted with permission from Levin et al.²⁸

CKD stage 5D. The Choices for Healthy Outcomes in Caring for End-Stage Renal Disease study is a large, prospectively collected national cohort of incident dialysis patients with repeated measures of laboratory values. In incident dialysis patients, serum levels of calcium and phosphorus at the start of dialysis were 9.35 mg/dl (2.34 mmol/l) and 5.23 mg/dl (1.69 mmol/l), respectively. Mean serum levels increased over the initial 6 months of renal replacement therapy (calcium 9.51 mg/dl or 2.38 mmol/l; phosphorus 5.43 mg/dl or 1.75 mmol/l).³⁰

Although there are numerous cross-sectional reports of serum levels of calcium, phosphorus, and PTH in CKD stage 5D population, the international Dialysis Outcomes and Practice Pattern Study provides the most comprehensive global view of the prevalence of disorders of calcium (corrected for albumin), phosphorus, and PTH.³³ Unfortunately, there is no standardization of PTH assays from around the world. Nevertheless, abnormalities were observed in parallel studies from large dialysis providers in the United States with central laboratories. Figure 5 provides a robust depiction of not only the distribution of abnormalities in laboratory values relevant to CKD-MBD but also a visual representation of changes in international practice patterns as well over the three phases of the Dialysis Outcomes and Practice Pattern Study observation (I=1996-2001, II=2002-2004, and III=2005-present).

Figure 5 | Changes in serum calcium, phosphorus, and iPTH with time in hemodialysis patients of DOPPS countries. Distribution of baseline serum calcium (a), phosphorus (b), and PTH (c) by country and the DOPPS phase. See text for details. DOPPS, Dialysis Outcomes and Practice Pattern Study; PTH, parathyroid hormone. Reprinted with permission from Tentori et al.³³

Recently, elevated serum total ALP (t-ALP) levels have been recognized as a possibly independent variable associated with an increase in the relative risk (RR) of mortality in patients with CKD stage 5D.^{31, 32} Regidor et al.³¹ have described an association of serum t-ALP levels with mortality among prevalent HD populations, in addition to U- or J-shaped curves for calcium, phosphorus, and PTH, further underscoring the complexity of the relationships of these laboratory abnormalities with outcomes. High levels of ALPs are associated with mortality, but there is no evidence that reducing these levels leads to improved outcomes. The use of ALPs to interpret other abnormalities of measured minerals within an individual (for example, as an indicator of bone turnover or as an indicator of other conditions such as liver disease, an so on) may be useful as detailed in Chapter 3.2.

Children. In children, one study showed that elevations in PTH occur as early as CKD stage 2, especially in children with slowly progressive kidney disease ^.34 Given the significant associations of biochemical abnormalities of CKD-MBD with growth and cardiac dysfunction ³⁵ in children, the Work Group felt it was reasonable to assess children for the biochemical abnormalities of CKD-MBD initially at CKD stage 2.

3.1.2 In patients with CKD stages 3-5D, it is reasonable to base the frequency of monitoring serum calcium, phosphorus, and PTH on the presence and magnitude of abnormalities, and the rate of progression of CKD (not graded).Reasonable monitoring intervals would be:

• In CKD stage 3: for serum calcium and phosphorus, every 6-12 months; and for PTH, based on baseline level and CKD progression.
• In CKD stage 4: for serum calcium and phosphorus, every 3-6 months; and for PTH, every 6-12 months.
• In CKD stages 5, including 5D: for serum calcium and phosphorus, every 1-3 months; and for PTH, every 3-6 months.
• In CKD stages 4-5D: for alkaline phosphatase activity, every 12 months, or more frequently in the presence of elevated PTH (see Chapter 3.2).

In CKD patients receiving treatments for CKD-MBD, or in whom biochemical abnormalities are identified, it is reasonable to increase the frequency of measurements to monitor for trends and treatment efficacy and side-effects (not graded).

There are no data showing that routine measurement improves patient-level outcomes. Nevertheless, suggestions can be made as to a reasonable frequency of measurement of these laboratory parameters of CKD-MBD. The clinician should adjust the frequency on the basis of the presence and magnitude of abnormalities, and on the rate of progression of kidney disease. The frequency of measurement needs to be individualized for those receiving treatments for CKD-MBD to monitor for treatment effects and adverse effects.

Table 12 provides reasonable guidance as to the frequency of monitoring, given the numerous caveats outlined above; clinical situations (stability and treatment strategies) and other factors will influence the frequency of testing, and this must be individualized. As with any long-term condition, longitudinal trends are important and some forms of systematic (for example, fixed interval) monitoring is likely to be of greater value than random monitoring.

3.1.3 In patients with CKD stages 3-5D, we suggest that 25(OH)D (calcidiol) levels might be measured, and repeated testing determined by baseline values and therapeutic interventions (2C). We suggest that vitamin D deficiency and insufficiency be corrected using treatment strategies recommended for the general population (2C).

The Work Group acknowledged that there is emerging information on the potential role of vitamin D deficiency and insufficiency in the pathogenesis or worsening of multiple diseases. In addition, vitamin D deficiency and insufficiency may have a role in the pathogenesis of secondary hyperpara- thyroidism (HPT) as detailed in Chapter 4.2. The potential risks of vitamin D repletion are minimal, and thus, despite uncertain benefit, the Work Group felt that measurement might be beneficial.

The prevalence of vitamin D insufficiency or deficiency varies by the definition used. Most studies define deficiency as serum 25(OH)D (calcidiol) values <10 ng/ml (25 nmol/l), and insufficiency as values >10 but <20-32 ng/ml (50-80 nmol/l).^{36, 37} However, there is no consensus on what defines 'adequate' vitamin D levels or toxic vitamin D levels,³⁸ although some believe a normal level is that which is associated with a normal serum PTH level in the general population, whereas others define it as the level above which there is no further reciprocal reduction in serum PTH upon vitamin D supplementation.^{39, 40} Numerous publications have found associations of vitamin D deficiency, usually defined as serum 25(OH)D values<10 or 15 ng/ml (<25 or 37 nmol/l), to be associated with various diseases.^{41, 42} In the general population^{43, 44} and in patients with CKD,⁴⁵ there is an association of low 25(OH)D levels with mortality. There is one prospective randomized controlled trial (RCT) in the general population that shows that vitamin D supplementation reduces the risk of cancer.⁴⁶ However, there are no data showing that the repletion of vitamin D to a specific 25(OH)D level reduces mortality.

Defining specific target or threshold levels in the current era is likely to be premature (see Recommendation 3.1.4),^{37, 42} and, in particular, using the criteria of a normal serum PTH level as vitamin D adequacy in CKD is problematic because of the multiple factors that affect PTH synthesis, secretion, target tissue response, and elimination in CKD. Studies in CKD patients and in the general population show widespread vitamin D deficiency; according to some definitions, almost 50% of those studied have suboptimal levels. In patients with CKD stages 3-4, some studies report lower 25(OH)D levels with more advanced stages of CKD.^{28, 47, 48} However, the Study To Evaluate Early Kidney Disease detailed above found no relationship between the stage of CKD and calcidiol levels. In the Study To Evaluate Early Kidney Disease, black individuals had lower levels of calcidiol and higher levels of PTH than did white individuals, despite higher levels of calcium and phosphorus.⁴⁹

Although position statements defining vitamin D deficiency exist, the definition of what level of vitamin D represents sufficiency is the subject of an ongoing debate. There are no data that the presence or absence of CKD would alter recommended levels. From a practical perspective, clinicians should also appreciate that--in the absence of knowing the optimum level, and with all the issues related to the measurement of serum levels of vitamin D sterols—the decision of whether to measure, when to measure, how often, and to what target level needs to be individualized. Furthermore, considerations as to how the information would impact management and treatment decisions should be considered on an individual patient basis, as well as by considering the impact on health-care resources/costs, where applicable. As detailed in Chapter 4.2, in patients with CKD stages 3 and 4, vitamin D deficiency may be an underlying cause of elevated PTH, and thus there is a rationale for measuring and supplementing in this population, although this approach has not been tested in a prospective RCT.

3.1.4 In patients with CKD stages 3-5D, we recommend that therapeutic decisions be based on trends rather than on a single laboratory value, taking into account all available CKD-MBD assessments (1C).

The interpretation of biochemical and hormonal values in the diagnosis of CKD-MBD requires an understanding of assay type and precision, interassay variability, blood sample handling, and normal postprandial, diurnal, and seasonal variations. Owing to these assay and biological variation issues, the Work Group felt that trends in laboratory values should be preferentially used over single values for determining when to initiate and/or adjust treatments

Table 13 describes the sources and magnitude of variation in the measurement of serum calcium, phosphorus, PTH, and vitamin D sterols. This table serves as a guide for clinicians and forms the basis for the recommendation that laboratory tests should be measured using the same assays, and at similar times of the day/week for a given patient. Health-care providers should be familiar with assay problems and limitations (discussed below). Furthermore, an appreciation of this variability further underscores the importance of utilizing trends, rather than single absolute values, when making diagnostic or treatment decisions.

3.1.5 In patients with CKD stages 3-5D, we suggest that individual values of serum calcium and phosphorus, evaluated together, be used to guide clinical practice rather than the mathematical construct of calcium-phosphorus product (Ca X P) (2D).

The mathematical construct of the calcium X phosphorus product (Ca X P) is of limited use in clinical practice, as it is largely driven by serum phosphorus and generally does not provide any additional information beyond that which is provided by individual measures.^{50, 51} The measurement of phosphorus is generally valid and reproducible, but is affected by diurnal and postprandial variation. Values may differ substantially (for example, up to 0.08 mg/dl; 0.026 mmol/l) in dialysis patients, depending on which shift or which interdialytic interval is chosen.³³ Furthermore, there are multiple situations in which a normal product is associated with poor outcomes, and the converse is similarly true. Thus, the Work Group advised against a reliance on this combined measurement in clinical practice.

3.1.6 In reports of laboratory tests for patients with CKD stages 3-5D, we recommend that clinical laboratories inform clinicians of the actual assay method in use and report any change in methods, sample source (plasma or serum), and handling specifications to facilitate an appropriate interpretation of biochemistry data (1B).

The use of biochemical assays for the diagnosis and management of CKD-MBD requires some understanding of assay characteristics and limitations, discussed by each assay below. The understanding of these sources of variability should allow clinicians and health-care providers to optimize the performance and interpretation of laboratory tests in CKD patients (for example, timing, location, laboratory used, and so on). Clinical laboratories should assist clinicians in the interpretation of data by reporting assay characteristics and kits used.

Calcium

Serum calcium levels are routinely measured in clinical laboratories using colorimetric methods in automated machines. There are quality control standards utilized by clinical laboratories. Thus, the assay is generally precise and reproducible. In healthy individuals, serum calcium is tightly controlled within a narrow range, usually 8.5-10.0 or 10.5 mg/dl (2.1-2.5 or 2.6 mmol/l), with some, albeit minimal, diurnal variation.⁵² However, the normal range may vary slightly from laboratory to laboratory, depending on the type of measurement used. In patients with CKD, serum calcium levels fluctuate more, because of altered homeostasis and concomitant therapies. In those with CKD stage 5D, there are additional fluctuations in association with dialysis-induced changes, hemoconcentration, and subsequent hemodilution. Moreover, predialysis samples collected from HD patients after the longer interdialytic interval during the weekend, as compared with predialysis samples drawn after the shorter interdialytic intervals during the week, often contain higher serum calcium levels. In the international Dialysis Outcomes and Practice Pattern Study, the mean serum calcium measured immediately before the Monday or Tuesday sessions was higher by 0.01 mg/dl (0.0025 mmol/l) than that measured before the Wednesday or Thursday sessions.³³

The serum calcium level is a poor reflection of overall total body calcium. Only 1% of total body calcium is measurable in the extracellular compartment. The remainder is stored in bone. Serum ionized calcium, generally 40-50% of total serum calcium, is physiologically active, whereas non-ionized calcium is bound to albumin or anions such as citrate, bicarbonate, and phosphate, and is therefore not physiologically active. In the presence of hypoalbuminemia, there is an increase in ionized calcium relative to total calcium; thus, total serum calcium may underestimate the physiologically active (ionized) serum calcium. A commonly used formula for estimating ionized calcium from total calcium is the addition of 0.8 mg/dl (0.2 mmol/l) for every 1 g decrease in serum albumin below 4g/dl (40 g/l). This 'corrected calcium' formula is routinely used by many dialysis laboratories and in most clinical trials. Unfortunately, recent data have shown that it offers no superiority over total calcium alone and is less specific than ionized calcium measurements.⁵³ In addition, the assay used for albumin may affect the corrected calcium measurement.⁵⁴ However, ionized calcium measurement is not routinely available and, in some instances, may require additional costs for measuring and reporting. Presently, most databases are already using the corrected calcium formula and there is an absence of data showing differences in treatment approach or clinical outcomes when using corrected vs total or ionized calcium. The Work Group did not recommend that corrected calcium measurements be abandoned at present. Furthermore, the use of ionized calcium measurements is currently not considered to be practical or cost effective.

Phosphorus

Inorganic phosphorus is critical for numerous normal physiological functions, including skeletal development, mineral metabolism, cell-membrane phospholipid content and function, cell signaling, platelet aggregation, and energy transfer through mitochondrial metabolism. Owing to its importance, normal homeostasis maintains serum concentrations between 2.5-4.5 mg/dl (0.81-1.45 mmol/l). The terms, phosphorus and phosphate, are often used interchangeably, but strictly speaking, the term phosphate means the sum of the two physiologically occurring inorganic ions in the serum, and in other body fluids, hydrogenphosphate (HPO42-) and dihydrogenphosphate (H2PO4-). However, most laboratories report this measurable, inorganic component as phosphorus. Unlike calcium, a major component of phosphorus is intracellular, and factors such as pH and glucose can cause shifts of phosphate ions into or out of cells, thereby altering the serum concentration without changing the total body phosphorus.

Phosphorus is routinely measured in clinical laboratories with colorimetric methods in automated machines. There are quality control standards used by clinical laboratories. Thus, the assay is generally precise and reproducible. Levels will be falsely elevated with hemolysis during sample collection. In healthy individuals, there is a diurnal variation in both serum phosphorus levels and urinary phosphorus excretion. Serum phosphorus levels reach a nadir in the early hours of the morning, increasing to a plateau at 1600 hours, and further increasing to a peak from 0100 to 0300 hours.^{55, 56} Similar results were found in patients with hypercalcuria and nephrolithiasis.⁵⁷ However, another study found no diurnal variation in patients on dialysis when studied on a non-dialysis day.⁵⁸ There are usually higher levels after a longer period of dialysis. In the international Dialysis Outcomes and Practice Pattern Study, samples collected from HD patients immediately before a Monday or Tuesday session vs a Wednesday or Thursday session were higher by 0.08 mg/dl (0.025 mmol/l).³³

Thus, the measurement of phosphorus is generally valid and reproducible, but may be affected by normal diurnal and postprandial variation. Again, trends of progressive increase or decrease may be more accurate than small variations in individual values.

Parathyroid hormone

PTH is cleaved to an 84-amino-acid protein in the parathyroid gland, where it is stored with fragments in secretory granules for release. Once released, the circulating 1-84-amino-acid protein has a half-life of 2-4 min. The hormone is cleaved both within the parathyroid gland and after secretion into the N-terminal, C-terminal, and mid-region fragments of PTH, which are metabolized in the liver and in the kidneys. Enhanced PTH synthesis/secretion occurs in response to hypocalcemia, hyperphosphatemia, and/or a decrease in serum 1,25-dihydroxyvitamin D (1,25(OH)2D), whereas high serum levels of calcium or calcitriol--and, as recently shown, of FGF-2359--suppress PTH synthesis/secretion. The extracellular concentration of ionized calcium is the most important determinant of the minute-to-minute secretion of PTH, which is normally oscillatory. In patients with CKD, this normal oscillation is somewhat blunted.⁶⁰

There has been a progression of increasingly sensitive assays developed to measure PTH over the past few decades (Figure 6). Initial measurements of PTH using C-terminal assays were inaccurate in patients with CKD because of the impaired renal excretion of C-terminal fragments (and thus retention) and the measurement of these probably inactive fragments. The development of the N-terminal assay was initially thought to be more accurate but it also detected inactive metabolites.

Figure 6 | PTH assays. The figure shows the entire parathyroid hormone molecule, composed of 84 amino acids. Mid/C-PTH, mid/carboxyl-terminus of parathyroid hormone; N-PTH, amino-terminus of parathyroid hormone; PTH, parathyroid hormone; RIA, radioimmunoassay. Reprinted with permission from Moe and Sprague.⁷⁰

The development of a second generation of PTH assays (Figure 6), the two-site immunoradiometric assay--commonly called an 'intact PTH' assay--improved the detection of full-length (active) PTH molecules. In this assay, a captured antibody binds within the amino terminus and a second antibody binds within the carboxy terminus.61 Unfortunately, recent data indicate that this 'intact' PTH assay also detects accumulated large C-terminal fragments, commonly referred to as '7-84' fragments; these are a mixture of four PTH fragments that include, and are similar in size to, 7-84 PTH.⁶² In parathyroidectomized rats, the injection of a truly whole 1- to 84-amino-acid PTH was able to induce bone resorption, whereas the 7- to 84-amino-acid fragment was antagonistic, explaining why patients with CKD may have high levels of 'intact' PTH but relative hypoparathyroidism at the bone-tissue level.^{63, 64 and 65} Thus, the major difficulty in accurately measuring PTH with this assay is the presence of circulating fragments, particularly in the presence of CKD. Unfortunately, the different assays measure different types and amounts of these circulating fragments, leading to inconsistent results.⁶⁶

More recently, a third generation of assays has become available that truly detect only the 1- to 84-amino-acid, full-length molecule: 'whole' or 'bioactive' PTH assays (Figure 6). However, they are not yet widely available and have not been shown convincingly to improve the predictive value for the diagnosis of underlying bone disease⁶⁷ or other serum markers of bone turnover,⁶⁸ in contrast to at least one report that suggested that levels of 1-84 PTH or the 1-84 PTH/large C-PTH fragment ratio may be a better predictor of mortality in CKD stage 5 than standard 'intact' PTH values.⁶⁹ Therefore, the Work Group felt that the widely available second-generation PTH assays should continue to be used in routine clinical practice at present.

There are a number of commercially available kits that measure so-called 'intact' PTH with second-generation assays. Much of the literature and recommendations from KDOQI Bone and Mineral guidelines ⁵ were based on the second-generation Allegro assay from Nichols, which is not currently available. A study evaluated these other assays in comparison with the Allegro kit, using pooled human serum, and found intermethod variability in results because of standardization and antibody specificity. The different assays measured different quantities of both 7-84 and 1-84 PTH (when added to uremic serum).⁶⁶ In addition, there are differences in PTH results when samples are measured in plasma, serum, or citrate, and depending on whether the samples are on ice, or are allowed to sit at room temperature.^{71, 72}

Thus, these data--which describe problems with sample collection and assay variability--raise significant concerns with regard to the validity of absolute levels of PTH and their strict use as a clinically relevant biomarker for targeting specific values. Nevertheless, the clinical consequences of not measuring PTH and treating secondary HPT are of equal concern. In an attempt to balance the methodological issues of PTH measurement with the known risks and benefits of excess PTH and treatment strategies, the Work Group felt that PTH should be measured, with standardization within clinics and dialysis units in the methods of sample collection, processing, and assay used. In addition, the Work Group felt that trends in serum PTH, rather than single values, should be used in the diagnosis of CKD-MBD and in the treatment of elevated or low levels of PTH. However, 'systematic' unidirectional trends observed in the majority of patients in a single center should prompt suspicion that the central laboratory may have changed the assay. The Work Group also felt that using narrow ranges of PTH defining an 'optimal' or 'target' range was neither possible nor desirable.

To ensure that the reader of this guideline is clear on the difference between these compounds, and to ensure the use of consistent nomenclature in clinical practice, Table 14 is provided. Following the table is an in-depth discussion relating to the assays and measurement of these compounds.

Assays of serum vitamin D metabolites

25(OH)D. The parent compounds of vitamin D—D₃ (cholecalciferol) or D₂ (ergocalciferol)--are highly lipophilic. They are difficult to quantify in the serum or plasma. They also have a short half-life in circulation of about 24 h. These parent compounds are metabolized in the liver to 25(OH)D₃ (calcidiol) or 25(OH)D₂ (ercalcidiol). Collectively, they are called 25(OH)D or 25-hydroxyvitamin D. The measurement of serum 25(OH)D is regarded as the best measure of vitamin D status, because of its long half-life of approximately 3 weeks. In addition, it is an assessment of the multiple sources of vitamin D, including both nutritional intake and skin synthesis of vitamin D. There is a seasonal variation in calcidiol levels because of an increased production of cholecalciferol by the action of sunlight on skin during summer months.

There are three types of assays for measuring calcidiol. Fortunately, unlike PTH, the specimen collection process is well standardized and the sample is stable over time. However, there are real differences in measurement methods. The gold standard of calcidiol measurement is high-performance liquid chromatography (HPLC), but this is not widely available clinically. This is because HPLC is time consuming, requires expertise and special instrumentation, and is expensive. In early 1985, Hollis and Napoli⁷³ developed the first radioimmunoassay (RIA) for total 25(OH)D, which was co-specific for 25(OH)D₂ and 25(OH)D₃. The values correlated with those obtained from HPLC analysis, and DiaSorin RIA became the first test to be approved by the Food and Drug Administration for use in clinical settings.⁷³ Subsequent developments led to the automation of the test. Nichols developed a fully automated chemiluminescence assay in 2001, allowing clinical laboratories the ability of rapid and large-volume detection. However, this assay was removed from the market in 2006. In 2004, DiaSorin (Stillwater, MN, USA) introduced its fully automated chemiluminescence assay, which, similar to its RIA, is co-specific for 25(OH)D₂ and 25(OH)D₃, reporting 'total' 25(OH)D concentration. This assay has recently been updated as a 'second-generation' assay with an improved assay precision.^{37, 74} Additional manufacturers, IDS (Fountain Hills, AZ, USA) and Roche Diagnostics (Burgess Hill, West Sussex, UK) also make automated RIAs and/or enzyme-linked immunosorbent assay tests, but there are only limited publications thus far. In the majority of reports in this field, the DiaSorin assay was used.

Another method now carried out is liquid chromatography-tandem mass spectrometry (LC-MS/MS). Similar to HPLC, the LC-MS/MS method also has the ability to quantify 25(OH)D₂ and 25(OH)D₃ separately, which distinguishes it from RIA and enzyme-linked immunosorbent assay technologies. This method is very accurate and has been shown to correlate well with DiaSorin RIA.^{75, 76} Next to DiaSorin assays, LC-MS/MS is the most frequently used procedure for the clinical assessment of circulating 25(OH)D.37 However, most clinical laboratories do not use this technique because of the substantial cost and need for highly trained operators. Only HPLC and LC-MS/MS can differentiate 25(OH)D₂ and 25(OH)D₃, whereas RIA and automated chemiluminescence technologies only measure total 25(OH)D--the sum of 25(OH)D₂ and 25(OH)D₃. There is controversy as to whether the ability to differentiate these metabolites is important, as they have similar biological effects.^{77, 78}

A recent study by Binkley et al.79 analyzed blood obtained from 15 healthy adults for 25(OH)D. Aliquots of serum from all volunteers and a calibrator (known to contain 30 ng/ml (75 nmol/l) 25(OH)D by HPLC) were sent to four laboratories. The methods used for 25(OH)D measurement included HPLC, LC-MS/MS in two laboratories, and RIA (DiaSorin). A good correlation was observed for 25(OH)D measurement among the laboratory using HPLC, the two laboratories using LC-MS/MS, and the laboratory using RIA (R2=0.99, 0.81, and 0.95, respectively). The classification of clinical vitamin D status as optimal or low was identical for 80% of the 15 individuals in all four laboratories. However, 20% would be variably classified depending on the laboratory used. A modest interlaboratory variability was noted, with a mean bias of the laboratories using LC-MS/MS and RIA being from +2.9 to +51 ng/ml (+7.2 to +127 nmol/l) when compared with the laboratory using HPLC. They found that a systematic bias led to 89% of values being higher in the non-HPLC laboratories, and that a correction of the 25(OH)D value using a single calibrator at all sites for all assays reduced the mean interlaboratory bias. This suggests that the use of a standard calibrator may increase agreement among laboratories.

Thus, the Work Group advises that clinicians should be aware of the assay methods when assessing vitamin D status. Currently, the assays for 25(OH)D are not well standardized, and the definition of deficiency is not yet well validated. At best, clinicians should ensure that patients use the same laboratory for measurements of these levels, if carried out. The most appropriate vitamin D assays presently available seem to be those that measure both 25(OH)D₂ and 25(OH)D₃. Presently, approximately 20-50% of the general population has low vitamin D levels, irrespective of CKD status. However, the benefits from replacing vitamin D have not been documented in patients with CKD, particularly if they are taking calcitriol or a vitamin D analog. Therefore, the utility of measurement is unclear, outside of clinical trial or research situations. Furthermore, there are no data indicating that the measurement is helpful in guiding therapy or in predicting outcomes in CKD, although vitamin D deficiency may be a treatable cause of secondary HPT, especially early in the course of CKD. The risk, benefit, and costs of testing in patients should be balanced with practical issues related to treatment trials.

1,25(OH)2D. 1,25(OH)2D is used to describe both hydroxylated D₂ (ercalcitriol) and D₃ (calcitriol) compounds, both of which have a short half-life of 4-6 h. Commercially available assays do not distinguish between 1,25(OH)2D₂ and 1,25(OH)2D₃, and there are insufficient data to support the different biological effects of these compounds. The gold standard for assessment of 1,25(OH)2D is HPLC, and only a small number of kits are available for routine measurement. Circulating levels of 1,25(OH)2D are approximately 1/1000th that of 25(OH)D. The measurement of 1,25(OH)2D will be affected by both the stores of 25(OH)D and the multiple factors that convert 25(OH)D to 1,25(OH)2D by the 25(OH)D-1alpha;-hydroxylase enzyme (CYP27B1), as well as its inactivation by the 24(OH)D hydroxylase enzyme (CYP24A1) to 1,24,25(OH)3D and other inactivation steps. The renal CYP27B1 is regulated by nearly every hormone involved in calcium homeostasis. Its activity is stimulated by PTH, estrogen, calcitonin, prolactin, growth hormone, low calcium, and low phosphorus, and is inhibited by its product 1,25(OH)2D, FGF-23, and metabolic acidosis. Recent data show that multiple other tissues and cells also have CYP27B1 activity, which is believed to have autocrine/paracrine functions.80 This extrarenal 1alpha;-hydroxylase does not seem to be regulated by factors related to calcium homeostasis, suggesting a role for the extrarenal production of 1,25(OH)2D other than that involved in mineral metabolism.

Furthermore, in patients with earlier stages of CKD and in the general population, mild-to-moderate vitamin D deficiency, or partly treated vitamin D deficiency, is frequently associated with increased levels of 1,25(OH)2D. Thus, even accurate levels can be misleading. The serum levels of 1,25(OH)2D are uniformly low in late stages of CKD-MBD, at least in patients not treated with vitamin D derivatives.

Thus, the Work Group did not recommend a routine measurement of 1,25(OH)2D levels, as the assays are not well standardized, the half-life is short, the measurement will be artificially altered by the exogenous administration of calcitriol and vitamin D analogs, and there are no data indicating that the measurement is helpful in guiding therapy or predicting outcomes.

Alkaline phosphatases

Alkaline phosphatases are enzymes that remove phosphate from proteins and nucleotides, functioning optimally at alkaline pH. Measurement of the level of t-ALP is a colorimetric assay that is routinely used in clinical laboratories in automated machines, with quality control standards routinely used. The enzyme is found throughout the body in the form of isoenzymes that are unique to the tissue of origin. Highest concentrations are found in the liver and bone, but the enzyme is also present in the intestines, placenta, kidneys, and leukocytes. Specific ALP isoenzymes to identify the tissue source can be determined after fractionation and heat inactivation, but these procedures are not widely available in clinical laboratories. Bone-specific ALP (b-ALP) is measured with an immunoradiometric assay. Elevated levels of t-ALP are generally due to an abnormal liver function (in which case, other tests are also abnormal), an increased bone activity, or bone metastases. Levels are normally higher in children with growing bones than in adults, and often are increased after fracture. In addition, t-ALP and b-ALP can be elevated in both primary and secondary HPT, osteomalacia, and in the presence of bone metastasis and Paget's disease.

The Work Group recommended that the measurement of t-ALP in the diagnosis and assessment of CKD-MBD may be used as an adjunct test, but if values are high, then liver function tests should be checked. t-ALP could reasonably be used as a routine test to follow response to therapy. The more expensive testing for b-ALP can be used when the clinical situation is more ambiguous. Relationships between b-ALP and bone turnover are discussed in the following chapter. However, testing for t-ALP is inexpensive and therefore may be helpful for following patients' response to therapy or determining bone turnover status when the interpretation of PTH is unclear. The use of b-ALP, an indicator of bone source, may provide additional and more specific information, although it is not readily available. Clinicians should consider the adjunct value of these tests in treating individual patients in the context of the caveats described above.

RESEARCH RECOMMENDATIONS

It is important to emphasize that CKD-MBD is a complex disorder affecting those at all stages of CKD. An understanding of the complex biology in combination with the complexity of measurement issues is of tantamount importance, if eventually the appropriate RCTs of treatment are to be conducted. Many different kinds of studies are required to further our knowledge. As it pertains to the recommendations and suggestions described in this chapter of diagnosis and monitoring, the key areas for research to address in the area of measurement and assay variability are listed below:

• To increase the understanding of inter- and intraindividual variations in the laboratory parameters of CKD-MBD, registries (for those in stages 3-4, on dialysis, and those with kidney transplants) should endeavor to collect serial data on CKD-MBD laboratory information.
• To ensure comparability between and within cohorts/facilities and countries and thus ensure the transferability of knowledge, there is a need to establish standards for all relevant laboratory parameters, including assays, handling, and timing of specimen collection.
• To conduct international trials (cohort, observational, or treatment), and to facilitate the appropriate uptake of study information, there is a need for the creation of an international registry to oversee and review the standardization of measurement methods. This group would necessarily work with pathology/laboratory medicine organizations to facilitate the implementation of these standards.
• To establish CKD cohort-specific ranges of normal and pathological values, there is a need to ensure the systematic collection of longitudinal prospective observational data and outcomes. Specific cohorts, about whom little is known about initial and serial 'expected' or acceptable values, include those initiating dialysis (with and without earlier CKD care), those receiving kidney transplants, and those on home-based therapies.

Chapter 3.2: Diagnosis of CKD-MBD: bone

INTRODUCTION

The bone-disease component of CKD-MBD may result in fractures (including asymptomatic fractures seen on vertebral radiographs), bone pain, deformities in growing children, reduced growth velocity, and abnormal height. Complications of hip fractures include bleeding, infection, loss of independence, and increased mortality. Vertebral fractures lead to height loss, reduced pulmonary function, gastrointestinal reflux, and chronic disability. In children, growth retardation and skeletal deformities reduce quality of life. In clinical studies of bone disease, surrogate outcomes are bone density and findings on bone biopsies. Potential surrogate outcomes are serum biochemical markers of bone resorption and bone formation.

It is important to recognize that most patients with postmenopausal or age-related osteoporosis also have early stages of CKD (stages 1 through, perhaps, to early stage 3). Patients with more advanced stages of CKD (stages 3-5D), in whom the biochemical abnormalities of mineral metabolism that define CKD-MBD are present, have renal osteodystrophy. Both idiopathic osteoporosis and renal osteodystrophy can lead to increased bone fragility and fractures, but these diseases have different pathophysiological backgrounds. Bone fragility is due to varying combinations of low bone mineral content and abnormal bone quality. CKD-MBD can lead to an abnormal bone quality even in the setting of a normal or high bone-mineral content, and the gold standard diagnosis for the bone component of CKD-MBD is bone biopsy-based histologic analysis. Osteoporosis is traditionally diagnosed as low BMD. Given these pathophysiological and diagnostic differences, the definition of 'osteoporosis' in adults is most appropriate only for those with CKD stages 1-3; in later CKD stages, those with low BMD should be designated as having 'CKD-MBD with low BMD.'

RECOMMENDATIONS

3.2.1 In patients with CKD stages 3-5D, it is reasonable to perform a bone biopsy in various settings including, but not limited to: unexplained fractures, persistent bone pain, unexplained hypercalcemia, unexplained hypophosphatemia, possible aluminum toxicity, and prior to therapy with bisphosphonates in patients with CKD-MBD (not graded).

3.2.2 In patients with CKD stages 3-5D, with evidence of CKD-MBD, we suggest that BMD testing not be performed routinely, because BMD does not predict fracture risk as it does in the general population, and BMD does not predict the type of renal osteodystrophy (2B).

3.2.3 In patients with CKD stages 3-5D, we suggest that measurements of serum PTH or bone-specific alkaline phosphatase can be used to evaluate bone disease because markedly high or low values predict underlying bone turnover (2B).

3.2.4 In patients with CKD stages 3-5D, we suggest not to routinely measure bone-derived turnover markers of collagen synthesis (such as procollagen type I C-terminal propeptide) and breakdown (such as type I collagen cross-linked telopeptide, cross-laps, pyridinoline, or deoxypyridinoline) (2C).

3.2.5 We recommend that infants with CKD stages 2-5D have their length measured at least quarterly, while children with CKD stages 2-5D should be assessed for linear growth at least annually (1B).

Summary of rationale for recommendations

• Patients with CKD stages 3-5, 5D, and 1-5T have an increased risk of fracture compared with the general population. These fractures are associated with increased morbidity and mortality.
• Fracture risk relates to bone mineral density and bone quality, together with risk for falling and trauma.
  Bone biopsies provide measurements of bone turnover, mineralization, and volume. These help to assess bone quality and the underlying physiology. The histology is variable and influenced by many factors, including stage of CKD, serum biochemistries, age, and treatments. The different types of renal osteodystrophy have only modest relationships with clinical outcomes.
• In patients with CKD stages 4-5D, BMD of the hip and radius is generally lower than that in the general population; lumbar spine BMD is similar to that in the general population.
• In the general population, a low BMD predicts fracture and mortality. The ability of BMD to predict fractures or other clinical outcomes in patients with CKD stages 4-5D is weak and inconsistent. BMD in patients with CKD stages 3-5D does not distinguish among types of renal osteodystrophy, as seen with bone histology.
• There are no longitudinal studies of changes in BMD in patients with CKD stages 4-5.
• PTH is one important factor that affects bone physiology. ALP may reflect osteoblast activity. Serum measurements of PTH and ALP are related to clinical outcomes, including relative risk of mortality. They also correlate with some of the histomorphometric measurements.
• Serum biochemical markers of bone turnover show correlations with findings on bone biopsies, but their diagnostic utility is limited and these serum tests have not been directly related to clinical outcomes, except ALPs and extreme values of PTH.
  An alteration in growth in infants and children is a sensitive indicator of the presence of CKD-MBD.

BACKGROUND: FRACTURES IN CKD PATIENTS

Prevalence

Abnormal bone quality and quantity can lead to increased bone fragility, resulting in fracture. In 1966, Pendras and Erickson81 reported their experience with the first 22 patients to receive long-term HD. Bone and mineral disorders emerged as one of the most troublesome complications; fractures occurred in 47% of the patients. Since then, several studies of fracture prevalence and incidence have been reported, with a prevalence from 10 to 40% in general dialysis populations and in approximately half of patients older than 50 years (Supplementary Table 4). The incidence rate of hip fractures in all patients who started dialysis in the United States from 1989 to 1996 was 4.4 times higher than that in the residents of Olmstead County.⁸² Fractures occur more commonly in elderly patients, in women, in diabetic patients, in those using glucocorticoids, and in those with a longer exposure to dialysis. Fractures are also common in elderly patients with CKD stages 3-4 (Supplementary Table 5). Hip fractures were seen two to three times more often than in persons without CKD.

Increased risk of another fracture

In the general population, previous fractures as an adult are strongly associated with the risk of a subsequent fracture. This is independent of age, bone density, or other identified risk factors.^{83, 84} and ⁸⁵ Among US women older than 65 years, those who had a vertebral fracture as seen on a spine radiograph were 5.4 times more likely to experience a new vertebral fracture in the next 3.7 years compared with women without a prevalent fracture. Even when adjusted for age and bone density, the risk was 4.1 times higher.⁸⁶ Similar findings are reported in several cohort studies and in the placebo groups of clinical trials.85 The World Health Organization fracture assessment tool includes earlier fracture after 50 years of age as one of the clinical risk factors, with a risk ratio for hip fracture of 1.⁸⁵ without BMD, and 1.62 including BMD in the model.⁸⁷ The risk of a new vertebral fracture increases with the higher number and severity of fractures seen on spine radiographs, but even a mild asymptomatic fracture of one vertebra is associated with a significantly increased risk.⁸³ However, it is important for clinicians to appreciate that these vertebral fractures do not cause increased back pain in about 60% of cases, and that a severe loss of vertebral height can be asymptomatic.⁸⁸

In patients with CKD stage 5D, one study⁸⁹ found that a vertebral fracture identified on a radiograph increased the risk of a new fracture by over sevenfold.

Mortality

Mortality in patients with CKD stage 5 who have had a hip fracture is about twice as high as that in patients of similar age and gender who have not had a fracture (Supplementary Table 7). Coco et al. followed up 1272 HD patients over 10 years and observed that the mortality for CKD stage 5 patients with a hip fracture was 2.7 times higher than that in fracture-free HD patients and 2.4 times higher than that in patients without CKD who had a fractured hip.⁹⁰ Three studies used data from the US Renal Data System. Mittalhenkle et al.⁹¹ recorded hip fracture cases over 5.5 years, and the mortality incidence was 2.15 times higher in cases than in controls matched for age, duration of dialysis, and cardiovascular risk scores. Adjusting for multiple risk factors resulted in an RR of 1.99 for mortality associated with hip fracture. Danese et al.⁸⁹ evaluated 9007 patients and found that a history of hip, vertebral, or pelvic fracture was associated with an age- and sex-adjusted mortality rate that was 2.7 times higher than that for the other dialysis patients. Kaneko et al.⁹² found that the adjusted hazard ratio for mortality was 1.95 in patients with long bone fractures, using data from 7159 individuals in the Dialysis Morbidity and Mortality Study.

This topic represents a comprehensive review of the literature of selected topics by the Work Group with assistance from the evidence review team to formulate the rationale for clinical recommendations. Thus, this should not be considered to be a systematic review.

RATIONALE

3.2.1 In patients with CKD stages 3-5D, it is reasonable to perform a bone biopsy in various settings, including, but not limited to: unexplained fractures, persistent bone pain, unexplained hypercalcemia, unexplained hypophosphatemia, possible aluminum toxicity, and prior to therapy with bisphosphonates in patients with CKD-MBD (not graded).

Abnormal bone histology, diagnosed by bone biopsy with histomorphometry, has been the primary tool used to diagnose and classify renal osteodystrophy. Although bone biopsy is invasive and thus cannot be performed easily in all patients, it is the gold standard for the diagnosis of renal osteodystrophy. As detailed below, renal osteodystrophy is a complex disorder and biochemical assays do not adequately predict the underlying bone histology. Thus, bone biopsy should be considered in patients in whom the etiology of clinical symptoms and biochemical abnormalities is not certain. Aluminum bone disease, although less common in the current era, also requires a bone biopsy for diagnosis in many individuals, as detailed in the KDOQI Bone and Mineral guidelines^.5 A bone biopsy should be considered in patients before treatment with bisphosphonates, because bone biopsy is the most accurate test for the diagnosis of adynamic bone disease, and the presence of adynamic bone disease is a contraindication to bisphosphonates. Thus, the Work Group encourages the continued training of nephrologists in the performance and interpretation of bone biopsies.

Classification of renal osteodystrophy by bone biopsy

Bone biopsies are performed to understand the pathophysiology and course of bone disease, to relate histological findings to clinical symptoms of pain and fracture, and to determine whether treatments are effective. The traditional types of renal osteodystrophy have been defined on the basis of turnover and mineralization as follows: mild, slight increase in turnover and normal mineralization; osteitis fibrosa, increased turnover and normal mineralization; osteomalacia, decreased turnover and abnormal mineralization; adynamic, decreased turnover and acellularity; mixed, increased turnover with abnormal mineralization.

A recent Kidney Disease: Improving Global Outcomes report² has suggested that bone biopsies in patients with CKD should be characterized by determining bone turnover, mineralization, and volume (TMV). Thus, in this guideline document, we have endeavored to examine data from published literature and report it using this TMV system.

Turnover. Patients with CKD display a spectrum of bone-formation rates from abnormally low to very high. Other measurements that help to define a low or high turnover (such as eroded surfaces, number of osteoclasts, fibrosis, or woven bone) tend to be associated with the bone-formation rate as measured by tetracycline labeling. This is the most definite dynamic measurement, hence it was chosen to represent bone turnover. It should be noted that an improvement of a bone biopsy cannot be determined on the basis of a simple change in the bone-formation rate, because the restoration of normal bone may require either an increase or a decrease in bone turnover, depending on the starting point.

Mineralization. The second parameter is mineralization, which reflects the amount of unmineralized osteoid. Mineralization is measured by the osteoid maturation time or by mineralization lag time, both of which depend heavily on the osteoid width as well as on the distance between tetracycline labels. The classic disease with an abnormality of mineralization is osteomalacia, in which the bone-formation rate is low and the osteoid volume is high. Some patients have a modest increase in osteoid, which is a result of high bone-formation rates. They do not have osteomalacia because the mineralization lag time remains normal. The overall mineralization, however, is not normal because unmineralized osteoid is increased. Patients with low bone-formation rates and a normal osteoid have adynamic disease (they do not even form the osteoid matrix, hence they do not manifest a problem with mineralization).

Volume. The final parameter is bone volume, which has not traditionally been included in previous schemes for describing renal osteodystrophy. Bone volume contributes to bone fragility and is separate from the other parameters. The bone volume is the end result of changes in bone-formation and resorption rates: if the overall bone formation rate is higher than the overall bone resorption rate, the bone is in positive balance and the bone volume will increase. If mineralization remains constant, an increase in bone volume would also result in an increase in BMD and should be detectable by dual-energy X-ray absorptiometry (DXA). Although both cortical and cancellous bone volumes decrease in typical idiopathic osteoporosis, these compartments are frequently different in patients with CKD. For example, in dialysis patients with high PTH levels, the cortical bone volume is decreased but the cancellous volume is increased.⁹³

Prevalence of abnormalities on bone biopsies

A systematic literature review of the prevalence of types of bone disease in CKD is shown in Figure 7. The review analyzed studies carried out between 1983 and 2006. Differing prevalences of bone disease types observed between studies are due to differing classification methods, in addition to differences related to geographical areas, genetic background, and treatment modalities. One of the most problematic differences in classification relates to the bone-formation rate. This requires tetracycline labeling, and thus normal ranges cannot be determined on autopsy or surgical series. The reported normal bone-formation rates show inconsistencies and variations.²⁰

The prevalence of bone histology types in children with CKD-MBD is similar to that observed in adults. Figure 8 shows the results from 325 children who had CKD stages 5-5D.^{18, 34, 94, 95} and ⁹⁶

Natural history of bone biopsy findings

The distribution of histological types in patients with CKD stage 5 was compared in studies before 1995 and after 1995 (Figure 9). The studies also revealed a decreased aluminum intoxication, from 40% of biopsies carried out before 1995 to 20% in patients biopsied after 1995.

The natural history of bone disease evaluated through bone histomorphometry is variable. The placebo groups from RCTs and from one longitudinal study are shown in Table 15. The overall trend is toward a worsening turnover (either getting too high or too low) and stable mineralization.The wide variability in the natural history of bone histology reflects the complex pathophysiology of CKD-MBD (Supplementary Table 6). Another way to evaluate the natural history of bone disease in CKD is to compare bone volume by bone biopsy in predialysis patients with that in dialysis patients. Studies dating from 1969 to 2007 show that bone volume/trabecular volume (BV/TV) is generally lower in dialysis patients compared with that in non-dialysis CKD patients across all renal osteodystrophy categories.^{105, 106, 107, 108, 109, 110, 111} and ¹¹²

Figure 7 | Prevalence of types of bone disease as determined by bone biopsy in patients with CKD-MBD. Bone formation (turnover) is high in those with osteitis fibrosa and mild disease, and low in those with osteomalacia and adynamic bone disease. Mineralization is abnormal in those with osteomalacia and mixed disease. AD, adynamic bone; OF, osteitis fibrosa; OM, osteomalacia.

 

Figure 8 | Prevalence of histological types of renal osteodystrophy in children with CKD stages 5-5D. AD, adynamic bone; OF, osteitis fibrosa; OM, osteomalacia.

 

Figure 9 | Types of renal osteodystrophy before and after 1995. OF, osteitis fibrosa; OM, osteomalacia.

Relationship between bone biopsy findings and clinical outcomes

Symptoms. A further analysis was carried out on 20 of the above studies conducted in HD patients to examine the relationship of bone biopsy histology findings to clinical symptoms and changing trends over time (Figure 10). Most of these patients had been referred for some clinical reason (6505 patients), whereas the remaining patients were apparently asymptomatic (863 patients). There did not seem to be differences in the prevalence of histological types between referred and asymptomatic patients.

Fractures. Most of the studies of bone histomorphometry have not been designed to fully evaluate the relationship between fractures and types of renal osteodystrophy. One study of 31 dialysis patients found that those with low-turnover osteodystrophy had fracture rates of 0.2 per year compared with 0.1 per year in those with osteitis fibrosa; this was because of a high number of rib fractures in the low-turnover patients.¹¹³ A review of 2340 biopsies carried out in Brazilian patients for clinical indications found that the frequency of fractures was significantly higher in those with osteomalacia compared with that in other forms. There were no differences in fracture history between those with adynamic bone disease, high bone turnover, or mixed bone disease.¹¹⁴ A study that followed up 62 patients for 5 years after bone biopsy found a higher rate of fractures in those with adynamic bone disease.¹¹⁵

Theoretically, we would expect that persons with a lower bone volume would be more likely to suffer fractures. However, we could locate no reports of prospective studies of patients with a low bone volume to determine the subsequent fracture rate.

 

Figure 10 | Prevalence of bone histology types by symptoms in patients with CKD stage 5D receiving HD treatment. CKD, chronic kidney disease; HD, hemodialysis; mixed, mixed renal osteodystrophy; OF, osteitis fibrosa; OM, osteomalacia.

Cardiovascular calcification. Several studies have examined this issue. London et al.¹¹⁶ found that aortic calcification was increased in HD patients with adynamic bone disease. They subsequently, with an expanded cohort, reported a significant interaction between the dosage of calcium-containing phosphate binders and bone activity, such that calcium load had a significantly greater influence on aortic calcifications and stiffening in the presence of adynamic bone disease.¹¹⁷ In contrast, Barreto et al.,¹¹⁸ in their series of 98 HD patients, did not observe an association between type of bone disease and coronary artery calcification (CAC) on cross sectional analysis. A more recent prospective study in HD patients found that lower trabecular bone turnover was associated with CAC development, whereas an improvement in bone turnover was associated with lower CAC progression in patients with both high- and low-turnover bone disorders at baseline.¹¹⁹

3.2.2 In patients with CKD stages 3-5D with evidence of CKD-MBD, we suggest that BMD testing not be performed routinely, because BMD does not predict fracture risk as it does in the general population, and BMD does not predict the type of renal osteodystrophy (2B).

Bone density does not predict fractures very well in patients with CKD stages 4-5. In addition, no treatments have been shown to reduce fracture risk in those patients with CKD stages 3-5 who have low BMD and biochemical abnormalities of CKD-MBD (discussed in Chapter 4.3). Spine BMD measurements can be misleading if there are anatomic abnormalities in the bone, if there is extensive osteophyte formation, or if there is aortic calcification; hip measurements also can have positioning errors. Although forearm measurements provide the least ability to predict fractures in older persons without CKD, the meta-analysis by Jamal et al.¹²⁰ found that the forearm was the most sensitive site in patients with CKD stage 5D. The Work Group acknowledges that having a low DXA or a decreasing DXA value is indicative of abnormal bones. However, as detailed below, the etiology of the abnormal bone in CKD-MBD is complex, and patients with CKD-MBD and osteoporosis should not be assumed to benefit from therapies such as bisphosphonates provided in the general population. Thus, the Work Group could not recommend the routine use of DXA in these patients.

BMD measurements

Noninvasive techniques for measuring BMD include DXA and quantitative computed tomography (CT). Other methods have been used in some studies, but they do not have the same extensive reference database or utility in clinical trials as does DXA.

The skeleton is composed of cortical and trabecular (cancellous) bone. The trabecular bone is very porous: about 20% of the tissue is bone and the rest is marrow or fat. DXA cannot differentiate between cortical and trabecular bone. Certain sites, however, contain higher percentages of trabecular bone (by weight). The forearm is almost all cortical bone, the vertebral body is 42% trabecular bone,¹²¹ the proximal femur is about 25% trabecular, and the total body about 80% cortical. These distinctions are important because bone remodeling in patients with CKD-MBD is different in trabecular bone compared with cortical bone. Quantitative CT can separately measure cortical and trabecular bone because it is a three-dimensional measurement.

Figure 11 | Distribution of osteoporosis, osteopenia, and normal bone density by creatinine clearance in general US population. Reprinted with permission from Klawansky et al.¹²²

DXA measurements of the spine may also be inaccurate because of height. In children or short adults, DXA measurements seem lower than those in larger adults because the volume of bone increases at a faster rate than does the projected area of the bone. Thus, the interpretation of DXA results in children with growth delays must take into account the size of the bone.

BMD in patients with CKD stages 3-4

The Third National Health and Nutrition Examination Survey, 1988-1994, measured BMD and serum creatinine in 13,831 adults older than 20 years. On the basis of the Cockcroft-Gault equation, 23% of adult women with CKD stages 3-4 had osteoporosis (BMD at total hip<0.64g/cm²).¹²² As seen in Figure 11, the percentage of people with low BMD was much greater in those with CKD than in those with normal kidney function.

Not only do patients with CKD stages 3-4 have a high prevalence of low bone density but elderly patients with osteoporosis usually have CKD stage 3 or 4 (Figure 12). In the US population, 61% of women with osteoporosis had CKD stage 3 and 23% had CKD stage 4. Most of this overlap is seen because both CKD and bone loss increase considerably with aging. In osteoporotic women younger than 60 years of age, the prevalence of CKD stage 4 was very low.¹²²

A cross-sectional and longitudinal study of 1713 older men and women found a significant linear association between estimated GFR and hip bone density. The bone loss over 4 years was associated with estimated GFR as measured by the Cockcroft-Gault equation, but not by the Modification of Diet in Renal Disease equation.¹²⁴

Clinical trials of postmenopausal osteoporosis therapy generally exclude patients with known kidney disease, hence the proportion of patients with CKD in the trials is lower than that in the general population. Measurement of estimated GFR was lower than 45 ml/min per 1.73 m² in 3.8% of the individuals in the teriparatide trial,¹²⁵ in 52% of the individuals in the pooled risedronate trials,¹²⁶ and in 9% of the individuals in the alendronate Fracture Intervention Trial.¹²⁷

Figure 12 | Overlap between osteoporosis and CKD stages 3-4. This graph shows the overlap between osteoporosis and CKD stages 3-4 in women from the United States, using data from the NHANES III survey. The kidney function was estimated using the Cockcroft-Gault equation, which results in a greater prevalence of CKD stage 3-4 than when other methods are used.¹²³ CKD, chronic kidney disease; NHANES III survey, The Third National Health and Nutrition Examination Survey.

BMD in patients with CKD stage 5D

Figure 13 shows the average values of BMD in studies of patients with CKD stage 5D. These values are expressed as Z-scores, which compare BMD in patients with BMD from the reference values of age- and gender-matched persons in the community. The prevalence of low BMD is influenced by the age of the cohort, the number of men, the proportion of non-Caucasians, the average duration of dialysis, and the skeletal sites used to define osteoporosis.

Figure 13 | Bone mineral density in patients with CKD stage 5D. The graph is a summary of studies arranged in chronological order; each point is the mean value for a study. When more than one skeletal site or gender was measured in a study, the points are connected by a vertical line. If data from men and women were reported separately, the points for women are in a lighter shade. The size of points is larger in studies with greater numbers of individuals. Data from studies that reported g/cm² were converted to Z-scores (hip and forearm) using the average age of the group of individuals and published normal reference ranges. Overall, the average cortical bone density for patients with CKD stage s.d. was about 0.5-1 s.d. below that expected for age and gender, but at the spine, the bone density measurements were closer to the average in persons without known CKD.^{115, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181 and 182 CKD, chronic kidney disease.}

BMD and fractures in the general population

In 1994, the World Health Organization proposed guidelines for the diagnosis of osteoporosis on the basis of measurements of BMD.¹⁸³ Osteoporosis was defined as BMD lower than 2.5 s.d. from that of a young white female. In 2005, they reported a meta-analysis of data from 39,000 persons and found that BMD strongly predicted fractures. For example, at the age of 50 years, the RR of a hip fracture was 3.68 for each s.d. of hip BMD.¹⁸⁴ Although BMD is an important factor that predicts fracture, it does have limitations and it is not the only significant factor. In patients with osteoporosis, the degree of trauma and the quality of the bone also determine whether bones will fracture. The World Health Organization recently developed a method of assessing fracture risk on the basis of BMD and clinical risk factors: age, gender, race, weight, previous adult fracture, parental history of hip fracture, history of cigarette smoking, alcohol use, rheumatoid arthritis, and glucocorticosteroid use.¹⁸⁵ The equations used to calculate the risk score are derived from international studies of 46,340 persons and were validated in 230,486 persons, with a mean age of 63 years. The risk of a hip fracture was 4.2 times higher for every s.d. increase in the risk score.¹⁸⁶ These calculations of absolute fracture risk will apply to patients with CKD stages 1-3 but have not been studied in patients with CKD stages 4 and 5.

A definition of osteoporosis based on BMD does not distinguish among different etiologies. The ability of a BMD measurement to diagnose osteoporosis is similar to that of a hematocrit measurement to diagnose anemia. Just as there are different causes of anemia (such as iron deficiency or hemolytic anemia), there are different causes of low BMD (such as corticosteroid-induced osteoporosis, osteomalacia, myeloma, or renal osteodystrophy).

A relationship between BMD by DXA and fractures has also been recently shown in children without CKD. In over 7000 10-year-old children, a low BMD adjusted for size parameters was associated with an 89% increased risk of fractures in the subsequent 2 years.¹⁸⁷ In young adults and middle-aged men and women, there are no large studies relating fractures to DXA results.

BMD and fractures in CKD patients

In patients with CKD stage 5, the relationship between BMD and fractures is not as strong as that in the general population. We identified 13 cross-sectional studies that measured BMD and prevalent fractures; there were no prospective studies. The results were variable: five studies found no relationship between BMD and fracture rate,^{113, 115, 181, 188, 189} whereas eight studies found a relationship in at least one skeletal site.^{153, 157, 169, 175, 180, 190, 191 and 192} A meta-analysis by Jamal et al.¹²⁰ included six of these studies and found no increased risk of hip fracture related to BMD at the hip. The spine and distal radius BMD values, however, were significantly lower in patients who had a fracture than in those who did not. In a study of 187 men, Atsumi et al.¹⁵⁷ found that each s.d. lowering of spine bone density increased the odds ratio of a spine fracture by 2.0. Elder and Mackun¹⁸⁰ studied 242 patients and found a lower BMD at the hip in cases with fragility fractures, and a trend toward a lower spine BMD. Ersoy et al.181 studied 292 patients receiving peritoneal dialysis and found no relationship between BMD and fractures.

The reasons for the poor performance of DXA in patients with CKD are not defined. Partially, this is because the measurements may overestimate BMD due to arthritic conditions, scoliosis, and aortic calcifications, but those would apply mainly to the lumbar spine and not to the total hip. Another reason is that CKD patients have poor bone quality that cannot be measured by absorptiometry. Abnormal microarchitecture, mineralization density, crystal deposition in the bone matrix, or abnormalities in the matrix itself could all contribute to the loss of bone strength. Patients with CKD, especially those with a high serum PTH, have increased cancellous bone volume but decreased cortical thickness;⁹³ this can alter the relationship between the overall bone strength and BMD findings. Furthermore, patients with CKD may experience more trauma to the skeleton if they have more frequent falls.^{193, 194, 195 and 196}

BMD and relationship with bone biopsy findings in CKD

The relationship between BMD and bone biopsy is not well defined. In patients with postmenopausal osteoporosis, there is a significant but weak correlation between bone volume on biopsy and BMD measured by DXA. In patients with CKD, Lindergard93 measured BV/TV on 71 biopsies from dialysis patients, and did not see a correlation with BMD at the radius. Similar results were seen by Gerakis et al.115 in a study of 62 patients. Torres et al.,¹⁹⁷ on the other hand, found a correlation coefficient of 0.82 between BV/TV and quantitative CT of the spine, and Van Eps et al.¹⁹⁸ found lower DXA values in patients with low BV on biopsy.

Is BMD different among the different types of renal osteodystrophy? Studies of 20-30 patients found similar BMD in all the types.^{146, 167, 199, 200 and 201} Boling et al.²⁰¹ examined 27 patients; the types had similar values for BMD measured by DXA, but the spine quantitative CT was 5% above the normal mean in patients with a high bone turnover and 30% below the mean in those with a low turnover. In a study of 62 patients, Gerakis et al.115 found that BMD by DXA was lower in osteitis fibrosa than in adynamic bone, but there were wide ranges in both types. The BMD by DXA was lower in those with severe osteitis fibrosa in the study by Fletcher et al.²⁰² in 73 patients, particularly at the proximal forearm, in which the BMD Z-score was -1.94 in severe osteitis fibrosa compared with -0.17 in mild disease. The patients with adynamic disease also had a low forearm BMD with a Z-score of -1.85. At the spine, those with mixed lesions were 2.85 s.d. higher than normal, compared with -0.77 s.d. lower in those with severe osteitis fibrosa.

BMD and mortality

In the general population, low BMD is associated with mortality. In CKD, low BMD was also associated with mortality, as shown in a single study by Taal et al.²⁰³ (Supplementary Table 7) in 88 HD patients. The risk was 4.3 times higher in those with hip BMD T-scores lower than -2.5 (the World Health Organization criteria for diagnosis of osteoporosis).

3.2.3 In patients with CKD stages 3-5D, we suggest that measurements of serum PTH or bone-specific alkaline phosphatase can be used to evaluate bone disease because markedly high or low values predict underlying bone turnover (2B).

HPT is one of the most important causes of bone disease in patients with CKD. The circulating PTH is related to bone biopsy findings, but a prediction of the type of renal osteodystrophy may be inaccurate. Bone biopsy remains the gold standard for the assessment of bone turnover, and as detailed below, measurements of circulating PTH or b-ALP have limited sensitivity and specificitiy, especially in detecting adynamic bone. In addition, as detailed in Chapter 3.1, the availability of various assay kits for PTH is another problem. However, bone biopsy is not practical in the majority of clinical patients, and when these serum markers are above or below thresholds, they can be used to estimate bone turnover. Large discrepancies between serum PTH and ALP measurements should prompt further investigation.

Serum PTH and ALPs and bone outcomes

Fractures. There have been several large prospective studies in CKD stage 5D patients relating serum PTH to fractures (Supplementary Table 8), with inconsistent results, as shown in Table 16.Several other cross-sectional studies^{157, 175, 180, 189, 191, 192, 208} have also evaluated this relationship and, in general, were negative. However, a case-controlled cohort study did find a 31% (95% confidence interval 0.57-0.83, P<0.001) reduction in global fracture risk after parathyroidectomy.²⁰⁹

An association between high serum t-ALP levels and the RR of fractures has been reported in dialysis patients by Blayney et al.³²

PTH and b-ALP relationship with bone histology. The classic findings of HPT in patients with CKD are high turnover with peritrabecular fibrosis, active osteoclasts and increased numbers of multi-nucleated osteoclasts, woven bone, blurry tetracycline labels, increased cancellous bone volume but decreased cortical thickness, and intratrabecular tunneling. The bone response to PTH, however, is not consistent, and there is evidence for skeletal resistance to PTH in patients with CKD-MBD.

The results of studies that reported correlations between PTH and bone-formation rates are shown in Figure 14, which shows the wide variabilities seen in different situations.^{69, 99, 108, 111, 146, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223 and 224} The older studies tended to find better correlations between PTH and bone-formation rates, whereas more recent studies show poor correlations. This follows a trend for associating findings of adynamic bone disease with high PTH levels. The reasons for poor correlations between PTH and bone formation are not clear, but could involve differences in the assays for PTH, secular changes in the dialysis population with more diabetic and elderly patients, differences in therapies, and differences in the racial composition of the studies. This figure also shows correlations with several bone turnover markers. Osteocalcin is generally no better than intact PTH (iPTH), whereas b-ALP shows a higher correlation with tetracycline-based bone-formation rates and has a better correlation to bone-formation rate than does PTH. b-ALP also has some predictive value for the diagnosis of high or low bone turnover (Table 17).

Figure 14 | Correlation coefficients between bone formation rate as seen on bone biopsies and serum markers of PTH, bone-specific ALP (BAP), osteocalcin (OC), and collagen cross-linking molecules (x-link) in patients with CKD stages 5-5D. Each point represents a study, and they are arranged in chronological order from 1981 to 2006 from left to right. Studies that measured more than one marker are joined by a vertical line. The small symbols are studies of 20-50 patients, medium symbols 51-100 patients, and large symbols >100 patients. CKD, chronic kidney disease; PTH, parathyroid hormone.

Even though there is usually a significant but weak correlation between serum PTH or other markers and bone formation rates, the ability to correctly classify an individual patient is limited. As with any diagnostic test, there is a trade-off between the sensitivity and the specificity of the test. The predictive value depends on the sensitivity and specificity, and on the overall prevalence of the condition. It also depends on which cutoff was used and how the diseases were defined. Some of the differences among studies could be caused by a different exposure to aluminum, which increases the skeletal resistance to PTH. The studies also used different PTH assays, which may confound interpretation as detailed in Chapter 3.1. Table 17 shows the results that were reported in studies that measured PTH and types of bone disease in patients with CKD stage 5D. The positive predictive value is the percentage of patients with a positive result on a test who actually have the disease (either high or low turnover), and the sensitivity is the percentage of patients with the disease who have a positive test result.

Much of the focus of renal osteodystrophy has been on bone turnover, but bone volume is another important factor in bone physiology. The correlations between BV/TV and PTH are not consistent among studies; four studies found no correlation,^{220, 230, 231 and 232} one reported a correlation coefficient of 0.51,219 and another of 0.56.²¹³ b-ALP showed no relationship with bone volume in three studies;^{220, 230, 231} in another, the correlation was 0.54;²¹⁹ and in one, the correlation between b-ALP and BV/TV was not significant, but the b-ALP was lower in those who had histological signs of osteopenia.²³²

PTH relationship with BMD. Table 18 shows the results from studies that measured BMD and serum markers in at least 50 patients with CKD-MBD. None of the studies found a positive effect of PTH on BMD; either the relationship was not significant or there was a significant inverse correlation.

PTH and combinations of biochemistries in the prediction of bone histology. None of the studies published to date in CKD patients have been adequately powered to assess if combinations of PTH and other bone-derived circulating biomarkers would be more predictive than individual markers. Kidney Disease: Improving Global Outcomes is coordinating an ongoing international collaborative effort to determine the predictive value of whole (1-84) PTH assays compared with currently used iPTH assays, with or without other biomarkers, to predict underlying bone histology using the TMV classification system.

3.2.4 In patients with CKD stages 3-5D, we suggest not to routinely measure bone-derived turnover markers of collagen synthesis (such as procollagen type I C-terminal propeptide) and breakdown (such as type I collagen cross-linked telopeptide, cross-laps, pyridinoline, or deoxypyridinoline) (2C).

Collagen-based markers of bone turnover, measured in the serum, have not been extensively evaluated in patients with CKD stages 4-5. The available studies show that these markers do not predict clinical outcomes or bone histology any better than does circulating PTH or b-ALP. Therefore, at this time, they are not recommended for diagnostic purposes in patients with later stages of CKD-MBD. In earlier stages of CKD, some of these markers seem promising for monitoring the treatment of osteoporosis, but they currently are not sufficiently validated to recommend their use.

Bone markers

Collagen based. Active osteoblasts secrete procollagen type I, and the propeptides at both C- and N-terminal ends are immediately cleaved and can be measured in the circulation (PICP and PINP). The collagen molecules are then covalently bonded through pyridinoline cross-linking. The fragments containing these pyridinoline links (at both the C- and N-terminal ends of the peptides) are released during bone resorption: carboxyterminal (CTX) and aminoterminal (NTX) cross-linking telopeptide of bone collagen, respectively. These collagen-based markers have been studied in normal populations, wherein there are significant but moderate correlations with bone-formation/resorption rates.²⁴⁰ The markers are increased after a fracture.²⁴¹

Other bone markers. Osteoblasts secrete other proteins that have been used to assess their function, including b-ALP (discussed in the previous section), osteocalcin, osteoprotegerin, and receptor activator for nuclear factor kappa;B ligand.²⁴² Osteoclasts secrete tartrate-resistant acid phosphatase. Osteocytes secrete FGF-23 in response to phosphate and calcitriol. High levels of FGF-23 are seen in patients with CKD, but this is a new measurement, and clinical significance remains to be determined. FGF-23 was recently shown to be associated with an increased RR of mortality in dialysis patients,²⁴³ but this may be related to phosphate or vitamin D metabolism and not to bone disease per se. Thus, although synthesized in bone, it seems premature to use FGF-23 as a bone biomarker.

Some of these markers are excreted by the kidneys, so in CKD, the serum concentrations may merely represent accumulation instead of bone turnover.

Markers of bone turnover and clinical outcomes. In cohorts of elderly women, most of whom have early stages of CKD, serum biochemical markers of bone turnover have been associated with fractures.^{244, 245 and 246} The utility of these markers in individual patients is uncertain, and they are currently not recommended in the routine evaluation of patients with postmenopausal osteoporosis. These markers, however, may be helpful in identifying those patients who respond to osteoporosis medications. In the fracture intervention trial of alendronate, the change in b-ALP and CTX was significantly related to the reduction in fracture incidence, and for hip fractures, the changes in markers predicted fractures better than did the BMD changes.²⁴⁷ Furthermore, in those women who had postmenopausal osteoporosis with low baseline PINP levels, alendronate did not reduce the risk of fractures.²⁴⁸ With raloxifene, the osteocalcin change predicted fracture incidence better than did the BMD change.²⁴⁹

In patients with CKD stages 4-5, there are limited data that relate serum markers to fractures. Urena et al.¹⁷⁵ found that cross-laps (C-terminal peptide) and b-ALP were not different between fracture and non-fracture cases in a survey of 70 dialysis patients.

A recent study evaluated patients with CKD stages 1-5 without known CVD and found that reduced tartrate-resistant acid phosphatase-5B and elevated b-ALP were both associated with an increase in the RR of cardiovascular mortality.²⁵⁰ These somewhat paradoxical findings suggest that much more work needs to be carried out to fully understand the clinical utility of such biomarkers.

Bone markers and bone histology

In CKD patients, a few studies show significant correlations between collagen cross-linking molecules and the bone formation rate (shown in Figure 14).

Bone volume was not related to these markers in two studies. Coen et al.²²⁰ measured a panel of circulating biomarkers (iPTH, osteocalcin, b-ALP, tartrate-resistant acid phosphatase, CTX, and deoxypyridinoline) in 41 patients with CKD stage 5, and none of them correlated with the BV/TV. Barreto et al.²³⁰ focused on factors that related to osteoporosis in 98 patients with CKD stage 5, half of whom had a BV/TV less than one s.d. from the normal mean. They found no relationship between the low BV/TV and serum iPTH, b-ALP, or deoxypyridinoline, but the tumor necrosis factor-alpha; and the osteoprotegerin/receptor activator for nuclear factor kappa;B ligand ratio was higher in those with a low BV/TV. Thus, at this point in time, there is insufficient evidence for the use of these markers. More research is clearly needed.

Bone markers and BMD

Predicting BMD at a single point in time. In the general population, observational studies of elderly people show that circulating bone turnover markers are not related to BMD at one point in time.²⁵¹ In clinical trials of osteoporosis medications, the baseline biochemical markers do not consistently predict the change in BMD. (As noted above, however, the baseline biochemical measurements may predict fractures in some cases, and this is more important than predicting BMD results.) The data in patients with CKD stages 4-5 are limited and inconsistent, as shown in Table 18.

Predicting change in BMD. In studies on osteoporosis, the changes in measurements of bone formation and resorption may be related to the changes in BMD with some treatments.²⁴⁷ On an individual level, it is not certain how reliable these markers will be in predicting BMD change. At present, there is no consensus with regard to the clinical utility of markers in individual patients with osteoporosis, but many ongoing studies are examining this issue, especially as anabolic drugs are being developed.

On a theoretical basis, bone markers should be able to predict the change in bone volume, which is determined by bone balance. Unfortunately, none of the current serum or urine markers of bone turnover are sensitive enough to allow the calculation of bone balance, and the interpretation of the measurements depends on the clinical situation. For example, the highest serum levels of turnover markers are found in patients with metastatic cancer, Paget's disease, and in healthy adolescent boys.

When interpreting bone turnover markers, it is important to remember the distinction between bone volume, as measured on bone biopsies, and BMD, as measured with a radiographic technique. Density depends on both the bone volume and the mineralization of the bone. Newly formed bone is not as dense as older bone, and patients with a high turnover have a greater proportion of newly formed bone with a low BMD. When bone turnover is decreased, the bone becomes 'older' and accumulates more minerals, increasing the DXA value without necessarily increasing the bone volume. In patients with CKD, the relationships are even more complicated because the mineralization is frequently abnormally low, so that BMD can be low even when bone volume is normal. Rapid increases in BMD can be observed when osteomalacia is treated, even without any formation of new bone, because the osteoid fills with mineral. The markers of bone formation that depend on the secretion of new collagen would not be able to detect this improved mineralization.

3.2.5 We recommend that infants with CKD stages 2-5D should have their length measured at least quarterly, while children with CKD stages 2-5D should be assessed for linear growth at least annually (1B).

In children with CKD stages 2-5D, abnormalities in statural growth are commonly observed. Such abnormalities may include a height below the 3rd percentile of the growth curve for normal children of the same gender; a negative statural growth curve against the genetic potential based on mid-parental height formulas even when on the normal growth curve; or a negative growth velocity, based on gender-specific curves of normal children. Growth should be assessed at least monthly in infants, quarterly in children below 2 years of age, and at least annually in older children and adolescents, and plotted accurately on the appropriate growth chart for either height, velocity, or ideally, both. This allows for an optimal understanding of the defects in linear growth that may occur with CKD in children. Growth velocity as rates and absolute changes in height is used as an end point in clinical trials of growth-hormone therapy in children and adolescents with CKD.

Linear height deficit (short stature) is one of the cardinal features of progressive CKD in pediatric patients. In normal children, the 50th percentile for height corresponds to a Z-score of 0. The 3rd percentile is a Z-score of -1.88. In children with CKD, over one-third of patients have Z-scores lower than -1.88.²⁵² Baseline kidney function, by height Z-score, shows that there are patients with severe height deficits, even though they have a moderate kidney function (>25 ml/min per 1.73 m²). In patients with a calculated clearance between 50 and 75 ml/min per 1.73 m², 18.2% (379 of 1720) had a height Z-score worse than -1.88. The mechanisms of linear growth failure include the presence of chronic metabolic acidosis, renal osteodystrophy, nutrient wasting, chronic inflammation, functional hypogonadism in some adolescents, and dysregulation of the growth hormone-insulin-like growth factor 1 endocrine axis. The latter has led to the development and use of a recombinant human growth hormone, which has been licensed by the Food and Drug Administration in the United States since 1988 for the treatment of linear growth failure in children with CKD, one of the measures of bone in CKD-MBD. However, using data from the North American Pediatric Renal Transplant Cooperative Study 2006 data report,²⁵² only 6.5% of all patients at entry into the registry were using recombinant human growth hormone. By 24 months of follow-up, 15.9% of patients being followed up were receiving recombinant human growth hormone. This low usage prompted an examination of the benefit and harm of recombinant human growth hormone in children (see Chapter 4.3).

RESEARCH RECOMMENDATIONS

Additional research is called for:

• A prospective study of BMD to determine fracture risk thresholds in CKD stages 3-5, 5D, and 1-5T.
  
• A prospective study of circulating biochemical markers (PTH, b-ALP, PINP, PICP, NTX, CTX, tartrate-resistant acid phosphatase, and osteoprotegerin) to determine if they can predict fractures or other clinical outcomes in CKD stages 3-5, 5D, and 1-5T.
• The development of an international standard for the assessment of renal osteodystrophy, particularly for dynamic measurements.

Chapter 3.3: Diagnosis of CKD-MBD: vascular calcification

INTRODUCTION

The diagnosis of CKD-MBD includes the detection of extraosseous calcification, including arterial, valvular, and myocardial calcification. It is generally well recognized that the prevalence of calcification increases with progressively decreasing kidney function and is greater than that in the general population. Cardiovascular calcification is associated with, and predictive of, adverse clinical outcomes, including cardiovascular events and death. However, there are some uncertainties with regard to the sensitivity and specificity of the different imaging tests available for detecting cardiovascular calcification. Further, there is also uncertainty as to whether altering the progression of cardiovascular calcification will impact patient outcomes (cause-effect relationship) in different stages of CKD. Finally, there is no clear evidence-based protocol or algorithm for the diagnostic and therapeutic strategies that need to be followed after yielding a positive calcification test result.

RECOMMENDATIONS

3.3.1 In patients with CKD stages 3-5D, we suggest that a lateral abdominal radiograph can be used to detect the presence or absence of vascular calcification, and an echocardiogram can be used to detect the presence or absence of valvular calcification, as reasonable alternatives to computed tomography-based imaging (2C).

3.3.2 We suggest that patients with CKD stages 3-5D with known vascular/valvular calcification be considered at highest cardiovascular risk (2A). It is reasonable to use this information to guide the management of CKD-MBD (not graded).

Summary of rationale for recommendations

• In the normal population, the magnitude of CAC as imaged by either electron beam CT (EBCT) or multislice CT (MSCT) is a strong predictor of cardiovascular event risk.
• In the CKD population, coronary artery and generalized vascular calcification is exceedingly more prevalent, more severe, and follows an accelerated course compared with that in the normal population.
• The reference standard in the detection of cardiovascular calcifications in CKD and in the general population is the CT-based CAC score, but other, more easily available techniques--for example, lateral abdominal X-ray, pulse wave velocity (PWV) measurements, and echocardiography (valvular calcification)--may yield comparable information.
• The presence and the severity of cardiovascular calcification strongly predict cardiovascular morbidity and mortality in patients with CKD.
• However, there is limited evidence from RCTs in CKD that the reduction of arterial calcification progression impacts mortality.
  
• A majority of Work Group members felt that inconsistencies remained among RCT reports aimed at showing that intervention improved patient level outcomes, and hence, indiscriminate screening in every patient with CKD-MBD was not recommended.
• However, there was consensus that known vascular/valvular calcification and its magnitude identify patients at high cardiovascular risk. Therefore, the presence of vascular/valvular calcification should be regarded as a complementary component to be incorporated into the decision making of how to individualize treatment of CKD-MBD.

BACKGROUND

Tissue calcification is a complex and highly regulated process in bone and teeth, and also at extraosseous sites. The most threatening localization of unwanted calcification is at vascular sites, where it may manifest as both medial and intimal calcification of arteries. In the general population, autopsy and imaging studies have identified calcification in >95% of atherosclerotic plaques. Calcification seems to be a part of the natural history of atherosclerotic plaques, with extensive calcification associated with late-stage (American Heart Association Stage Va and VII) atherosclerosis. In the general population, atherosclerotic plaque calcification is associated with cardiovascular events such as myocardial infarction, symptomatic angina pectoris, and stroke.^{253, 254 and 255} Medial calcification causes arterial stiffness, resulting in an elevated pulse pressure and increased PWV, thereby contributing to left ventricular hypertrophy, dysfunction, and failure. Furthermore, an advanced calcification of the heart valves may lead to dysfunction contributing to heart failure and an increased risk of endocarditis. Cardiovascular calcifications are usually progressive, and their extent and severity are highest in patients with CKD. Recent reports suggest an increased prevalence of cardiovascular calcification in patients at early stages of CKD. Thus, a considerable percentage of CKD patients are at risk of cardiovascular events from vascular calcification.

As mentioned above, two patterns of vascular calcification have been described: predominantly intimal and predominantly medial calcification. There is, however, an ongoing debate with regard to the differential role of intimal (atherosclerotic) vs medial (arteriosclerotic) calcification in CVD in CKD patients.^{256, 257} In the general population, an elevated coronary artery calcium score almost exclusively reflects the atherosclerotic disease burden. In two small autopsy studies, it became apparent that, in dialysis patients, CAC is also predominantly localized in the coronary intima, whereas the medial calcifications observed in a minority of such patients seemed to be adjacent to plaque areas just beneath the internal elastic lamina.^{258, 259} Although the coronary vascular bed may differ considerably from other arteries with regard to the calcification process and its manifestations, the same group observed a 'pure' medial calcification in the coronary arteries during the early stages of CKD.²⁵⁷ A 'pure' medial calcification, in the absence of intimal disease, was also observed in epigastric arteries obtained from dialysis patients at the time of renal transplantation.²⁶⁰ An older study identified both intimal and medial calcifications in iliac arteries of such patients.²⁶¹ Thus, there is neither definitive evidence to suggest that isolated medial calcification is distinct from the calcification that occurs in the natural history of atherosclerosis nor is there definite proof against it.

Arterial calcification assessed by all the available imaging studies cannot accurately differentiate calcification that is localized to the intima from calcification in the media adjacent to the internal elastic lamina, or in the medial layer. Experimental and ex vivo studies suggest that the vascular smooth muscle cell may be critical in the development of calcification by transforming into an osteoblast-like phenotype.²⁶² In addition, the pericyte in the media and adventitia may have a role in the secretion of vascular calcification-inducing factors. The stimulus for such a transformation may depend on the location of calcification within the artery wall. For example, in intimal lesions, atherosclerosis may be the most important stimulus. However, in patients with CKD and medial calcification, there may be additional, or additive, factors potentially explaining why medial calcification of the peripheral arteries can be seen without intimal changes and is more common in CKD than in the non-CKD population.²⁶⁰ Elevated phosphorus, elevated calcium, oxidized low-density lipoprotein cholesterol, cytokines, and elevated glucose, among others, stimulate this transformation of vascular smooth muscle cells into osteoblast-like cells in vitro using cell-culture techniques. These factors likely interact at the patient level to increase and/or accelerate calcification in CKD. Given the potential complexity of the pathogenesis and the inability of radiological techniques to differentiate the location of calcification, the approach to all patients with calcification should be to minimize atherosclerotic risk factors and control biochemical parameters of CKD-MBD. In vivo animal studies have shown less arterial calcification with non-calcium-based binders compared to that with calcium-based binders.^{263, 264} Unfortunately, trials in dialysis patients evaluating such strategies to treat either atherosclerosis or CKD-MBD have not conclusively shown that such an intervention positively affects patient-level outcomes.^{265, 266 and 267} Despite this, given the high cardiovascular burden in CKD, the majority of the Work Group felt that the treatment approaches to limit the calcium intake from phosphate binders in CKD patients with known vascular/valvular calcification were appropriate until definitive studies are conducted, as detailed in Chapter 4.1.

Extraosseous calcification in patients in advanced stages of CKD has been observed since the early days of dialysis,^{268, 269} but was originally thought to result predominantly from a supersaturation of serum with calcium and phosphate ions, that is, passive precipitation. However, in recent years, it became evident that vascular calcification is also an active cellular process. As already pointed out above, the presence or upregulation of inducers of cellular osteogenic transformation and hydroxyapatite formation (among which high phosphate probably has a central role)²⁶² causes the differentiation of vascular smooth muscle cells into an osteoblast-like phenotype of vascular smooth muscle cells. Newly discovered calcium-regulatory factors, including fetuin-A, matrix Gla protein, osteoprotegerin, and pyrophosphates--all of which possess properties of systemic or local calcification inhibitors--may have a key role in fine-tuning protection against unwanted calcification, and some of these factors may be dysregulated in uremia.²⁷⁰ A seminal paper by Murshed et al., however, showed that even complex pathological mineralization disorders can be prevented by modulating extracellular phosphate concentration.²⁷¹ Therefore, it is biologically plausible that the calcification process develops from unique stimuli and progresses in an accelerated manner in CKD patients. As epidemiological studies suggest a direct relationship between calcification and impaired clinical outcomes, cardiovascular calcification is thus regarded as a relevant clinical end point by most investigators mirroring cardiovascular event risk. However, it cannot yet be used as a reliable surrogate marker for interventions, as the link between intervention and patient-level outcomes when calcification is ameliorated has not been shown conclusively.

Finally, a rare but very severe form of medial calcification of small (cutaneous) arteries is calciphylaxis, also called calcific uremic arteriolopathy. This complication is strongly associated with CKD-related disturbances of mineral metabolism, including secondary HPT, in approximately one-third of cases. It is characterized by ischemic, painful skin ulcerations followed by superinfections, and is associated with high mortality. Relationships with dysregulated calcification inhibitors (fetuin-A and matrix Gla protein) have been implicated in the pathogenesis of calciphylaxis, but because of the relatively low incidence of the disease, no conclusive data are available to firmly comment on the nature of the disease process or to allow generalizable treatment options to be recommended.

This topic represents a comprehensive review of the literature of selected topics by the Work Group with assistance from the vidence review team to formulate a rationale for clinical recommendations. Thus, this should not be considered as a systematic review.

RATIONALE

3.3.1 In patients with CKD stages 3-5D, we suggest that a lateral abdominal radiograph can be used to detect the presence or absence of vascular calcification, and an echocardiogram can be used to detect the presence or absence of valvular calcification, as reasonable alternatives to computed tomography-based imaging (2C).

Diagnostic tests

Most studies examining calcification in CKD reported on the use of CT-based techniques (EBCT and MSCT) in the detection of cardiovascular calcification in patients with CKD-MBD (Supplementary Table 9). EBCT and MSCT are currently regarded as the most sensitive methods for the detection and quantification of cardiovascular calcification. One study explicitly evaluated the sensitivity and specificity of several imaging tests and functional/hemodynamic measures for detecting CAC compared with EBCT.²⁷² This analysis focused on pulse pressure measurements, valvular calcification (by echocardiography), and abdominal aortic calcification (by lateral abdominal X-ray), respectively, according to the severity of CAC scores as assessed by EBCT scores of 30-99, 100-399, 400-999, and >1000. No meaningful correlation was found between pulse pressure and CAC scores. In contrast, a strong correlation was detected between abdominal aortic calcification by plain radiograph and CAC scores. Valvular calcification, detected by echocardiography, was another good predictor of CAC.

We reviewed six additional studies which carried out correlation analyses comparing CT-based imaging techniques of assessing CAC with other measures of calcification. These latter measures included pulse pressure, abdominal aortic calcification by lateral X-ray, PWV, echocardiography (valvular calcifications), intimal-media thickness (IMT) of the carotid arteries, and MSCT of the thoracic and abdominal aorta.^{273, 274, 275, 276, 277 and 278} PWV and abdominal aortic calcifications seemed to be reasonably good predictors of CAC scores, whereas the value of IMT, valvular calcification, and especially pulse pressure was limited. However, these studies were not designed to test sensitivity and specificity in this regard. The majority of the reported data referred to the CKD stage 5D population, whereas some studies included patients in different CKD stages.^{273, 275} Only one study evaluated children (CKD stage 4).²⁷⁸ Thus, EBCT and MSCT remain the gold standard. However, a plain X-ray examination allows the detection of vascular calcification, and echocardiography allows the detection of valvular calcification, with reasonable sensitivity, as compared with the more expensive CT-based techniques. Thus, the Work Group felt that plain X-ray and echocardiography were reasonable alternatives to the gold standard of CT-based imaging.

3.3.2 We suggest that patients with CKD stages 3-5D with known vascular/valvular calcification be considered at highest cardiovascular risk (2A). It is reasonable to use this information to guide the management of CKD-MBD (not graded).

To recommend widespread global screening for the diagnosis of vascular calcification in all patients with CKD, the Work Group felt that the following was needed: (i) There should be an accurate and reliable diagnostic test (see above); (ii) vascular calcification should be prevalent enough to warrant screening; (iii) the tests should prompt a specific intervention; and (iv) the intervention should impact hard clinical end points. The Work Group felt that the data to support (i) and (ii) were strong, the data to support (iii) were somewhat inconsistent, and the data to support (iv) were limited. Thus, the Work Group did not recommend indiscriminate screening in all patients with CKD at this time, although this was a split decision. However, vascular calcification is an important component of CKD-MBD, and animal, epidemiological, and observational studies support that vascular/valvular calcification is a likely cause of cardiovascular morbidity and mortality in patients with CKD-MBD; thus, an assessment for vascular calcification is warranted in some patients. These may include, but are not limited to, patients with significant hyperphosphatemia requiring a differentiated high-dose phosphate-binder therapy, patients on a transplant waiting list, and any patient in whom the caring physician decides that a knowledge of the presence of vascular calcification may impact therapeutic decision making.

Prevalence

Twenty-five reports including information on the baseline prevalence of vascular or valvular calcification were evaluated (Supplementary Table 10). The studies included a total of more than 4000 patients in different stages of CKD, the majority being in CKD stage 5D. In adult patients on dialysis, CAC has been detected in 51-93% of the study populations; prevalent dialysis patients had a higher likelihood of having detectable CAC scores than did incident ones. Valvular calcification was present in 20-47% of patients in CKD stage 5D. The prevalence of calcifications was variable at other vascular sites and was dependent on the sensitivity of the method used.

In CKD stages 3-5, published information related mostly to CAC scores showed that 47-83% of patients had cardiovascular calcification. In children with CKD stage 5D, the prevalence of a positive CAC was found to be 20% in one study.²⁷⁸ In young adults receiving dialysis treatment (age ranges: 20-30 years in one study, 19-39 years in a second study) with childhood-onset CKD, CAC prevalence was 87.5 and 92%, respectively.^{279, 280} Valvular (aortal or mitral) calcifications were present in 20-25% of 653 patients with CKD stages 3-5 in the Multi-Ethnic Study of Atherosclerosis,²⁸¹ whereas the degree of renal dysfunction was only modestly associated with valvular calcification. In patients on dialysis, valvular calcification is more common, with one series reporting the presence of valvular calcification in 32% of patients.²⁸²

Eight studies investigating the natural history of calcification in a predefined prospective longitudinal approach in CKD were examined (Supplementary Table 11). Follow-up periods ranged from 1 to 3 years; detection methods were MSCT, EBCT, X-ray of pelvis and calves, and in one study, IMT. The major finding in this context is that once calcification is established, it follows a progressive course. In contrast, non-calcified patients with CKD have a high likelihood of remaining free of cardiovascular calcification over months to years. Compared with the non-CKD population, the progression of cardiovascular calcification is enhanced in patients with CKD. Furthermore, there is a strong relationship between the magnitude and severity of calcification and pre-existing coronary artery disease.

Risk relationships

We reviewed 10 reports on the risk relationships between cardiovascular calcification and mortality in patients with CKD (Supplementary Tables 12 and 13). Most of these studies were again conducted in dialysis patients, including one in peritoneal dialysis patients, but there is also information on renal transplant recipients and patients in CKD stages 4-5. EBCT, MSCT, ultrasound, echocardiography, and several X-ray techniques (pelvis, abdomen and hands) were used as diagnostic tests. In all but one study, cardiovascular calcification or progression of calcification were identified as independent risk predictors for cardiovascular and all-cause mortality. In only one study²⁸³ did valvular calcification lose its significance in predicting death after a multivariate adjustment.

In some of these studies (as well as in others that primarily addressed the natural history of calcification), risk associations were reported between the development and progression of calcification and epidemiological and biochemical parameters. Age was the most consistent risk factor for severe or progressive calcification, whereas diabetes, time on dialysis, male gender, high serum iPTH and/or ALP levels, inflammation (C-reactive protein levels), calcium intake, hyperphosphatemia, and increased Ca X P were identified in some studies, but the latter relationships could not be uniformly reproduced. No studies of adequate quality reported on the relationship between cardiovascular calcification and bone outcomes in CKD patients.

Management of patients with vascular/valvular calcification

Cardiovascular calcification development and progression can be influenced by treatment. Given that vascular calcification is associated with increased cardiovascular risk, and that the pathogenesis seems to be related to CKD-MBD (biochemical and bone) abnormalities and atherosclerosis, it is appropriate to evaluate and modify both.

CKD-MBD. Longitudinal studies have also shown that the progression of vascular calcification seems to be modifiable by the choice of phosphate binders. Five studies compared the effects of different phosphate-binder therapies on the progression of CAC scores in chronic HD patients^{284, 285, 286, 287 and 288} (see Chapter 4.1). The Treat-to-Goal study (n = 200) compared sevelamer-HCl to calcium-containing phosphate binders, analyzing the progression of coronary artery and aortic calcification (by EBCT) in prevalent HD patients over 1 year. Although calcification scores progressed with calcium-containing phosphate binders, treatment with sevelamer-HCl was associated with a lack of calcification progression. A similar design was used, and the results showed more calcification progression in patients treated with calcium-based binders compared with sevelamer-HCl in the Renagel in New Dialysis Patients study (n = 129), which studied incident HD patients who were randomized within 90 days after starting dialysis treatment. The Calcium Acetate Renagel Evaluation-2 study (n = 203) showed that the use of sevelamer-HCl and calcium acetate was associated with equal progression of CAC when statins were used to achieve a similar control of the serum low-density lipoprotein cholesterol in the two study arms.²⁸⁷ Interestingly, in Calcium Acetate Renagel Evaluation-2, the combination of sevelamer-HCl and atorvastatin was actually associated with a higher progression rate of CAC than that in Treat-to-Goal,²⁸⁴ instead of showing an amelioration of CAC progression with the combination of calcium acetate and statin. It is difficult to reconcile these differences, although one potential explanation is that the Calcium Acetate Renagel Evaluation-2 study patient population had a higher number of cardiovascular risk factors than did that of the Treat-to-Goal study.²⁸⁹ The Bone Relationship with Inflammation and Coronary Calcification study (n = 101) investigated the effects of calcium acetate vs sevelamer-HCl on CAC progression and bone histomorphometry in HD patients. Although CAC progression rates did not differ between both phosphate-binder arms, this study was hampered by a much smaller sample size and several significant confounders: imperfect matching of baseline CAC scores between the two study arms; the use of high dialysate calcium concentrations (1.75 mmol/l (3.5 mEq/l)) in most patients, resulting in a positive calcium balance; and multiple interventions during the course of the study aimed at improving adynamic bone disease.²⁸⁸ It is possible that these confounders 'neutralized' any potential advantage of sevelamer-HCl being a calcium-free phosphate binder. Finally, Russo et al. examined CAC score progression in patients with CKD stages 3-5 (n = 90). Patients were treated with either low-phosphate diet alone, low-phosphate diet plus calcium carbonate, or low-phosphate diet plus sevelamer-HCl. Calcification progression was lowest in the sevelamer-HCl-treated group compared with the calcium- and diet-only groups.²⁹⁰ These studies are discussed in more detail in Chapter 4.1.

There were no studies investigating the effect of parathyroidectomy on calcification progression or regression that met the inclusion criteria for review. There was one study addressing the issue of vascular calcification progression in renal transplant recipients (CKD stages 1-5T; n = 55) by measurements of IMT by high-resolution B-mode ultrasound at 3, 6, and 12 months after transplantation.²⁹¹ Regression of IMT was observed in association with a decline in serum iPTH levels. One question regarding this study is whether IMT indirectly reflects carotid artery calcification or other vascular remodeling processes induced by atherosclerosis or hypertension.

To date, there are no prospective studies in humans that have evaluated the impact of calcimimetics or calcitriol and vitamin D analogs on arterial calcification. However, a recent observational study showed a U-curve type of relationship between serum 1,25(OH)2D₃ and arterial calcification in children and adolescents with CKD stage 5D.²⁹² No such association existed between serum 25(OH)D and arterial calcification. In one study in adult patients with CKD stage 5, no independent association of serum 25(OH)D or 1,25(OH)2D₃ levels with arterial calcification was observed,²⁹³ although the authors of another report identified an association between 25(OH)D deficiency and the magnitude of vascular calcification.²⁹⁴ It is noteworthy that, in the two latter studies, there was an association of arterial calcification with arterial PWV. Experimental studies showed differential effects of calcimimetics and calcitriol on extraosseous calcification, the former being neutral or protective, the latter being a dose-dependent risk factor for calcification.^{295, 296 and 297} The experimental data supporting less toxicity of vitamin D analogs compared with calcitriol are not completely consistent across studies, but, in general, support the claim that there is reduced calcification with equivalent PTH lowering with different vitamin D analogs^{.295, 298, 299 and 300}

Atherosclerosis. CAC is a strong predictor of atherosclerotic disease in the general population. An evidence-based review of cardiovascular calcification in the general population was not carried out by the Work Group. However, it was recognized that most population studies measuring CAC did not necessarily exclude individuals on the basis of kidney function and thus include variable numbers of CKD patients. These studies have been summarized in the American College of Cardiology/American Heart Association 2007 Clinical Expert Consensus Document on Coronary Artery Calcium Scoring by Computed Tomography.³⁰¹

In general, this literature evaluating the general population supports the view that CAC is part of the development of atherosclerosis and occurs almost exclusively in atherosclerotic human arteries. Calcification occurs early in the atherosclerotic process; however, the amount of calcification per lesion has a variable relationship with the associated severity of luminal stenosis. The relationship between the degree of calcification in an individual lesion and the probability of plaque rupture is unknown. In the general population, the overall coronary calcium score can be considered as a measure of the overall burden of coronary atherosclerosis. The American College of Cardiology/American Heart Association document indicates that the relationship between CAC and cardiovascular events in the CKD population is less clear than that in the non-CKD population because of a relative lack of informative studies and the possibility that medial calcification may not be indicative of atherosclerotic disease severity. In the non-CKD population, high-risk patients were not considered appropriate for this form of testing and so other approaches to clinical assessment and risk-reducing therapies were suggested. This latter suggestion may or may not be applicable to CKD patients, as the standard approaches for clinical assessment (Framingham risk-factor ranking) may be inappropriate for the kind of vascular disease in CKD patients. The almost exclusive relationship between magnitude of calcification and atherosclerosis burden is controversial in CKD patients^{,256, 257} in contrast to the situation in the general population. For example, whereas >50% of cardiovascular events are classical myocardial infarctions in the general population, this figure is below 20% in the CKD population, despite a higher absolute number of cardiovascular events.³⁰²

Antiatherosclerotic strategies using statin treatment have been shown to have a beneficial impact on the atherogenic profile, atheroma progression, and cardiovascular events in patients with no known CKD.^{303, 304 and 305} However, at the same time, they do not seem to protect against the progression of arterial calcification when studied in the general population.^{306, 307} In CKD patients, there are no data on the effects of statins on arterial calcification, as compared with those of placebo. Even worse, the 4D study failed to show a benefit of atorvastatin treatment on the outcome of diabetic dialysis patients. Studies in progress like SHARP (Study of Heart and Renal Protection) and AURORA (A Study to Evaluate the Use of Rosuvastatin in Subjects on Regular Hemodialysis: An Assessment of Survival and Cardiovascular Events) may help to gain a better understanding of the benefits of correcting atherosclerotic risk factors on cardiovascular events and mortality in patients with CKD stages 3-5 and 5D, respectively.^{308, 309} (Note added in proof: In AURORA, rosuvastatin failed to show a significant effect on the composite primary end point of death from cardiovascular causes, nonfatal myocardial infarction, or nonfatal stroke in chronic hemodialysis patients.) In the interim, we currently extrapolate the approach to atherosclerosis-related cardiovascular calcification from the general population, but there is some skepticism as to whether this approach may indeed apply to the CKD population, especially in CKD stage 5D.

RESEARCH RECOMMENDATIONS

• To determine the efficacy of different pharmacological agents for the prevention or delay of arterial calcification in patients with hyperphosphatemia, a prospective, randomized, placebo-controlled trial evaluating different phosphate-binder regimens in CKD stages 4-5D, should be conducted. The primary end point should be cardiovascular and all-cause mortality, with parallel assessments of cardiovascular and aortic calcification.
• Studies are needed to determine the role of screening for cardiovascular calcification and validate its usefulness for individual prognosis, risk reduction, and therapeutic decision making in patients with CKD. Such studies should address the question of whether a knowledge of vascular calcification may prospectively impact patient outcomes, and whether a broad approach of routine testing in patients with CKD should be considered for recommendation in the future.
• Studies are needed that compare patient outcomes of specified treatment strategies in response to the presence or absence of vascular calcification.
• To determine the efficacy of different pharmacological agents in the prevention or delay of arterial calcification in patients with secondary HPT, a prospective, randomized, placebo-controlled trial comparing calcitriol, vitamin D analogs, and calcimimetics in CKD stages 4-5D should be conducted. The primary end point should be cardiovascular and all-cause mortality, with parallel assessments of cardiovascular and aortic calcification.
• To determine the efficacy of different pharmacological agents for the prevention or delay of arterial calcification in patients with vitamin D deficiency, a prospective, randomized, placebo-controlled trial evaluating the administration of cholecalciferol or ergocalciferol in CKD stages 4-5D should be conducted. The primary end point should be cardiovascular and all-cause mortality, with parallel assessments of cardiovascular and aortic calcification.
• To determine the efficacy of calcification inhibitors in the prevention or delay of arterial calcification, a prospective, randomized placebo-controlled trial evaluating the administration of vitamin K in CKD stages 4-5D should be conducted. The primary end point should be cardiovascular and all-cause mortality, with parallel assessments of cardiovascular and aortic calcification.
• To understand the pathophysiology of arterial calcification, additional case-control pathological studies should be conducted to evaluate the histological presence of intimal and medial calcification in the aorta and other non-coronary arteries in CKD patients compared with non-CKD patients.
• To understand the pathophysiology of calciphylaxis, epidemiological or registry studies should be conducted on individuals with calciphylaxis, either based on the clinical assessments (painful livedo and/or ulcerations and exclusion of differential diagnoses such as diabetic ulcers, vasculitis, or cholesterol emboli) or, preferably, based on biopsy results. The study should evaluate exposure to candidate risk factors (calcification inhibitor levels, CKD-MBD treatment, dialysis mode, vitamin K status, and mineral parameters such as PTH, calcium, phosphorus, and ALP) and the natural history of the disease on the basis of pathology and risk factors.

SUPPLEMENTARY MATERIAL

Supplementary Table 4. Prevalence and incidence of fractures in patients with CKD 5D.
Supplementary Table 5. Fractures in patients with CKD stages 3-4.
Supplementary Table 6. Overview table of selected studies of the natural history of bone disorders.
Supplementary Table 7. Overview table of selected studies demonstrating the risk relationship between bone measurements and mortality in CDK stage 5D.
Supplementary Table 8. Overview table of selected studies demonstrating the risk relationship between hormonal parameter, PTH, and fractures in CKD stage 5D.
Supplementary Table 9. Overview table of selected studies of diagnostic tests: studies for vascular and valvular calcification techniques in CKD.
Supplementary Table 10. Overview table of selected studies presenting data on calcification prevalence.
Supplementary Table 11. Overview table of selected studies demonstrating the natural history of vascular and valvular calcifications in CKD.
Supplementary Table 12. Overview table of selected studies demonstrating the risk relationship between vascular calcification and mortality in CKD.
Supplementary Table 13. Overview table of selected studies demonstrating the risk relationship between valvular calcification and mortality in CKD stage 5D.
Supplementary material is linked to the online version of the paper at http://www.nature.com/ki