Useful biochemical markers for diagnosing renal osteodystrophy in predialysis end-stage renal failure patients

Useful Biochemical Markers for Diagnosing Renal Osteodystrophy in Predialysis End-Stage Renal Failure Patients An R.J. Bervoets, MD, Goce B. Spasovski, MD, PhD, Geert J. Behets, Geert Dams, Momir H. Polenakovic, MD, PhD, Katica Zafirovska, MD, PhD, Viviane O. Van Hoof, MD, PhD, Marc E. De Broe, MD, PhD, and Patrick C. D’Haese, PhD ● Background: Various biochemical markers have been evaluated in dialysis patients for the diagnosis of renal osteodystrophy (ROD). However, their value in predialysis patients with end-stage renal failure (ESRF) is not yet clear. Methods: Bone histomorphometric evaluation was performed and biochemical markers of bone turnover were determined in serum of an unselected predialysis ESRF population (N ⴝ 84). Results: Significant (P < 0.005) differences between the five groups with ROD (ie, normal bone [N ⴝ 32], adynamic bone [ABD; N ⴝ 19], hyperparathyroidism [N ⴝ 8], osteomalacia [OM; N ⴝ 10], and mixed lesion [N ⴝ 15]) were noted for intact parathyroid hormone, total (TAP) and bone alkaline phosphatase (BAP), osteocalcin (OC), and serum calcium levels. Serum creatinine and (deoxy)pyridinoline levels did not differ between groups. For the diagnosis of ABD, an OC level of 41 ␮g/L or less (<7.0 nmol/L) had a sensitivity of 83% and specificity of 67%. The positive predictive value (PPV) for the population under study was 47%. The combination of an OC level of 41 ng/L or less (<7.0 nmol/L) with a BAP level of 23 U/L or less increased the sensitivity, specificity, and PPV to 72%, 89%, and 77%, respectively. ABD and normal bone taken as one group could be detected best by a BAP level of 25 U/L or less and TAP level of 84 U/L or less, showing sensitivities of 72% and 88% and specificities of 76% and 60%, corresponding with PPVs of 89% and 85%, respectively. In the absence of aluminum or strontium exposure, serum calcium level was found to be a useful index for the diagnosis of OM. Conclusion: OC, TAP, BAP, and serum calcium levels are useful in the diagnosis of ABD, normal bone, and OM in predialysis patients with ESRF. Am J Kidney Dis 41:997-1007. © 2003 by the National Kidney Foundation, Inc. INDEX WORDS: Adynamic bone disease (ABD); renal osteodystrophy (ROD); biochemical markers; osteocalcin (OC); bone alkaline phosphatase (BAP); end-stage renal failure (ESRF).

C

HRONIC RENAL failure has a repercussion on bone, known as renal osteodystrophy (ROD), even before dialysis treatment is started.1 It therefore is important to evaluate bone status early in the evolution of chronic renal failure. Patients with mild or moderate chronic renal failure (glomerular filtration rate of 20 to 50 mL/min) rarely experience symptoms. Nevertheless, it has been shown in recent studies that more than 50% of patients with only moderate renal failure have abnormal bone histological characteristics.2 To date, the gold standard for the diagnosis of ROD remains histomorphometric and histochemical examination of a bone biopsy specimen.2 Five types of bone disease can be distinguished: osteomalacia (OM), adynamic bone disease (ABD), hyperparathyroidism (HPTH; either mild HPTH or osteitis fibrosa), mixed lesion (MX), and normal bone. The question remains whether bone biopsy is justified in asymptomatic patients. These are mainly patients with either normal bone or ABD. Because bone biopsy is invasive, alternative methods have been evaluated during the last 20 years. These include bone mineral densitometry,3,4 bone radiography,5,6 and determination of various biochemical blood parameters reflecting

bone turnover either directly or indirectly.5,7-9 With regard to the latter, most investigators measure intact parathyroid hormone (iPTH), total or ionized serum calcium, phosphate, alkaline phosphatase (AP), vitamin D status, and markers of either bone formation or bone resorption.10-12 Among these, osteocalcin (OC), carboxy-terminal propeptide of type I collagen, and bonespecific AP (BAP) reflect bone formation, whereas carboxy-terminal telopeptide of type I collagen and pyridinolines (PYD) and deoxypyridinolines (DPYD) reflect bone resorption.3 Some of these parameters correlate well with bone histomorphometric data, but do not always From the Department of Nephrology-Hypertension, University of Antwerp, Belgium; and the Department of Nephrology, University of Skopje, Skopje, Macedonia. Received October 1, 2002; accepted in revised form January 13, 2003. A.R.J.B. was the recipient of a grant from the Foundation for Scientific Research, Flanders, Belgium. Address reprint requests to Marc E. De Broe, MD, PhD, Department of Nephrology, University Hospital Antwerp, Wilrijkstraat 10, B-2650, Edegem, Belgium. E-mail: [email protected] © 2003 by the National Kidney Foundation, Inc. 0272-6386/03/4105-0010$30.00/0 PII: S0272-6386(03)00197-5

American Journal of Kidney Diseases, Vol 41, No 5 (May), 2003: pp 997-1007

997

998

BERVOETS ET AL

reach a high enough diagnostic performance to substitute for bone histomorphometry.2 Moreover, most of the studies reported to date have been undertaken in patients treated by dialysis and in selected populations.10 Studies evaluating the diagnostic performance of biochemical markers11,13-19 in patients with end-stage renal failure (ESRF) not yet on dialysis therapy are limited to the studies of Dahl et al,6 Hutchison et al,5 and Mazzaferro et al.20 In recent years, an evolution in the spectrum of ROD in the renal failure population has been noticed. This is characterized mainly by a decrease in the frequency and severity of osteitis fibrosa (previously the most common type of ROD), concomitantly with an increased prevalence of ABD. In addition, the so-called aluminum-related bone disease has now disappeared. Removal of aluminum from water used for the preparation of dialysis fluid by the introduction of adequate water treatment installations and replacement of aluminum-based phosphate binders have resulted in an important decrease in severe chronic aluminum intoxication. In the present study, in a search for biochemical markers that could reach an acceptable diagnostic performance predicting the histomorphometric and/or histological picture of a bone biopsy specimen, we examined an unselected population consisting of 84 Macedonian patients with ESRF not yet on dialysis therapy. Patients were recruited from a well-defined geographic area during a relatively short period (9 months). For each patient, bone histomorphometric findings were correlated with serum levels of various biochemical parameters, after which the diagnostic performance was evaluated for the most relevant ones. PATIENTS, MATERIAL, AND METHODS

Patient Selection From September 1998 to June 1999, all patients with ESRF not yet on dialysis therapy who presented at any of the 18 Macedonian ambulatory nephrology centers covering the whole country were asked to participate in this study. There were no particular exclusion or inclusion criteria. Of 103 patients, 93 patients gave their informed consent and underwent blood sampling and transiliacal bone biopsy after double tetracycline labeling. Bone biopsy specimens were obtained using a 5-mm Bordier-Meunier needle. The bone biopsy specimen was divided into two pieces. The first part

was fixed in Burkhardt’s solution for 24 hours, after which it was transferred to 96% ethanol and stored at 4°C until histomorphometric quantification for diagnostic classification. The second piece was weighed directly after sampling and used for bulk analysis. Because nine biopsy specimens were inadequate to make a pathological diagnosis, histomorphometric analysis was performed in 84 of 93 bone biopsy specimens. At the time of bone sampling, a serum sample was collected, centrifuged, and stored at ⫺80°C until analysis. Only serum samples of patients for whom a histomorphometric bone examination was performed were analyzed (N ⫽ 84). None of the patients was administered vitamin D or erythropoietin. The only phosphate binder administered was calcium carbonate, three times daily (70% of patients at the time of study). None of the patients had clinical symptoms or overt bone disease. None of the patients had undergone parathyroidectomy or renal transplantation or had liver disease. Underlying renal diagnoses were chronic glomerulonephritis (15%), interstitial kidney disease (19%), nephrosclerosis (27%), insulin-dependent diabetes mellitus (12%), polycystic disease (4%), amyloidosis (2%), multicystic disease, and obstructive nephropathy and uric nephropathy (1% each). Renal diagnosis was undefined in 17% of patients. The study was approved by the local ethical committees of participating hospitals.

Histomorphometric and Histochemical Analysis of Bone Undecalcified bone samples were dehydrated and embedded in 100% methylmetacrylate. Five-micron thick slices were Goldner stained for histological analysis (static parameters) and stained with Aluminon (Aldrich-Europe, Beerse, Belgium) for the detection and localization of aluminum. In case of aluminum positivity, iron staining also was performed to eliminate false-positive results caused by costaining of the latter element.21,22 Ten-micron thick slices were used for visualization of tetracycline labels under fluorescent microscopy (dynamic parameters). Histological classification of ROD was performed according to international standards.23 Different types of ROD were diagnosed according to amount of osteoid, presence of fibrosis, and bone formation rate (BFR) as follows: normal histological characteristics, osteoid area less than 12%, no fibrosis, and BFR of 97 to 613 ␮m2/mm2/d; HPTH, osteoid area less than 12% either without fibrosis and with a BFR greater than 613 ␮m2/mm2/d (mild) or with fibrosis and a BFR greater than 97 ␮m2/mm2/d (osteitis fibrosa); OM, osteoid area greater than 12%, no fibrosis, and a BFR less than 97 ␮m2/mm2/d; ABD, osteoid area less than 12%, no fibrosis, and a BFR less than 97 ␮m2/mm2/d; and MX (OM and HPTH component), osteoid area greater than 12% either with fibrosis and a BFR greater than 0 ␮m2/mm2/d or without fibrosis and a BFR greater than 97 ␮m2/mm2/d. Biopsy specimens were examined and classified without knowledge of biochemical and clinical findings. Concentrations of aluminum and strontium in bone, both known to have a role in ROD,24-33 were determined by means of electrothermal atomic absorption spectrometry

USE OF BONE MARKERS IN PREDIALYSIS PATIENTS

with Zeeman background correction (Perkin-Elmer Zeeman 3030; graphite furnace HGA 600; Perkin-Elmer, Norwalk, CT). Calcium concentration, used to correct for differences in bone density, was determined using atomic absorption spectrometry with flame atomization (Perkin-Elmer model 3110). Methods for the measurement of aluminum and strontium in bone have been described previously in detail by our group.34-36 In our laboratory, bone aluminum and strontium levels of dialysis patients of 14 ␮g/g or less (ⱕ0.52 ␮mol/g) wet weight (w/w) and 60 ␮g/g or less (ⱕ0.68 ␮mol/g) w/w are not considered to be associated with aluminum- or strontium-related bone disease, respectively.37

Biochemical and/or Chemical Analysis of Serum iPTH in serum was measured by means of an immunoradiometric assay (IRMA) using the PTH-120 MIN-IRMA kit of Biosource Europe SA (Nivelles, Belgium). Normal values with this kit range between 13 and 66 pg/mL (ng/L). Intraassay and interassay coefficients of variation (CVs) are 4.9% and 5.7%, respectively. Serum OC was measured by means of the Medgenix Osteocalcin IRMA kit (Biosource Europe SA). Normal values range between 5 ␮g/L (0.86 nmol/L) and 25 ␮g/L (4.23 nmol/L). Reported intra-assay and interassay CVs with this method are 2.7% and 5.8%, respectively. PYDs and DPYDs were measured using an in-house– developed method.38 In summary, serum is hydrolyzed overnight using hydrochloric acid at 95°C, centrifuged, and filtered through a Millex GS 0.22-␮m filter (Millipore NV, Brussels, Belgium). From this step onward, all reagents used are part of the Bio-Rad Crosslinks kit (Bio-Rad, Munich, Germany). Ultrafiltrates are washed in wash and conditioning reagent and put on conditioned cellulose columns in a vacuum box. Columns are washed three times, then eluted twice under slow pressure. One hundred microliters of eluate is transferred to a Bio-Rad high-performance liquid chromatography apparatus (Bio-Rad, Richmond, CA) and passed through a reversed-phase C18 ion pair column. Standards and controls are from Bio-Rad. Normal values for pyridinolines are at the lower limit of detection (4 and 2 nmol/L for PYD and DPYD, respectively), whereas interassay CVs are 3.0% and 5.0%, respectively. Total AP (TAP) level was determined by means of a kinetic method at 25°C on a SLT Lab Instruments (Austria). Normal values with this method range between 63 and 166 U/L. Isoenzymes were quantified after electrophoretic separation on an agarose gel (ISOPAL kit, Analis, Namen, Belgium), described previously by Van Hoof et al.39 Aluminum and strontium in serum were measured using the same technique as for their determination in bone.34,36,40 In our laboratory, threshold values for these parameters in dialysis patients are 30 ng/mL or less (ⱕ1.11 ␮mol/L) and 100 ng/mL or less (ⱕ1.14 ␮mol/L), respectively.24,37 Interassay CVs of aluminum and strontium measurements are 2.9% and 3.2%, respectively. All these parameters were determined in single batches in one laboratory. Serum calcium and serum phosphorous were measured by means of standard automated techniques. Normal values for serum calcium are 8.5 to 10.2 mg/dL (2.13 to 2.55 mmol/L),

999

and for serum phosphorous, 2.6 to 4.6 mg/dL (0.84 to 1.49 mmol/L).

Statistics For statistical analysis, we used SPSS 10.0 for Windows 2000 (SSPS Inc, Chicago, IL). We performed nonparametric analysis on all data obtained during this study. Values are expressed as median and range unless otherwise stated. Relationships between histomorphometric bone parameters and various biochemical markers were assessed using Spearman’s rank order correlation. Spearman’s rank order correlation also was used to assess the interrelationship between various markers under study. Bonferroni adjustment was applied to correct for the number of correlations. P less than 0.05 is considered significant at a two-tailed level. KruskalWallis was used for comparison of more than two groups, followed by Mann-Whitney U test, in combination with Bonferroni test to correct for the number of comparisons when pairwise comparison of various biochemical markers in various types of ROD was performed. The diagnostic performance of tests showing a high enough discriminative power between the various groups was assessed further by using receiver operator characteristic (ROC) curves of the different markers studied. By varying the decision thresholds (cutoff values) over the complete range of test results, the ROC plot graphically presents all possible sensitivity/specificity pairs for a particular test and is useful for comparing the ability of tests to discriminate between alternative states of health, in this case ROD. Sensitivity (the true-positive fraction) is defined as the number of true-positive test results divided by the sum of the number of true-positive and false-negative test results. Specificity (the ability of a test to correctly exclude the presence of the disease) is defined as the number of true-negative results divided by the sum of the number of true-negative and false-positive results. Areas under the ROC curves (AUCs) also were calculated. Possible values range from 1.0 (perfect separation of test values into two groups) to 0.5 (no distributional difference). An AUC greater than 0.7 indicates a discriminating strength of statistical significance; an AUC greater than 0.8 indicates excellent discriminating power for the test.41 The optimal discrimination limit for each test was determined at the maximum of the Youden’s index (Y ⫽ sensitivity ⫹ specificity ⫺ 1). Bayes’ theorem was applied to determine positive (PPV) and negative predictive value (NPV) in function of the prevalence of a particular type of ROD.

RESULTS

Characteristics and spectrum of ROD of the study population have been described previously in detail (G. Spasovski, PhD thesis, University of Skopje, Macedonia, 2002). An overview of demographics is listed in Table 1. Mean age of the study population was 54.2 ⫾ 12.1 years; 44 patients were men. Routine blood analysis showed the typical portrait of a patient with end-stage chronic renal insufficiency. All pa-

1000

BERVOETS ET AL Table 1.

Demographic Description of the Population Under Study Bone Histological State

No. of patients Prevalence (%) Men/women Mean age (y) Diabetes Serum creatinine (mg/dL) Serum phosphorus (mg/dL) Total protein (g/dL)

ABD

Normal

HPTH

MX

OM

19 23 13/6 52.0 ⫾ 15.3 8/19 9 (5-19) 6.9 (1.1-10.8) 6.2 (3.4-8.2)

32 38 16/16 53.8 ⫾ 10.7 4/32 11 (6-20) 6.7 (1.9-11.1) 6.6 (4.5-8.4)

8 9 7/1 49.3 ⫾ 6.9 0/8 10 (5-15) 6.3 (4.7-8.7) 6.6 (5.6-7.4)

15 18 5/10 54.7 ⫾ 13.8 2/15 9 (6-16) 7.4 (2.1-12.8) 7.3 (5.1-8.2)

10 12 3/7 63.2 ⫾ 5.7* 1/10 10 (6-17) 6.8 (4.8-12) 7.3 (5.1-7.9)

NOTE. Values expressed as mean ⫾ SD or median (range) unless noted otherwise. To convert serum creatinine in mg/dL to ␮mol/L, multiply by 88.4; to convert serum phosphorus in mg/dL to mmol/L, multiply by 0.323; to convert serum total protein in g/dL to g/L, multiply by 10. *Mann-Whitney P ⬍ 0.05 versus other groups.

tients had a creatinine clearance less than 5 mL/min (0.08 mL/s). Serum creatinine and total protein levels did not differ significantly between the various ROD groups (Table 1). Analysis of Goldner-stained slices and fluorescent labeling showed that in the present study population and using the proposed histomorphometric criteria, 38% of patients still had normal bone histological characteristics despite their advanced renal failure and relatively increased iPTH levels (63% had iPTH ⬎ 150 pg/mL [ng/L]). Twenty-three percent of patients had ABD, 12% had OM, 9% presented with mild HPTH, and 18% had MX. No patient presented with marrow fibrosis. Aluminon staining was negative in all patients, as was bulk analysis of bone for aluminum and strontium content, two factors known to cause OM30,31,42 or ABD.26,27,42-47 Overall bone concentrations were 4.5 ⫾ 5.6 ␮g/g (0.17 ⫾ 0.21 ␮mol/g) w/w for aluminum and 46.3 ⫾ 25.7 ␮g/g (0.53 ⫾ 0.29 ␮mol/g) w/w for strontium. Overall mean serum concentrations were 12.8 ⫾ 18.2 (SD) ng/mL (0.47 ⫾ 0.67 ␮mol/L) for aluminum and 63.6 ⫾ 51.7 ng/mL (0.73 ⫾ 0.59 ␮mol/L) for strontium. These values all range within normal limits for patients with renal failure.24 None of the biochemical markers under study (iPTH, OC, TAP, BAP, PYD, DPYD) correlated with sex, age, or underlying renal disease. Only PYD and DPYD levels correlated with serum creatinine levels (r ⫽ 0.54; r ⫽ 0.51, respectively; P ⬍ 0.01). Some of the markers significantly differed between different groups of ROD.

Kruskal-Wallis showed significant differences between groups for iPTH, BAP, TAP, OC, and serum calcium levels (P ⬍ 0.005), whereas PYD or DPYD levels did not differ (Table 2). Furthermore, Mann-Whitney U showed a significant difference (P ⬍ 0.05) in serum calcium levels between both ABD (higher) and OM (lower) versus all other groups. Significant associations between biochemical and histological parameters assessed by Spearman’s correlation matrix are listed in Table 3. In contrast to BAP and TAP levels, OC, iPTH, DPYD, or PYD levels did not correlate significantly with any histomorphometric or histodynamic parameter. A detailed analysis showed OC and BAP levels to have the best diagnostic performance to differentiate ABD from other types of ROD. An overview of the sensitivity and specificity for the different markers is listed in Tables 4 and 5. Using a cutoff value of 41 ␮g/L or less (ⱕ7.0 nmol/L), OC level showed a sensitivity of 83% and specificity of 67% for the diagnosis of ABD versus other types of ROD (Fig 1A). For the prevalence noted in the present study, this corresponds with a PPV and NPV of 47% and 94%, respectively (Fig 2A). In combination with BAP level (cutoff value ⱕ 23 U/L), sensitivity decreases to 72%, but specificity increases to 89% in the diagnosis of ABD versus other types of ROD, resulting in a PPV and NPV for the population under study of 77% and 93% (Figs 1 and 2A). iPTH level is less useful, with a sensitivity of 78% and specificity of only 53% at a cutoff value

USE OF BONE MARKERS IN PREDIALYSIS PATIENTS Table 2.

1001

Concentration of Biochemical Markers in the Study Population

Biochemical Marker

ABD

Normal

HPTH

MX

OM

iPTH (pg/mL)* OC (␮g/L)* TAP (U/L)* BAP (U/L)* PYD (nmol/L) DPYD (nmol/L) Calcium in serum (mg/dL)*

128 (2-377) 28.6 (0-244) 59 (5-116) 18.5 (1-42) 131 (12-467) 23 (2-63) 9 (6.4-10.5)†

227 (17-611) 58.2 (7-117) 61 (18-106) 21 (7-56) 145 (69-3,640) 28.5 (14-279) 8.5 (3.1-12.0)

200.5 (38-1,790) 76 (11-183) 91 (50-223) 41.5 (20-126) 159.5 (58-536) 35 (16-112) 8.9 (6.7-11.4)

309 (53-1,714) 41.4 (11-183) 116 (23-178) 46 (16-115) 162.5 (48-374) 42 (12-67) 7.8 (4.8-10.0)

376 (55-549) 36.6 (28-64) 104 (40-126) 33 (17-90) 147 (70-363) 35 (17-74) 6.8 (4.9-11.0)‡

NOTE. Values expressed as median (range). To convert serum iPTH in pg/mL to ng/L, multiply by 1.0; to convert serum OC in ␮g/L to nmol/L, multiply by 0.171; to convert serum TAP or BAP in U/L to U/L, multiply by 1.0; to convert serum calcium in mg/dL to mmol/L, multiply by 0.25. *Kruskall-Wallis P ⬍ 0.005. †Mann-Whitney P ⫽ 0.037 versus all other groups. ‡Mann-Whitney P ⫽ 0.022 versus all other groups.

of 237 pg/mL (ng/L) or less in the diagnosis of ABD versus other types of ROD. Figure 1 also shows AUCs of the various markers in the diagnosis of ABD. These were 0.796, 0.785, and 0.752 for OC, BAP, and iPTH, respectively, thus indicating all three markers to have a significant discriminating power. Because many investigators do not consider normal bone a criterion for classification,5,48-52 we combined the ABD group with the group with normal bone versus other types of ROD (Fig 1B). BAP level was found to be a good marker to distinguish this combination from other types of ROD, showing a sensitivity of 72% and specificity of 76%, corresponding with a PPV and NPV of 89% and 82% at a cutoff level of 25 U/L or less (Fig 2B). TAP level was an even better marker than BAP level to differentiate this combined group from other types of ROD. Using a Table 3.

cutoff value of 84 U/L or less, sensitivity was 88% and specificity was 60%, resulting in a PPV and NPV for the population under study of 85% and 89%, respectively (Fig 2B). OC level is less powerful to detect ABD and normal bone taken as one group, shown by an AUC of only 0.522 compared with those for BAP (0.819), TAP (0.759), and iPTH (0.712; Fig 1B). In the absence of aluminum or strontium overload, the prevalence of OM within the study population is relatively high (12%). Patients with OM had significantly lower levels of serum calcium (6.98 ⫾ 1.86 mg/dL [1.75 ⫾ 0.47 mmol/ L]) compared with all other groups (Table 2). Using serum calcium level, we are able to differentiate between OM and all other types of ROD with both a sensitivity and specificity of 78% at a cutoff value of 7.15 mg/dL or less (ⱕ1.79 mmol/ L), whereas the combination of a high serum

Correlation Matrix of Biochemical and Histological Parameters

Correlation

r

P

Correlation

r

P

iPTH v TAP iPTH v BAP TAP v BAP PYD v DPYD iPTH v OC iPTH v PYD iPTH v DPYD PYD v creatinine DPYD v creatinine

0.45 0.55 0.76 0.93 0.39 0.35 0.49 0.54 0.51

⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001

BAP v BFR BAP v OWI BAP v OAR TAP v OWI TAP v OAR iPTH v OWI iPTH v BFR OC v DLPM OC v MAR OC v AJAR OC v BFR

0.42 0.46 0.54 0.38 0.44 0.44 0.24 0.46 0.34 0.38 0.42

⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.05 ⬍0.001 ⬍0.001 ⬍0.001 ⬍0.001

Abbreviations: OWI, osteoid width; OAR, osteoid area; DLPM, double-labeled perimeter; MAR, mineral apposition rate; AJAR, adjusted apposition rate.

1002

BERVOETS ET AL

Table 4. Sensitivity and Specificity of Biochemical Markers in the Diagnosis of ABD Versus All Other Types of ROD

BAP ⱕ 23 U/L TAP ⱕ 66 U/L OC ⱕ 41 ␮g/L iPTH ⱕ 237 pg/mL BAP ⱕ 23 U/L and OC ⱕ 41 ␮g/L

Sensitivity (%)

Specificity (%)

Youden’s Index

83 74 83 78 72

66 54 67 53 89

0.49 0.28 0.50 0.31 0.61

NOTE. To convert serum TAP or BAP in U/L to U/L, multiply by 1.0; to convert serum iPTH in pg/mL to ng/L, multiply by 1.0; to convert serum OC in ␮g/L to nmol/L, multiply by 0.171.

calcium level (cutoff value ⬎ 7.4 mg/dL [1.85 mmol/L]) and low OC level (cutoff value ⱕ 35 ␮g/L [5.98 nmol/L]) allows further differentiation between ABD and OM (which essentially are two low bone turnover diseases), with a sensitivity of 68% and specificity of 89%. DISCUSSION

Much is known about ROD in dialysis patients. However, data on this issue in patients with predialysis ESRF are limited to the studies of Dahl et al,6 Hutchison et al,5 Mazzaferro et al,20 and Coen et al.53 In the present study, we had the opportunity to assemble a relatively large number of patients (N ⫽ 93) who consented to undergo bone biopsy before entering a dialysis program. Patients were recruited from 18 different centers in one geographic region encompassing 80% of all patients who presented with ESRF during a 10-month period. Hence, the population under study is unique in its nonselective character of patient inclusion in the study protocol. Serum concentrations of various biochemical markers studied (OC, BAP, TAP, and iPTH) differed significantly between groups, with the exception of DPYD or PYD. Because PYD and DPYD levels correlate well with serum creatinine levels, an explanation could be that these rather small molecules, in contrast to the other biochemical markers (molecular weights: BAP and TAP, 140 kd in their dimeric form; OC, 5,700 d; PTH, 9,500 d) under study, can still be filtered through the kidney. Hence, serum PYD and DPYD concentrations will depend greatly on renal function, making them less suitable for use

in populations with varying degrees of renal failure, as is the case to a certain extent in the population under study. In dialysis patients, it has been suggested that PYD and DPYD levels are good markers of bone resorption and have a substantial diagnostic performance within this population.9,54-56 First, we assessed whether any of the biochemical parameters was of value in the diagnosis of ABD. We found BAP was a good biochemical marker for the detection of ABD in predialysis patients with ESRF. Our group also previously showed low serum levels of BAP (ⱕ27 U/L) to be a good index for the presence of ABD in hemodialysis patients57 because sensitivity and specificity were 78% and 86%, corresponding to a PPV of 75% for the population under study, respectively (35% of patients had ABD). We also found OC level to be a valuable marker for the detection of ABD, with a diagnostic performance close to that of BAP. Although in dialysis patients, OC levels have been found to be considerably higher than those in subjects with normal renal function with both high- and low-turnover ROD, this parameter has been reported to discriminate well between high- and low-turnover types of ROD.10 Moreover, OC level correlated well with TAP and iPTH levels.11 Conversely, we previously reported OC level in dialysis patients to be inferior to BAP level in the noninvasive diagnosis of ROD.57 Concerns also have been raised about the compound’s instability, resulting in the accumulation of degradation products with decreasing renal function; an issue that, with the introduction of new assays, can be solved by measuring the intact molecule. In the present population, the most accurate Table 5. Sensitivity and Specificity of Biochemical Markers in the Diagnosis of ABD and Normal Bone as One Group Versus Other Types of ROD

BAP ⱕ 25 U/L TAP ⱕ 84 U/L iPTH ⱕ 259 pg/mL BAP ⱕ 25 U/L and iPTH ⱕ 259 pg/mL

Sensitivity (%)

Specificity (%)

Youden’s Index

72 88 76 86

76 60 63 58

0.48 0.48 0.39 0.44

NOTE. To convert serum TAP or BAP in U/L to U/L, multiply by 1.0; to convert serum iPTH in pg/mL to ng/L, multiply by 1.0.

USE OF BONE MARKERS IN PREDIALYSIS PATIENTS

1003

Fig 1. (A, B) ROC curves for BAP, TAP, iPTH, and OC levels in the diagnosis of ABD versus other types of ROD and the diagnosis of ABD and normal bone considered one group versus other types of ROD.

diagnosis of ABD was made by the combined use of BAP and OC at their respective cutoff values of 23 U/L or less and 41 ␮g/L or less (ⱕ7.0 nmol/L), yielding a PPV and NPV of 77% and 93%, respectively. The diagnostic value of iPTH level in the detection of ABD (AUC, 0.752) is not as good as OC (AUC, 0.796) and BAP levels (AUC, 0.785). Recent data in the literature58 suggest that com-

mercially available assays to determine iPTH levels also measure other fragments of PTH, including the truncated 7-84 PTH peptide. New assays that measure whole PTH, specific for 1-84 PTH, appear to be more accurate and probably may hold a better diagnostic performance in the differentiation of various types of ROD. Whole PTH values appear to range from 20% to 90% of iPTH values.59

Fig 2. (A, B) PPVs and NPVs of various biochemical markers at a fixed cutoff level in the diagnosis of patients with ABD and those with ABD and normal bone versus other types of ROD.

1004

In a similar group of 30 patients with ESRF not yet on dialysis therapy, Hutchison et al5 evaluated iPTH as a discriminating marker between ABD and osteitis fibrosa. In their study, an iPTH level of 65 pg/mL (ng/L) or less had a sensitivity of 88% and specificity of 91% in identifying patients with ABD. For the diagnosis of osteitis fibrosa, an iPTH level greater than 200 pg/mL (ng/L) had a sensitivity of 83% and specificity as high as 88%. We found that iPTH at a cutoff value of less than 65 pg/mL (ng/L) had a sensitivity of only 40% and specificity of 89% in identifying patients with ABD, whereas 237 pg/mL (ng/L) or less corresponded with a sensitivity of 78% and specificity of only 53%. The discrepancy between their and our results can be explained at least in part because the majority of patients in the study by Hutchison et al5 had severe HPTH (27%); osteitis fibrosa was present in 15 of the 30 patients. Hence, one can reasonably expect serum iPTH levels to be distinctly increased in these patients compared with those not presenting with high bone turnover. Conversely, in our population, a relatively high number of patients presented with low to normal bone turnover, whereas those with HPTH expressed only the mild form (no study patient had marrow fibrosis), which was accompanied by only moderately increased iPTH levels, resulting in a substantial overlap with the other ROD groups. In this context, it is worth mentioning that serum iPTH levels in the range measured in the present study population repeatedly have been reported to be of limited diagnostic value.19,60 Therefore, Qi et al60 showed that no definite diagnosis of ROD can be made on the basis of iPTH levels between 65 and 450 pg/mL (ng/L). Uren˜ a and De Vernejoul9 reviewed the diagnostic value of different biochemical markers and found that iPTH levels in patients with ESRF should be kept between 120 and 400 pg/mL (ng/L), corresponding to values less than 65 pg/mL (ng/L) in subjects with normal renal function. In agreement with the present data and those from our previous studies in dialysis patients, results of a very recent study comparing histological patterns and diagnostic predictivity of iPTH levels in predialysis versus hemodialysis patients also indicate the predictive value of iPTH level

BERVOETS ET AL

in the noninvasive diagnosis of renal bone disease to be greater in dialyzed patients than predialysis patients. In this study, significantly lower slopes of regression curves correlating iPTH levels with various histological and histodynamic parameters noticed in predialysis patients versus hemodialysis patients made the investigators put forward the interesting suggestion that bone resistance to PTH might be more pronounced in predialysis patients.61 We also evaluated the diagnostic performance of the various biochemical markers in the detection of ABD and normal bone as one group versus other types of ROD. The reason for doing so is in keeping with the suggestion made by Fournier et al62 that only aluminum-related ABD bone is actually a bone disease. Hence, in the absence of significant aluminum exposure, ABD may be considered a bone histological pattern that has only discrete clinical relevance.63 Moreover, nonaluminic ABD does not seem to be associated with decreased bone volume or decreased bone density62,64 or be characterized by bone pain or an increased incidence of fractures.63 In this regard, bone biopsy in patients with either ABD or normal bone would not be justified. Therefore, considering ABD and normal bone as a single group to discriminate against the clinically relevant types of ROD, such as HPTH, OM, and MX, seems to be an interesting approach. Again, BAP level was found to be the most powerful marker, with a PPV and NPV of 89% and 82% at a cutoff level of 25 U/L or less, respectively. In the diagnosis of OM in the group with low bone turnover disease, possible aluminum or strontium exposure should be assessed first because this is well recognized as a causal factor for the development of OM. If there has been no aluminum or strontium exposure, as in the population under study, hypovitaminosis D is a more probable cause, and serum calcium level is a helpful marker for the detection of OM. Moreover, in the ABD group, a serum calcium level of 7.4 mg/dL or less (ⱕ1.85 mmol/L) and OC level of 35 ␮g/L or less (ⱕ5.98 nmol/L) allows the detection of false-positive results in patients with OM. In the context of hypovitaminosis and the development of OM, data from Ghazali et al65 are of particular interest because in their study, osteomalacic lesions, diagnosed by the presence

USE OF BONE MARKERS IN PREDIALYSIS PATIENTS

of Looser’s zones, were associated with decreased levels of 25-(OH) vitamin D3, (not 1␣,25(OH)2 vitamin D3). In conclusion, we found BAP level to be a useful diagnostic marker to differentiate between ABD/normal bone and the other types of ROD. Combined with OC level, BAP level is a reliable index for the presence of ABD as a single entity. In the absence of increased aluminum or strontium exposure, a low serum calcium level allows one to differentiate between OM and the other types of ROD. Within the group of patients diagnosed as having ABD using BAP, false positives for ABD having OM can be detected by the combination of serum calcium and OC. ACKNOWLEDGMENT The authors thank Dirk De Weerdt for his excellent artwork.

REFERENCES 1. Llach F, Coburn JW: Renal osteodystrophy and maintenance dialysis, in Maher JF (ed): Replacement of Renal Function by Dialysis. Dordrecht, The Netherlands, Kluwer, 1989, pp 911-952 2. Malluche H, Faugere MC: Renal bone disease 1990: An unmet challenge for the nephrologist. Kidney Int 38:193211, 1990 3. Rix M, Andreassen H, Eskildsen P, Langdahl B, Olgaard K: Bone mineral density and biochemical markers of bone turnover in patients with predialysis chronic renal failure. Kidney Int 56:1084-1093, 1999 4. Atsumi K, Kushida K, Yamazaki K, Shimizu S, Ohmura A, Inoue T: Risk factors for vertebral fractures in renal osteodystrophy. Am J Kidney Dis 33:287-293, 1999 5. Hutchison AJ, Whitehouse RW, Boulton HF, et al: Correlation of bone histology with parathyroid hormone, vitamin D3, and radiology in end-stage renal disease. Kidney Int 44:1071-1077, 1993 6. Dahl E, Nordal KP, Attramadal A, Halse J, Flatmark A: Renal osteodystrophy in predialysis patients without stainable bone aluminum. A cross-sectional bone-histomorphometric study. Acta Med Scand 224:157-164, 1988 7. Nilsson P, Melsen F, Malmaeus J, Danielson BG, Mosekilde L: Relationships between calcium and phosphorus homeostasis, parathyroid hormone levels, bone aluminum, and bone histomorphometry in patients on maintenance hemodialysis. Bone 6:21-27, 1985 8. Salusky IB, Ramirez JA, Oppenheim W, Gales B, Segre GV, Goodman WG: Biochemical markers of renal osteodystrophy in pediatric patients undergoing CAPD/ CCPD. Kidney Int 45:253-258, 1994 9. Urena P, De Vernejoul MC: Circulating biochemical markers of bone remodeling in uremic patients. Kidney Int 55:2141-2156, 1999 10. Malluche HH, Faugere MC, Fanti P, Price PA: Plasma levels of bone Gla-protein reflect bone formation in patients

1005

on chronic maintenance dialysis. Kidney Int 26:869-874, 1984 11. Charhon SA, Delmas PD, Malaval L, et al: Serum bone Gla-protein in renal osteodystrophy: Comparison with bone histomorphometry. J Clin Endocrinol Metab 63:892897, 1986 12. Cohen-Solal ME, Sebert JL, Boudailliez B, et al: Non-aluminic adynamic bone disease in non-dialyzed uremic patients: A new type of osteopathy due to overtreatment? Bone 13:1-5, 1992 13. Sebert JL, Fournier A, Leflon P, et al: Comparative evaluation of bone aluminum content and bone histology in patients on chronic hemodialysis and hemofiltration. Nephron 42:34-40, 1986 14. Hamdy NA, Risteli J, Risteli L, et al: Serum type I procollagen peptide: A non-invasive index of bone formation in patients on haemodialysis? Nephrol Dial Transplant 9:511-516, 1994 15. Joffe P, Heaf JG, Hyldstrup L: Osteocalcin: A noninvasive index of metabolic bone disease in patients treated by CAPD. Kidney Int 46:838-846, 1994 16. Mazzaferro S, Pasquali M, Ballanti P, et al: Diagnostic value of serum peptides of collagen synthesis and degradation in dialysis renal osteodystrophy. Nephrol Dial Transplant 10:52-58, 1995 17. Urena P, Hruby M, Ferreira A, Ang KS, De Vernejoul MC: Plasma total versus bone alkaline phosphatase as markers of bone turnover in hemodialysis patients. J Am Soc Nephrol 7:506-512, 1996 18. Urena P, Ferreira A, Kung VT, et al: Serum pyridinoline as a specific marker of collagen breakdown and bone metabolism in hemodialysis patients. J Bone Miner Res 10:932-939, 1995 19. Gerakis A, Hutchison AJ, Apostolou T, Freemont AJ, Billis A: Biochemical markers for non-invasive diagnosis of hyperparathyroid bone disease and adynamic bone in patients on haemodialysis. Nephrol Dial Transplant 11:24302438, 1996 20. Mazzaferro S, Coen G, Ballanti P, et al: Osteocalcin, iPTH, alkaline phosphatase and hand X-ray scores as predictive indices of histomorphometric parameters in renal osteodystrophy. Nephron 56:261-266, 1990 21. Verbueken AH, Van de Vyver FL, Van Grieken RE, et al: Ultrastructural localization of aluminum in patients with dialysis-associated osteomalacia. Clin Chem 30:763-768, 1984 22. Visser WJ: Aluminum induced bone disease: Histology, in De Broe ME, Coburn JW (eds): Aluminum and Renal Failure. Dordrecht, The Netherlands, Kluwer, 1989, pp 241247 23. Salusky IB, Coburn JW, Brill J, et al: Bone disease in pediatric patients undergoing dialysis with CAPD or CCPD. Kidney Int 33:975-982, 1988 24. Schrooten I, Elseviers MM, Lamberts LV, De Broe ME, D’Haese PC: Increased serum strontium levels in dialysis patients: An epidemiological survey. Kidney Int 56:1886-1892, 1999 25. D’Haese PC, Schrooten I, Goodman WG, et al: Increased bone strontium levels in hemodialysis patients with osteomalacia. Kidney Int 57:1107-1114, 2000 26. Andress DL, Maloney NA, Endres DB, Sherrard DJ:

1006

Aluminum-associated bone disease in chronic renal failure: High prevalence in a long-term dialysis population. J Bone Miner Res 1:391-398, 1986 27. Parisien M, Charhon SA, Arlot M, et al: Evidence for a toxic effect of aluminum on osteoblasts: A histomorphometric study in hemodialysis patients with aplastic bone disease. J Bone Miner Res 3:259-267, 1988 28. Dahl E, Nordal KP, Halse J, Flatmark A: The early effects of aluminium deposition and dialysis on bone in chronic renal failure: A cross-sectional bone-histomorphometric study. Nephrol Dial Transplant 5:449-456, 1990 29. Felsenfeld AJ, Machado L, Bover J, Trinidad P, Rodriguez M: Effect of aluminium on the development of hyperparathyroidism and bone disease in the azotaemic rat. Nephrol Dial Transplant 8:325-334, 1993 30. Goodman WG, Henry DA, Horst R, Nudelman RK, Alfrey AC, Coburn JW: Parenteral aluminum administration in the dog: II. Induction of osteomalacia and effect on vitamin D metabolism. Kidney Int 25:370-375, 1984 31. Hodsman AB, Sherrard DJ, Alfrey AC, et al: Bone aluminum and histomorphometric features of renal osteodystrophy. J Clin Endocrinol Metab 54:539-546, 1982 32. Quarles LD, Dennis VW, Gitelman HJ, Harrelson JM, Drezner MK: Aluminum deposition at the osteoid-bone interface. An epiphenomenon of the osteomalacic state in vitamin D-deficient dogs. J Clin Invest 75:1441-1447, 1985 33. Rodriguez M, Felsenfeld AJ, Llach F: The evolution of osteomalacia in the rat with acute aluminum toxicity. J Bone Miner Res 4:687-696, 1989 34. D’Haese PC, Van de Vyver FL, de Wolff FA, De Broe ME: Measurement of aluminum in serum, blood, urine, and tissues of chronic hemodialyzed patients by use of electrothermal atomic absorption spectrometry. Clin Chem 31:2429, 1985 35. Liang L, D’Haese PC, Lamberts LV, De Broe ME: Direct calibration for determining aluminum in bone and soft tissues by graphite furnace atomic absorption spectrometry. Clin Chem 37:461-466, 1991 36. D’Haese PC, Van Landeghem GF, Lamberts LV, Bekaert VA, Schrooten I, De Broe ME: Measurement of strontium in serum, urine, bone, and soft tissues by Zeeman atomic absorption spectrometry. Clin Chem 43:121-128, 1997 37. D’Haese PC, Couttenye MM, Goodman WG, et al: Use of the low-dose desferrioxamine test to diagnose and differentiate between patients with aluminium-related bone disease, increased risk for aluminium toxicity, or aluminium overload. Nephrol Dial Transplant 10:1874-1884, 1995 38. Van Hoof VO, Couttenye MM, Van Brussel KA, Wendelen LC, Waegemans SE, De Broe ME: Biochemical evaluation of bone metabolism in dialysis patients. Nierenund Hochdruckkrankheiten 28:S402-S406, 1999 (suppl 10) 39. Van Hoof VO, Lepoutre LG, Hoylaerts MF, Chevigne R, De Broe ME: Improved agarose electrophoretic method for separating alkaline phosphatase isoenzymes in serum. Clin Chem 34:1857-1862, 1988 40. D’Haese PC, Couttenye MM, Lamberts LV, et al: Aluminum, iron, lead, cadmium, copper, zinc, chromium, magnesium, strontium, and calcium content in bone of end-stage renal failure patients. Clin Chem 45:1548-1556, 1999 41. Hanley JA, McNeil BJ: The meaning and use of the

BERVOETS ET AL

area under a receiver operating characteristic (ROC) curve. Radiology 143:29-36, 1982 42. Ott SM, Maloney NA, Coburn JW, Alfrey AC, Sherrard DJ: The prevalence of bone aluminum deposition in renal osteodystrophy and its relation to the response to calcitriol therapy. N Engl J Med 307:709-713, 1982 43. Andress DL, Maloney NA, Coburn JW, Endres DB, Sherrard DJ: Osteomalacia and aplastic bone disease in aluminum-related osteodystrophy. J Clin Endocrinol Metab 65:11-16, 1987 44. Charhon SA, Chavassieux PM, Chapuy MC, Traeger J, Meunier PJ: High bone turnover associated with an aluminum-induced impairment of bone mineralization. Bone 7:319-324, 1986 45. Plachot JJ, Cournot-Witmer G, Halpern S, et al: Bone ultrastructure and x-ray microanalysis of aluminum-intoxicated hemodialyzed patients. Kidney Int 25:796-803, 1984 46. De Vernejoul MC, Belenguer R, Halkidou H, Buisine A, Bielakoff J, Miravet L: Histomorphometric evidence of deleterious effect of aluminum on osteoblasts. Bone 6:1520, 1985 47. De Vernejoul MC: Renal osteodystrophy: Aluminium and secondary hyperparathyroidism. Rev Prat 40:613-618, 1990 48. Sherrard D, Baylink D, Wergedal J: Bone disease in uremia. Trans Am Soc Artif Intern Organs 18:412-415, 421, 1972 49. Llach F, Felsenfeld AJ, Coleman MD, Keveney JJ Jr, Pederson JA, Medlock TR: The natural course of dialysis osteomalacia. Kidney Int Suppl 18:S74-S79, 1986 50. Moriniere P, Cohen-Solal M, Belbrik S, et al: Disappearance of aluminic bone disease in a long term asymptomatic dialysis population restricting A1(OH)3 intake: Emergence of an idiopathic adynamic bone disease not related to aluminum. Nephron 53:93-101, 1989 51. Hernandez D, Concepcion MT, Lorenzo V, et al: Adynamic bone disease with negative aluminium staining in predialysis patients: Prevalence and evolution after maintenance diaysis. Nephrol Dial Transplant 9:517-523, 1994 52. Torres A, Lorenzo V, Hernandez D, et al: Bone disease in predialysis, hemodialysis, and CAPD patients: Evidence of a better bone response to PTH. Kidney Int 47:1434-1442, 1995 53. Coen G, Mazzaferro S, Ballanti P, et al: Renal bone disease in 76 patients with varying degrees of predialysis chronic renal failure: A cross-sectional study. Nephrol Dial Transplant 11:813-819, 1996 54. Ibrahim S, Mojiminiyi S, Barron JL: Pyridinium crosslinks in patients on haemodialysis and continuous ambulatory peritoneal dialysis. Nephrol Dial Transplant 10: 2290-2294, 1995 55. Vaccaro F, Gioviale MC, Picone FP, Buscemi G, Romano M: Procollagen type I C-propeptide in kidney transplant recipients. Minerva Med 87:269-273, 1996 56. Fournier A, Oprisiu R, Said S, et al: Invasive versus non-invasive diagnosis of renal bone disease. Curr Opin Nephrol Hypertens 6:333-348, 1997 57. Couttenye MM, D’Haese PC, Van Hoof VO, et al: Low serum levels of alkaline phosphatase of bone origin: A good marker of adynamic bone disease in haemodialysis patients. Nephrol Dial Transplant 11:1065-1072, 1996

USE OF BONE MARKERS IN PREDIALYSIS PATIENTS

58. Martin KJ, Gonzalez EA: The evolution of assays for parathyroid hormone. Curr Opin Nephrol Hypertens 10:569574, 2001 59. Slatopolsky E, Finch J, Clay P, et al: A novel mechanism for skeletal resistance in uremia. Kidney Int 58:753761, 2000 60. Qi Q, Monier-Faugere MC, Geng Z, et al: Predictive value of serum parathyroid hormone levels for bone turnover in patients on chronic maintenance dialysis. Am J Kidney Dis 26:622-631, 1995 61. Coen G, Ballanti P, Bonucci E, et al: Renal osteodystrophy in predialysis and hemodialysis patients: Comparison of histologic patterns and diagnostic predictivity of intact PTH. Nephron 91:103-111, 2002

1007

62. Fournier A, Moriniere P, Cohen Solal ME, et al: Adynamic bone disease in uremia: May it be idiopathic? Is it an actual disease? Nephron 58:1-12, 1991 63. Fournier A, Moriniere P, Oprisiu R: Adynamic bone disease in uremia, in Drueke T, Salusky IB (eds): The Spectrum of Renal Osteodystrophy. New York, NY, Oxford University Press, 2001, pp 227-268 64. Fournier A, Yverneau PH, Hue P, et al: Adynamic bone disease in patients with uremia. Curr Opin Nephrol Hypertens 3:396-410, 1994 65. Ghazali A, Fardellone P, Pruna A, et al: Is low plasma 25-(OH)vitamin D a major risk factor for hyperparathyroidism and Looser’s zones independent of calcitriol? Kidney Int 55:2169-2177, 1999