Serum Osteocalcin Assay by the Elecsys 2010 is Robust and Suitable for Monitoring Bone Turnover

E.T. Leary, T.K. Aggoune, T.H. Carlson. Pacific Biometrics, Inc., Seattle, WA

ASBMR 26th Annual Meeting Oct 1-5, 2004 Seattle, WA
JBMR 2004; Vol 19:S240


Serum osteocalcin (OC) has demonstrated clinical utility as a biomarker of bone formation. However, OC is commonly believed to be unstable and to require special sample handling. OC fragments especially produced in some disease states may interfere with many existing assays. Therefore, we evaluated the long-term performance of the N-MID Osteocalcin assay on the automated Elecsys 2010 analyzer (Roche Diagnostics) and the stability of different sample types under a variety of conditions. The Elecsys OC is specific for intact OC (amino acids 1-49) and the main N-terminal fragment (amino acids 1-43) resulting from proteolytic cleavage after blood collection.

OC assay precision was evaluated by commercial and in-house quality control samples. To evaluate the stability of OC by Elecsys, matched serum, K3 EDTA plasma (EDTA) and Li heparin plasma (Heparin) were collected from volunteers. Sample stability was evaluated at room temperature, at 4°C for 7 days and at -70°C for 2 yr. To measure OC in a specimen matrix likely to accumulate fragment, matched serum, EDTA, and Heparin samples were collected from 44 subjects on kidney hemodialysis. OC analyzed within one month of collection and again after 4 yr of storage at -70°C were compared. To demonstrate that samples stored long-term were suitable for monitoring clinical response, OC was measured in 103 osteoporotic postmenopausal women on antiresorptive therapy, at baseline and after 12 weeks of treatment. Values determined immediately after the treatment period were compared with those after 3 yr of storage at – 70°C.

Within-run precision of OC was 1.4 -3.6 % CV at 72 – 7.9 ng/mL. Among-run precision was 3.7-5.9% CV at 21.3 – 110 ng/mL. OC concentrations in serum, EDTA and Heparin did not differ significantly. EDTA was a more stable matrix at room temperature than either serum or Heparin (3-4 days versus 24 hours or less). Change in OC at 24 mo at -70°C was less than 10% for all three sample types, with no trend. Assays of samples from subjects on renal dialysis with greatly elevated values (17- 1133 ng/mL, SD = 300 ng/mL) recovered well in dilution studies. OC values agreed after 4 yr of storage in serum, EDTA, and Heparin and correlated well with the initial results (all r > 0.99, slopes 0.92 – 1.01 and intercepts -7.2 – +1.3 ng/mL). Initial OC results for 103 osteoporotic women on antiresorptive therapy exhibited a mean decrease of 20.2% from baseline (baseline mean = 32.2 ng/ml) compared to -21.1 % (baseline mean = 30.4 ng/mL) measured in samples following 3 yr of storage.

We conclude the N-MID Osteocalcin assay on the Elecsys is stable and robust, and may be included in clinical trials with usual precautions.


Serum osteocalcin (OC) has demonstrated clinical utility as a biomarker of bone formation. However, the use of OC for assessing bone turnover and monitoring therapy has been complicated by analyte instability, both short-term and long-term, and assay non-specificity. The N-MID Osteocalcin (N-MID OC)* on the automated Elecsys analyzer (Roche Diagnostics, IN) is specific for both the intact OC (amino acids 1-49) and large N-terminal mid-region fragment (amino acids 1-43) that results rapidly from proteolytic cleavage after blood collection. The combination of assay automation and potentially more robust analyte and assay stability may increase the usefulness of OC in clinical studies and patient care. *Pending FDA Clearance

We evaluated the performance of N-MID OC with respect to:

  • Short-term and long-term consistency of the N-MID OC assay
  • Stability of OC in different sample types and in samples from different disease states (osteoporosis and renal dialysis) under various storage conditions
  • Suitability of samples stored for long-terms for clinical trials


Short-term and Long-term Assay Imprecision

  • Short-term (both within-run and among-run) and long-term (includes multiple lots of reagents and calibrators) imprecision were evaluated using (1) lyophilized control material (Roche Diagnostics, IN) and individual human sera or pooled sera; (2) frozen matched sets of serum, K3EDTA plasma (EDTA) and lithium heparin plasma (Heparin).
  • Serum, EDTA and Heparin samples from three subjects and two frozen serum pools, all prepared from freshly collected blood, were followed over 24 months. All samples were separated from cells within 2 hr of collection and aliquots frozen at -70°C within 8 hr. N-MID OC was analyzed fresh (within 8 hr) and after 1, 3, 6, 12, 18 and 24 months of storage at – 70°C.

Long-term OC Stability 

Long-term OC stability was assessed as follows:

  • Matched sets of serum, EDTA and Heparin samples described above were monitored for 24 months.
  • Matched sets of serum, EDTA and Heparin samples from 44 subjects on renal hemodialysis were analyzed within one month of collection and again after 4 years of storage at -70°C, using a second aliquot of sample.

Comparison of Clinical Response in Samples Stored Long-term

  • Sera from 103 osteoporotic postmenopausal subjects obtained at baseline and 3 months after initiation of an anti-resorptive therapy were analyzed at the end of the treatment period and again after 3 years of storage at -70°C, using a second sample aliquot.


Table 1.  Long-term Precision of N-MID OC in Lyophilized Quality Control Samples*

[table “table1” not found /]

*Multiple lots of reagent and calibrators were used over several 9-10 months periods

  • Table 1 illustrates representative among-run imprecision data on lyophilized controls using multiple lots of reagents and calibrators. The same level of performance was observed in serum pools during multiple time periods.
  • With-in run imprecision of N-MID OC was less than long-term imprecision at 3.6 – 1.4 % CV at 7.9 – 72 ng/mL (reported previously )

Figure 1. 2 Year Individual Sample Stability


Table 2. 2 Year Individual and Pooled Serum Sample Precision

[table “table2” not found /]

Figure 1 and Table 2

  • The subject means represent duplicate analyses at each time point (fresh, 1, 3, 6, 12, 18 and 24 mo) for each sample type. The order of the analysis was randomized.
  • Means of the frozen in-house serum pools represent duplicate analyses measured at the same time points as samples from the three subjects.
  • Four lots of reagents and calibrators were used during the 24 mo period.
  • Serum, EDTA and Heparin samples stored at – 70°C were stable for 24 mo. Except for the EDTA baseline value of Subject A (which appeared to be an outlier compared to all other time points and sample types), recoveries were within 10% of baseline values
  • OC concentrations and – 70°C stability in the three sample types were similar (EDTA values were not corrected for the dilution by the anticoagulant). EDTA was a more stable matrix at room temperature than either serum or Heparin; 3-4 days versus 24 hours or less (reported previously)
  • Total imprecision (CV) of all sample types over 24 months was 5.7%, similar to long-term imprecision in lyophilized controls. Lyophilized controls were prepared throughout the study and therefore did not include imprecision contributed from long-term sample instability. These data support the conclusion that N-MID OC is stable in all sample types for at least 24 mo at -70°C.

Figure 2.   4 Year Renal Sample Stability

  • Serum, EDTA and Heparin samples from 44 renal dialysis subjects were analyzed within 1 month of collection and again after storage for 4 years at – 70°C in batch mode. The correlation of serum, EDTA and Heparin (17 – 1133 ng/mL) gave r > 0.99, slopes from 0.92 to 1.01 and intercepts from – 7.2 to +1.3. (The highest 4 values were excluded in Figure 2. The regressions with or without the high values were comparable). Serum is shown in Figure 2.
  • There was no statistical difference among the sample types in recovery compared to the original values. However, all sample types indicated a negative bias compared to the original values (P <0.005). Serum had the largest negative bias (mean % bias = -12.4%).
  • The bias between the 4 yr and original values may be because of the instability of the renal sample matrices; although differences in the N-MID OC assay during the two measurement periods can not be excluded. Quality control sample data covering the entire 4-year period was unavailable.

Figure 3.
OC Response to Anti-Resorptive Therapy: Comparison of Values Measured at End-of- Treatment Period and in 3-Yr Old Samples

  • OC response to an anti-resorptive therapy in 103 postmenopausal osteoporotic women was measured at the end of the 3-month treatment period and again in samples stored at – 70°C for 3 years.
  • Regression of the 3-year and original samples was: y = 0.964x – 0.699 R2 = 0.924 N = 103
  • Initial response was -20.2% (from baseline mean of 32.2 ng/mL) compared to -21.1% (from baseline mean of 30.4 ng/mL). The response was not statistically different.


  • The Elecsys N-MID OC assay is robust and precise, both in the short- and long-term
  • N-MID OC may be measured in serum, K3 EDTA plasma and Li Heparin plasma
  • N-MID OC is stable at – 70°C in all three sample types for at least 2 yr
  • In samples from hemodialysis patients, where accumulation of OC fragments is expected, comparison of original OC values and values in samples stored at – 70°C for 4 yr was excellent (r> 0.99 for all three sample types), although a small negative bias was observed in the 4-yr old samples.
  • OC response to an anti-resorptive agent assessed in 3 year-old samples is statistically the same as response in the original samples.
  • Samples stored long-term (at least 3 yr) are suitable for clinical trials. However, to minimize all potential variation, including changes in samples because of improper storage and changes in assay performance, samples from all visits of each subject should be analyzed together in a batch. This is especially important when small clinical response is expected.