Adjustment of ionized calcium concentration for serum pH is not a valid marker of calcium homeostasis: implications for identifying individuals at risk of calcium metabolic disorders

Abstract

Background

Ionized calcium (iCa) is the biologically active form of this micronutrient. Serum determination of iCa is measured via ion-electrode potentiometry (IEP) and reporting iCa relative to pH 7.4 is normally utilized to avoid the potential confounding effects of ex vivo changes to serum pH. Adjustment of iCa for pH has not been adequately justified.

Methods

In this study, utilizing carefully standardized protocols for blood collection, the preparation of serum and controlling time of collection-to-analysis, we determined serum iCa and pH utilizing an IEP-analyser hosted at an accredited diagnostic laboratory.

Results

Regression analysis of unadjusted-iCa (iCa_raw) concentration versus pH was described by linear regression and accounted for 37% of serum iCa_raw variability. iCa_raw was then expressed at pH 7.4 by either adjusting iCa_raw based on the linear regression equation describing the association of iCa with serum pH (iCa_regr) or using IEP coded published normative equations (iCa_pub). iCa_regr was comparable to iCa_raw, indicating that blood collection and processing methodologies were sound. However, iCa_pub yielded values that were significantly lower than iCa_raw. iCa_pub did not identify 15% subjects who had greater than desirable serum concentration of iCa based on iCa_raw. Sixty percent of subjects with low levels of iCa_raw were also not detected by iCa_pub. Determination of the kappa value measure of agreement for iCa_raw versus iCa_pub showed relatively poor concordance (κ = 0.42).

Conclusions

With simple protocols that avoid sampling artefacts, expressing iCa_raw is likely to be a more valid and physiologically relevant marker of calcium homeostasis than is iCa_pub.

Introduction

Calcium has multiple biological effects influencing cell function and systems physiology.^1,2 Plasma calcium concentration is tightly regulated and in healthy subjects, is kept within a range of between 2.2 and 2.6 mmol/L. Half of the total plasma calcium is bound to proteins or anions. Ionized or ‘diffusable’ calcium (iCa) is the biologically active fraction.^1–3 The plasma concentration of iCa in healthy subjects is ordinarily maintained between 1.12 and 1.30 mmol/L. However, disruptions to this homeostatic system lead to disorders of calcium metabolism that may precipitate, or exaggerate progression of several chronic disease conditions.^1–5

Primary hyperparathyroidism (h-PTH) is the third most commonly reported endocrine disorder and is characterized by frank clinical features including markedly elevated serum iCa.⁶ However, some h-PTH subjects are reported with normal or mildly elevated serum calcium.⁷ Elevated PTH in the absence of hypercalcaemia may be a consequence of secondary mediators such as vitamin D deficiency, insufficient dietary intake of calcium or hypercalciuria.⁷ However, when secondary causes of h-PTH have been excluded, serum total calcium may still paradoxically remain within normal limits.^7,8 While it is presently unclear whether there is an obligatory increase in total serum calcium in response to PTH hypersecretion, a greater proportion of serum calcium nonetheless tends to be ionized, emphasizing the importance of regulating the concentration of the biologically active form.

Chronically low levels of serum total calcium are also a risk factor for chronic disorders including osteoporosis, nephrolithiasis, cardiovascular and neurodegenerative conditions.⁹ Hypocalcaemia is commonly a consequence of inadequate dietary intake, or insufficient vitamin D, which stimulates PTH secretion to achieve normocalcaemia through calcium reabsorption via kidney, bone resorption and calcium absorption via the bowel.^3,6

The physiological relevance of determining iCa was established decades ago.¹⁰ However, until the advent of ion-selective electrode direct potentiometry (IEP), measures of serum iCa were not common because of technical difficulties in adjusting for confounders.¹¹ The advent of IEP analysers with contemporary measures of endocrine modulators of calcium metabolism and vitamin D homeostasis provide a platform for more detailed distinction of calcium metabolic disorders.¹²

The measure of iCa in serum can be significantly affected by pH due to the competition for protein binding sites between calcium and hydrogen ions.^11,13 Therefore, to avoid potential confounding effects due to pre-analytical collection and transport of blood specimens, iCa is normally referenced relative to a serum pH of 7.4. The iCa value is typically adjusted based on published algorithms which describe the correlation of iCa and serum pH.^14,15

The validity and clinical relevance of reporting iCa adjusted to pH 7.4 has not been justified. In controlled animal models, Wolfe and Weir¹⁶ showed significant intergenerational strain differences in mean serum pH, suggesting that variability in serum pH is a naturally occurring phenomenon. Several studies including that of Barth et al.¹⁷ also reported significant non-linear and regression differences between laboratories when adjusting total serum calcium for albumin concentration according to published normative data. Moreover, others have reported that up to one-third of subjects measured via IEP-based iCa analysis may have venous-serum samples that fall outside the pH reference limits utilized in published algorithms to adjust iCa relative to a serum pH of 7.4.¹⁸

The ABL800 FLEX auto-analyser produced by Radiometer is a reputable product used by many accredited diagnostic laboratories. When calculating iCa, the ABL800 FLEX auto-analyser utilizes a linear regression correction for pH, claimed to be valid for pH values between 7.2 and 7.6.^14,19 In this study, we determined the putative validity of the calculated corrected value for iCa generated by the ABL800 FLEX auto-analyser (iCa_pub). The findings suggest that a linear adjustment of iCa relative to a pH of 7.4 may result in an under-estimation of subjects with hyper- or hypo-calcaemia. Uncertainty in the validity of expressing iCa adjusted for pH may contribute to paradoxical observations of calcium homeostasis in subjects with parathyroid hormone disorders and our understanding of the role of iCa in physiological function and disease risk.

Methods

Subjects

A total of 285 otherwise healthy participants aged between 18 and 80 y of age were recruited for the study approved by the Curtin University Human Research Ethics Committee (approval number HR97/2011). Potential participants completed a medical history and medications questionnaire and were interviewed to confirm the information provided. Subjects who had major surgery or a clinical event ‘in the last six months’; haemophilia; cancer/chemotherapy or with HIV were not included in the study.

Blood sampling

Participants arrived for provision of fasting blood samples between 8:00 and 10:00 in collection rooms adjacent to the accredited diagnostic centre following an overnight fast (minimum 8 hours). Participants were asked to avoid prolonged strenuous exercise and to limit alcohol intake to two standard drinks 24 h prior blood sampling.

Fasting blood samples were collected and handled under standardized preanalytical conditions to prevent interference of sampling artefacts that may affect pH and iCa values.^20,21 A venous sample was drawn from each subject after being seated for 10 min into serum separator Vacutainer™ tubes (Becton Dickinson, Franklin Lakes, NJ, USA). All specimens were collected under consistent conditions where tourniquet use was avoided where possible, care was taken to avoid muscular contraction in the limb and blood tubes were filled to capacity without the ‘ingress’ of air into the tube at the end of sampling. In cases where a tourniquet was used, a discard tube was drawn before the sample after release for 30 s prior sampling. Samples were allowed to clot for 30 min and immediately centrifuged at 3800 rpm for eight minutes at room temperature and thereafter analysed on the ABL800 FLEX series auto-analyser as described by the manufacturer. The average total time of sampling-to-analysis was 60 min. Samples exceeding two hours since collection were excluded.

iCa determination

The analyser algorithm for pH adjustment to 7.4 programmed into the ABL800 FLEX auto-analyser is based on algorithm provided by Fogh-Andersen et al.¹⁵ and commonly used in many IEP devices. The equation is: Corrected iCa²⁺ (pH 7.4) = Measured iCa²⁺ [1–0.53 × (7.40–measured pH)]. Note that this equation is only applicable to samples with pH values within 7.2–7.6, otherwise the conversion of the measured iCa concentration to pH 7.4 is considered invalid. Samples outside the indicated reference range were not included in the analyses presented.

Statistical analysis

Regression analysis was performed using unadjusted-iCa (iCa_raw) as the dependent variable and pH as the independent variable. In addition to the linear effect of pH, the deviations from the linear were assessed by adding in the square of pH into the regression model. The linear model was considered to be an adequate fit if the deviations or the quadratic effect was not significant (P > 0.05). The predicted values from the regression model were determined (iCa_regr). Adjusted R² was calculated to represent the percentage of variance in iCa_raw that was accounted by the independent variable pH.

Comparisons between the iCa_raw, iCa_regr and iCa_pub were performed using paired t-tests, since each subject had all three measurements. Cohen's kappa measure of agreement between the two measures iCa_raw and iCa_pub was determined. The equation for κ is $κ = \frac{⪻ (a) - ⪻ (e)}{1 - ⪻ (e)}$ : where Pr(a) is the relative observed agreement among measures, and Pr(e) is the hypothetical probability of chance agreement. If the measures are in complete agreement, then κ = 1. If there is no agreement among the measures other than what would be expected by chance (as defined by Pr(e)); κ = 0.

A frequency distribution of iCa_raw and iCa_pub were each expressed into three categories: below 1.20 mmol/L (Hypo-iCa); within the desirable reference range of 1.20 mmol/L ≤ iCa ≤ 1.29 mmol/L (Normal-iCa); or with serum iCa ≥ 1.30 mmol/L, were compared. The three reference ranges indicated were selected for consistency with other studies in the absence of consensus-based values.^6,22 P-values less than 5% were considered as statistically significant. All analyses were conducted using SPSS version 10.

Results

The cohort of 285 subjects investigated had a mean serum pH of 7.367, but despite strict blood collection protocols to avoid sample collection artefacts, the pH range was nonetheless significant (7.08–7.75; Figure 1). The association of unadjusted serum ionized calcium (iCa_raw) versus pH of serum measured by the ABL800 FLEX auto-analyser is depicted in Figure 1. The association of iCa_raw with pH was significant and best described by linear regression analysis. However, it was notable that serum pH only described 37% of subject iCa_raw variability.

Figure 1

Correlation of unadjusted ionized calcium (iCa_raw) and serum pH in 285 otherwise healthy subjects. The regression equation describing the association of iCa_raw with serum pH in this cohort is indicated. The mean of iCa_pub was significantly different from iCa_raw at P < 0.0001

Table 1 lists the mean + SD, the median and range for iCa_raw; following adjustment based on the linear regression analysis depicted in Figure 1 (iCa_regr); or based on the published normative corrections utilized in the adjustment settings of the ABL800 FLEX auto-analyser (iCa_pub). The mean and median of iCa_raw versus iCa_regr were not significantly different for this cohort. However, in contrast the mean and median of iCa adjusted to pH 7.4 according to published normative equations was significantly lower than the unadjusted value (iCa_raw), or iCa_regr. Figure 2 depicts the subject frequency distribution of the delta difference between iCa_pub versus iCa_raw. Paired analysis found that iCa_pub was significantly different from iCa_raw by approximately −0.22 mmol/L. In contrast, the frequency distribution of the delta difference between iCa_regr versus iCa_raw shown in Figure 2 frame B, was not significantly different (0.00018 mmol/L).

Figure 2

Subject frequency plot of the delta between the alternate measures of ionized calcium (iCa). The top frame depicts a subject frequency plot of the delta between the alternate measures of iCa. iCa_raw adjusted to pH 7.4 based on published normative corrections (iCa_pub), or the equation describing the association of iCa_raw versus serum pH (shown in Figure 1) (iCa_regr) are indicated. The table depicts the mean delta difference and significance versus iCa_raw

Table 1

Mean and standard deviation for measures of ionized calcium (iCa)

	pH	iCa_raw (mmol/L)	iCa_regr(mmol/L)	iCa_pub(mmol/L)
Mean ± SD	7.367	1.278 ± 0.045	1.278 ± 0.036	1.255 ± 0.036*
Minimum	7.080	1.130	1.162	1.136
Median	7.360	1.280	1.275	1.253
Maximum	7.750	1.420	1.437	1.412

Mean and standard deviation, median and range for measures of ionized calcium (iCa_raw) adjusted to a serum pH value of 7.4 by published normative algorithms (iCa_pub), or utilizing the regression equation which described the association of iCa_raw with pH (shown in Figure 1) (iCa_regr). Asterix (*) indicates that the mean iCa_pub was significantly different from the mean of iCa_raw or iCa_regr at P < 0.0001

Consideration of iCa adjusted to pH 7.4 is illustrated in Figure 3 as a scatter plot relative to individuals iCa_raw value. The data demonstrated that irrespective of an individual's serum pH, iCa_pub consistently underestimates the serum concentration of iCa.

Figure 3

Scatter plot depicting the difference in iCa_pub and iCa_regr relative to iCa_raw. Scatter plot for subjects depicting the difference in ionized calcium adjusted to a serum pH value of 7.4 by published normative algorithms (iCa_pub), or utilizing the regression equation which described the association of iCa_raw with pH (shown in Figure 1) (iCa_regr), relative to the unadjusted value (iCa_raw)

Table 2 shows the distribution of subjects with iCa_raw versus iCa_pub stratified according to a serum concentration of iCa (<1.20 mmol/L); a serum concentration of iCa (≥1.20 and ≤1.29 mmol/L); or above the upper reference limit of ≥1.30 mmol/L. Of the 285 subjects studied, 70 (24.6%) were identified as having serum iCa ≥1.30 mmol/L based on iCa_raw. In contrast, iCa_pub identified 26 individuals (9.1% of subjects) with iCa ≥ 1.30 mmol/L. Conversely, six of 10 subjects with iCa below the lower desirable concentration of 1.20 mmol/L based on iCa_raw were classified as normocalcaemic when reported according to published normative corrections (iCa_pub). Determination of Cohen's measure of agreement between iCa_raw and iCa_pub showed relatively poor concordance between two measures iCa_raw and iCa_pub (κ measure of agreement = 0.42).

Table 2

Frequency of subjects identified as having serum iCa_raw and iCa_pub below (Hypo), within (Normal); or above (Hyper) the normal reference range

	iCa_pub			Total
	Hypo-iCa	Normal-iCa	Hyper-iCa	Total
iCa_raw
Hypo-iCa	4 (1.4%)	6 (2.1%)	0 (0%)	10 (3.5%)
Normal-iCa	9 (3.2%)	196 (68.8%)	0 (0%)	205 (71.9%)
Hyper-iCa	0 (0%)	44 (15.4%)	26 (9.1%)	70 (24.6%)
Total	13 (4.6%)	246 (86.3%)	26 (9.1%)	285 (100%)

Table 2 depicts the frequency of subjects identified as having serum iCa_raw or iCa_pub below 1.20 mmol/L (Hypo-iCa); within the reference range of 1.20 mmol/L ≤ iCa ≤ 1.29 mmol/L (Normal-iCa); or with serum iCa ≥ 1.30 mmol/L. Shaded cells show where there is agreement between the two measures. The highlighted cell shows significant divergence between iCa_raw and iCa_pub and identifies subjects in this cohort with notionally elevated levels of iCa

Discussion

This study determined the best-fit association of uncorrected iCa versus serum pH, generated by a commonly used IEP-analyser, the ABL800 FLEX auto-analyser by Radiometer. The association of iCa versus the independent variable (pH) was significant but only accounted for approximately 37% of the measures recorded in a cohort of 285 subjects. The findings reiterate that serum iCa is significantly regulated by factors other than serum pH.

In this study, the cohort had an unsurprising mean serum pH of 7.367, but despite controlling for potential sample collection confounders, variability around the mean serum pH nonetheless was as high as 5.2% (pH 7.08–7.75). Serum samples with 7.1 > pH > 7.6 were not included in the analyses; however, if variability in serum pH does indeed represent natural biological variation between individuals, then in absolute terms this would reflect in this cohort potential differences of iCa_raw of 0.51 mmol/L (estimated on the linear regression equation generated). The findings suggest that if there is confidence in the integrity of blood sample collection and analysis, then adjustment of measured iCa_raw relative to an ex vivo pH cannot be justified. Indeed, in a physiological context it is the absolute tissue exposure to iCa_raw that needs to be considered.

Stratification of subjects based on iCa_raw identified that approximately 25% of subjects (70 of 285 individuals) had a serum iCa above the upper reference limit of 1.29 mmol/L, whereas the adjusted value based on published algorithms (iCa_pub) identified just 9.1%. All of the latter individuals were confirmed as being in the hyper-iCa_raw category. Although fewer in number than the hyper-iCa_raw group, six out of 10 individuals identified by iCa_raw as having lower than desirable levels of iCa, were within the normal range of 1.20–1.29 mmol/L according to the iCa_pub adjustment. Clearly, utilizing iCa_raw rather than iCa_pub may alter clinical management of individuals. Moreover, utilizing a robust measure of iCa_raw may alter our understanding of how iCa influences physiological systems and disease risk.

Sustained hyper- or hypocalcaemic disorders have significant physiological effects that can contribute to onset and progression for a range of chronic disorders.^6,9 Confidence in the validity of iCa measures are therefore pivotal to optimize clinical management. Our data suggest that adoption of published generalized linear models to adjust the raw measured value of iCa is not appropriate. Rather, standardizing blood sampling and management protocols will provide confidence that IEP measures of iCa_raw are biologically relevant. If an adjusted value of iCa relative to serum pH 7.4 is to be considered, then the data suggest that adjustment should be at the very least reiterated per IEP-unit to ensure comparability of research findings. In this study, the linear regression equation that described iCa_raw versus pH (iCa_regr) was significantly different from the commonly utilized algorithm (iCa_pub). Nowadays, it would be relatively simple to analyse the association of iCa_raw with sample pH and provide an iCa_regr that is specific for the model of IEP used and site of testing.

The findings of this study do not support the utilization of iCa adjusted to a mean serum pH of 7.4. Moreover, utilization of published normative equations describing the correlation of serum iCa with serum pH is a historic practice that may lead to significant confounders when considering research reports. Thode et al.¹⁴ suggested otherwise and found consistent agreement between iCa adjusted based on published algorithms and actual values with relatively careful sampling techniques. While a comparison of iCa determined based on regression analysis versus sample pH was not reported by Thode et al., the difference in findings emphasizes the uncertainty of normalization between alternate IEP devices, utilizing the same algorithm. Limitations of this study include only a single measure per subject. However, the coefficient of variation for single measures was found to be <2% per sample (data not shown). The Australian population cohort was multi-ethnic and this may have contributed to the variability in serum pH reported.

Footnotes

DECLARATIONS

Competing interests: None to declare.

Funding: This research was funded by the National Health and Medical Research Council of Australia.

Ethical approval: The ethics committee of Curtin University approved this study (REC number: HR97/2011).

Guarantor: JM.

Contributorship: VL and JM researched literature and conceived the study. VL and JM were involved in protocol development, gaining ethical approval and patient recruitment. VL and JM wrote the first draft of the manuscript. VL, JM and SD were responsible for data and statistical analysis. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

Acknowledgements: The authors would like to thank Linda Gregory and phlebotomists from PathWest of Royal Perth Hospital for their support in facilitating this study.

References

Mundy

Guise

. Hormonal control of calcium homeostasis. Clin Chem 1999;45:1347–52

Costanzo

. Regulation of calcium and phosphate homeostasis. Adv Physiol Educ 1998;20:206–16

Baird

. Ionized calcium. Clin Chim Acta 2011;412:696–701

Peacock

. Calcium metabolism in health and disease. Clin J Am Soc 2010;5:S23–30

Fujita

. Calcium paradox: consequences of calcium deficiency manifested by a wide variety of diseases. J Bone Miner Metab 2000;18:234–6

Fraser

. Hyperparathyroidism. Lancet 2009;374:145–58

Bilezikian

Silverberg

. Normocalcemic primary hyperparathyroidism. Arq Bras Endocrinol Metabol 2010;54:106–9

Ladenson

Lewis

McDonald

Slatopolsky

Boyd

. Relationship of free and total calcium in hypercalcemic conditions. J Clin Endocrinol Metab 1979;48:393–7

Fujita

Palmieri

. Calcium paradox disease: calcium deficiency prompting secondary hyperparathyroidism and cellular calcium overload. J Bone Miner Metab 2000;18:109–25

10.

McLean

Hastings

. The state of calcium in the fluids of the body. J Biol Chem 1935;108:285–322

11.

Boink

Buckley

Christiansen

IFCC recommendation on sampling, transport and storage for the determination of the concentration of ionized calcium in whole blood, plasma and serum. J Automat Chem 1991;13:235–9

12.

O'Neill

Gordon

Guo

Zhu

McCudden

. Multivariate analysis of clinical, demographic, and laboratory data for classification of disorders of calcium homeostasis. Am J Clin Pathol 2011;135:100–7

13.

Wang

McDonnell

Sedor

Toffaletti

. pH effects on measurements of ionized calcium and ionized magnesium in blood. Arch Pathol Lab Med 2002;126:947–50

14.

Thode

Holmegaard

Transbol

Fogh-Andersen

Siggaard-Andersen

. Adjusted ionized calcium (at pH 7.4) and actual ionized calcium (at actual pH) in capillary blood compared for clinical evaluation of patients with disorders of calcium metabolism. Clin Chem 1990;36:541–4

15.

Fogh-Andersen

. Ionized calcium analyzer with built-in pH correction. Clin Chem 1981;27:1264–7

16.

Wolfe

Weir

. High and low blood-pH selected lines of mice. The fate of pH and sex ratio following relaxed selection with intensive breeding. J Hered 1972;63:109–12

17.

Barth

Fiddy

Payne

. Adjustment of serum total calcium for albumin concentration: effects of non-linearity and of regression differences between laboratories. Ann Clin Biochem 1996;33:55–8

18.

Bowers

Jr Brassard

Sena

. Measurement of ionized calcium in serum with ion-selective electrodes: a mature technology that can meet the daily service needs. Clin Chem 1986;32:1437–47

19.

Brauman

Delvigne

Deconinck

Willems

. Factors affecting the determination of ionized calcium in blood. Scand J Clin Lab Invest Suppl 1983;165:27–31

20.

Clase

Norman

Beecroft

Churchill

. Albumin-corrected calcium and ionized calcium in stable haemodialysis patients. Nephrol Dial Transplant 2000;15:1841–6

21.

Arkin

Bessman

Calam

Procedures for the Collection of Diagnostic Blood Specimens by Venipuncture; Approved Standard. Vol. 23, 5th edn. Clinical and Laboratory Standards Institute, 2003:1–52

22.

Tfelt-Hansen

Brown

. The calcium-sensing receptor in normal physiology and pathophysiology. Crit Rev Clin Lab Sci 2005;42:35–70