Calcitonin is a peptide hormone comprising of 32 amino acids with a molecular weight of 3,418 D and is mainly secreted by the C cells of the thyroid gland. In normal endocrine processes, calcitonin inhibits calcium reabsorption from the bones, thereby main6338taining bone strength [1, 2]. The plasma calcitonin concentration is relatively high in infants, and decreases rapidly until adolescence, leveling off during adulthood [3]. On average, its concentration is higher in men than in women [4, 5].

Calcitonin concentration abnormally increases in medullary thyroid cancer [6, 7]. A cut-off concentration of 10 pg/mL is most commonly used to differentiate between C-cell hyperactivity diseases (including medullary thyroid cancer) and other thyroid diseases [8-10]. In addition, a follow-up of plasma calcitonin concentration after treatment for medullary thyroid cancer is useful for monitoring recurrence and predicting patient prognosis [11, 12]. The diagnostic utility of plasma calcitonin concentration further increases when reviewed in combination with the carcinoembryonic antigen (CEA) concentration, RET protooncogene mutations, and needle aspiration biopsy findings [13, 14]. The measurement of serum calcitonin level is markedly significant as it serves as a quick and cost-effective method for diagnosing medullary thyroid cancer [6, 15-19]. Serum calcitonin concentration testing, like most tests conducted in modern clinical laboratories, is becoming increasingly advanced and automated [5, 9, 20-22].

However, standardization and harmonization of calcitonin measurements using various assays have not been achieved to date. Only few studies have evaluated the equivalence of different calcitonin assays [5, 23]. In this context, we compared the serum calcitonin levels measured using three different automated immunochemical analyzers, namely the LIAISON (Diasorin, Saluggia, Italy), Atellica IM 1600 (Siemens Healthcare Diagnostics Inc., New York, USA), and Cobas e801 (Roche Diagnostics GmbH, Mannheim, Germany), to evaluate the equivalence of the test results and determine whether the current clinical calcitonin cut-offs apply to the available calcitonin immunoassays used in clinical laboratories.

Previously, the serum calcitonin levels were primarily measured using a radioimmunoassay. However, the differences in assay specificity and sensitivity, matrix-related and nonspecific serum effects, and heterogeneity in the circulating calcitonin levels contributed to contradicting results and discrepancies in the calcitonin concentrations measured using different assays. In modern clinical laboratories, several highly sensitive two-site immunometric methods, including chemiluminescence immunoassay (CLIA) are used. However, the methods for measuring calcitonin levels are not currently fully standardized [25]. Therefore, we conducted direct comparisons of the currently available instruments used for calcitonin measurement to empirically demonstrate the level of harmonization in clinical laboratories.

Initially, when all data points were included in the comparison, both the Cobas e801 and Atellica IM 1600 analyzers showed limited positive bias compared with the LIAISON analyzer, with acceptable correlation coef-cients for both comparisons. The comparison of the Atellica system with the Cobas indicated better correlation and non-significant bias. Similarly, the bias noted upon comparing the Cobas and the Atellica with the LIAISON analyzer led to clinically significant differences in the interpretation of the calcitonin levels. Based on the clinical cut-off of 10 pg/mL, of the 39 samples determined positive by LIAISON analyzer, 9 samples were determined negative by Cobas analyzer and 7 samples were determined negative by the Atellica analyzer, thereby revealing substantial discordance with the results from LIAISON system. After excluding the data points for concentrations >20 pg/mL, the discordance between the measured values around 10 pg/mL was elucidated. We observed some aberrant high concentrations when using the LIAISON analyzer compared to the other two analyzers. Comparison of the Atellica to the Cobas analyzer showed relatively high correlation, although some positive bias was observed. Based on the clinical cut-off of 10 pg/mL, the calcitonin levels of 39 of the 104 samples (37.5%) were above the threshold level when using the LIAISON analyzer, while the calcitonin levels of 9 of the 39 samples were below the threshold level when using the Cobas analyzer, revealing significant discordance with the results obtained using the LIAISON analyzer. Similarly, the calcitonin levels of 7 of the 39 samples were below the threshold level when using the Atellica analyzer, which also showed substantial discordance with the results obtained using the LIAISON system. After excluding the data points for concentrations >20 pg/mL (to elucidate the discordance between the values measured around the 10 pg/mL level), the results obtained using the Cobas e801 and Atellica IM 1600 analyzers showed weak correlations with those obtained using the LIAISON analyzer. The discordance in the calcitonin concentration measured near 10 pg/mL seemed to be the major contributor to this weak correlation. We observed some aberrantly high concentrations using the LIAISON analyzer compared to concentrations obtained for identical samples analyzed using the other analyzers. Comparison of the Atellica to the Cobas analyzer revealed higher correlation, although some positive bias was observed.

The Royal College of Pathologists of Australasia suggests an acceptable difference ±2 pg/mL calcitonin level for concentrations that are the same as, or less than 20 pg/mL and ±10% for concentrations above 20 pg/mL [26]. The allowable difference band was drawn on relative bias plots in Figs. 1 and 2. When all the data points were analyzed, Cobas e801 and Atellica IM1600 showed some positive bias over LIAISON between 20 pg/mL and 600 pg/mL, while for data points below 20 pg/mL, the two analyzers assigned a profoundly lower value for some samples than the LIAISON. The relative difference for these samples, even exceeding -100% sometimes, may have critically attributed to the weak correlation between LIAISON and the other two analyzers. For these samples, factors involved in pre-analytical or analytical phases could be considered as causes of the discrepancy. When the samples were being measured, there were no cases of recognized hemolysis, icterus, or lipidemia. Moreover, the principles of the three assays invariably resort to mouse anti-hCT (human calcitonin) antibody, albeit minor variations exist among reagent compositions. Characteristic reagent components of the three assays are indicated in Table 1. LIAISON assay included a small amount of polyclonal mouse IgG in its assay buffer to prevent the interference from heterophilic antibody. However, the Cobas and Atellica assays used a biotinylated antibody. This may result in interference from any intake of a biotin substance. All these minor differences among the assays might attribute to the discordance to some extent.

Due to the lack of a reference method for measuring the calcitonin concentration, we cannot conclude that results obtained using one of these analyzers may be closer to the actual values than those obtained using the other analyzers. However, because there were some aberrant high concentrations around 10 pg/mL when using the LIAISON analyzer in our study, we attempted to verify whether false-positive results using this analyzer were reported previously. Our results may be more informative when reviewed in combination with those reported in previous studies [5, 23]. Bieglmayer et al. [23] compared four calcitonin assays: the first version of a CLIA from LIAISON (Stillwater, MN, USA), a CLIA from Nichols Institute Diagnostics (San Juan Capistrano, CA, USA), a CLIA from Diagnostic Products Corporation (Los Angeles, CA, USA), and an immunoradiometric assay from Scantibodies (Santee, CA, USA). Comparisons of these assays revealed implausible detection of calcitonin levels in thyroidectomized patients using the LIAISON system and substantial discordance in the measured concentrations among all assays assessed, especially at low calcitonin levels. Cavalier et al. [5] compared the LIAISON Calcitonin II-gen to the cisbio h-CT (CIS bio international, Gif-sur-Yvette, France) to validate the analytical performance of the second-generation LIAISON calcitonin assay. The measurement of serum calcitonin concentrations in 250 consecutive patients showed no significant systematic bias or significant differences, as noted by the Wilcoxon test. They concluded that the specificity of the new version of the LIAISON analyzer was markedly improved, although 1 of the 14 thyroidectomized patients showed detectable calcitonin concentrations.

Currently, there is no consensus on the reference method for calcitonin measurement. To standardize the calcitonin assay, the World Health Organization (WHO) established an international reference preparation, IRR 70/234, which was replaced by the IS 89/620 standard. All three assays addressed in this study are traceable to this IS 89/620 standard. However, this standard is not commutable (that is, not proven to be equivalent for representative authentic clinical samples) for any specific assay [27]. Therefore, the possibility of significant inter-assay differences could not be disregarded, although most assays are currently calibrated to a common international reference standard. In addition, it remains uncertain whether the current clinical calcitonin cut-off applies to all available calcitonin immunoassays or if monitoring using different methods is clinically reliable when patients are transferred between healthcare facilities. There is a scarcity of studies on the concordance among various calcitonin test methods. The results of the present study indicated the differences in the measured calcitonin values for each of the three analyzers, especially at low values near the clinical cut-off. These differences are probably attributable to the use of different calibrators among the manufacturers, which is an inevitable challenge as commutable reference material is not available. Therefore, joint efforts by the expert scientific community, manufacturers, and regulatory organizations are warranted to establish a reference method and commutable reference material for improved standardization and harmonization of the calcitonin assays.

Discordant results for calcitonin levels can lead to misclassification of patients and unnecessary additional examinations or delayed diagnosis of serious disease. Due to the high sensitivity and specificity of calcitonin to medullary thyroid cancer, routine calcitonin measurements are performed in many countries, including major parts of Europe [16]. In patients with elevated basal serum calcitonin levels >10 pg/mL, pentagastrin stimulation testing is performed to distinguish among the false-positive findings and to more accurately estimate the possibility of medullary thyroid cancer [28]. Stimulated calcitonin concentrations >200 pg/mL suggest medullary thyroid cancer, which requires thyroidectomy. With respect to the impact of early diagnosis, a study on the screening of 10,864 patients reported elevated calcitonin levels in 44 patients, all confirmed to have medullary thyroid cancer based on pentagastrin-stimulated calcitonin tests. Among these patients, 59% achieved complete remission compared to 2.7% of patients with medullary thyroid cancer before the implementation of calcitonin screening [17]. A common cut-off of 10 pg/mL is conventionally used for the initial screening of C-cell hyperplastic conditions [8-10]. However, additional studies are needed to achieve a sophisticated and evidence-based consensus on new criteria considering the inter-method variability. Sex-specific cut-offs should also be considered as the reference intervals of normal calcitonin levels for women and men significantly differ [4, 5, 29]. In the current situation where there is an apparent lack of harmonization among the calcitonin measurement methods, the same instrument should be used to obtain the follow-up calcitonin concentration for a patient. Moreover, it is recommended that each laboratory use method-specific reference interval rather than universal cut-off for clinical decision.

This study had several limitations. First, we compared only three assays that use the CLIA method. Thus, further studies evaluating the equivalence among more assays using different test principles are warranted. Second, as the clinical information of patients was not available, we could not accurately identify the assay that provided clinically reliable results when there were discrepancies in the classification based on concentration.

The results of this study showed that, although three automated calcitonin immunoassays showed acceptable correlations for the full analytical measurement range, the correlation and agreement were limited at low concentration ranges, especially at the clinical decision threshold (10 pg/mL). Hence, continuous joint efforts by the expert scientific community, manufacturers, and regulatory organizations are imperative for improved standardization and harmonization of the calcitonin tests to avoid unnecessary and costly additional testing or absence of testing that may lead to a missed diagnosis of cancer. A single instrument should be consistently used when following-up a patient to avoid the effect of systemic bias among the instruments.