Large-scale comparison of two immunoassays for adrenocorticotropic hormone in human plasma

Abstract

Accurate measurement of plasma adrenocorticotropic hormone (ACTH) is critical in the diagnosis and management of conditions that affect pituitary or adrenal gland function. Herein, we assessed the agreement and the extent of discordance between immunoassays from Roche and Siemens in two large collections of specimens. In a collection of 602 de-identified specimens (n = 561 from patients and n = 41 from healthy volunteers), the ACTH results were highly correlated with a Spearman correlation coefficient of 0.87. For all specimen comparisons, 56 (9.3%; 95% confidence interval [CI], 7.2%–11.9%) had a discordance with > 50% difference in the ACTH levels (including 2 healthy donors). In Bland-Altman analysis, 37 (6.1%; 95%CI, 4.5%–8.4%) specimens fell outside the limits of agreement (lower limit: -82% in difference, and upper limit: 75%), the majority of which (33 [89.2%]) had higher ACTH results (including 2 healthy donors) using the Siemens assay than the Roche assay (p < 0.05). Similar results were observed in another collection of 100 de-identified patient specimens. HAMA/heterophile absorption experiments did not explain these phenomena. These observations demonstrate that a substantial fraction of patient specimens had significant differences in the ACTH results using the Roche and Siemens assays (p < 0.001), primarily due to higher levels from the Siemens assay. Therefore, abnormally high levels of ACTH should be interpreted with caution, particularly when the Siemens assay is used.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-26501-3.

Introduction

Adrenocorticotropic hormone (ACTH) measurements are a component of paradigms to evaluate adrenal function. In individuals with adrenal hyperfunction (Cushing’s syndrome), the measurement of ACTH provides an important clue as to the cause of this hyperfunction¹. In individuals with measurable ACTH levels (normal or high), an ACTH-dependent source is implicated: either Cushing’s disease (a pituitary source) or ectopic ACTH sources. By contrast, suppressed ACTH levels are indicative of adrenal steroid production that is independent of ACTH (i.e., an adrenal adenoma or carcinoma)¹. A plasma ACTH level below 10 pg/mL suggests an ACTH-independent cause of Cushing’s syndrome, while a level above 20 pg/mL suggests an ACTH-dependent cause². Further testing is needed to differentiate the two forms if the level is between 10 and 20 pg/mL. Thus, ACTH measurements may trigger additional testing and imaging.

Several FDA-cleared tests are available in the United States for measuring ACTH, including those produced by Roche Diagnostics and Siemens Healthineers. Published reports have implicated potentially misleading results from the Siemens ACTH assay in specific patients identified during the evaluation for Cushing’s syndrome and other conditions^3–7. In these reported instances, the Siemens assay yielded ACTH levels which suggested a pituitary or ectopic source. Repeat or parallel testing in these patients using the Roche assay yielded lower ACTH levels, implicating overactivity of the adrenal gland itself. Of note, other authors described a case in which the Roche assay provided ACTH results inconsistent with the clinical picture and which was clarified by ACTH testing using the Siemens assay⁸. The inconsistencies in ACTH test results have also been discussed more broadly in the context of aberrations occasionally encountered in immunoassays in general⁹.

A previous study has compared the Roche and Siemens ACTH assays in a small collection of samples¹⁰, but the frequency and basis of discordance between their results have not been broadly examined. The current study was undertaken to define frequencies of the discordance in two large collections of specimens and examine the contribution of HAMA/heterophile antibodies.

Results

ACTH test results in a collection of 602 specimens

We first compared ACTH test results using the Roche Cobas e602 and Siemens Immulite 2000 assays in a collection of 602 specimens, including 561 de-identified patient plasma specimen discards from daily clinical testing and 41 specimens from healthy donors. ACTH levels for these specimens ranged from 1.0 to 1,215 pg/mL using the Roche assay with median (interquartile range [IQR]) of 35 (17–64) and from 5.0 to 1,273 pg/mL using the Siemens assay with median (IQR) of 38 (16–64). The measurements using these assays were highly correlated with a Spearman correlation coefficient of 0.87 (Fig. 1). The measurements using the Roche assay increased by 0.76 pg/mL per 1 pg/mL increase using the Siemens assay according to the Deming regression analysis.

Fig. 1.

Dot plots comparing ACTH measurements using the Roche and Siemens assays. (a) All specimens tested. (b) Specimens with ACTH ≤ 200 pg/mL. The red dots denote specimens that had a discordance with > 50% difference in ACTH measurements. The dashed diagonal line has a slope of 1. ACTH, adrenocorticotropic hormone.

Among all 602 specimens, the distribution of the difference in ACTH measurements between the Roche and Siemens assays had a median (IQR) of 1.6 (-4.0–4.9) pg/mL. There were 229 (38%) specimens that had higher values using the Siemens assay than the Roche assay (group 1) and 373 (62%) specimens that had higher values using the Roche assay than the Siemens assay (group 2). The variability of the difference in ACTH measurements was higher in group 1 than in group 2—median (IQR) difference was 7.6 (2.5–29.0) pg/mL vs. 3.7 (2.0–7.4) pg/mL. The two groups significantly differed in terms of the relative difference in ACTH measurements—median (IQR) difference was 19% (7%–44%) in group 1 vs. 15% (7%–24%) in group 2 (p < 0.001).

For all specimen comparisons, 56 (9.3%; 95% confidence interval [CI], 7.2%–11.9%) had a discordance with > 50% difference in ACTH measurements (including 2 healthy donors). Of these discordant measurements, there were a total of 4 (1.1%) among the 373 where Roche assay reported higher values and 52 (22.7%) among the 229 where Siemens assay reported higher values (Fig. 1 and supplementary Fig. 1). Those with higher values using the Siemens assay than the Roche assay were more likely to be discordant than those with higher values using the Roche assay than the Siemens assay (odds ratio, 27.0; 95% CI, 9.7–104.3; p < 0.001). Notably, for many of these specimens with discordant measurements, their ACTH levels were in the normal range using one assay but were elevated using the other assay (Fig. 1), which could lead to different interpretations of the test results.

In Bland-Altman analysis evaluating agreement between the measurements, the mean difference was − 3%, and the lower and upper limits of agreement were at difference of -82% and 75%, respectively (Fig. 2). Five hundred sixty-five (93.9%) of the 602 samples were within the limits. Of the 37 (6.1%; 95%CI, 4.5%–8.4%) specimens that fell outside the limits, 33 specimens had significant higher ACTH results (including 2 healthy donors) using the Siemens assay than the Roche assay (p < 0.05) and 4 specimens had significant higher ACTH results (including no healthy donors) using the Roche assay than the Siemens assay (p < 0.05).

Fig. 2.

Bland-Altman plots comparing ACTH measurements using the Roche and Siemens assays. (a) All specimens tested. (b) Specimens with mean ACTH ≤ 200 pg/mL. The solid horizontal line marks the mean difference, and the dashed horizontal lines mark the lower and upper limits of agreement. The red dots denote specimens that fall outside of the lower and upper limits. ACTH, adrenocorticotropic hormone.

ACTH test results in a separate collection of 100 specimens

We next compared ACTH test results using the Roche and Siemens assays in a separate collection of 100 specimens. The measurements using these assays were also highly correlated with a Spearman correlation coefficient of 0.72 (supplementary Fig. 2). For all specimen comparisons, 36 (36%; 95% CI, 27.3%-45.8%) had a discordance with > 50% difference in ACTH measurements, all of which were higher using the Siemens assay than the Roche assay. In Bland-Altman analysis, the mean difference between the measurements was − 34%, and the lower and upper limits of agreement were at difference of -139% and 70%, respectively (supplementary Fig. 3). Three (3%) specimens had significantly higher ACTH results using the Siemens assay than the Roche assay (p < 0.05).

Effect of HAMA (human anti-mouse antibody) /heterophile antibodies on ACTH test results

Finally, we examined the possibility that the observed discordances were attributed to the interference of HAMA/heterophile antibodies. We assessed the effect of pre-treatment of specimens with HAMA/heterophile absorption tubes on assay results. Specimens that showed a difference between the Roche and Siemens assays greater than 30% or 10 pg/mL were subjected to HAMA/heterophile absorption. Of the 602 specimens in the initial sample collection, 74 were available for evaluating the effect of HAMA/heterophile antibodies.

Of the 26 specimens that were initially identified to be discordant, 9 specimens that had high ACTH measurements using the Siemens assay had more than 30% lower ACTH measurements by the same assay after the treatment of the HAMA/heterophile antibodies, whereas 5 other specimens that had high ACTH measurements using the Siemens assay had even higher ACTH measurements (> 30% increase) after the treatment of the HAMA/heterophile antibodies (Fig. 3A). ACTH measurements by the Roche assay remained largely unaffected by the treatment of the HAMA/heterophile antibodies (Fig. 3B).

Fig. 3.

The effect of pre-treatment of the 74 specimens with HAMA/heterophile absorption tubes on assay results. (a) ACTH measurements using the Siemens assay in specimens with and without pre-treatment with HAMA/heterophile absorption tubes. (b) ACTH measurements using the Roche assay in specimens with and without pre-treatment with HAMA/heterophile absorption tubes. (c) The difference in ACTH measurements, expressed as (ACTH_Roche ─ ACTH_Siemens)/[(ACTH_Roche + ACTH_Siemens)/2], was compared for specimens with and without HAMA/heterophile absorption treatment. ACTH, adrenocorticotropic hormone. HAMA, human anti-mouse antibody.

Among all 74 specimens, the overall difference in ACTH levels was not significant for those with vs. without HAMA/heterophile absorption (1.0 pg/mL, 95% CI, 0.90–1.10; p = 0.998). The agreement of the discordant and concordant ACTH results between the specimens with and without HAMA/heterophile absorption was 83.8% (62 of the 74 specimens) (Fig. 3C). In a conditional logistic regression model that assessed the HAMA effect on the ACTH discordance rate, the specimens with HAMA/heterophile absorption did not have a significantly higher discordance rate than those without (odds ratio, 1.84; 95% CI, 0.67–5.07; p = 0.24).

Discussion

When viewed in aggregate, the ACTH results using the Roche and Siemens assays in our study are highly correlated in both specimen collections. When focused on samples with substantial differences in results between the assays, although discordances were noted to occur between the two assays in both directions (i.e., Siemens high but Roche low or Siemens low but Roche high), discordances with Siemens high but Roche low were observed more frequently than the inverse (Siemens low but Roche high).

The ACTH assays from Roche and Siemens differ in methods and antibodies used. The primary difference lies in the specific epitopes of the 39-amino acid ACTH molecule that the antibodies are designed to target, with the Roche assay using two monoclonal antibodies specific for residues 9–12 and residues 36–39 while the Siemens assay using a murine monoclonal and rabbit polyclonal antibodies¹⁰. Antibodies can recognize the full-length ACTH molecule and smaller fragments, and ACTH precursors found in ectopic ACTH syndrome can affect test results¹¹. Since ACTH is an unstable peptide (particularly in blood), its degradation from inappropriate sample handling could affect test results although recent studies show that “ACTH in plasma remains viable for analysis under processing and storage conditions less stringent than commonly practiced”^12–14. Furthermore, the lack of a gold-standard international reference for ACTH may contribute to the variability between assays¹⁵. Indeed, prior to our study, a series of reports have identified instances where testing using the Siemens assay yielded results (normal or elevated ACTH levels) which suggested an ACTH-dependent process – findings that were not confirmed when testing was repeated using the Roche assay^3–7.

The variations seen between different ACTH immunoassays could also be due to analytical interference caused by substances present in patient plasma. In some studies, investigators have suggested that discrepant results might reflect the effect of HAMA antibodies that may be present in some specimens^5,8. In our study, analyses of HAMA/heterophile interference were performed using HAMA/heterophile absorption tube. While HAMA/heterophile antibodies were found in several samples run on the Immulite, their frequency was not significant in comparison to the overall Siemens/Roche discordance. This observation raised the possibility that some of the discordances observed might not be attributable to effects caused by the presence of HAMA/heterophile antibodies. However, for those specimens with discrepant test results and no effect of HAMA/heterophile absorption tube, other complementary analyses are needed to further assess potential involvement of heterophilic antibodies^6,7,16. Such analyses include polyethylene glycol (PEG) precipitation to remove interfering HAMA and other heterophile antibodies or serial dilutions of plasma where nonlinearity of test results would be expected in the presence of heterophilic antibodies.

A statistically different result does not necessarily affect the diagnosis and management of the patient’s condition (when, for example, two test results are both within or both out of reference ranges), but it might for some patients. Therefore, abnormally high levels of ACTH should be interpreted with caution, particularly when the Siemens assay is used. In a previous analysis¹⁰, for discordant results between the Roche and Siemens assays, the Roche results were more closely correlated with the concentration of intact, biologically active ACTH detected by liquid chromatography-tandem mass spectrometry.

Materials and methods

Study design

This is a comparative study of test results from two commercially available ACTH assays using deidentified medical discard plasma specimens from patient samples submitted for diagnostic testing, as well as samples from healthy volunteers. The ACTH assays were performed at Quest Diagnostics Nichols Institute laboratories at San Juan Capistrano, CA and Chantilly, VA. The study protocol involving remnant specimens was reviewed by the WCG Institutional Review Board, an independent ethical review board (www.wcgclinical.com), and the need for authorization was waived by the IRB as this study was deemed exempt based on federal regulation 45 CFR Parts 46, 160, and 164. In accordance with ethical requirements and US Department of Health and Human Services guidelines for the use of deidentified specimen remnants in research studies, specimens were deidentified prior to the study and were limited to remnant specimens previously submitted for commercial testing. The healthy control specimens used in this study were obtained under IRB protocol #BR13-002, WCG IRB #20,121,940, approved by the Western IRB (now known as WIRB-Copernicus Group), which permits the collection of blood samples, medical, and demographic information for use in development, validation, and assessment of laboratory tests. Samples used in this study were obtained after informed consent was obtained from all the healthy volunteers participating in the study.

Samples and assays

Two collections of plasma specimens, preferably in EDTA (lavender-top) tubes although also acceptable in EDTA PPT (white-top) tubes, were used for this study, one with 602 samples and the other with 100 samples. Since ACTH is unstable in whole blood due to proteolytic degradation, collection procedures were recommended including immediate centrifugation after blood collection to separate plasma from cells, transfer of plasma to separate, plastic, specimen transport containers or vials, and immediate freeze. De-identified patient plasma specimens submitted to Quest Diagnostics for clinical testing (3/2019 to 9/2020; all received frozen, -20 to -45 °C) were sequestered daily following clinical testing if sufficient volume remained for comparison (n = 561). Additional plasma specimens were obtained from healthy donors (n = 41; 20 males aged 22 to 61 years and 21 females aged 23 to 64 years). In addition, a second set of patient specimens collected in the afternoon was used (n = 100). The specimens were frozen in -70 °C freezer within 12 h of clinical testing being completed and stored within − 20 °C freezer while awaiting placement within the − 70 °C. For measurement of ACTH levels, the specimens were thawed and run on the Immulite 2000 followed by the Roche Cobas e602 assay. All specimens were then refrozen. Specimens with discrepant results were then thawed, treated to remove HAMA/heterophile immunoglobulins, and assayed again by both assays. These procedures were carried out to minimize potential impact of sample handling on ACTH stability. We note that repeated freeze–thaw cycling could result in a small decrease in ACTH concentration—a decrease of 10.8% was detected using the Siemens assay following 3 cycles of freeze-thaw¹⁷. The reference range for adults (≥ 18 years) was 7.2 to 63.3 pg/mL using the Roche ACTH assay and 6 to 50 pg/mL using the Siemens ACTH assay. The assays were regularly calibrated using calibrators/standards from the assay manufacturers, and passing rate of proficiency testing was 100%.

HAMA/heterophile absorption

Specimens showing a difference between assays (Siemens compared to Roche) greater than 30% or 10 pg/mL were subjected to HAMA/heterophile absorption using the HAMA/heterophile absorption tube from Scantibodies Laboratory Inc. (Santee, CA 92071). The manufacturer’s instructions were followed as indicated in the Instructions for Use of the absorption tube. Untreated specimens and their corresponding HAMA/heterophile-treated counterparts were then assayed together on the same instrument. An individual specimen is considered positive for HAMA/heterophile interference if the difference between the neat specimen and the HAMA-treated specimen exceeds 30%. Of the 602 specimens in the initial study, 74 were available for evaluating the effect of HAM/heterophile antibodies.

Statistical analysis

Correlation between the ACTH results was determined using Spearman correlation coefficient and the linear relationship was assessed using Deming regression. The difference was calculated as (ACTH_Roche ─ ACTH_Siemens)/[(ACTH_Roche + ACTH_Siemens)/2]. The agreement was evaluated by Bland-Altman analysis on the difference (%) and the average of the results from the two assays. Discordance was defined as difference outside the limits of agreement at the mean difference ± 1.96 standard deviation in Bland-Altman analysis or difference > 50%. The distribution of difference in percentage between groups was assessed using Wilcoxon rank sum test. The odds ratio of having a discordance between groups was evaluated using the Fisher’s Exact test. The overall difference in ACTH levels between those with vs. without HAMA/heterophile absorption was assessed using a linear mixed model with adjustments for assay method (Siemens vs. Roche) and discordance status to account for both between-subject and within-subject variances. The HAMA effect on the ACTH discordance rate was assessed using a conditional logistic regression model. All analyses were performed in R software.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

MJM and TEB conceived the study and collected data. YL and JZL analyzed the data. All authors interpreted the data. YL and JZL wrote the manuscript. TEB, LAB and MJM edited the manuscript. All authors approved final manuscript.

Data availability

The data that support the findings of this study are not openly available due to reasons of sensitivity and are available from the corresponding author upon reasonable request.

Declarations

Competing interests

All authors (YL, JZL, TEB, LAB and MJM) were employed by Quest Diagnostics during the study.