Calcitonin gene‐related peptide and headache: Comparison of two commonly used assay kits highlights the perils of measuring neuropeptides with enzyme‐linked immunosorbent assays

Abstract

Objectives/Background

This study was undertaken to test the ability of a widely used enzyme‐linked immunosorbent assay (ELISA) kit to detect calcitonin gene‐related peptide (CGRP) isoforms to better understand currently published clinical data. There is significant interest in measuring CGRP as a biomarker in headache and migraine research, with ELISAs being the preferred detection method. ELISAs use antibodies that have been raised against an antigen to allow selective quantification of an analyte in a sample. Understanding the specificity of these antibodies is crucial to interpreting results. One commercially available kit (Cusabio CSB‐E08210h, Kit A) is purported to specifically detect β‐CGRP (one of the two CGRP isoforms) over α‐CGRP and has been used to investigate the potential for CGRP to be used as a biomarker for migraine. We investigated the ability of this kit to detect multiple isoforms of CGRP to better interpret published clinical results. We used a second ELISA kit (Bertin Bioreagent, A05481, Kit B) that is nonselective for the different CGRP isoforms as a control to ensure the CGRP we used could be detected in ELISAs.

Methods

We performed ELISAs according to the manufacturer's instructions, testing concentrations within the advertised range of each kit. At least three independent experiments were conducted for each kit.

Results

Kit B was able to detect both human and mouse α‐CGRP and β‐CGRP with a high degree of reproducibility. In contrast, Kit A did not detect bioactive forms of human α‐CGRP or β‐CGRP, nor mouse α‐CGRP or β‐CGRP.

Conclusion

Kit A may not be reliable for future studies, as it does not appear to detect mature bioactive CGRP. Importantly, conclusions from previous studies that used this kit may need to be reevaluated, as it is not clear what analyte the kit has detected. Our findings also highlight the importance of understanding research tools to ensure accurate interpretation of results.

Plain Language Summary

Calcitonin gene‐related peptide (CGRP) is strongly linked to migraine, and accurate measurement in patient samples could lead to better diagnosis and treatment; however, there is currently a lack of consensus as to its reliability as a biomarker. We tested a commercially available method that has been used to detect CGRP in patient samples (Kit A) and compared this to a control (Kit B) that was known to accurately detect CGRP. Kit A could not detect the mature form of CGRP; as such, previous research using Kit A may need to be reevaluated in light of these results.

Abbreviations

ACN

acetonitrile

CGRP

calcitonin gene‐related peptide

ELISA

enzyme‐linked immunosorbent assay

QC

quality control

INTRODUCTION

The success of calcitonin gene‐related peptide (CGRP) pathway therapeutics for migraine management raises the prospect of measuring this peptide as a biomarker. CGRP measurement in patient samples could enable better understanding of molecular mechanisms, predict and monitor treatment response, or guide diagnosis. ¹

There are now numerous studies reporting measurement of CGRP in migraine, but also cluster headache, idiopathic intracranial hypertension, and trigeminal neuralgia. ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ Despite this wealth of data, there is no clear consensus as to whether CGRP has utility as a prognostic or diagnostic factor. ² , ⁸ , ¹¹ There are likely to be many contributing factors such as variation in experimental methodologies, ⁸ differences between population groups (which may have different baseline levels of CGRP), and differences in the samples tested (as each bodily fluid will contain its own suite of proteins that could affect detection ¹² ). Sample handling, including the exact reagents used during sample collection (e.g., protease inhibitors), the timeframe of sample collection and its relationship to peptide half‐life, whether samples are diluted before detection, and the time between sampling and running the assay can each have an impact on the data, and thus the conclusions drawn from results. ⁸ , ¹³ However, an issue that does not seem to have been widely addressed is the potential contribution of the different CGRP assays in use, many of which have had little validation, and which may detect different forms of CGRP (Supporting Information, Sections 1–3).

Common approaches to peptide quantification are radioimmunoassay and enzyme‐linked immunosorbent assay (ELISA). ELISAs are now the most heavily used due to their convenience. However, a seemingly underappreciated consideration of this methodology is that these assays rely on antibodies, which have limitations. The antigens/immunogens used to generate these antibodies, the epitope(s) the antibodies actually detect, as well as how many recognition antibodies a given kit uses will each influence precisely what is detected and the conclusions that can be drawn from the assay (see Supporting Information, Section 1 for a summary of different ELISA formats, and how they relate to interpretation of results). Hence, understanding what the antibodies in ELISAs actually detect is critical to being able to correctly interpret results.

The CALCA and CALCB genes encode precursor proteins that are sequentially processed to produce the mature, bioactive α‐CGRP and β‐CGRP neuropeptides, as well as other peptides. Figure 1 illustrates CGRP processing, with the Supporting Information (Section 2) providing additional detail. The bioactive forms of CGRP are of particular interest as biomarkers, because these are the forms that are most likely to be released from neurons in response to neuronal activation, and are the forms that current CGRP‐targeting antibody therapeutics target. Although there are many kits marketed for the detection of CGRP, the term “CGRP” itself is ambiguous, as are “CALCA” and “CALCB,” as they could refer to any of the peptides created during the lifecycle of CGRP. For instance, the antibodies could recognize the mature bioactive peptide, the pre‐pro‐peptide (or other intermediates along the biosynthetic pathway), or a peptide sequence that is present in the mature peptide as well as in degradation products.

FIGURE 1.

Calcitonin gene‐related peptide (CGRP) processing from the 128/127 amino acid full pre‐pro‐peptides to the 37 amino acid mature bioactive peptides. The black brackets indicate a disulfide bond. The signal sequence is shown in green. All sequences are human. CPE, carboxypeptidase E; PAM, peptidylglycine alpha‐amidating monooxygenase; PC, prohormone convertase. [Colour figure can be viewed at wileyonlinelibrary.com]

Consequently, commercial CGRP ELISAs (Table S1) have the potential to detect a range of CGRP forms (Supporting Information, Sections 1 and 2). Hence, to properly interpret data from any given “CGRP” ELISA, it is important to understand which form(s) of the peptide the assay detects, and to ensure that validation includes specificity testing (https://www.ema.europa.eu/en/ich‐m10‐bioanalytical‐method‐validation‐scientific‐guideline), because the limited information provided by manufacturers is usually insufficient (Supporting Information, Section 3).

A CGRP ELISA kit produced by Cusabio (Cat# CSB‐E08210h), that herein we refer to as “Kit A,” has now been used by researchers in a range of patient samples, including plasma, tear fluid, saliva, and serum. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ¹² , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ These samples are being investigated across a range of conditions, such as chronic and episodic migraine, inflammatory bowel disease, and cluster headache. In some cases, it is reported that there are differences in analyte levels between healthy controls and disease groups, or in the case of migraine, between interictal and ictal phases. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ¹² , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ Detection of the analyte varies between sample types, being ~5–9 pg/mL (plasma), ~400–2200 pg/mL (tear fluid), ~70–250 pg/mL (saliva), and 0.5–5 pg/mL (serum). Based on the information given by the supplier, it could be concluded that Kit A selectively detects β‐CGRP (Supporting Information, Section 4), and this matter has been discussed in the literature. ⁶ , ⁸ , ¹⁸

Selective identification of α‐CGRP or β‐CGRP would be of considerable value to help delineate the role of each peptide in various conditions, including migraine and headache disorders. However, α‐CGRP and β‐CGRP are very similar in sequence (Supporting Information Figure S4), so a careful strategy is needed to develop an assay that selectively detects each peptide. Hence, it is essential to establish whether Kit A selectively detects β‐CGRP over α‐CGRP. We therefore tested the ability of this kit to detect mature, functionally active α‐CGRP and β‐CGRP. As a control, to ensure that our bioactive CGRP could be detected using ELISA methodology, we tested a second CGRP ELISA kit (Bertin Bioreagent, A05481), which herein we refer to as “Kit B,” that is reported to detect mature α‐CGRP and β‐CGRP, and that has also been used to test clinical samples. ² , ¹³ , ¹⁹

METHODS

Peptides

Peptides used in this study represent the mature, bioactive forms that correspond to the sequences presented in Figure S4. Further detail on these peptides is provided in Table 1. Peptides were diluted in H₂O and stored at −30°C as 1‐mM aliquots in protein LoBind tubes (Eppendorf, Hamburg, Germany). Individual aliquots were limited to two freeze–thaw cycles. These batches of peptides have been shown to be functionally active at CGRP‐responsive receptors (Figure S5). In some cases, we used the exact same peptide aliquot in each experiment; however, as the two kits have different incubation times and could not be run concurrently, this involved freezing and rethawing peptide. There were no differences in results generated between two sources of the human and mouse α‐CGRP, hence we have not differentiated between sources in this article.

TABLE 1.

Origin and details relating to CGRP peptides used.

Peptide	Species	Manufacturer
α‐CGRP	Human	Bachem (Bubendorf, Switzerland) Cat# 4013281.1000 Lot# 1000031490
α‐CGRP	Human	University of Auckland 32
α‐CGRP	Mouse/rat	Bachem Cat# 4025897.1000 Lot# 100014400
α‐CGRP	Mouse/rat	University of Auckland 33
β‐CGRP	Human	Bachem Cat# 4015500.1000 Lot# 1000031674
β‐CGRP	Mouse	University of Auckland 33

Abbreviation: CGRP, calcitonin gene‐related peptide.

General ELISA information

Two batches of each ELISA kit were tested to ensure that external factors such as those encountered during shipping and storage did not affect function (Table 2). ²⁰ Kits were stored as per manufacturer's protocols and used within the expiry dates provided.

TABLE 2.

Details of the CGRP ELISA kits used.

	Kit A	Kit B
Kit name	Human calcitonin gene related peptide, CGRP ELISA kit	CGRP (human) ELISA kit
Catalogue number	CSB‐E08210h (Cusabio; Houston, TX, USA)	A05481 (Bertin Bioreagent; Montigny le Bretonneux, France)
Batch/lot numbers used	I09229672, N01228970	0223, 0323

Abbreviations: CGRP, calcitonin gene‐related peptide; ELISA, enzyme‐linked immunosorbent assay.

Kit A ELISA methods

The kit was prepared as per manufacturer's instructions. Briefly, the kit‐provided biotin antibody and horseradish peroxidase–avidin were individually diluted 1:100 with their respective diluent. Wash buffer was warmed to room temperature, mixed, then diluted 1:25 in Milli‐Q H₂O. The standard vial was centrifuged in a benchtop mini microcentrifuge (Cat# C1601‐B; Labnet, NY, USA) for 30 s at 6000 rpm. The standard curve was then prepared as per kit protocol. Known concentrations of human or mouse α‐CGRP and β‐CGRP were diluted from 1‐mM stocks in kit sample diluent. One hundred microliters of the standard curve (1.56–100 pg/mL), sample diluent alone (0 pg/mL), or CGRP samples was added to the wells in duplicate. A plate sealer (Cat# 100‐SEAL‐PLT; Excel Scientific, CA, USA) was applied to the plate, and the plate was incubated for 2 h at 37°C. Wells were aspirated, and 100 μL of biotin antibody was added to each well. A new plate sealer was applied, and the plate was incubated for 1 h at 37°C. Wells were aspirated and washed three times for 2 min with 200 μL of wash buffer. After the final wash, the buffer was aspirated and 100 μL of horseradish peroxidase–avidin was added to each well; a plate sealer was applied, and the plate was incubated for 1 h at 37°C. Wells were then aspirated and washed five times with 200 μL wash buffer. After the final wash was aspirated, 90 μL of kit‐provided tetramethylbenzidine substrate was added to each well, and the plate was incubated for 30 min at 37°C in the dark. Kit‐provided stop solution (50 μL) was then added to each well, and the plate was gently tapped to stop the reaction. The absorbance was then immediately read at 450 nm and 570 nm using a ClarioStar plate reader (BMG Labtech; Ortenberg, Germany). A quality control is not included in the kit, and there has been no independent verification to identify a peptide that the kit can detect, hence no positive control was able to be included in experiments with this kit.

Data analysis was performed as described by the manufacturer (https://www.cusabio.com/m‐225.html). For each well, the absorbance at 570 nm was subtracted from the absorbance at 450 nm. The average absorbance of the zero standard (or “blank”) wells included in each experiment was then subtracted from all wells. The resulting readings for the standard curve were then averaged, and the standard curve was plotted using CurveExpert 1.4. The average absorbance was plotted on the x‐axis against the concentration of standard on the y‐axis. The best‐fit curve was then determined by the software and used to interpolate the concentration of analyte in each well. Each condition was tested in duplicate; these duplicates were averaged to give a final concentration of analyte per experiment.

Kit B ELISA methods

As for Kit A, multiple batches of Kit B were tested (Table 2). On the experimental day, the kit‐provided wash buffer, ELISA buffer, and Ellman reagent were prepared as per the manufacturer's instructions. Briefly, the kit was warmed to room temperature; the ELISA buffer was then reconstituted with 50 mL of Milli‐Q H₂O and left to stand for 5 min. The kit‐provided standard, quality control (QC), and CGRP tracer were prepared by adding 1 mL (standard and QC) or 10 mL (CGRP tracer) ELISA buffer. In addition, known concentrations of human and mouse α‐CGRP and β‐CGRP were diluted in ELISA buffer from 1‐mM stocks. Wash buffer was prepared by diluting 1 mL of kit‐provided wash buffer concentrate and 200 μL kit‐provided Tween 20 in 400 mL Milli‐Q water, followed by mixing with a magnetic stirrer until the reagents dissolved. ELISA wells were washed five times with 300 μL of wash buffer, then 100 μL of the standard curve (7.61–1000 pg/mL), QC, ELISA buffer alone, and CGRP samples were added to the wells in duplicate; two wells per run were left without addition to serve as a “blank” as directed by the kit. CGRP tracer (100 μL) was added to each well (excluding the blank wells); the plate was then sealed and incubated overnight (16–20 h) at 4°C. Wells were aspirated and washed three times with wash buffer (300 μL); the third wash was performed on an orbital shaker for 2 min at 300 rpm in a clockwise motion. During this time, Ellman reagent was prepared by reconstituting the powdered reagent in 49 mL Milli‐Q H₂O and adding 1 mL of concentrated wash buffer. Following the 2‐min shake, the plate was then washed a further three times with 300 μL wash buffer per well, with the third wash including a 2‐min shake. The wash buffer was then aspirated, and 200 μL of Ellman reagent was added. Plates were sealed and covered with light‐proof aluminium foil, then incubated at room temperature on an orbital shaker (300 rpm) in the dark. The plate was then read at 414 nm using a ClarioStar plate reader. In initial experiments, the plate was read periodically, being returned to the shaker in the dark between each read. The plate was read at 30, 45, 60, 90, and 120 min. From this, we chose to proceed with the 45‐min time point in subsequent experiments, as this allowed good separation between the highest concentration of the standard and the blank (>0.5 absorbance units).

Data analysis was performed according to the manufacturer's instructions. The average absorbance of the blank wells was subtracted from all wells. Data analysis was performed in Prism 10 (GraphPad, San Diego, CA, USA). The absorbance was plotted on the y‐axis, and the concentration of the standard was plotted on the x‐axis. As recommended by the manufacturer, a straight line was then fit to the data, and this line was used to interpolate the concentration of analyte within each well. Each condition was assayed in duplicate; these duplicates were averaged to give a final concentration of analyte per experiment.

Experimental design and data presentation

Each independent experiment was performed on a separate day, with reagents either being freshly prepared on the day of each experiment (such as peptides and standards) or used within the recommended storage time (for kit reagents that are prepared in bulk, such as the ELISA buffer in Kit B). Each independent experiment involved its own standard curve, nonspecific binding condition, and for Kit B, a blank and QC condition (as indicated by the manufacturer). All data are presented as the mean ± standard error of the mean, with each independent experiment producing one “n”.

Mass spectrometry on the Kit A standard

We performed two separate mass spectrometry runs using the kit‐provided standard. The first sample was the kit standard diluted to 100 pg/mL in Kit A sample diluent (the highest concentration of the provided standard), whereas the second run used the powdered, undiluted standard as the starting material. We differentiate between the two samples by referring to the former as diluted and the latter as powdered sample.

When working with the powdered sample, the sample was dissolved in 20 μL of 100 mM triethylammonium bicarbonate by repeated vortex and sonication. Once the sample was visibly dissolved, it was centrifuged at 30,000 × g for 30 min to settle any undissolved particles. From this point on, the powdered and diluted samples were treated identically. A bicinchoninic acid assay was performed to determine the protein concentration. The samples were mixed with an equal volume of 2× S‐Trap lysis buffer (100 mM triethylammonium bicarbonate, 10% [vol/vol] sodium dodecyl sulfate) and sample processed through the S‐Trap micro spin column (ProtiFi; Fairport, NY, USA) according to the manufacturer's protocol. A tryptic digest was performed by adding trypsin in a 1:20 (wt/wt) ratio with trypsin to protein. The digestion was performed overnight at 37°C in a humid chamber. The following day, tryptic peptides were recovered according to manufacturer's protocol and were dried in a SpeedVac (Savant, France).

The dried peptides were solubilized in 20 μL of solubilization buffer composed of 5% (vol/vol) acetonitrile (ACN), 0.1% (vol/vol) formic acid. A 2.5 μL sample volume was injected on to the Orbitrap Exploris 240 mass spectrometer (Thermo Fisher Scientific; Waltham, MA, USA) coupled to an Ultimate 3000 nano‐flow uHPLC system (Thermo Fisher Scientific). The liquid chromatography method length was of 120‐min duration and consisted of following ACN gradient steps, 5% to 25% ACN in 84 min, 25% to 40% ACN in 7 min, and 45% to 98% ACN in 13 min. The MS1 was scanned from 400 to 2000 m/z range. The +2 and higher charge state MS1 ions were allowed for MS2 fragmentation. The higher collisional dissociation stepped collision energy was set at 28, 30, 32, 35. The dynamic exclusion was enabled for 1 repeat count with an exclusion duration of 11 s. The total cycle time was set at 1.5 s.

The resulting MS data were analyzed with Sequest HT (Thermo Fisher Scientific) search engine node within Proteome Discoverer software (version 2.5, Thermo Fisher Scientific). The data were queried on the SwissProt and Uniprot human (both downloaded in 2023) protein sequence databases. The search was set up for tryptic peptides with a maximum of three missed cleavage sites. Within the data analysis, deamidation of asparagine and glutamine, and oxidation of methionine were allowed as dynamic modification. Furthermore, carbamidomethylation of cysteines was selected as static modification. The maximum precursor mass tolerance threshold was set at 10 parts per million, and the MS2 fragment ion mass error was set at 0.02 Da. The resulting peptide/protein list was exported as a spreadsheet for further interpretation.

RESULTS

The standard curves for each kit were as expected based on information provided by the manufacturer's (Figure 2), and where provided in the kit (Kit B only), the QC sample was within the range indicated by the kit. The expected concentration for the Kit B QC was 125 pg/mL, and we achieved a mean value of 117.3 ± 7.8 pg/mL (n = 5).

FIGURE 2.

Detection of human α‐calcitonin gene‐related peptide (α‐CGRP) and β‐CGRP by Kit B (Bertin, A05481; A, B) and lack of detection of human α‐CGRP and β‐CGRP by Kit A (Cusabio, CSB‐E08210h; C, D). In panels A and B, the gray‐filled circles indicate expected and observed results; in panels C and D, the black circles indicate expected results, and the black‐dashed circles indicate observed results. Panels B and D are the same data as presented in panels A and C, respectively. Panel B shows a zoom in on the Kit B standard curve to match the detection range of Kit A; panel D has the x‐ and y‐axes swapped relative to panel C to allow easier comparison to Kit B. In each case, an example standard curve is provided. Results are indicative of our experiments in which mature, bioactive human or mouse α‐CGRP or β‐CGRP was diluted to 100 pg/mL and tested in the kits (Kit B detection range = 7.81–1000 pg/mL, Kit A detection range = 1.56–100 pg/mL). Further results are provided in Tables 3 and 4.

When testing detection of bioactive α‐CGRP and β‐CGRP, we focused on the 100 pg/mL concentration. We reasoned that this was an appropriate concentration to test, as it is very close to the concentration of the QC provided by Kit B and is within the standard curve of both kits. In some instances, we also investigated higher and lower concentrations to explore whether there were concentration‐dependent effects. Kit B detected human α‐CGRP and β‐CGRP, with the interpolated value being close to the expected value in all cases (Figure 2A, Table 3). When testing 100 pg/mL of human peptides, we obtained a mean value of 128.6 ± 7.77 pg/mL for α‐CGRP and 92.53 ± 9.56 pg/mL for β‐CGRP (both n = 3); for individual values and results from other concentrations, see Table 3.

TABLE 3.

Comparison of detection of human α‐CGRP and β‐CGRP by Kit A (Cusabio, CSB‐E08210h) and Kit B (Bertin Bioreagent, A05481) human CGRP ELISA kits.

Concentration tested, pg/mL	Concentration observed, pg/mL
	Human α‐CGRP		Human β‐CGRP
	Kit A	Kit B	Kit A	Kit B
38,000	<1.56	–	<1.56	–
1000	1.69, <1.56, <1.56	–	<1.56, 2.59, <1.56	717.5
100 a	1.85, <1.56, <1.56, <1.56	142.2, 115.6, 128	<1.56. <1.56, <1.56, 3.84	111.5, 80.9, 85.2
25	2.60	–	<1.56	–

Note: Each value is from an independent experiment comprising a separate standard curve and sample dilutions, and assays performed on different days. Each value is the mean of two technical replicates. “–” indicates experiments not performed. Human peptides from multiple sources were tested (Table 1) to ensure that a lack of detection was not due to external factors such as conditions encountered during manufacture, shipping, or storage. Similar results were obtained with all batches, hence these are not delineated.

Abbreviations: CGRP, calcitonin gene‐related peptide; ELISA, enzyme‐linked immunosorbent assay.

^a 100 pg/mL is the highest concentration on the Kit A standard curve, and within the range of Kit B; hence we performed most testing on this concentration.

In contrast, Kit A did not detect either human β‐CGRP or α‐CGRP at or near the expected concentration (Figure 2C, Table 3). Additionally, we tested Kit A with another concentration that was within the standard curve (25 pg/mL) and similarly did not detect human β‐CGRP or α‐CGRP near or at the expected concentration. Occasionally, an individual 25‐ or 100‐pg/mL sample exceeded the lower limit of detection; however, this did not correlate with the expected concentration in any case (Table 3). To investigate whether the kit was less sensitive than advertised, we performed further testing with higher CGRP concentrations (1000 and 38,000 pg/mL); however, once again we were unable to obtain CGRP detection (Table 3).

Two separate lots of the kits were tested with the same results. As we were able to generate a standard curve that matched the manufacturer's data (Figure 2C), it appears that our assays were performed correctly, as the wells used in the assay detected the standard, and we could measure a change in absorbance. Therefore, our lack of detection was not due to technical error. Kit A does not provide an internal QC, as the manufacturer suggests that successful generation of the standard curve is evidence that the kit is functional; therefore, we are forced to conclude that we have successfully performed this experiment (Supporting Information Figure S7). To our knowledge, there has been no report of a positive control of known sequence that is recognized by this kit.

As the mature human and mouse CGRP peptides have a high degree of similarity in amino acid sequence, we anticipated that any kit that detects one mature, bioactive CGRP would also detect other related mature, bioactive CGRPs. We therefore also tested the ability of the two CGRP ELISA kits to detect mouse α‐CGRP and β‐CGRP, as preclinical testing will often use rodents, and CGRP measurements can be incorporated into these studies. ²¹ Although the antibodies used in Kit A are reportedly designed to target human CGRP, the mature human and mouse CGRP peptides have a high degree of similarity in amino acid sequence (Figure S4), and for Kit B, the manufacturer states that the kit can detect rat α‐CGRP and rat β‐CGRP. As for human peptides, we primarily tested 100 pg/mL, as this is within the standard curve range of both kits. Kit B detected mouse α‐CGRP and β‐CGRP (100 pg/mL), with mean values of 100.7 ± 9.5 and 128.1 ± 15.0 pg/mL, respectively (both n = 3). For individual values, see Table 4. As with the human peptides, Kit A did not appear to detect α‐CGRP or β‐CGRP at concentrations that matched the expected concentrations, with most values being below the lower limit of detection (Table 4). We performed additional Kit A experiments using concentrations of 1000 pg/mL for both mouse α‐CGRP and β‐CGRP; however, in neither case did this produce robust detection (Table 4). Statistical testing of the results from Kit B found that there was no significant difference in the kit's ability to detect the four tested forms of CGRP (one‐way analysis of variance, comparing human α‐CGRP, human β‐CGRP, mouse α‐CGRP, and mouse β‐CGRP). No statistical testing could be performed on the Kit A data.

TABLE 4.

Comparison of detection of mouse α‐CGRP and β‐CGRP by Kit A (Cusabio, CSB‐E08210h) and Kit B (Bertin Bioreagent, A05481) human CGRP ELISA kits.

Concentration tested, pg/mL	Concentration observed, pg/mL
	Mouse α‐CGRP		Mouse β‐CGRP
	Kit A	Kit B	Kit A	Kit B
1000	<1.56, <1.56, 4.67, <1.56	–	<1.56, <1.56, <1.56	1022
100 a	1.67, 3.14, <1.56	83.06, 115.6, 103.3	<1.56, <1.56	145.1, 141.1, 98.11

Note: Each value is from an independent experiment comprising a separate standard curve and sample dilutions, and assays performed on different days. Each value is the mean of two technical replicates. “–” indicates experiments not performed. Peptides from multiple sources were tested (Table 1) to ensure that a lack of detection was not due to external factors such as conditions encountered during manufacture, shipping, or storage. Similar results were obtained with all batches, hence these are not delineated.

Abbreviations: CGRP, calcitonin gene‐related peptide; ELISA, enzyme‐linked immunosorbent assay.

^a 100 pg/mL is the highest concentration on the Kit A standard curve, and within the range of Kit B; hence we performed most testing on this concentration.

Kit A does not appear to detect mature bioactive β‐CGRP but could instead have been developed using a different CALCB protein, such as the pre‐pro‐peptide sequence (Figure 1; Supporting Information, Section 2). We therefore performed two separate mass spectrometry runs using the kit standard in an effort to identify which form of CGRP it contained. Theoretically, our mass spectrometry protocol could detect peptides in the low picogram range, assuming the standard was a pure sample. The kit did not state the composition of the standard, hence we proceeded with our experiment on the basis that it may be pure peptide. In neither case were we able to detect CGRP/CALCA/CALCB‐derived peptides, and instead we detected primarily bovine serum albumin. This lack of detection is likely due to the low concentration of standard in the sample, relative to bovine serum albumin within the sample that likely obscured the signal from the standard.

DISCUSSION

Our findings have major implications for headache and headache research. We demonstrate that the widely used Kit A may not be suitable for detecting mature bioactive α‐ or β‐CGRP in clinical samples or otherwise. Although the kit is clearly able to detect an analyte in clinical samples based on the literature, based on our results, this analyte does not appear to be a mature form of CGRP. This affects the conclusions drawn from the clinical studies that have reported data using this kit. ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ¹² , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ The potential impact of our findings warrants further consideration and discussion within the field.

We investigated a reportedly β‐CGRP‐specific kit (Kit A) and compared it to a kit that has an explicit inability to distinguish α‐CGRP and β‐CGRP (Kit B). An essential starting point with any immunoassay is to test whether the assay actually detects the substance of interest, before conducting other key validation experiments, such as determining the effect of biological fluid matrix on results (https://www.ema.europa.eu/en/ich‐m10‐bioanalytical‐method‐validation‐scientific‐guideline). ⁸ , ¹³ All of these validation details should then be reported when publishing. Additionally, it is important to report all data handling steps, as there are some values in the literature that are beyond the range of the standard curve provided by the supplier. For instance, analyte concentrations have been reported as high as 2200 pg/mL, ⁷ although the Kit A standard curve only provides a maximum concentration of 100 pg/mL. It is therefore presumed that values exceeding the range of the standard curve have been achieved through multiplication, such as to account for an initial sample dilution, or because of a low volume of starting sample. However, methods are usually too brief, and information such as this is not explicitly stated. We urge reporting of all such data handling steps to ensure data transparency.

In our hands, Kit B detected all forms of mature bioactive CGRP, whereas this was not the case for Kit A. We tested concentrations of 25 and 100 pg/mL, which represent values on the linear portion and the high end of the Kit A standard curve, respectively. In neither case were we able to detect CGRP reliably. Although there were instances where the kit reported a value that was above the detection threshold, in no case did it correspond to the expected concentration of CGRP. We were able to generate a standard curve with Kit A; thus, to better understand what the antibodies in Kit A could be detecting, we performed mass spectrometry on the Kit A standard. We were unable to detect any CGRP‐related peptides in the sample. Our inability to detect a CGRP peptide with mass spectrometry is not problematic in itself, given that the standard is provided in a low amount (100 pg per kit), sample processing could result in protein loss, and our results show that the standard contains a complex matrix of other proteins, notably bovine serum albumin. However, the kit's inability to detect mature CGRP, together with the exact sequence of protein used to develop Kit A being unknown, creates a situation in which the ability of Kit A to detect bioactive CGRP is brought into question. Previously published results that use Kit A to rule in/out CGRP as a biomarker should be reevaluated, as this kit does not appear to detect mature CGRP.

Our approach applies to the myriad of other commercially available CGRP ELISAs and to the quantification of other neurology‐relevant peptides such as pituitary adenylate cyclase activating peptide (PACAP), amylin, or other substances like hormones. ²² , ²³ We provide a framework that others could use to test the specificity of ELISA kits to ensure the reliability of their results. However, we believe the onus should not only be on researchers, and we urge commercial suppliers to provide more information and more validation.

We acknowledge that our study has limitations, but it was important to urgently bring these data to the attention of the field. With further testing of additional batches of Kit A, we may have obtained different results. Even if this were the case, it would be important to test mature bioactive CGRP in each batch of kit; this applies to any ELISA kit as an independent positive control. Although we used mass spectrometric analysis of the kit standard, we were not able to resolve the question of what this kit detects, and therefore this remains an open question. As outlined in the Supporting Information (Section 4), our efforts to obtain further information from the supplier did not provide clarity, and we can only speculate that the kit was actually developed to detect the CALCB pre‐pro‐peptide sequence, rather than mature β‐CGRP, which would explain our results (Figure S3). Further experiments, such as more detailed mass spectrometry, may be able to shed light on what the manufacturer has provided as a standard; however, the onus should not entirely be on the researchers to perform this sort of experiment. Instead, we urge manufacturers of these antibody‐based products to provide more information and perform more validation as to what their kits detect (see Supporting Information, Section 1 for information relating to antibody production). Although it is understandable that some information needs to remain proprietary, given the ambiguity of the term CGRP, it is essential to provide more detail. This also applies more widely to all antibody‐based detection methodologies, as there is growing evidence that antibodies are not always specific for, or even able to detect, their advertised target. ²⁰ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸

There is also a need for the manufacturer to be clear in advertising products. For instance, Abcam supplies a kit that is advertised to detect S100A12/CGRP, and this kit appears to have been used in the literature to detect “CGRP” levels in human samples. ²⁹ However, to our best understanding, it would be extremely difficult to develop an ELISA kit that could recognize both CGRP and S100A12. The more likely scenario therefore is that the company has a typographical error on the website, as S100A12 is also known as calgranulin C, or CCRP (Supporting Information Figure S8). Therefore, although this ELISA kit is found when performing a search for CGRP ELISAs, it is very unlikely that this kit was ever intended to detect CGRP. This adds to the confusion in the literature and highlights the importance of validating tools before they are used.

When reexamining past work using Kit A with this knowledge, it is imperative to keep in mind that although there may be statistically significant differences in analyte detection between patient conditions (e.g., interictal vs. ictal samples from patients with migraine) when using this kit ³ , ⁴ , ⁵ , ⁶ , ⁷ , ¹² , ¹⁴ , based on our results, the detected analyte does not appear to be a mature form of CGRP. It is possible that the kit is instead detecting an immature form of CGRP; however, we were not able to independently verify this. If it were an immature peptide being detected, this could be interesting and suggest that patients with migraine have higher circulating levels of that substance; however, without confirming what the kit detects, this is speculative. Until either the manufacturer confirms what exactly the kit detects, and/or a research group is able to independently validate a peptide that the kit can recognize, it is impossible to say with confidence what this kit detects.

Our experience with an ELISA kit not detecting the stated analyte is not unique, ²² and therefore it is important to proceed with caution when considering using ELISAs. Drawing attention to this issue may help move headache research forward by resolving questions about biomarkers and their relevance to headache. Biomarkers feature as important headache research priorities in a recent report from an international multistakeholder group, and therefore this is important to get right. ³⁰ This matter extends far beyond headache, given extensive links between CGRP and other disorders. ³¹ With validated assays, and results that can be compared, we will have a far stronger case for ruling in, or out, peptides like CGRP as biomarkers in headache disorders and beyond.

CONCLUSION

The Kit A CGRP ELISA kit did not detect mature bioactive CGRP, and conclusions from prior studies using this kit may need to be reevaluated. Our data therefore add another facet for consideration, as researchers attempt to study CGRP as a biomarker in migraine and beyond.

AUTHOR CONTRIBUTIONS

Michael L. Garelja: Conceptualization; data curation; formal analysis; investigation; validation; visualization; writing – original draft; writing – review and editing. Tayla A. Rees: Conceptualization; visualization; writing – original draft; writing – review and editing. Debbie L. Hay: Conceptualization; funding acquisition; project administration; resources; supervision; writing – original draft; writing – review and editing.

FUNDING INFORMATION

Michael L. Garelja and Debbie L. Hay were supported by a Marsden Fund grant. Tayla A. Rees acknowledges support from the International Headache Society.

CONFLICT OF INTEREST STATEMENT

There are no conflicts of interest for Michael L. Garelja, Tayla A. Rees, and Debbie L. Hay relevant to this article. None of the authors holds any interest in the manufacture or supply of CGRP ELISA kits. Debbie L. Hay is or has been a consultant or speaker for AbbVie, Nxera Pharma, Lilly, Amgen, Lundbeck, and Teva, and has received research funding from Pfizer, AbbVie, and Solros Therapeutics in the past 3 years. Debbie L. Hay is a member of the International Union of Pharmacology Transparency & Reproducibility Committee.

Supporting information

Figures S1–S8 and Table S1.

Garelja ML, Rees TA, Hay DL. Calcitonin gene‐related peptide and headache: Comparison of two commonly used assay kits highlights the perils of measuring neuropeptides with enzyme‐linked immunosorbent assays. Headache. 2025;65:1744‐1753. doi: 10.1111/head.15011