Rapid Serological Detection of Anti-GM-CSF Autoantibodies in Autoimmune Pulmonary Alveolar Proteinosis Using a Novel Immunochromatographic Test.

Chika Narita: data curation (lead), formal analysis (lead), investigation (lead), methodology (equal), validation (equal), writing – original draft (equal), writing – review and editing (equal). Kenjiro Shima: formal analysis (equal), investigation (equal), project administration (equal), resources (lead), writing – review and editing (equal). Jun‐ichiro Sekiguchi: conceptualization (equal), funding acquisition (equal), investigation (equal), validation (equal). Yasuyuki Ozeki: conceptualization (equal), formal analysis (equal), funding acquisition (lead), investigation (equal), project administration (lead), visualization (equal). Makoto Akiwa: data curation (equal), formal analysis (equal), software (equal), validation (equal). Shohei Koike: data curation (equal), formal analysis (equal), validation (equal). Sachiko Yuki: data curation (equal), investigation (equal), methodology (equal), resources (equal), validation (equal). Toshiaki Kikuchi: resources (equal), writing – review and editing (equal). Keiichi Akasaka: investigation (equal), resources (equal), validation (equal). Nobutaka Kitamura: data curation (equal), formal analysis (equal), methodology (equal), supervision (equal), validation (equal). Koh Nakata: data curation (equal), formal analysis (equal), funding acquisition (equal), investigation (equal), methodology (equal), project administration (equal), resources (lead), supervision (equal), validation (equal), writing – original draft (lead).

The samples used in this study were collected for the development of a GMAbs measurement ELISA previously conducted at Niigata University, and were reused in this study. An ethical application for reuse was submitted to Niigata University and Kohjin Bio's Ethical Committee and was approved (2023–0126 for Niigata University, 02602–2302 for Kohjin Bio). In addition, a portion of the samples was submitted for drug approval to The Pharmaceutical and Medical Devices Agency and was approved (30600EZX00015000).

All serum samples used in a previous study [ 13 ] were reused, except for several that had been previously depleted. For the training study, 139 serum samples including 72 from healthy subjects and 67 from APAP patients were used in this study. There was no significant difference in the age or sex ratio between patients with APAP and healthy subjects (Table  1 ). For the validation study, another 382 samples were obtained from 211, 128, 5, and 38 patients with APAP, diffuse lung diseases other than PAP, hereditary PAP, and secondary PAP (Table  2 ). To address the ethical aspects of the sample reuse, we applied to the ethics committees of Niigata University and Kohjin Bio., and after obtaining approval (2023‐0126 for Niigata University, 02602–2302 for Kohjin Bio.), we used the samples in a state that enabled the subjects to refuse (opt‐out) upon the request of the subject. We informed the subjects of the purpose of the research. The samples were blinded by the Niigata University Clinical Translational Research Center, and background information was kept confidential until the study results were obtained. Demographic data of subjects in the training study. Abbreviations: APAP, autoimmune pulmonary alveolar proteinosis; DSS, disease severity score; GMAbs, anti‐granulocyte‐macrophage colony‐stimulating factor autoantibody; SD, standard deviation. Determined by ELISA kit. The definition of the concentration unit U/mL is detailed in Reference [ 13 ]. Demographic data of cases in the validation study. 50.0 (42.5–60.0) 60.0 (50.0–64.0) 52.0 (40.0–65.0) 68.0 (62.8–75.0) 66.8 (29.6–125.8) 0.12 (0.08–0.17) 0.10 (0.07–0.17) 0.12 (0.09–0.20) Note : Connective tissue disease (CTD) cases included one amyopathic dermatomyositis cases, three rheumatoid arthritis cases, one Sjögren syndromes, and two other diseases. Infectious disease cases included two Pneumocystis pneumonias, one influenza pneumonias, one Legionella and one non‐tuberculous mycobacterial disease. Miscellaneous diseases included one alveolar haemorrhages, and seven other diseases. Abbreviations: APAP, autoimmune pulmonary alveolar proteinosis; DIPD, drug induced pulmonary disease; GMAbs, anti‐granulocyte‐macrophage colony‐stimulating factor autoantibody; HPAP, hereditary pulmonary alveolar proteinosis; IIPs, idiopathic interstitial pneumonias; IPF, idiopathic pulmonary fibrosis; SD, standard deviation; SPAP, secondary pulmonary alveolar proteinosis. Determined by ELISA kit. ICT strips were made by conjugating a mouse‐derived anti‐human IgG antibody (MilliporeSigma, Burlington, MA, USA) that captures GMAbs labeled with self‐made colloidal gold nanoparticles. A conjugation pad was prepared by soaking glass fiber in the conjugation solution (Glassfiber Diagnostic Pad, Merck Ltd., Tokyo, Japan) and then drying it. E. coli ‐derived recombinant human GM‐CSF (rHuGM‐CSF, Amoytop Biotech, Fujian, China) was immobilized on the test line position of the nitrocellulose membrane, and horse‐derived anti‐mouse IgG antibody (Horse Anti‐Mouse IgG antibody, Vector Laboratories California, USA) was immobilized on the control line position. Sample pad, absorption pad, nitrocellulose membrane (Lateral Flow Membrane, Advantech Toyo Kaisha Ltd., Tokyo, Japan), and glass fiber were attached to a backing sheet (backing sheet for immunochromatography polystyrene, Lohmann, Neuwied, Germany) and formed into strips. After cutting, it was inserted into a custom‐made plastic cassette and used as an ICT prototype. The principle of the ICT prototype (Figure  1A ) is to relies on capillary action and antigen–antibody interactions. For detection, the first step is to apply 5 μL of a sample into the container (Sin Figure  1A,B ). Next, by dropping 100 μL of the developing solution onto developer pad D, the antibodies in the serum migrate along with the developing solution over the test strip due to capillary action. The colloidal gold‐labelled anti‐human IgG antibody impregnated into the conjugation pad moves through the development area while forming an immune complex with the GMAbs in the serum (E, Figure  1B ). If GMAbs exist in the dropped serum, they will be captured by the recombinant human GM‐CSF immobilized on the test strip (F, Figure  1B ), and the test line will be detected as a reddish‐purple line appearing at this position. Immune complexes that are not captured by the recombinant human GM‐CSF migrate further through the development section and are captured by the anti‐mouse immunoglobulin antibody immobilized on the control determination section (G, Figure  1B ), and a reddish‐purple line will be visible at this position. Schematic diagram of immunochromatographic assay for detecting anti‐GM‐CSF autoantibodies. A: Immunochromatographic prototype. The sample position (S) is used to apply 5 μL of serum. The drop position developer pad (D) is used to add 100 μL of loading solution to the device. Appearance of two lines at both the control (C) and test (T) positions indicates positive result for anti‐GM‐CSF autoantibodies; B: Structural diagram of Immunochromatography test strip. The test uses the conjugation pad containing the colloidal gold‐labelled anti‐IgG monoclonal antibody (E), a nitrocellulose strip onto which recombinant human GM‐CSF (T) and horse anti‐mouse IgG antibodies (C) are immobilized, and an absorption pad. The standard polyclonal antibody used in this study was described in a previous report [ 13 ]. Briefly, the polyclonal autoantibody purified from pooled sera of 20 APAP patients was used as a standard antibody to adjust the intensity of the ICT strips immobilized with varying amounts of recombinant GM‐CSF and colloidal gold‐labelled anti‐human IgG antibody were prepared. The appearance of a test line was examined using a dilution series of the above antibody (200, 100, 50, ……1.56, 0.78, 0.39, 0 U/mL). The ICT, which was optimized to approximate the cutoff value between APAP and healthy controls with 1.65 U/mL, was used as the prototype for the subsequent experiments (Figure  S1 ). Among the 382 validation study samples, 50 samples from APAP patients and 50 samples from non‐APAP patients were extracted while matching age and sex. The serum sample was diluted 201‐fold with a diluting buffer (PBS/0.1% BSA/0.1% polyoxyethylene) and then serially diluted 1‐, 2‐, 4‐, 8‐, 16‐, …, 512‐, 1024‐, 2048‐fold, and the end point dilution fold was determined using the ICT prototype. The titre was defined as the highest dilution fold that yielded a visible test line. In addition, for serum samples with GMAbs concentration less than 60 U/mL, the serum samples were directly diluted at 1, 2, 4, 8, 16, …, 8192, 16,384, 32,768‐fold and applied to the prototype. The titre of serum with a negative test line was defined as 0. Continuous variables were expressed as median (IQR). If the difference of the mean value between two groups was normally distributed, the Student's t test was used; otherwise, the Mann–Whitney U test was used. Based on the clinical performance test results, sensitivity, specificity, positive, and negative predictive values were calculated. Inter‐examiner agreement was assessed using the k efficient. Correlation between GMAbs concentration and the antibody titre was performed using Spearman's correlation analysis. All statistical analyses were performed using SAS software, version 9.4 (SAS Institute, Cary, NC, USA). A p value < 0.05 was considered statistically significant.

The training study samples included samples from 72 healthy subjects and 67 patients with APAP (Table  1 ). The concentrations of GMAb were 0.12 (0.09–0.17) U/mL and 63.7 (29.2–106.8) U/mL, respectively. Both distributions appeared lognormal and showed no overlap between healthy subjects and patients. When the training study samples were tested using the prototype kit, all 67 of 67 patients with APAP were positive, whereas none were positive in 72 healthy subjects. Among the patient samples, the five sera with the lowest GMAb concentrations (3.50, 5.72, 7.52, 8.10, and 8.63 U/mL) were selected and the test line was evaluated independently by 5 examiners (Figure  2 ). As shown in Figure  2 , all five examiners judged positive for all 5 sera. In contrast, six serum samples from healthy individuals with the highest GMAbs concentrations (0.36, 0.37, 0.40, 0.44, 0.46, and 0.71 U/mL) were selected and applied to the prototype kit. All five examiners judged all six samples to be negative. Thus, the LDL of 3.5 U/mL for the ICT prototype was confirmed to be appropriate based on the training study results (Figure  S2 ). However, the intensity of the visible test line did not necessarily exhibit a linear relationship with the serum GMAb concentrations, as shown in Figure  2 . This seemed to suggest that serum components may interfere with the reaction between GMAb and immobilized GM‐CSF. Estimation of the lower detection limit for anti‐GM‐CSF autoantibody measurement using serum samples from six healthy subjects and five APAP patients. The serum in the six healthy subjects ranged from 0.36 to 0.71 U/mL, and all five examiners judged the test line to be negative. The serum anti‐GM‐CSF autoantibody concentrations in the five patients ranged from 3.50 to 8.63 U/mL, and all five examiners judged the test line to be positive. Of the 420 validation study samples reported previously [ 13 ], 382 were used the present study. Of these, the numbers of samples in the present (numbers in parentheses are those in the previous study [ 13 ]) study from patients with APAP, SPAP, HPAP, and other lung diseases other than PAP were 211 (213), 38 (40), 5 (5), and 128 (162), respectively (Table  2 ). The median (IQR) concentration of GMAb in the serum was 66.8 (29.6–125.8), 0.12 (0.08–0.17), 0.10 (0.07–0.17), and 0.12 (0.09–0.20) U/mL, respectively. Samples from other lung diseases included 46, 16, 36, 6, 29, 13, and 17 serum samples from patients with connective tissue disease (CTD), drug‐induced pulmonary disease (DIPD), idiopathic interstitial pneumonias (IIPs), idiopathic pulmonary fibrosis (IPF), other IIPs, infectious diseases, and miscellaneous lung diseases, respectively. All 211 sera from patients with APAP tested positive for GMAb using the prototype kit. Figure  3 shows the positive line of 4 samples with the lowest GMAb concentration ranging from 5.59 to 6.47 U/mL, all of which were visually judged as positive by all five examiners. Again, the color intensity of the positive test line did not appear to correlate linearly with GMAb concentration. Figure  3 also shows the results of applying eight samples with high GMAb concentrations ranging from 0.71 to 2.85 U/mL from diseases other than APAP to the prototype kit, and all of them were negative for GMAb. Namely, of the five examiners, all five gave negative visual results. Importantly, four samples with GMAb concentrations of 1.90, 2.56, and 2.85 U/mL above the ELISA cutoff value of 1.65 U/mL were negative. Specifically, all five examiners recorded negative visual results. All these negative ICT results had GMAbs concentrations below the LDL at 3.50 U/mL, confirming the LDL value was appropriate. Appearance of the test line when 12 validation study samples with anti‐GM‐CSF autoantibody concentrations near the ELISA cutoff value (1.65 U/mL) were applied to the ICT. All five examiners judged the serum from APAP patients with concentrations of 5.59–6.47 U/mL to be positive. Eight non‐APAP serum samples, including those with concentrations of 1.9, 2.56, 2.85, and 2.85 U/mL, tested negative. Of the 217 samples that were 1.65 U/mL or higher by ELISA, 213 were positive using the ICT prototype kit, yielding a positive concordance rate of 98.16%. Of the 165 samples that were less than 1.65 U/mL by ELISA, all 165 tested negative by the ICT prototype kit, yielding a negative concordance rate of 100%. Overall, the concordance rate between the two methods was 98.95%. The four samples with discordant results were derived from one SPAP patient (1.90 U/mL), two IIP patients (2.56 and 2.85 U/mL), and one CTD patient (2.85 U/mL) (Table  S1 ). Of the 382 validation study samples, 171 samples were serum samples from patients with lung disease other than APAP. Two of those with GMAb at 48.57 and 36.24 U/mL showed positive test lines of the prototype kit. One was from a patient with CTD and another was from a patient with IIP. When the clinical performance of this kit was evaluated based on the results of applying the validation study sample to the prototype kit, all 211 of APAP samples were positive, giving a sensitivity of 100%. Furthermore, among the 171 samples other than APAP, 169 samples were negative, so the specificity was 98.83% (Table  S2 ). Of note, of those, there were 2 samples that had a cutoff value of 1.65 U/mL or higher in the ELISA method, but all 5 examiners judged the test line negative using the prototype kit. These two samples were from a SPAP patient (1.90 U/mL) and a CTD patient (2.85 U/mL). The accuracy of the ICT was also evaluated based on the correlation between the GMAbs concentration measured by ELISA and the antibody titre as the endpoint dilution titre. A scatter plot of the correlation between the antibody titres of each of the 50 serum samples randomly selected from validation study samples with APAP and other lung diseases and GMAbs concentration determined by the conventional ELISA method is shown in Figure  4 . There was a strong correlation in all samples (ρ = 0.925, p  < 0.001) (Figure  4A ). There was a tendency for high‐concentration samples to deviate from the regression line, but most of the titres for samples that deviated were within a single dilution step (two‐fold) from the line. However, the correlation was weaker for samples with GMAb concentrations below 60 U/mL (ρ = 0.670, p  < 0.001) (Figure  4B ). This weak correlation in the low concentration range reaffirmed that the accuracy of ICT decreases at low concentrations. Scatter plot of the correlation between the end point for serial dilution rate of serum samples obtained from 50 patients and 50 controls determined by the immunochromatography, and the antibody concentration derived from the conventional ELISA method. A: All 100 samples, B: Samples with antibody concentration less than 60 U/mL by the ELISA method. There was a strong correlation in all samples (ρ = 0.925, p  < 0.001) (A), but the correlation was weaker for samples with GMAbs concentrations below 60 U/mL (ρ = 0.670, p < 0.001) (B).

In this study, we developed a prototype of ICT for detecting GMAbs, and a clinical performance test was conducted using the validation study samples previously analyzed in the ELISA development study [ 13 ]. By setting the test line intensityto approximate the cutoff value of conventional ELISA, we succeeded in developing a prototype ICT that could quickly perform serum diagnosis in clinical settings. To date, no studies have reported the use of ICT for serum diagnosis of APAP. Commercialization of this newly developed kit is expected to facilitate rapid diagnosis in clinical practice. ICT that can rapidly determine the presence or absence of autoantibodies in a short time by simply dropping serum and developing solution is extremely useful in clinical settings. However, because the determination is done visually, there is uncertainty that arises near the LDL. As the LDL value increases, the specificity inevitably decreases. Using a highly purified polyclonal antibody derived from an APAP patient, the ICT was adjusted so that the detection limit corresponded to the ELISA cutoff value of 1.65 U/mL, but when actual serum was tested, GMAb concentrations between 1.65 and 3.5 U/mL could not be detected. Based on a previous report [ 13 ], the estimated proportion of APAP cases falling within the grey zone (1.65–3.5 U/mL) is 0%–3.8%. This highlights a potential limitation of ICT in rare borderline cases, although the overall diagnostic impact is minimal. Although the purified polyclonal antibodies diluted with PBS could be detected to 3.12 U/mL, GMAbs in serum may have difficulty binding to immobilized GM‐CSF due to various interfering substances contained in serum. Lipemia may interfere with the antigen–antibody precipitation, leading to unrealistically high results and may also interfere in some immunoassays since lipoproteins may interfere with the antigen–antibody reaction by blocking the binding sites of antibodies [ 21 , 22 , 23 ]. Serum proteins, especially autoantibodies such as rheumatoid factor and human anti‐mouse antibodies, can interfere with immunochromatographic detection of autoantibodies, leading to false‐positive and false‐negative results [ 24 , 25 ]. Furthermore, even if GMAbs in serum were to bind to immobilized GM‐CSF, nonspecific IgG bound to colloidal gold anti‐IgG antibodies attached to the nitrocellulose membrane surrounding immobilized GM‐CSF would make it impossible to determine whether the line is positive or not. ICT provides rapid results and exhibits high sensitivity and specificity for some antigens, such as the detection of HBs antigen [ 26 ]. The ELISA method has been shown to be particularly reliable in detecting anti‐neutrophil cytoplasm antibodies [ 27 ]. However, performance in detecting autoantibodies may vary depending on the isotype of the antibody. ELISA and ICT for detecting GMAbs both use the principle of capturing autoantibodies with a labelled secondary antibody and capturing the autoantibodies with immobilized GM‐CSF, but the detection sensitivity and specificity were different between the two. The isotype of the antibody of interest must be considered depending on clinical requirements. It is also expected that using both together will improve diagnostic accuracy. In particular, the grey zone revealed in this study, namely, samples with GMAb concentrations of 1.65–3.5 U/mL, cannot be diagnosed without using ELISA. For cases in which PAP is strongly suspected clinically, but ICT is negative, it is recommended to apply ELISA for confirmation. Moreover, when evaluating the progression or therapeutic response in which GMAb concentrations fluctuate, there will likely require quantitative evaluation. As highlighted in this study, regardless of the performance of ICT, the specificity of APAP serological diagnosis is under 100%. This is because serum GMAbs levels are elevated in some patients with diseases other than APAP. Even if the study was limited to lung diseases in which GGO was observed on HRCT, GMAbs levels were elevated in one patient with CTD and one patient with IIP. Broadening the discussion of GMAbs measurement beyond respiratory diseases increases in GMAbs have been reported in many more types of diseases. Nylund reported that Crohn's disease (CD) patients with high GMAbs levels had higher intestinal permeability than CD patients with low GM‐CSF Ab levels, even in the absence of differences in systemic or intestinal inflammation [ 28 ]. Salvator et al. reported that the plasma dilution (IC50) that caused inhibition of GM‐CSF‐induced STAT5 phosphorylation was similar among serum samples in PAP, cryptococcosis, and nocardiosis patients [ 12 ]. Toullec et al. reported that serum from endometriosis patients contained high levels of GMAbs, which correlated with the severity and number of lesions [ 29 , 30 ]. The ICT platform may potentially be adapted for serological detection of autoantibodies in other immune‐mediated diseases; however, further disease‐specific validation will be required. In conclusion, we successfully developed and validated an ICT kit for detecting GMAbs in APAP. This assay demonstrates high sensitivity and specificity and is suitable for point‐of‐care use. While ELISA remains essential in ambiguous cases, this ICT will likely enhance diagnostic efficiency and accessibility.

Pulmonary alveolar proteinosis (PAP) is a rare lung disease characterised by excessive accumulation of waste materials composed of phospholipids, cholesterol, and proteins in the alveoli, leading to respiratory failure due to impaired gas exchange [ 1 , 2 , 3 ]. PAP is classified into autoimmune pulmonary alveolar proteinosis (APAP), secondary pulmonary alveolar proteinosis (SPAP), and congenital/hereditary pulmonary alveolar proteinosis (CPAP/HPAP). APAP accounts for approximately 92% of adult PAP, SPAP accounts for 7%–9%, and CPAP/HPAP accounts for less than 1% [ 1 , 4 ]. GMAbs are polyclonal antibodies with high avidity [ 5 , 6 ] present in high concentrations in the lungs of APAP patients. As such, GMAbs in APAP exhibited strong neutralizing ability [ 6 , 7 ]. These properties also suggested that GMAbs neutralize the biological activity of GM‐CSF produced in the patient's lungs [ 1 , 4 , 6 ]. Thus, since GMAbs in APAP are highly specific, detection that reveals their presence in serum is crucial for the diagnosis. Shortly after discovering the presence of GMAbs in the patient's lungs, we also confirmed that they were also present at high concentrations in the serum [ 5 ]. Therefore, we demonstrated that serological diagnosis was possible by measuring GMAb concentrations [ 8 ]. Later, the detection of serum GMAbs became an essential test for the serological diagnosis of APAP [ 4 , 9 ]. However, GMAbs are present in trace amounts in the serum of healthy individuals [ 10 ], and low concentrations of autoantibodies have been detected in rare cases such as pneumoconiosis, sarcoidosis, hypersensitivity pneumonitis [ 11 ], cryptococcal meningitis, and severe nocardiosis [ 12 ]. Therefore, to distinguish APAP from healthy subjects and other lung diseases, it is important to establish a cutoff value separating the serum GMAb concentration of healthy subjects and APAP patients. When using the ELISA kit, this value is affected by the slope of the calibration curve using the standard antibody and also by the binding of nonspecific IgG other than GMAbs to GM‐CSF coated on the plate [ 13 ]. The lower this cutoff value is, the lower the false‐positive rate for healthy sera and also the lower the false negative rate for APAP serum, and the cutoff value can be lowered by reducing this nonspecific binding as much as possible [ 14 , 15 ]. In the newly developed ELISA reported in 2020, we succeeded in minimising nonspecific serum IgG binding to the coated GM‐CSF as much as possible by using 33‐8F‐H as the standard antibody and improving the coating GM‐CSF and blocking reagent [ 13 ]. Based on the dissociation in the distribution of antibody concentrations between the patients and healthy subjects, the cutoff value, which is the boundary between the two, could be calculated using a logistic model to be 1.65 U/mL. Although the ELISA method allows quantitative measurement and can measure many samples at once, it is time‐consuming and requires specialised measurement equipment such as an absorbance reader, so it is not suitable for rapid diagnosis in clinical settings. For this reason, the practical application of a simple immunochromatography test (ICT) has been eagerly awaited. Such antibody detection systems by ICT are widely applied to diagnose autoimmune diseases such as celiac disease and thalassemia and to diagnose past infections in the field of infectious diseases. Specifically, the systems can detect anti‐tissue transglutaminase antibodies associated with celiac disease, an autoimmune disease induced by wheat gluten and related proteins [ 16 ]. For thalassemia, the presence of Hb Bart's antibodies is used to diagnose the disease [ 17 ]. Furthermore, for many infectious diseases, by detecting antibodies such as IgG and IgM that are specific to the infectious disease, it is possible to determine whether there is a history of infection or whether there are antibodies in the body, and whether or not the person has been infected with the relevant infectious disease [ 18 , 19 , 20 ]. ICT has been applied to detect autoantibodies in various diseases, but it is necessary to evaluate the sensitivity and specificity of the kit for each disease to determine whether serological diagnosis is possible through qualitative judgement. In 2024, GM‐CSF inhalation therapy was approved in Japan ahead of the rest of the world, and medical treatment has begun under national insurance system in combination with the Designated Incurable Disease Medical Expense Subsidy System. At the same time, there was a need to develop test and diagnostic kits that would be covered by the insurance. Concurrently, we developed an ICT to detect GMAbs, which was approved for insurance coverage. By using a colloidal gold‐labelled anti‐human IgG antibody as a capture antibody and immobilizing GM‐CSF, we can detect GMAbs contained in serum with sensitivity and specificity comparable to the conventional ELISA method using serum samples from APAP/other lung diseases.

Chika Narita, Jun‐ichiro Sekiguchi, Yasuyuki Ozeki, Makoto Akiwa, and Shohei Koike are employees of Kojin Bio. Co. Ltd., which sponsored this research. Keiichi Akasaka has received payment for expert testimony from Chugai Pharmaceutical Co. Ltd. unrelated to the present manuscript, and support for attending meetings/travel from GlaxoSmithKline Pharmaceuticals Ltd. unrelated to the present manuscript. Sachiko Yuki, Kenjiro Shima, and Nobutaka Kitamura have no conflicts of interest to declare. Koh Nakata has received funding from Nobelpharma and consulting fees from Nobelpharma Co. Ltd. unrelated to the present manuscript. Toshiaki Kikuchi is Editor‐in‐Chief of the journal and co‐author of this article. He was excluded from the peer‐review process and all editorial decisions related to the acceptance and publication of this article. Peer review was handled independently by Co‐Editor in Chief Paul Reynolds and acting co‐Editor in Chief David Lam to minimise bias.

Data S1: resp70137‐sup‐0001‐Supinfo.docx.