Immunogenicity and safety of homologous and heterologous ChAdOx1-S and mRNA-1273 vaccinations in healthy adults in Taiwan.

BACKGROUND: In Taiwan, the vaccination program started in March 2021, with ChAdOx1-S being the first available WHO-approved COVID-19 vaccine, followed by Moderna vaccine. This study aimed to investigate the immunogenicity and safety of homologous and heterologous prime-boost regimens with ChAdOx1-S and mRNA-1273.
METHODS: From March to November 2021, homologous or heterologous regimens with ChAdOx1-S and mRNA-1273 vaccination (ChAdOx1-S/ChAdOx1-S, mRNA-1273/mRNA-1273, ChAdOx1-S/mRNA-1273) were given to 945 healthy participants. Serum samples were collected at designated time points. The anti-RBD/S1 antibody titers and neutralizing ability were measured by three different immunoassays: Elecsys® Anti-SARS-CoV-2 S (Roche Diagnostics, Mannheim, Germany), AdviseDx SARS-CoV-2 IgG II (Abbott Diagnostics Division, Sligo, Ireland), and cPass™ SARS-CoV-2 Neutralization Antibody Detection Kit (GenScript, New Jersey, USA).
RESULTS: We found that heterologous vaccination with ChAdOx1-S/mRNA-1273 had an acceptable safety profile and induced higher total anti-RBD/S1 antibody production (p < 0.0001), yet lower anti-RBD/S1 IgG titer (p < 0.0001) and neutralizing ability (p = 0.0101) than mRNA-1273/mRNA-1273 group. Both regimens showed higher antibody titers and superior neutralizing abilities than ChAdOx1-S/ChAdOx1-S. An age-dependent antibody response to ChAdOx1-S/mRNA-1273 was shown after both the priming and the booster doses. Younger age was associated with higher antibody production and neutralizing ability.
CONCLUSIONS: Heterologous ChAdOx1-S/mRNA-1273 vaccination regimen is generally safe and induces a robust humoral immune response that is non-inferior to that of mRNA-1273/mRNA-1273.

Introduction

The devastating coronavirus disease 2019 (COVID-19) outbreak, caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is a global threat that has resulted in over 5000,000 mortalities [1,2]. The capricious nature of the virus, with increasing numbers of variants, posed great challenges to the healthcare system and caused detrimental effects to economy and social life [3,4]. To overcome the pandemic, vaccines were developed and manufactured using novel techniques. At the time of writing this article, three World Health Organization (WHO)-approved vaccines were available in Taiwan, including ChAdOx1-S (AstraZeneca, Oxford, UK), mRNA-1273 (ModeRNA, Cambridge, MA USA, hereafter referred to as mRNA-1273 or Moderna vaccine), and BNT162b2 (BioNTech Manufacturing, ComiRNAty, BioNTech, Mainz, Germany, hereafter referred to as BNT vaccine) [5].

As an area with low COVID-19 prevalence, vaccine acceptance was relatively low at the beginning because of vaccine hesitancy [6], [7], [8]. However, an outbreak in mid-2021 urged the need for immunization. Heterologous vaccination program was thus been considered with possible benefits of supply chain flexibility and avoidance of serious adverse effects [9], [10], [11], [12]. In this study, we aimed to evaluate the immunogenicity and safety of heterologous vaccines with the prime-boost sequence of ChAdOx1-S/mRNA-1273, which were compared with those of homologous regimens of ChAdOx1-S/ChAdOx1-S and mRNA-1273/mRNA-1273. Both ChAdOx1-S and mRNA-1273 have well-established safety profile data and evidence of clinical efficacy [13,14]. Multiple platforms including several antibody-detecting immunoassays, one competitive enzyme-linked immunosorbent assay (ELISA) and one neutralizing assay were used to evaluate humoral immune responses.

Methods

Study design and enrollment of participants

This was a prospective and multiple-center study. We enrolled healthy participants in this study conducted at the National Taiwan University (NTUH), China Medical University Hospital (CMUH), and National Cheng Kung University Hospital (NCKUH) from May 2021 (Fig. 1 ). Each participant was subjected to venipuncture up to three times: right before the first dose (Day 0, immediately before the second dose (V2), and four weeks after the second dose (V4) (Fig. 2 ). The participants were categorized into one of the following three vaccination programs depending on the availability of vaccines at the time of enrollment: ChAdOx1-S/ChAdOx1-S, with an eight-week interval between doses; mRNA-1273/mRNA-1273, with a four-week interval; or a heterologous prime-boost combination of ChAdOx1-S/mRNA-1273, with an eight-week interval.

Fig. 1

Flow of participants and volunteers through the study.

Fig. 2

Timeline of blood sampling.

Twenty volunteers formed the sentinel group to evaluate the performance of immunoassays and weekly changes of antibody titers. Seven of them received ChAdOx1-S/ChAdOx1-S vaccination and thirteen received mRNA-1273/mRNA-1273. Volunteers in the sentinel study were subjected to weekly venipuncture from Day 0 to V4 (Fig. 2). Patient characteristics were collected on the day of enrollment (Table 2). The study was approved by the institutional review board of the NTUH (202101064RINB), CMUH (CMUH110-REC1–090), and NCKUH (A-BR-110–029), and informed consent was obtained from all participants.

Table 2

Basic demographic data of all participants.

Regimen	ChAdOx1-S/ChAdOx1-S	mRNA-273/mRNA-1273	ChAdOx1-S/mRNA-1273
Sample size (n)	225	353	367
Female (%)	56.89	47.59	53.13
Age (years), mean (95% CI)	49.52 (47.25–51.79)	53.71 (52.16–55.26)	52.04 (50.46–53.62)

CI, confidence interval.

Safety evaluation

All study participants were asked to complete an online health questionnaire to report local and systemic adverse events within seven days after the first and second dose. The participants were allowed to report severe or unbearable symptoms directly to the research members at any time during the study, with timely response from the medical professionals.

Immunoassays and neutralization assays

All blood samples were collected using anti-coagulant-free serum-separating blood tubes. After venipuncture, the tubes were centrifuged, and collected sera were stored at no higher than −20 °C if not tested immediately. Seven automated immunoassays, a competitive enzyme-linked immunosorbent assay (ELISA) and a microneutralization assay were used for measure antibody responses. Details of the assays are presented in Table 1 and the Supplementary Materials.

Table 1

The immunoassays used for evaluating the antibody response after the first dose of the ChAdOx1 COVID-19 vaccine.

Kit	Elecsys Anti-SARS-CoV-2 S	ACCESS SARS-CoV-2 II IgG		ACCESS SARS-CoV-2 IgG 1st IS	ADVIA Centaur® SARS-CoV-2 IgG (sCOVG) assay	AdviseDX SARS-CoV-2 IgG II	EliA SARS-CoV-2-Sp1 IgG P2 Research		EliASARS-CoV-2-Sp1 IgM P2 Research
Company (city, country)	Roche Diagnostics GmbH (Mannheim, Germany)	Beckman Coulter Diagnostics, Inc. (Brea, USA)			Siemens Healthcare Diagnostics Inc. (Tarrytown, USA)	Abbott IrelandDiagnostics Division (Sligo, Ireland)	Thermo Fisher Scientific, Inc.(MA, USA)
Targeting antibody	high affinity antibodies (IgG included)	IgG			IgG	IgG	IgG		IgM
Immunoassay	ECLIA	CLIA			CLIA	CMIA	FEIA
Analyzer	Cobas e411, e601 & e602	Access 2 Immunoassay System analyzer			Atellica® IM Analyzer	Architect i system	Phadia 250
Protein targeting	Recombinant RBD of S1 protein	Recombinant RBD of S1 protein			Biotinylated S1 RBD antigen	Purified SARS-CoV-2 recombinant antigen	Recombinant S1 protein
Specimen;amount required	Serum or plasma20 μL	Serum or plasma20 μL			Serum or plasma40 μL	Serum or plasma25 μL	Serum or plasma20 μL
Unit conversion	1 U/ml = 0.972 BAU/ml	1 IU/ml = 1 BAU/ml			1 index = 21.8 BAU/ml	1 AU/ml = 0.142 BAU/ml	NA
Positive result cutoffs and units	≥ 0.80 U/mL	≥ 10 AU/mL	≥ 10 IU/mL		≥ 1.00 Index (U/ml)	≥ 50.0 AU/mL	> 10 U/ml(7–10 equivocal)
Testing time	18 min	NA			NA	NA	1 min interval after first test
Reported best sensitivity/PPA (95% C.I.)	98.8%(98.1–99.3%)	98.9%(92.7–100%)	100% (91.4–100%)		96.41%(92.74–98.54%)	Functional sensitivity.	100%(85.8–100%)	NA
Timing of best sensitivity	–	15–60 days	≥ 14 days		≥ 21 days	does not apply	> 8 days	–
Reported best specificity/NPA (95% C.I.)	99.98%(99.91–100%)	99.9%(99.5–100%)	99.8%(99.4–99.9%)		99.90%(99.64–99.99%)	Not addressed	100%(99.5–100%)	NA
Confirmed cross-reactivity
Non-coronaviruses	None	None	None		None	None	None	None
Other seasonal coronaviruses	No cross reactivity with antibodies to MERS-CoV and Common Coronavirus panel	May cause positive results	No cross reactivity to 229E, NL63, HKU1 or OC43		No cross reactivity with antibodies against 29 human coronaviruses	No cross reactivity with antibodies against 229E, HKU1, NL63, OC43	No cross reactivity with antibodies against 229E, HKU1, NL63, OC43	NA

BAU, binding antibody units; CLIA, chemiluminescent immunoassay; CMIA, chemiluminescent microparticle immunoassay; ECLIA, electrochemiluminescence immunoassay; FEIA, fluorescence enzyme immunoassay; LFA, lateral flow assay; MERS, Middle East respiratory syndrome; NPA, negative percent agreement; NTD, N-terminal domain; PPA, positive percent agreement; RBD, receptor-binding domain.

Statistical analysis

All data were analyzed using STATA software. Nonparametric tests (Mann–Whitney test and Kruskal–Wallis test) were used to compare antibody levels between different groups. Correlation analysis between assays was done with Pearson's correlation, and statistical significance was set at p-values less than 0.05. Univariate logistic regression was used for the analysis between quantitative and semi-quantitative measures and plotting of receiver operating characteristic (ROC) curves.

Results

Participants and samples

From March to October 2021, blood samples were obtained from 20 volunteers in the sentinel study (7 for ChAdOx1-S/ChAdOx1-S and 13 for mRNA-1273/mRNA-1273), and 945 participants (225 for ChAdOx1-S/ChAdOx1-S, 353 for mRNA-1273/mRNA-1273, and 367 for ChAdOx1-S/mRNA-1273). The mean age of mRNA-1273/mRNA-1273 group was significantly higher than that of ChAdOx1-S/ChAdOx1-S group, mainly because more elderly individuals received mRNA-1273 vaccination under the government's priority policy. There was no age difference between other groups.

Approximately 95.43% of all participants experienced mild to moderate adverse events within seven days after vaccination. The most reported adverse events include tenderness at the injection site, fatigue, muscle soreness, fever, and headache (Fig. 3 ). The event rates for the prime/booster dose were 84.82%/71.65% for ChAdOx1-S/ChAdOx1-S, 82.75%/89.34% for mRNA-1273/mRNA-1273, and 70.46%/98.54% for ChAdOx1-S /mRNA-1273 group. No serious or life-threatening adverse events were reported in the present study.

Fig. 3

Percentage of reported adverse events in the seven days following each dose of vaccine injection. (A) ChAdOx1-S/ChAdOx1-S, (B) mRNA-1273/mRNA-1273, (C) ChAdOx1-S/mRNA-1273.

Sentinel study

The weekly changes of antibody titers from 20 volunteers are shown in Fig. 4 (ChAdOx1-S/ChAdOx1-S) and Fig. 5 (mRNA-1273/mRNA-1273). Almost all volunteers had an exponential surge in anti-S1/RBD antibody titers after each vaccination, followed by a gradual decrease in the slope. The only exception was the Thermo Fisher IgM assay, in which the levels of anti-RBD/S1 IgM antibodies increased and declined immediately after the first dose of ChAdOx1-S for most volunteers. Most samples taken after vaccination showed a positive neutralizing ability using the GenScript cPass™ assay (≥ 30% inhibition), while the mRNA-1273/mRNA-1273 regimen seemed to induce stronger neutralization effect than ChAdOx1-S/ChAdOx1-S.

Fig. 4

Individual trends in anti-RBD/S1 antibody levels in 7 sentinel study volunteers receiving ChAdOx1-S/ChAdOx1-S regimen, plotted with logarithmic vertical axis. The vertical dotted line denotes the duration of the second vaccination. The horizontal dashed line with gray rectangle shade indicates the cutoff value of each assay according to the respective package inserts. (A) Roche Elecsys® Anti-SARS-CoV-2 S, (B) Beckman Coulter ACCESS SARS-CoV-2 II IgG, (C) Beckman Coulter ACCESS SARS-CoV-2 IgG 1st IS, (D) Siemens ADVIA Centaur® SARS-CoV-2 IgG (sCOVG) assay; (E) Abbott AdviseDX SARS-CoV-2 IgG II; (F) Thermo Fisher EliA SARS-CoV-2-Sp1 IgG P2 Research, (G) Thermo Fisher EliA SARS-CoV-2-Sp1 IgM P2 Research.

Fig. 5

Individual trends in anti-RBD/S1 antibodies in 13 sentinel study volunteers receiving mRNA-1273/mRNA-1273 regimen, plotted with logarithmic vertical axis. The vertical dotted line denotes the duration of the second vaccination. The horizontal dashed line with gray rectangle shade indicates the cutoff value of each assay according to the respective package inserts. (A) Roche Elecsys® Anti-SARS-CoV-2 S, (B) Beckman Coulter ACCESS SARS-CoV-2 II IgG, (C) Beckman Coulter ACCESS SARS-CoV-2 IgG 1st IS, (D) Siemens ADVIA Centaur® SARS-CoV-2 IgG (sCOVG) assay, (E) Abbott AdviseDX SARS-CoV-2 IgG II.

Correlation analysis

The correlation of measures between different immunoassays is shown in Supplemental Fig. 1. The correlation between IgG-measuring assays (Abbott, both Beckman assays, Siemens, and Thermo Fisher) was good, with coefficients higher than 0.7. The correlation between the Roche kit and the other kits varied. On the other hand, the correlation coefficients between TCID50 and other assays were generally low, with the GeneScript cPass™ assay being the highest (R = 0.6358). The results are shown in the Supplementary Materials. Univariate logistic regression and ROC curves were plotted using GeneScript cPass™ as the reference method (cutoff: 30%). Most of the immunoassays had optimal predictive values (area under the curve (AUC) > 0.9), except for the Thermo Fisher IgM assay (Supplemental Fig. 2). Based on the performance and the accessibility for operation of different assays, three assays were chosen for completing subsequent analysis, including the Roche, the Abbott and the GeneScript cPass™ assays.

Screening for previously occult SARS-CoV-2 infection

All samples collected from participants at Day 0 and V2 visits were negative for anti-N antibody (< 1.0 cutoff index [COI]), except the samples from two participants. The samples were consistent positive at low levels at both Day 0 and V2. The results were judged to be false positive reactions because both participants had no detectable anti-S protein antibody at Day 0 [15,16]. These findings indicated no evidence of occult and unidentified SARS-CoV-2 infections among the study participants before entering the study and before V2.

Antibody responses of different vaccine regimens

The analysis of antibody response was done using three assays: the Roche and the Abbott assays for measuring antibody titers and GeneScript cPass™ for evaluation of neutralizing ability. For Roche assay, mRNA-1273 generated significantly higher titers of antibodies at V2 than those receiving ChAdOx1-S as first dose. At V4, the heterologous vaccines (ChAdOx1-S/mRNA-1273) generated the highest titer, followed by mRNA-1273/mRNA-1273 with ChAdOx1-S/ChAdOx1-S being the lowest of the three (Fig. 6 ). The results of V2 samples measured by the Abbott assay showed similar trends as those by Roche assay. However, for V4 samples, mRNA-1273/mRNA-1273 vaccination generated the highest anti-RBD/S1 IgG titers, followed by ChAdOx1-S/mRNA-1273 group, and the difference was statistically significant. The antibodies generated by mRNA-1273 vaccination seemed to have a dominant effect on the neutralizing ability, compared with that by ChAdOx1-S vaccination.

Fig. 6

Antibody titers of all participants and volunteers with different vaccination regimens. The whiskers denote the median (long) and the first and third interquadrant (short) values of all measurements. The analysis was done using Mann-Whitney U test. The vertical axes of Fig. 4(A) and Fig. 4(B) are scaled in a logarithmic manner. (A) Measurements by Roche Elecsys® Anti-SARS-CoV-2 S, (B) Measurements by Abbott AdviseDX SARS-CoV-2 IgG II, (C) Measurements by GenScript cPass™ SARS-CoV-2 Neutralization Antibody Detection Kit.

Antibody response in different age groups

The participants were divided into three age groups: 20–40 years old, >40–60 years old, and >60 years. For V2 samples, antibody production and neutralizing ability were inversely correlated with age (p-values < 0.05 for all regimens), implying a better immune response in younger people after the first dose of vaccine. For V4, however, the samples of ChAdOx1-S/ChAdOx1-S group showed an opposite trend compared with that of V2 by anti-RBD/S1 antibody measured by the Roche assay. In another word, older age generated higher serum levels of anti-RBD/S1 antibody after booster dose (p = 0.0129). However, this V4 trend was not seen in the Abbott assay (p = 0.4447), nor in the GeneScript cPass™ assay (p = 0.5969). As for the V4 samples from the mRNA-1273/mRNA-1273 group, no age-related variation was found in terms of antibody production or neutralizing ability. In contrast, the V4 samples of ChAdOx1-S/mRNA-1273 group showed an inverse correlation of antibody levels to age using all three assays (p = 0.0112, 0.0228, and 0.0264 for the Roche, Abbot, and GeneScript cPass™ assays, respectively) (Fig. 7 ).

Fig. 7

Age-dependent anti-RBD/S1 antibody titers of different vaccination groups, measured by three different immunoassays (Roche Elecsys® Anti-SARS-CoV-2 S, Abbott AdviseDX SARS-CoV-2 IgG II, GenScript cPass™ SARS-CoV-2 Neutralization Antibody Detection Kit). The whiskers denote the median (long) and the first and third interquadrant (short) values of all measurements. The analysis was done using Kruskal–Wallis test. Fig. 5(A) to Fig. 5(F) were plotted using logarithmic vertical axes. (A)(B)(C) Antibody titers of different age groups with ChAdOx1-S/ChAdOx1-S regimen, measured by Roche, Abbot, and GenScript assays, respectively. Antibody titers of different age groups with mRNA-1273/mRNA-1273 regimen, measured by Roche (D), Abbot (E), and GenScript (F) assay, respectively. Antibody titers of different age groups with ChAdOx1-S/mRNA-1273 regimen, measured by Roche (G), Abbot (H), and GenScript (I) assays, respectively.

Discussion

In this study, we evaluated the antibody responses and safety profiles of the heterologous vaccine regimen of ChAdOx1-S/mRNA-1273 in addition to ChAdOx1-S/ChAdOx1-S and mRNA-1273/mRNA-1273 homologous regimens in Taiwan. Three immunoassays were selected for the evaluation based on the sentinel study results: the Roche, Abbott, and GenScript cPass™ assays. We found that the heterologous regimen ChAdOx1-S/mRNA-1273 regimen was safe, with no reported serious adverse effects, and elicited a robust antibody response.

The first proposed heterologous regimen, with the prime-boost combination of ChAdOx1-S/BNT vaccine, was assessed by Borobia et al. in a randomized controlled trial with 676 participants [12]. Several studies have been conducted in different countries to investigate the safety and immunological profile of heterologous vaccines against COVID-19 since then [17,18]. The combination of ChAdOx1-S/mRNA-1273, while less studied than ChAdOx1-S/BNT, has also demonstrated good immunogenicity in a Denmark-based observational study and a Swedish study [19,20]. Safety profile of the ChAdOx1-S/mRNA-1273 regimen was evaluated in a German study of 96 healthy participants. The reported rates of adverse events after ChAdOx1-S/mRNA-1273 was comparable to that after mRNA-1273/mRNA-1273, and was higher than that after ChAdOx1-S/ChAdOx1-S [21]. Our study also shows similar findings in that all reported adverse effects were mild or moderate [16, 22]. Although being lack of protective efficacy of effectiveness data, this “miss-match” vaccination strategy has been accepted or officially recommended by a growing number of countries, including Germany, Canada, and Thailand [23].

A comprehensive evaluation of vaccine-induced immune response can be complicated. Both B cell and T cell responses are important for vaccine-induced protection [12, 22, [24], [25], [26]]. Previous studies have shown an association between measured anti-RBD/S1 antibody titers and clinical protective effects against SARS-CoV-2 [27]. However, the measured antibody levels may differ depending on the methods and target antigen used in different immunoassays [28], [29], [30]. In our study, we used three different assays to measure the levels of anti-RBD/S1 antibodies and neutralizing ability following vaccination, which is one of the strengths of this study. One of our findings is that the V4 samples of the ChAdOx1-S/mRNA-1273 group yielded significantly higher levels of anti-RBD/S1 antibodies than homologous regimens using the Roche assay, while in the Abbott assay, the mRNA-1273/mRNA-1273 group showed higher IgG titers than the others. Considering that antigen-specific IgM usually plays a minor role, it is reasonable to infer that the difference is caused mainly by the presence of IgA, which can be detected by the Roche assay but not Abbott [31]. In previous studies, anti-SARS-CoV-2 S IgA antibodies were found in patients with COVID-19 as well as vaccinated individuals, and these antibodies also play an important role in mucosal defense against the disease [32,33]. It is likely that adenovirus-vectored vaccines have a comparable or even superior inducibility of anti-RBD/S1 IgA antibodies compared to mRNA vaccines or vaccines from traditional platforms [12,17,18,[34], [35], [36]]. Further research on the humoral responses to vaccines from different platforms may provide more information for optimizing vaccination strategy.

Another finding of our study is that the immunological response to the first dose is inversely age-dependent for both ChAdOx1-S and mRNA-1273, with higher antibody titers and better neutralization abilities among younger people. After the booster dose, however, the difference was less significant among different ages, except for the ChAdOx1-S/mRNA-1273 group (Fig. 6). This result suggests a crucial role of a booster dose, especially for elderly people, to achieve a robust immunity against SARS-CoV-2 [37], [38], [39]. Interestingly, in the ChAdOx1-S/ChAdOx1-S group, the antibody titers after the booster dose (V4) were higher in older participants when measured by the Roche assay, but not by the Abbott assay (Fig. 6). A possible explanation is that more anti-vector antibodies were produced in the younger age group after the prime dose [40]. The hypothesis that anti-vector immunity after a prime dose of vectored vaccine possibly hinders the efficacy of the subsequent booster dose(s) was implied by the result from earlier trials of ChAdOx1-S in which different dosage and intervals were tested [41], [42], [43]. As repeated vaccination has been advocated for a better protection against COVID-19, the role of anti-vector immunity is an important issue for vaccine development in the future.

One limitation of this study is that humoral immunity against variant strains has not been tested. The vaccines and recombinant antigens used in the immunoassays were developed based on the original strain. The effect of vaccine-induced antibodies toward variants is questionable, especially for variants of concern that have extensive mutation sites, such as the Delta and Omicron variant [44]. A live virus or pseudovirus neutralizing assay is required to obtain such information. The neutralizing assay TCID50 was performed in some of our samples using the Alpha variant strain (B 1.1.7/GRY clade, UK variant) as the target. Despite previous studies showed that antibodies induced by the Wuhan virus or vaccine containing Wuhan strain still retain neutralizing ability against the B 1.1.7 variant, the results in our study somehow showed otherwise [44], [45], [46], [47]. Samples tested with TCID50 demonstrated inadequate neutralizing ability against the B 1.1.7 variant, while the other assays suggested robust neutralizing ability and antibody titers against the original Wuhan variant. These discrepant results require further investigation. However, this indicates that no single test could be representative enough for the evaluation of humoral immunity. A combination of multiple assays may be important to provide unbiased information.

This study is the first one in Taiwan to evaluate the serological response to different regimens of SARS-CoV-2 vaccination. We found that heterologous vaccination with ChAdOx1-S/mRNA-1273 is generally safe, well-tolerated, and induces an antibody response that is non-inferior to that of mRNA-1273/mRNA-1273. Age-dependent response was seen after the prime dose, but the differences were less significant following a booster dose for both the homologous and heterologous regimens. This result suggests that boosting is crucial for a better protection, especially in elderly people.

Funding

This work was supported by the Taiwan Centers for Disease Control and Prevention, Minister of Health and Welfare, (YH110001).

Declaration of Competing Interest

The authors declare no conflict of interest.