Effects of simultaneous administration of local anesthetics on seizure induced threshold: an experimental study in rats

Abstract

Background

The effect of the simultaneous administration of local anesthetics, whether synergistic, additive, or antagonistic, has not yet been established. This study investigated the interaction between concurrently administered short- and long-acting local anesthetics.

Methods

A total of 160 male Sprague–Dawley rats were divided into 16 groups (n = 10 each): 4 local anesthetic alone groups (lidocaine, mepivacaine, ropivacaine, and levobupivacaine) and 12 simultaneous administration groups that received adjusted concentrations of short- and long-acting local anesthetics. Local anesthetics were administered at a constant rate until seizure onset, which was determined by electroencephalographic monitoring. The cumulative dose required to induce seizures was calculated in each group. A linear regression analysis was performed to assess the relationship between the doses of the combined agents, with additivity inferred from an adjusted R² of > 0.7. In addition, pharmacodynamic interactions were evaluated using an interaction index based on the Loewe additivity model.

Results

For all combinations of local anesthetics, R² was > 0.7. A linear regression analysis indicated an additive effect for each combination. Furthermore, an analysis of the interaction index indicated a tendency toward additivity.

Conclusions

This study revealed no clear evidence of antagonism or synergism for the simultaneous administration of short-acting (lidocaine or mepivacaine) and long-acting anesthetics (ropivacaine or levobupivacaine). An additive effect was suggested in terms of the seizure threshold when short- and long-acting local anesthetics were administered simultaneously. This finding may therefore support the safe and effective combined use of local anesthetics in clinical regional anesthesia.

Introduction

Ultrasound-guided peripheral nerve blocks are widely used in the perioperative period. Local anesthetics, such as peripheral nerve blocks and epidural anesthesia, are used for regional anesthesia. Amide local anesthetics are divided into short-acting anesthetics (e.g., lidocaine, mepivacaine) and long-acting anesthetics (e.g., ropivacaine and levobupivacaine). Generally, long-acting local anesthetics are used during regional anesthesia, but sometimes short-acting and long-acting anesthetics are administered simultaneously to achieve rapid onset and prolonged postoperative analgesia [1–3]. However, the simultaneous administration of short- and long-acting local anesthetics has been reported to cause local anesthetic systemic toxicity (LAST) [4–6]. LAST can cause critical organ damage; therefore, it should be avoided. The toxicity of the simultaneous administration of local anesthetics has been investigated in previous animal studies [7–12]. Synergistic [7]additive [8, 9]invariable [10]and antagonistic [11] effects have also been reported. However, these reports did not examine ropivacaine or levobupivacaine, long-acting local anesthetics that are currently in use, or mepivacaine, a short-acting local anesthetic. The interactions of these local anesthetics should also be examined.

In addition to these local anesthetic interactions, lidocaine has also been reported to have an anticonvulsant effect [13–17]. If this effect is also effective against other local anesthetic-induced seizures, it may help to control LAST. In this study, we validated several lidocaine concentrations.

Therefore, we investigated whether the simultaneous administration of short-acting and long-acting local anesthetics, the combination of lidocaine and ropivacaine, lidocaine, and levobupivacaine, mepivacaine, ropivacaine, mepivacaine, and levobupivacaine, could have a synergistic, additive, or antagonistic effect on the seizure-induced threshold.

Materials and methods

The protocol for this experiment was approved by the Animal Experiment Institution Review of Fukushima Medical University (approval number: 2022071; Fukushima, Japan) on April 18, 2022. This study was conducted in accordance with the Animal Experiment Guidelines of Fukushima Medical University and the guidelines of the National Institute of Health Sciences of Japan (Kawasaki, Japan). In this study, all measures were taken to reduce animal suffering.

Rats and grouping

We used 160 male Sprague–Dawley rats (body weight, 265–438 g; age, 8–12 weeks [Japan SLC, Inc., Shizuoka, Japan]). The animals received the same food and water and were kept in standard cages. All experiments were conducted between 8:30 a.m. and 17:00 p.m. The 160 rats were divided into 16 groups, 4 groups with administration of local anesthetic alone, and 12 groups with simultaneous administration of short- and long-acting local anesthetics (n = 10, per group). The grouping and total amount of local anesthetic per mL administered to each group were as follows: group Li, lidocaine (10 mg/mL) alone; group Me, mepivacaine (10 mg/mL) alone; group Ro, ropivacaine (2.5 mg/mL) alone; group Le, levobupivacaine (2.5 mg/mL) alone. The 12 simultaneous administration groups were simultaneously administered a 1:1 mixture of short- and long-acting local anesthetics: group LR, lidocaine and ropivacaine; group LL, lidocaine and levobupivacaine; group MR, mepivacaine and ropivacaine; and group ML, mepivacaine and levobupivacaine. The grouping and total amount of local anesthetics per ml administered each group are listed below: group LRa, lidocaine (10 mg/mL) and ropivacaine (2.5 mg/mL); group LRb, lidocaine (10 mg/mL) and ropivacaine (1.25 mg/mL); group LRc, lidocaine (5 mg/mL) and ropivacaine (2.5 mg/mL); group LLa, lidocaine (10 mg/mL) and levobupivacaine (2.5 mg/mL); group LLb, lidocaine (10 mg/mL) and levobupivacaine (1.25 mg/mL); group LLc, lidocaine (5 mg/mL) and levobupivacaine (2.5 mg/mL); group MRa, mepivacaine (10 mg/mL) and ropivacaine (2.5 mg/mL); group MRb, mepivacaine (10 mg/mL) and ropivacaine (1.25 mg/mL); group MRc, mepivacaine (5 mg/mL) and ropivacaine (2.5 mg/mL); group MLa, mepivacaine (10 mg/mL) and levobupivacaine (2.5 mg/mL); group MLb, mepivacaine (10 mg/mL) and levobupivacaine (1.25 mg/mL); and group MLc, mepivacaine (5 mg/mL) and levobupivacaine (2.5 mg/mL).

Anesthesia and operation

All rats were placed in a plastic box and administered 5% sevoflurane with oxygen. After anesthesia, a 24 G intravenous cannula was inserted and fixed in the tail vein. A drill was used to perforate the skull to avoid damaging the dura mater with 5% sevoflurane for sedation and pain relief. We then inserted custom-made electroencephalogram (EEG) electrodes into the skull and fixed them in place using dental cement. Each rat was subjected to a single experiment and was not reused. EEG electrodes were surgically implanted in all rats and seizure induction was performed once per subject. The EEG electrodes were connected to PowerLab/8sp (ML785; ADInstruments), and EEG signals were recorded using chart5 (ADInstruments). After the operation, anesthesia was terminated, and all rats were awakened. After 60 min, a continuous infusion of local anesthetic was started in each group at a constant rate of 0.4 mL/kg/min using a syringe pump (pump 11 elite; Harvard Apparatus) until seizure onset. Analgesics were not administered after the procedure in this study, considering the possibility that they could affect seizure onset.

Toxic endpoint

The toxic endpoint was seizure onset. With reference to previous studies [9, 12] seizure activity was defined as the occurrence of repeated multiple sharp waves of > 100 µV measured using EEG. We judged that a seizure occurred when an EEG of ≥ 100 µV or higher appeared 1 time per 1.0 s. We defined the time when the initial multiple waves were observed on the EEG as the time at which the seizure started. We recorded the duration of infusion and calculated the cumulative dose of the local anesthetics. After confirming seizures on the EEG, we stopped continuous infusion. Once a seizure was reliably induced, it was considered the humane endpoint of the experiment.

In cases where seizures led to cardiac arrest, animals died during the procedure. All other animals were euthanized immediately after the experiment using carbon dioxide inhalation during the unconscious period due to seizures for pain relief.

Statistical analysis and sample size calculation

Evaluation of model fit with simple linear regression

We plotted the cumulative dose of local anesthetics against the onset of seizures. We analyzed the data using a simple linear regression analysis and created a regression line for each combination of local anesthetics. We examined whether the effects of short-acting and long-acting local anesthetics were additive, synergistic, or antagonistic, based on a simple linear regression analysis [18]. Simple linear regression was performed to examine the dose relationship between short- and long-acting local anesthetics at the time of seizure onset.

The regression model was defined as follows:

where Y is the cumulative dose of the long-acting agent (mg/kg) and X is the cumulative dose of the short-acting agent (mg/kg). “a” is the slope, and “b” is the y intercept. These values were estimated for each drug combination. Groups Li, Ro, LRa, LRb, and LRc were used to analyze the effects of the combination of lidocaine and ropivacaine. Groups Li, Le, LLa, LLb, and LLc were used to analyze the effect of the combination of lidocaine and levobupivacaine. The Me, Ro, MRa, MRb, and MRc groups were used to analyze the combination of mepivacaine and ropivacaine. The Me, Le, MLa, MLb, and MLc groups were used to analyze the combination of mepivacaine and levobupivacaine. The additive effect was determined with a coefficient of determination (R²) of > 0.7. If the R² was ≤ 0.7, the effects were considered synergistic or antagonistic, rather than additive.

Evaluation of additivity using the Loewe additivity model

We evaluated the pharmacodynamic interaction between short- and long-acting local anesthetics using the interaction index [19] based on the Loewe additivity model. The threshold dose of each local anesthetic when administered alone was defined as the standardized value of “1.0” for each drug.

For each combination group, we calculated the fractional dose of each agent by dividing the cumulative dose at seizure onset by the respective single-agent threshold dose. The interaction index (γ) is defined as the sum of the fractional doses:

where 𝐷1 and 𝐷2 are the observed doses in combination, and 𝑇1, 𝑇2 are the single-agent threshold doses. A γ value close to 1.0 indicates an additive interaction, while values below or above suggest synergism or antagonism, respectively.

We also calculated the expected additive cumulative dose for each combination using the Loewe principle (i.e., where the sum of the fractional doses equals 1.0) to allow for comparison with the observed total dose. We adopted the interpretive criteria commonly used for the Chou–Talalay Combination Index (CI) [20] which is conceptually analogous to the Loewe interaction index.

With reference to the relationship between CI values and modes of drug interaction as shown in Table 1, γ values between 0.90 and 1.10 were considered indicative of additive interactions, values below 0.90 suggest synergism, and values above 1.10 suggest antagonism [20].

Table 1.

Relationship between the combination index values and the mode of drug interaction

Range of CI values	Mode of drug interaction
< 0.1	Very strong synergism
0.1–0.3	Strong synergism
0.3–0.7	Synergism
0.7–0.85	Moderate synergism
0.85–0.90	Slight synergism
0.90–1.10	Nearly additive
1.10–1.20	Slight antagonism
1.20–1.45	Moderate antagonism
1.45–3.3	Antagonism
3.3–10	Strong antagonism
> 10	Very strong antagonism

CI combination index

All statistical analyses were conducted using EZR (ver. 1.55, Saitama Medical Center, Jichi Medical University), a graphical user interface for R (ver. 2.7-1, R Foundation for Statistical Computing) [21].

We calculated the sample size for the simple linear regression analysis with a correlation coefficient (r) of 0.8, an α error of 0.05, a power of 0.8, and a two-sided test, which resulted in 10 rats being required for each group.

Results

Toxic endpoint

During the surgical procedure, anesthesia was maintained with 5% sevoflurane with oxygen. No gross movement or withdrawal responses were observed at this concentration, suggesting a sufficiently deep anesthetic plane to minimize procedural pain. We show an example of the EEG activity before the start of local anesthetic administration in Fig. 1A and typical seizure activity on EEG in Fig. 1B. We considered that a seizure occurred when multiple waves appeared, as shown in Fig. 1B. EEG seizure activity was detected in all rats that received each local anesthetic. Table 2. shows the mean time from the start of drug infusion to the onset of seizures in each group.

Fig. 1.

Representative electroencephalographic traces showing the baseline activity before local anesthetic infusion and seizure activity following administration. (A) Baseline activity before local anesthetic infusion. (B) Seizure activity as the occurrence of repeated multiple sharp waves of > 100 µV

Table 2.

The mean duration from local anesthetic administration to seizure onset in each group

Group

Li

Me

Ro

Le

Duration ±SD (minutes)

21.88 ± 1.12

31.63 ± 1.18

30.33 ± 1.82

28.62 ± 0.50

Group

LRa

LRb

LRc

LLa

LLb

LLc

MRa

MRb

MRc

MLa

MLb

MLc

Duration ±SD (minutes)

13.02 ± 0.77

15.08 ± 0.97

17.12 ± 0.93

12.65 n± 0.33

13.68 ± 0.68

18.62 ± 1.23

15.83 ± 0.57

18.43 ± 1.02

20.53 ± 0.70

14.73 ± 0.60

17.88 ± 1.13

18.57 ± 0.52

SD standard deviation

Change in the threshold of local anesthetic toxicity

Evaluation of model fit with simple linear regression

Figure 2 presents simple linear regression lines representing the dose relationship between the two local anesthetics in each combination. These plots demonstrate that increasing the dose of one agent allows a reduction in the dose of the other agent to reach the same endpoint (seizure onset), rather than showing changes in seizure threshold per se. The results for the combination of lidocaine and ropivacaine are shown in Fig. 2A. R² for the combination of lidocaine and ropivacaine was 0.76 (P < 0.001). The results for the combination of lidocaine and levobupivacaine are shown in Fig. 2B. R² for the combination of lidocaine and levobupivacaine was 0.80 (P < 0.001). The results for the combination of mepivacaine and ropivacaine are shown in Fig. 2C. R² for the combination of mepivacaine and ropivacaine was 0.83 (P < 0.001). The results for the combination of mepivacaine and levobupivacaine are shown in Fig. 2D. R² for the combination of mepivacaine and levobupivacaine was 0.86 (P < 0.001). For each local anesthetic combination, the regression line for any combination showed R² > 0.7 (P < 0.001). An additive effect was suggested owing to the fitting of the regression lines for each combination.

Fig. 2.

Regression line for the combinations of local anesthetics. (A) Lidocaine and ropivacaine. (B) Lidocaine and levobupivacaine. (C) Mepivacaine and ropivacaine. (D) Mepivacaine and levobupivacaine. Each data plot represents the combination dose at which seizure occurred in an individual rat.R2: adjusted R-squared

Evaluation of additivity using the Loewe additivity model

The mean cumulative dose of lidocaine alone was 87.54 mg/kg. The cumulative dose of mepivacaine alone was 126.55 mg/kg, ropivacaine alone was 30.33 mg/kg, and levobupivacaine alone was 28.63 mg/kg. The doses were as defined as a standardized value of “1.0” (i.e., 1.0 lidocaine was 87.54 mg/kg, 1.0 mepivacaine was 126.55 mg/kg, 1.0 ropivacaine was 30.33 mg/kg, and 1.0 levobupivacaine was 28.63 mg/kg). The γ for each group are listed in Table 3.

Table 3.

Expected and observed cumulative doses and interaction index (γ) for the combinations of local anesthetics

Group	Expected short LA (mg/kg)	Expected long LA (mg/kg)	Observed short LA (mg/kg) median (IQR)	Observed long LA (mg/kg) median (IQR)	γ median (IQR)
LRa Lidocaine 10 mg/mL Ropivacaine 2.5 mg/mL	43.77	15.17	52.00 (45.15 – 57.13)	13.00 (11.29 – 14.28)	1.02 (0.89 – 1.12)
LRb Lidocaine 10 mg/mL Ropivacaine 1.25 mg/mL	65.66	7.58	64.70 (51.70 – 67.17)	8.09 (6.46 – 8.40)	1.01 (0.80 – 1.04)
LRc Lidocaine 5 mg/mL Ropivacaine 1.25 mg/mL	21.89	22.75	31.82 (31.17 – 35.19)	15.91 (15.58 – 17.60)	0.89 (0.87 – 0.98)
LLa Lidocaine 10 mg/mL Levobupivacaine 2.5 mg/mL	43.77	14.32	51.00 (47.23 – 53.32)	12.75 (11.81 – 13.33)	1.03 (0.95 – 1.08)
LLb Lidocaine 10 mg/mL Levobupivacaine 1.25 mg/mL	65.66	7.16	54.73 (47.55 – 58.38)	6.84 (5.94 – 7.30)	0.86 (0.75 – 0.92)
LLc Lidocaine 5 mg/mL Levobupivacaine 2.5 mg/mL	21.89	21.47	34.32 (32.32 – 40.32)	17.16 (16.16 – 20.16)	0.99 (0.93 – 1.17)
MRa Mepivacaine 10 mg/mL Ropivacaine 2.5 mg/mL	63.28	15.17	60.90 (59.12 – 69.43)	15.23 (14.78 – 17.36)	0.98 (0.95 – 1.12)
MRb Mepivacaine 10 mg/mL Ropivacaine 1.25 mg/mL	94.91	7.58	69.03 (64.52 – 84.78)	8.63 (8.06 – 10.60)	0.83 (0.78 – 1.02)
MRc Mepivacaine 5 mg/mL Ropivacaine 2.5 mg/mL	31.64	22.75	39.28 (38.16 – 43.38)	19.64 (19.08 – 21.69)	0.96 (0.93 – 1.06)
MLa Mepivacaine 10 mg/mL Levobupivacaine 2.5 mg/mL	63.28	14.32	58.93 (54.87 – 61.40)	14.73 (13.72 – 15.35)	0.98 (0.91 – 1.02)
MLb Mepivacaine 10 mg/mL Levobupivacaine 1.25 mg/mL	94.91	7.16	65.70 (62.05 – 76.68)	8.21 (7.76 – 9.59)	0.81 (0.76 – 0.94)
MLc Mepivacaine 5 mg/mL Levobupivacaine 2.5 mg/mL	31.64	21.47	37.87 (35.78 – 39.94)	18.93 (17.89 – 19.97)	0.96 (0.91 – 1.01)

LA local anesthesia, IQR interquartile range

Discussion

In this study, we indicated that the simultaneous administration of short- and long-acting local anesthetics has an additive effect rather than a synergistic or antagonistic effect. The findings of previous studies on several local anesthetic combinations suggest that the simultaneous administration of local anesthetics is synergistic [7] or antagonistic [11]. However, our study revealed no evidence of synergism or antagonism for the simultaneous administration of short-acting, lidocaine, or mepivacaine, and long-acting anesthetics, ropivacaine or levobupivacaine, from a simple linear regression analysis and interaction index. Most γ values ranged from 0.90 to 1.10, which is consistent with pharmacological additivity. A few combinations exhibited slightly lower values (LRc = 0.89, LLb = 0.86, MRb = 0.83, and MLb = 0.81), which may suggest mild or moderate synergistic interactions. These results suggested that this interaction is unlikely to be antagonistic. Some groups showed mild or moderate synergistic interactions based on their γ values; however, these findings should be interpreted with caution. For example, previous studies that have reported synergy in other drug combinations often showed considerably lower CI values (e.g., 0.69 or even 0.29) [22] which are clearly more indicative of synergism than the values observed in our study. If mild or moderate synergy is considered pharmacologically close to additivity, it is also possible that the γ values observed in our study reflect biological variability in pharmacokinetics and pharmacodynamics rather than true synergistic interactions. While we cannot fully exclude the possibility of mild or moderate synergistic effects in certain combinations, the data support the likelihood of additive rather than antagonistic or strongly synergistic interactions in practical terms.

The findings of previous studies [8, 9] suggest that the simultaneous administration of a local anesthetic combination results in additive effects on LAST. The results of the combination of local anesthetics in the current study support those of these studies. All local anesthetics block sodium (Na) channels. When local anesthetics are administered simultaneously, the number of blocked Na + channels is expected to increase. However, based on the results of this study, we could not explain the mechanism associated with this additive effect.

Previous studies have laid the foundation for understanding pharmacodynamic interactions between local anesthetics, reporting primarily additive effects. Our findings are consistent with this prior work and extend the principle of additivity to combinations of agents with differing durations of action, specifically, short-acting (lidocaine, mepivacaine) and long-acting (ropivacaine, levobupivacaine) local anesthetics. This suggests that the additive nature of central nervous system toxicity remains robust, even in more contemporary anesthetic practices.

While earlier research often focused on lidocaine and bupivacaine or tetracaine, clinical practice has shifted toward newer amide-type agents, such as ropivacaine and levobupivacaine, which are now frequently used for regional anesthesia. These agents are considered to have lower cardiotoxic and neurotoxic potentials in comparison to bupivacaine [23, 24]. Nonetheless, our results indicate that when used in combination with short-acting agents, such as lidocaine or mepivacaine, the total dose of local anesthetics should still be carefully managed to avoid LAST. Further studies are necessary to define safe combined dosing strategies in a clinical context.

No anticonvulsant effect of lidocaine was observed in this study. Lidocaine showed an additive effect, similar to that of the short-acting local anesthetic, mepivacaine. The anticonvulsant effect of lidocaine is due to the Na channel blockade. However, a previous report [17] demonstrated that lower-level concentrations of lidocaine at plasma concentrations < 5 µg/mL can suppress seizures, whereas higher-level concentrations are proconvulsant, thereby causing seizures. Since we did not measure the plasma concentration of lidocaine, it was not known whether it was greater or less than 5 µg/mL in this study. The concentrations of lidocaine mixed with ropivacaine and levobupivacaine in this experiment were 1% and 0.5%, respectively. A lower concentration of lidocaine (0.25%, 0.125%, or even lower) might have the above-described anticonvulsant effects. Further investigation of the anticonvulsant effect of lidocaine is needed, including not only the total amount of lidocaine, but also plasma concentrations.

In this study, local anesthetics were administered intravenously. The increase in the plasma concentration of local anesthetics differs between regional anesthesia and intravenous administration. In general, intravenous administration increases the plasma concentration faster than administration at other sites, such as the epidural and perineural spaces. However, during regional anesthesia, the plasma concentration of the local anesthetic gradually increases. To approximate the increased concentration of local anesthetics during regional anesthesia, we administered local anesthetics continuously via the intravenous route, rather than by bolus administration. In regional anesthesia, the change in the plasma concentration of local anesthetics varies depending on the method of administration, patient’s condition, administration site, and other factors. We demonstrated that the simultaneous continuous intravenous administration of local anesthetics has an additive effect, but we could not determine whether the same interaction occurred during regional anesthesia.

Although we acknowledge that the lack of data on plasma and brain concentrations limits the mechanistic interpretation of our results, our experimental design focused on evaluating pharmacodynamic endpoints, particularly seizure thresholds, in vivo. Continuous intravenous infusion was selected to simulate gradual systemic absorption of local anesthetics during regional anesthesia. Previous studies evaluating local anesthetic toxicity and interactions have used infusion models [8, 9]. Therefore, while this model does not replicate the exact pharmacokinetics of regional anesthesia, it provides a controlled and reproducible method to assess systemic toxicity and drug–drug interactions. We believe that this approach offers clinically meaningful insights into the additive effects of local anesthetic combinations on central nervous system toxicity.

Although further studies are needed to determine precise dosing strategies, our findings suggest that the simultaneous use of short- and long-acting local anesthetics may have additive effects on central nervous system toxicity. Clinically, this could inform safer practices when combining local anesthetics during regional anesthesia, potentially allowing for flexible agent selection without increasing the seizure risk.

Nevertheless, this study has several limitations. First, this study was conducted in rats, and it is not known whether similar results would be obtained in humans. Second, in this study, we did not calculate the median effective dose (ED50) and did not create an isobologram calculated using the ED50. Although the ED50 is typically determined from a sigmoid dose–response curve based on bolus administration, continuous infusion was used in this study, following a previously reported method [9]. Therefore, the ED50 values could not be calculated in this study. As discussed above, when local anesthetics are used in regional anesthesia, plasma concentrations gradually increase. Therefore, we considered continuous intravenous infusion to be more appropriate than bolus administration following a previous study. Third, this study was conducted using a relatively simple method, and we did not measure the plasma or brain concentrations of local anesthetics. As the simultaneous administration of local anesthetics may have an additive effect, the next step is to include plasma or brain concentrations. Fourth, we only studied seizures, a symptom of LAST. Other symptoms of LAST, such as arrhythmia and cardiac arrest, also need to be investigated. The simultaneous administration of local anesthetics may have an additive effect on the central nervous system; however, for other symptoms such as arrhythmia and cardiac arrest, it may have a different effect.

Conclusions

The findings of our study suggest that the simultaneous administration of local anesthetics causes an additive effect on the central nervous system symptoms of LAST. However, we could not determine the mechanism associated with the additive effect or determine the safe total amount. Therefore, further studies are warranted.

Authors’ contributions

YN designed the study, acquired the data, and prepared the manuscript. MM designed the study, contributed to coordination of the study, and assisted in the preparation of the manuscript. SO assisted statistical analysis. KY and AH assisted experiment. SI assisted in the preparation of the manuscript. All authors have read and approved the final manuscript.

Funding

This work was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI [grant number 19K18248].

Data availability

The data that support the fndings of this study are available on request from the corresponding author, YN.

Declarations

Competing interests

The authors declare no competing interests.