Assessment of volume and depth loss of incompletely set hydraulic calcium silicate cements used as root-end fillings in a simulated surgical model

Abstract

Root-end filling materials often face environmental exposure before completely setting, which may alter their properties compared to completely set samples in most studies. This in-vitro study evaluated the volumetric changes and depth of material loss of three incompletely set hydraulic calcium silicate cements in a simulated root-end filling model. Thirty prepared human mandibular premolars with retrograde cavities were randomly divided into three groups (n = 10), each filled with ProRoot MTA, Biodentine, or Endosequence root repair material fast-set putty (ERRM-FSP), and submerged in phosphate-buffered saline. Micro-CT assessed changes at 15 min, 24 h, 7 days, 14 days, and 30 days after root-end filling. All materials exhibited volumetric reduction and depth loss over time. At 24 h, Biodentine exhibited the highest volumetric change. At 30 days, ProRoot MTA and Biodentine had significantly higher depth loss than ERRM-FSP, which showed the least change in both parameters. This study suggests ERRM-FSP offers superior volumetric stability as a root-end filling material in surgical situations, making it a potentially better choice compared to ProRoot MTA and Biodentine.

Introduction

Endodontic microsurgery is considered a predictable treatment for post-treatment apical periodontitis. The procedure typically involves the removal of infected periapical tissue, root-end resection, preparation of a retrograde cavity, and placement of a root-end filling material. The prognosis in endodontic microsurgery can be influenced by several factors, one of which is the ability of the root-end filling material to effectively seal the root canal system. The root-end filling materials must possess low solubility and good dimensional stability to prevent the leakage of microorganisms and their byproducts, ensuring long-term success of the procedure^1,2.

Hydraulic calcium silicate cements (HCSCs) demonstrated favorable clinical outcomes when used as root-end filling materials, as evidenced by numerous studies^3–5. Currently, HCSCs in endodontics exhibit diverse types distinguished by their clinical applications and chemical compositions. One notable aspect is the type of vehicle used, which can be either aqueous or non-aqueous vehicles⁶. In the context of formulation, an aqueous vehicle refers to a water-based medium in which other substances are dissolved or suspended. In contrast, non-aqueous vehicles utilize solvents that do not contain water as their primary component; instead, they often rely on oils, alcohols, or organic liquids to dissolve or suspend active ingredients. The choice of vehicle can influence the hydration process of HCSCs. Aqueous HCSCs, such as ProRoot Mineral Trioxide Aggregate (ProRoot MTA, Dentsply, Tulsa, OK, USA) and Biodentine (Septodont Inc., Saint-Maur-des-Fossés, France), require powder-liquid mixing to initiate their setting reactions. On the other hand, the hydration reaction of non-aqueous HCSCs, such as Endosequence Root Repair Material Fast Set Putty (ESRRM-FSP, Brasseler USA, Savannah, GA), begins upon contact with environmental moisture.

The loss of material thickness observed in follow-up cases is reflected in both the volumetric change and the depth of material loss. These volumetric changes can arise from various mechanisms, such as expansion, contraction, disintegration, material solubility, and washout by tissue fluids. Most prior studies on this topic have predominantly focused on solubility values derived from various testing methods such as (ISO 9917-1; 2007)⁷ and (ISO 6876; 2012)⁸. However, the results have exhibited considerable variability. This variation may be influenced not only by the diversity of testing methods used but also by differences in material setting conditions, as most tests are typically conducted on materials that have completely set^1,2,9–11. In endodontic surgical procedures, root-end filling materials are frequently exposed to fluids such as periapical tissue fluid or blood prior to complete setting. This exposure may disrupt the setting process, potentially altering the material’s properties. Understanding the characteristics of each material in an incompletely set condition is essential for guiding clinicians in selecting the appropriate materials. To simulate clinical conditions, the volumetric stability of root-end filling materials was assessed by evaluating both volumetric changes and the depth of material loss using micro-computed tomography (micro-CT), which enables non-destructive, three-dimensional tracking of material properties over time. Investigating both volumetric changes and depth of material loss provides a more comprehensive assessment that considers all influencing factors, offering deeper insights than focusing solely on solubility.

Therefore, the aims of this study were to evaluate the volumetric changes and depth of material loss of ProRoot MTA, Biodentine, and ERRM-FSP in a simulated root-end filling model, under conditions that mimic the surgical environment, over a 30-day period. The null hypothesis proposed that there would be no significant differences in these characteristic changes among the materials or immersion periods.

Methods

Sample size calculation

For volumetric change and depth of material loss evaluation, based on the results obtained from our pilot study, an effect size of 0.21 was added to a statistical power of 0.95 and a probability level (α) of 0.05 inputs into F test family for one-way analysis of variance (G*power 3.1 for Macintosh). To detect differences among these materials, the estimated minimum sample size for each group was determined to be nine. However, to increase the power of the study, we utilized ten samples for each experimental group.

Sample selection

Thirty human mandibular premolars with single straight root canals were obtained from a private dental clinic. These teeth were extracted for orthodontic purposes as part of the patients’ dental treatment plans, not specifically for research. The study protocol was approved by the Human Research Ethics Committee of the Faculty of Dentistry, Chulalongkorn University, Thailand (HREC-DCU 2023-004). Since only de-identified extracted teeth with no traceable donor information were used, the requirement for informed consent was waived by the Human Research Ethics Committee of the Faculty of Dentistry, Chulalongkorn University, Thailand. All methods in this study were performed in accordance with relevant guidelines and regulations [International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) and Good Clinical Practice (GCP)].

All teeth were stored in a 0.1% thymol solution until use. Preoperative radiographs were taken to confirm that the dentin thickness at the 3 mm level from the root apex were similar in both the buccolingual and mesiodistal directions by measuring the distances from the outer to inner root surfaces. The mean and standard deviation values for the recruited teeth were 1.17 ± 0.27, 1.18 ± 0.29, 1.68 ± 0.33 and 1.57 ± 0.30 mm. respectively, for buccal, lingual, mesial, and distal thicknesses.

Sample preparation

Root canal preparation

Following the access opening, the root canals were shaped using R40 Reciproc rotary files (VDW, Munich, Germany), with the instrument termination 1 mm short of the apical foramen. Irrigation was carried out using conventional needle irrigation, delivering a total of 10 mL of 2.5% sodium hypochlorite (NaOCl) for each tooth. The final rinse involved 5 mL of 17% EDTA, followed by an additional 5 mL of 2.5% NaOCl. The root canals were then filled with gutta-percha and AH Plus sealer (Dentsply De Trey GmbH, Konstanz, Germany) using the lateral condensation technique. Radiographs were taken to verify the quality of the obturation. Samples that exhibited inconsistencies in the root filling were discarded and replaced. The access cavities were sealed with Caviton (GC, Tokyo, Japan). All samples were stored in a humidity chamber (CONTHERM 1200, CONTHERM Scientific LTD., New Zealand) set to 100% humidity and 37 °C for 7 days to ensure complete setting of the sealer.

Root-end preparation

Fingernail polish was applied twice to the external root surface to limit the permeable area to the resected root surface only. To simulate apical root resection, 3 mm of the root tip was removed perpendicularly to the long axis of the root using an Isomet (IsometTM 1000, Buehler, IL, USA). The root was then prepared for a retrograde cavity using a high-speed diamond cylinder bur with a 1.2 mm diameter and a 3 mm depth, under a dental operating microscope (OPMI Pico, Carl Zeiss Meditec Inc). Following this, all samples underwent micro-CT scanning (SkyScan 1272, Bruker, Kontich, Belgium) at a voxel size of 15 μm, 70 kVp, and 114 µA under high resolution (1024 × 1024). This scanning allowed the acquisition of three parameters: volume of the retrograde cavity and resected root surface area (both dentine and cavity areas). Any samples exhibiting crack lines from micro-CT were excluded and replaced with new ones. The remaining samples were then randomly assigned to one of three groups as follows:

Group I: ProRoot MTA.

White ProRoot MTA (Dentsply Tulsa Dental, Tulsa, OK) was prepared by a single drop of liquid from a single-dose container with 0.12 g of powder. The mixture was delivered into the cavity using a plastic instrument and condensed with fitted pluggers.

Group II: Biodentine.

Biodentine (Septodont, Saint Maur des Fosses, France) was prepared by dispensing 5 drops of liquid from a single-dose container into the powder-containing capsule. The mixture was activated using a Biodentine Capsule Mixer (Septodont, Saint Maur des Fosses, France) at 4,000 rounds per minute for 30 s. Subsequently, it was delivered into the cavity using a plastic instrument and condensed with fitted pluggers.

Group III: ERRM-FSP.

ERRM-FSP was extruded from the syringe onto a glass slab and rolled into small cylinder. Each cylinder was delivered into the cavity using a plastic instrument and condensed with fitted pluggers.

The homogeneity of the root-end filling was confirmed once again by scanning using micro-CT. Any samples showed heterogeneity of filling was discarded and replaced with a new one.

Volumetric change: micro-CT evaluation

All samples were placed in a customized silicone jig positioned on translucent plastic tubes, ensuring sample stability during the scanning process. Only the 3 mm resected root immersed in 1.5 mL of phosphate-buffered saline (PBS). The models were immersed in PBS 15 min after root-end filling. The models were then scanned using micro-CT, with the roots continuously immersed in PBS throughout the study period to simulate immediate and continuous exposure to tissue fluid.

Micro-CT data was used to assess volumetric stability of the root-end filling materials by evaluating their volumetric changes and the depth of material loss following exposure to PBS. Each sample was scanned at five different time points: 15 min after root-end filling (T1); 24 h (T2); 7 days (T3); 14 days (T4); and 30 days (T5). Each scanning session utilized micro-CT settings consistent with those described previously.

After each scanning session, all models were stored in a 37 °C chamber until the subsequent scanning period. The images obtained from different time periods were saved in the coronal, sagittal, and axial planes by using the Data Viewer program (V1.5.2.4; Bruker-MicroCT). All datasets were reconstructed using NRecon software (v2.2.0.6; Bruker, Kontich, Belgium). After the image reconstruction, CTAn (ver. 1.20.8.0, Skyscan, Bruker, Kontich, Belgium) software was used for the three-dimensional (3D) volumetric visualization and analysis.

The percentage of volumetric changes of the materials were calculated using following formula:

Tx = T2, T3, T4, and T5.

The depth of material loss was the distance between the outermost apical level of material surface at T1 (A) and the most coronal level at which material loss could be observed (B) as shown in Fig. 1a.

Fig. 1.

Representative images of 3D reconstruction of root-end filling materials obtained from micro-CT (a) Images showing the positions for measuring depth of material loss; A: the outermost apical level of material surface at T1; B: the most coronal level at which material loss could be observed (b) Images demonstrating depth of material loss and cross sections at positions A and B after immersion in PBS at various time points.

T1: at 15 min after root-end filling, T2: at 24 h, T3: at 7 days, T4: at 14 days, T5 at 30 days.

Statistical analysis

The normal distribution of all data was confirmed by the Shapiro–Wilk test. One-way ANOVA were used to compare among groups regarding the parameters of volume of the retrograde cavity, resected root surface area: both dentine and cavity. A two-way repeated measures ANOVA was conducted to analyze differences in the percentage of volumetric change and depth of material loss among immersion periods and materials, followed by Bonferroni correction for post hoc tests to assess group differences at a significance level of 0.05.

Results

The mean volume of the retrograde cavity, resected root surface area of dentine, and resected root surface area of the cavity showed no significant differences among tested groups (Table 1).

Table 1.

Mean and standard deviation of retrograde cavity volume and resected root surface (both cavity area and dentine area) before retrograde filling.

	Parameters	ProRoot MTA	Biodentine	ERRM-FSP	p-value
Resected root surface	Cavity area (mm2)	1.89 ± 0.14	1.86 ± 0.17	1.96 ± 0.07	0.192
	Dentine area (mm2)	7.97 ± 2.91	7.89 ± 2.31	6.45 ± 2.33	0.332
Retrograde cavity	Volume (mm3)	5.59 ± 0.40	5.30 ± 0.27	5.64 ± 0.50	0.195

Volumetric change

Mean initial volumes of root-end filling materials at T1 were 5.22 ± 0.35 mm³, 5.30 ± 0.70 mm³, and 5.36 ± 0.36 mm³ for ProRoot MTA, Biodentine, and ERRM-FSP respectively which showed no significant difference among tested groups (p = 0.815). The mean percentage of volumetric change for each group after different immersion periods was shown in Fig. 2. There were significant differences in the percentage of volumetric change among “materials”, “immersion periods”, and “interaction between material and immersion period”.

Fig. 2.

Means ± standard deviations of the percentage of volumetric changes of the three different materials after immersed in PBS between immersion periods, analyzed by two-way repeated measures ANOVA. T1: at 15 min after root-end filling, T2: at 24 h, T3: at 7 days, T4: at 14 days, T5 at 30 days. Different upper-case letter in the same column indicated statistically significant differences among materials (p < 0.05). Different lower-case letter in the same row indicated statistically significant differences between time intervals (p < 0.05).

The volumetric change values of all tested material tended to gradually increase over time. For materials comparison, Biodentine showed a higher percentage of volumetric change than ProRoot MTA (p = 0.008) and ERRM-FSP (p = 0.002) at 24 h. However, there was no statistically significant difference between Biodentine and ProRoot MTA from 7 days onward (p > 0.05). For all immersion periods, ERRM-FSP exhibited a significantly lower percentage of volumetric change compared to other materials, except at 24 h, where there was no significant difference between ProRoot MTA and ERRM-FSP (p = 0.572).

Depth of material loss

From T1 to T5, the mean depth of material loss was observed in the range of 0.40–0.56 mm for ProRoot MTA, 0.16–0.28 mm for Biodentine, and 0.00–0.08 mm for ERRM-FSP. As indicated in Table 2, the mean depth of material loss after immersion in PBS was significantly higher for ProRoot MTA compared to Biodentine and ERRM-FSP. However, there was no significant difference between ProRoot MTA and Biodentine at 24 h (p = 0.056).

Table 2.

Mean and standard deviation of depth of material loss (mm.) in different types of HCSCs between immersion periods.

	T1	T2	T3	T4	T5
ProRoot MTA	0.23 ± 0.13Aa	0.28 ± 0.12Ab	0.31 ± 0.11Ac	0.35 ± 0.08Ad	0.35 ± 0.05Ad
Biodentine	0.16 ± 0.03Aa	0.18 ± 0.03Bb	0.24 ± 0.03Bc	0.27 ± 0.04Bd	0.31 ± 0.04Bc
ERRM-FSP	0.02 ± 0.02Ba	0.05 ± 0.17Ca	0.09 ± 0.02Cb	0.11 ± 0.02Cc	0.13 ± 0.02Cd

T1: at 15 min after root-end filling, T2: at 24 h, T3: at 7 days, T4: at 14 days, T5 at 30 days.

Different upper-case letters in the same column indicated statistically significant differences among materials (p < 0.05). Different lower-case letter in the same row indicated statistically significant differences between immersion periods (p < 0.05).

For comparison within the same material, the depth of material loss increased significantly over time, except for ERRM-FSP over a 24-h period and ProRoot MTA after 14 days (p > 0.05). Figure 1b showed representative images of 3D reconstruction of the different root-end filling materials before and after immersion in PBS at various time points.

Discussion

The three-dimensional hermetic seal is one of the most important requirements of the root-end filling material¹². These materials should ideally demonstrate minimal solubility and maintain dimensional stability when exposed to tissue fluids^2,11,13. There is only one study that examined the volumetric change of root-end filling using micro-CT. However, this study was conducted with completely set materials in blocks, using only two types of materials, which does not fully replicate the conditions observed in clinical settings¹⁴. Therefore, this study is the first to investigate the volumetric stability of root-end filling materials in an incompletely set condition within human roots when exposed to simulated tissue fluids, aiming to more accurately reflect clinical conditions. Our findings reveal significant differences in volumetric change and depth of material loss among tested materials, highlighting the superior volumetric stability of ERRM-FSP. Thus, the null hypothesis was rejected.

The volumetric stability of root-end filling materials is influenced not only by their intrinsic physical and chemical properties of materials but also by other factors such as the contact surface area of root-end filling material, number of exposed dentinal tubules on resected root surface both in area and angle of resection¹⁵. To ensure the accuracy of results, several factors that could influence the outcome were controlled. Firstly, the volume of retrograde cavities and the area of resected dentin and cavity surfaces were standardized (Table 1). Secondly, the root surfaces were sealed with nail polish to mitigates the influence of varying permeability, addressing another potential confounding factor across the groups. Additionally, testing in this real root model limits the material’s surface contact with tissue fluid to only one side, accurately replicating the conditions of a true root-end filling in surgical root canal treatment and/or apexification. This approach also takes into account the effect of dentinal tubule permeability from the resected root surface, which can impact the volumetric stability of the materials, as opposed to relying on artificial conditions.

Preparing retrograde cavities using different techniques, such as ultrasonics or burs, can produce varied dentin surfaces. Ultrasonic preparation is considered essential in modern microsurgery; however, repeatedly grinding with a small ultrasonic tip to create larger cavities may lead to unstandardized samples. In contrast, preparing a retrograde cavity with a 1.2 mm diameter cylindrical bur in a single stroke yields more consistent and comparable cavities across groups. Thus, the use of burs for retrograde cavity preparation in this study was solely for standardization purposes to ensure a more consistent methodology and is considered one of the limitations. Additionally, the dissolution of root-end filling materials is more heavily influenced by the surface area in direct contact with tissue fluid than by the interface between the material and dentin. Therefore, controlling the major factor—the exposed surface area—is of greater importance.

PBS, a synthetic tissue fluid, is widely used to simulate periapical fluid in in-vitro studies for testing the physical properties of HCSCs^16–18. PBS has demonstrated components and properties similar to periapical tissue fluid, which is crucial for assessing materials used in dental applications¹⁹. Therefore, our study conducted a volumetric stability test for root-end filling materials using a simulated root-resected model in PBS. However, blood exposure can alter the properties of HCSCs²⁰. Furthermore, studies comparing the effects of PBS and blood on hydraulic cement have shown a negative impact on compressive strength²¹ and surface microhardness²² in the blood-contaminated group. However, other studies have observed positive material expansion²³ when the cement is contaminated with blood. Therefore testing materials under blood-contaminated conditions may better simulate clinical situations and may potentially yield different results. Previous studies^24–26involving blood contamination utilize either a small amount of freshly drawn blood incorporated into the samples or contact with blood for a short duration (contaminated the environment or exposed to blood within minutes). In contrast, this study investigated the materials over an extended observation period. Incorporating fresh human blood under such conditions would be impractical without anticoagulants like heparin^27,28or sodium citrate²⁹which are not typically present in clinical environments. A notable limitation of this study is the use of a static PBS model, which lacks the dynamic pulsatile blood flow characteristic. A more comprehensive understanding of root-end conditions may be achieved through future in animal study or in vitro incorporating dynamic blood flow. Additionally, the micro-CT method was employed to assess the volumetric stability of the materials by measuring the volumetric change and depth of material loss³⁰. This approach offers the versatility to conduct longitudinal assessments over different timeframes without damaging the samples. Furthermore, PBS did not introduce any artifacts during micro-CT imaging due to the absence of radiopacifiers that could cause beam hardening effects. The samples were also stabilized in the silicone jig to prevent movement artifacts, resulting in clear 3D images, as shown in Fig. 1.

From the results of this study, Biodentine exhibited a significantly higher volumetric change at 24 h after immersion compared to other materials, which is consistent with its higher solubility reported in previous studies^2,31,32. Despite the study design ensuring all materials were exposed to fluid 15 min after root-end filling which allows 12 min initial setting for Biodentine, it still experienced considerable dissolution. This could be attributed to the faster initiation of the hydration reaction that begins immediately upon mixing and is accelerated with calcium chloride as its water component. During the reaction, the major components, di- and tri-calcium silicates, are converted into calcium silicate gel and calcium hydroxide. This process results in an initial loss of material mass, and since calcium hydroxide is soluble in tissue fluid, this further contributes to material dissolution³¹. Consequently, Biodentine exhibited greater volume and depth loss within the first 24 h. In contrast, although MTA also reacts to form calcium hydroxide, its reaction occurs more slowly, leading to a more gradual dissolution process. As a result, a continuous increase in volume loss was observed up to 7 days, as shown in Fig. 2, which is consistent with previous studies conducted according to ISO standards³³. Furthermore, after 24 h, there was no significant difference in volumetric change between ProRoot MTA and Biodentine. This increased volumetric change may be attributed to MTA’s extended setting time, which prolongs its exposure to a moist environment. These findings suggested that the different compositions between ProRoot MTA and Biodentine may play a significant role in the early phases of dissolution.

ERRM-FSP, which utilizes a non-aqueous vehicle, exhibited the lowest volumetric change compared to MTA and Biodentine over the entire study period (Fig. 2). This finding suggests that non-aqueous type may dissolve less than aqueous type, possibly due to the proportion of thickening agents which could enhance dimensional stability. Moreover, the short final setting time of ERRM-FSP (20-min), compared to MTA (180-min) and Biodentine (110-min)³⁴, may reduce the material dissolution from retrograde cavity³⁵.

The rate of volumetric change was represented by the slope of the graph shown in Fig. 2. Biodentine exhibited a higher rate of volumetric change during the initial phase, followed by ProRoot MTA. However, after 14 days, the rate of volumetric change of all materials stabilized at the same level. This stabilization may indicate the complete setting of all materials.

Based on the study findings, it is also the first to clearly evaluate the depth of material loss, which is a significant concern, particularly regarding material thickness in clinical applications. Our study found a significantly greater depth of material loss in ProRoot MTA compared to Biodentine and ERRM-FSP (Fig. 1), consistent with the observed percentage of volumetric change. These findings indicate that varying thicknesses of HCSCs may be required to optimize sealing properties, particularly for groups with significant material dissolution, where a greater material thickness may be necessary to achieve adequate sealing. The initial seal achieved with hydraulic cement is purely mechanical. However, over time, a hydration reaction occurs at the material-dentin interface, leading to the formation of hydroxyapatite crystals that establish chemical bonds, thereby providing a durable, long-lasting seal. The loss of volume and reduction in the depth of the material may indirectly suggest a potential compromise of the bond strength and sealing ability, especially at the material-dentin interface³⁶, although this has not been directly tested.

Employing an in vitro methodological setup that closely mimics clinical scenarios can enhance our understanding of the physical properties of endodontic materials. However, there were limitations in that a full replication of real clinical conditions could not be achieved, such as the use of PBS instead of periapical fluid, the absence of fluid circulation, similar to the dynamic fluid flow in the human body, may contribute to greater changes in volumetric reduction, and the application of nail varnish on the external root surface to limit permeability. Furthermore, variations in root canal anatomy and/or retrograde cavity size may influence the volumetric stability of these materials. Therefore, caution is warranted when extrapolating our findings to different clinical scenarios.

At the end of the observation period in this study, all tested materials continued to demonstrate volumetric change, indicating that dissolution processes may continue beyond the time frame examined.

However, the early dimensional stability of ERRM-FSP may hypothesize a superior long-term clinical success compared to other materials that exhibit greater volumetric and depth loss. Unfortunately, the long-term outcomes of surgical endodontics comparing these materials are still unavailable. Furthermore, the dissolution properties tested using micro-CT cannot indicate the adaptation of the material to the cavity walls or detect material leakage. Therefore, future studies with longer observation periods, along with research on marginal adaptation, leakage testing, push-out bond strength and interfacial integrity assessment, under more clinically relevant conditions are recommended to better understand the long-term behavior and clinical performance of these materials. These improvements would lead to a more comprehensive understanding of how these materials perform under conditions that more closely resemble clinical practice.

In conclusion, the findings of the present study highlight the differences in material volumetric stability between aqueous-based HCSCs and non-aqueous-based formulations. Specifically, ProRoot MTA and Biodentine, both aqueous-based HCSCs, exhibited greater volumetric changes and depth loss, whereas ERRM-FSP, a non-aqueous vehicle, showed significantly less volumetric change and depth loss. These results emphasize the importance of considering formulation differences when selecting materials for root-end filling, as well as for other applications such as apexification and perforation repair. The greater volumetric stability observed with ERRM-FSP may provide an advantage in maintaining material integrity and stability over extended periods.

Conclusions

Under condition of incompletely set root-end filling in simulated surgical environment, ERRM-FSP exhibited significantly lowest volumetric change and depth of material loss compared to Biodentine and ProRoot MTA. All materials exhibited high rate of volumetric change during the first 24 h, after which the rate gradually decreased until 30 days.

Author contributions

C.S.: Project administration, conceptualization, methodology, investigation, data curation, formal analysis, visualization, resources, writing-original draft preparation, funding acquisition. T.J.: Methodology, investigation, data curation, formal analysis, supervision, writing-review and editing. S.H.: Conceptualization, methodology, investigation, resources, formal analysis, supervision, writing-review and editing.

Data availability

Data can be provided upon reasonable request. Please contact corresponding author at Sirawut.h@chula.ac.th.

Declarations

Competing interests

The authors declare no competing interests.