In Silico and In Vitro Anti-Helicobacter Pylori Effects of Combinations of Phytochemicals and Antibiotics

Helicobacter pylori infection is a WHO class 1 carcinogenic factor of gastric adenocarcinoma. In the past decades, many studies have demonstrated the increasing trend of antibiotic resistance and pointed out the necessity of new effective treatment. This study was aimed at identifying phytochemicals that can inhibit H. pylori and possibly serve as adjuvant treatments. Here, in silico molecular docking and drug-like properties analyses were performed to identify potential inhibitors of urease, shikimate kinase and aspartate-semialdehyde dehydrogenase. These three enzymes are targets of the treatment of H. pylori. Susceptibility and synergistic testing were performed on the selected phytochemicals and the positive control antibiotic, amoxicillin. The in-silico study revealed that oroxindin, rosmarinic acid and verbascoside are inhibitors of urease, shikimate kinase and aspartate-semialdehyde dehydrogenase, respectively, in which, oroxindin has the highest potency against H. pylori, indicated by a minimum inhibitory concentration (MIC) value of 50 μg/mL. A combination of oroxindin and amoxicillin demonstrated additive effects against H. pylori, as indicated by a fractional inhibitory concentration (FIC) value of 0.75. This study identified phytochemicals that deserve further investigation for the development of adjuvant therapeutic agents to current antibiotics against H. pylori.


Introduction
Helicobacter pylori infections may cause chronic gastritis, peptic ulcer, gastric cancer and other non-gastric related disorders, such as lymphoma [1]. The World Health Organisation has considered H. pylori as a class 1 carcinogen [2]. A recent study reported that until 2015, around 4.4 billion people were infected with H. pylori, and the prevalence rate in Africa, Latin America and Asia is substantially increasing [3]. Thus, H. pylori is a global public health concern. The main treatment option for H. pylori is the standard triple therapy, combining two antibiotics with one proton pump inhibitor, such as clarithromycin and amoxicillin with omeprazole [4]. Due to the development of a drug-resistant strain, the failure rate of triple therapy has increased to more than 20% in many parts of the world [5]. This causes the use of higher doses or more drugs, such as the quadruple therapy, and this has led to a higher risk of side effects. To solve this problem, some researchers began to combine phytomedicines with triple therapy [6,7]. Some of their results showed the ability of phytomedicine to reduce side effects and decrease the treatment failure rate; however, their pharmacological mechanism of action is unclear.
Many pharmacological targets against an H. pylori infection have been identified, and they are generally related to H. pylori's morphological structure, survival mechanisms and toxic factors. One well-known example of these targets is urease. H. pylori requires urease and the H + -gated urea

Results and Discussion
This study performed molecular docking and drug-like properties analysis to identify phytochemicals that may inhibit the three pharmacological target enzymes of H. pylori: urease, shikimate kinase and aspartate-semialdehyde dehydrogenase. Phytochemicals were selected from the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP). GOLD v5.5 was used as the docking suite. The identified inhibitors were oroxindin, rosmarinic acid and verbascoside, respectively. In vitro susceptibility and synergistic testing against H. pylori were then performed on these three phytochemicals and the parallel positive control antibiotic (amoxicillin) to calculate their minimum inhibitory concentration (MIC) and fractional inhibitory concentration (FIC) values.

In Silico Simulations
The accuracy of the docking procedures varies substantially between different docking suites. Here, we validated our docking procedures on urease and shikimate kinase using receiver operating characteristic (ROC) analysis; their area under the curve (AUC) values were 0.90 and 0.77, respectively ( Figure 1). An AUC value of 0.7 or above indicates a reliable docking procedure [21,22]. Hence, our docking approaches have reliable predictive power. Molecular docking and drug-like properties analysis were performed on the three target enzymes to select phytochemicals for in vitro studies. The drug-like properties were classified into three categories: ADME (absorption, distribution, metabolism and excretion), physicochemical and drug safety (see Section 4.3). We selected one phytochemical with the finest balance between the predicted binding affinity and drug-like properties for each target enzyme. The three selected phytochemicals were oroxindin, verbascoside and rosmarinic acid ( Figure 2). Oroxindin obtained a high urease binding score of 84.9, which is comparable to most of the known urease inhibitors with binding scores ranging from 57.1 to 111.8. The botanical source of oroxindin is Radix Bupleuri, which has been shown to have an anti-H. pylori effect [23]. However, the active ingredients of Radix Bupleuri responsible for the anti-H. pylori effect have not been identified. Molecular docking and drug-like properties analysis were performed on the three target enzymes to select phytochemicals for in vitro studies. The drug-like properties were classified into three categories: ADME (absorption, distribution, metabolism and excretion), physicochemical and drug safety (see Section 3.3). We selected one phytochemical with the finest balance between the predicted binding affinity and drug-like properties for each target enzyme. The three selected phytochemicals were oroxindin, verbascoside and rosmarinic acid ( Figure 2). Molecular docking and drug-like properties analysis were performed on the three target enzymes to select phytochemicals for in vitro studies. The drug-like properties were classified into three categories: ADME (absorption, distribution, metabolism and excretion), physicochemical and drug safety (see Section 4.3). We selected one phytochemical with the finest balance between the predicted binding affinity and drug-like properties for each target enzyme. The three selected phytochemicals were oroxindin, verbascoside and rosmarinic acid ( Figure 2). Oroxindin obtained a high urease binding score of 84.9, which is comparable to most of the known urease inhibitors with binding scores ranging from 57.1 to 111.8. The botanical source of oroxindin is Radix Bupleuri, which has been shown to have an anti-H. pylori effect [23]. However, the active ingredients of Radix Bupleuri responsible for the anti-H. pylori effect have not been identified. Oroxindin obtained a high urease binding score of 84.9, which is comparable to most of the known urease inhibitors with binding scores ranging from 57.1 to 111.8. The botanical source of oroxindin is Radix Bupleuri, which has been shown to have an anti-H. pylori effect [23]. However, the active ingredients of Radix Bupleuri responsible for the anti-H. pylori effect have not been identified. Oroxindin could be one of Radix Bupleuri's active phytochemicals against H. pylori. Oroxindin has similar chemical structures with similar functional groups to quercetin and baicalin, which have been experimentally demonstrated with H. pylori inhibitors [24][25][26]. Regarding drug-like properties, oroxindin was predicted to be a non-inhibitor to all Cytochrome P450 enzymes and hERG, as well as a non-central nervous system (CNS) penetrant (Table 1). This means oroxindin is unlikely to have drug-drug interactions, cardiotoxicity and CNS side effects. Oroxindin also has high water solubility and poor human intestinal absorption (HIA, Table 1). These results indicate that oroxindin can dissolve, spread and reach H. pylori in the human gastric region without too much systematic absorption into the bloodstream.  1 The docking scores of oroxindin, verbascoside and rosmarinic acid corresponded to urease, aspartate-semialdehyde dehydrogenase and shikimate kinase, respectively. MW: molecular weight; log P: octanol-water partition coefficient at 25°C under standard conditions (optimal value: −1.00 to 4.20); aqueous solubility was calculated at pH 6.4 (>0.1 indicates soluble); Caco-2 predicts passive intestinal permeability (≤1.00 indicates poorly permeable); PPB represents plasma protein binding; central nervous system (CNS) values of ≤ −3.50 indicates non-central nervous system penetrant; HIA is human intestinal absorption (≤30% indicates poorly absorbed); Ames estimates mutagenic potential (≤0.33 indicates non-mutagenic, 0.33-0.67 is undefined, >0.67 is mutagenic); hERG values of less than 0.33 indicates non-inhibitor of hERG channel and has low risk of cardiotoxicity; CYP is Cytochrome P450 and NI means non-inhibitor.
The docking score of rosmarinic acid on shikimate kinase was 79.1, which is comparable to the scores of many known inhibitors which range from 57.8 to 89.2. Rosmarinic acid achieved an excellent drug-like profile, with appropriate ADME, physicochemical and drug safety properties (Table 1). Similar to oroxindin, rosmarinic acid is highly water soluble, has poor HIA, and is unlikely to produce drug-drug interactions, cardiotoxicity or CNS side effects ( Table 1). The botanical sources of rosmarinic acid are Melissa officinalis and Ocimum basilicum; Melissa officinalis exhibits a gastroprotective effect against gastric ulcers in animal studies [27], and Ocimum basilicum has been shown to significantly inhibit H. pylori in an in vitro study [28]. Again, the pharmacological mechanism of these two botanical sources is not clear and the active ingredients are unknown. Rosmarinic acid has demonstrated its antimicrobial effects on many bacteria strains, including Enterobacter species [29], Escherichia coli [30], Aspergillus niger [31], and many more [32]. Rosmarinic acid also has synergistic effects with amoxicillin, vancomycin and ofloxacin against Staphylococcus aureus [33]. Hence, rosmarinic acid has a higher possibility of inhibiting H. pylori than many other phytochemicals and was selected for further in vitro studies.
Verbascoside was the phytochemical selected to target aspartate-semialdehyde dehydrogenase. It obtained a docking score of 82.31 and ranked 11th out of 4450 herbal compounds. Similar to the other two phytochemicals, verbascoside has acceptable drug-like properties, such as good water solubility, non-CNS penetrant and non-inhibitor to all CYPs and hERG. ( Table 1) High contents of verbascoside were shown in the leaf extracts of Aloysia triphylla, which exerted antibacterial effects against Proteus mirabilis [34], Staphylococcus aureus [35] and H. pylori [36] in in vitro studies. A recent study also pointed out the synergistic effects of verbascoside and gentamicin against Staphylococcus aureus and Escherichia coli [37]. Verbascoside demonstrated gastroprotective effects in an animal study by inhibiting the excretion of gastric acid through blocking H + /K + -ATPase. This indicates that verbascoside has the potential to mimic the action of a proton pump inhibitor in H. pylori treatment [38].
Due to the above-mentioned high docking scores, favourable drug-like properties and antibacterial effects suggested by the literature, we selected oroxindin, verbascoside and rosmarinic acid for further in vitro studies. According to our knowledge and literature searches in various databases, including the Cochrane Library, Embase, Medline and the China National Knowledge Infrastructure database, there is no experimental proof on the inhibitory effects of these three phytochemicals against H. pylori.

In Vitro Susceptibility and Synergistic Testing
All the three selected phytochemicals, oroxindin, verbascoside and rosmarinic acid had anti-H. pylori effects ( Table 2). Among them, oroxindin had the highest potency, indicated by the lowest MIC value of 50 µg/mL. However, it is still less potent than the positive control antibiotic, amoxicillin, which obtained a MIC value of 0.250 µg/mL ( Table 2). Regarding the synergistic testing, oroxindin demonstrated additive effect with each of the other two phytochemicals and amoxicillin, indicated by the value of 0.75 (Table 3). There were no synergistic or antagonistic effects observed in all the other combinations.  The MIC value of oroxindin was 50 µg/mL, which is comparable with some well-known anti-H. pylori phytochemicals, such as scopolin (50-100 µg/mL) [39], chelerythrine (25-100 µg/mL) [40], and protopine (25-100 µg/mL) [40]. However, the MIC value of oroxindin is larger than some high potency phytochemicals, such as berberine (0.78-25 µg/mL) [40], fuscaxanthone (16.3-131.2 µg/mL) [41], palmatine (3.12-6.25 µg/mL) [42] and allicin (6 µg/mL) [43]. Certainly, MIC values can be affected by many factors, such as experimental procedures and reagents; thus, it is not reasonable to directly compare MIC values obtained from different studies. However, comparing these values provides a rough overview of their potency, signifying the anti-H. pylori effect of oroxindin.
Although many phytochemicals have suggested anti-H. pylori effects, many of them have undesirable side effects or unfavourable drug-like properties [44]. For example, allicin has superior MIC values (6 µg/mL) [43], but it can cause gastric side effects, such as heartburn, diarrhoea and nausea. It also has antithrombotic properties and can prolong bleeding time [45]. Another example is palmatine, which can significantly inhibit H. pylori and has antiviral, anticancer and antihyperlipidemic effects [42]. However, it has noticeable DNA toxicity and complex interactions with liver metabolic enzymes [46]. In contrast, our validated in silico study suggested that oroxindin has poor human intestinal absorption, indicating the amount of oroxindin that can be absorbed into the bloodstream is low. Hence, the risk of causing systematic side effects and the risk of interacting with metabolic enzymes in other organs is low. Also, an H. pylori infection can irritate the stomach and cause gastritis, and oroxindin has demonstrated its gastroprotective effects by suppressing the inflammatory response and conserving the gastric barrier function [47].
The MIC values of oroxindin, verbascoside and rosmarinic acid are higher than the positive control, amoxicillin ( Table 2). This indicates the phytochemicals are less potent than one of the current antibiotic drugs of H. pylori. The current recommended treatment of H. pylori is triple therapy, which contains two antibiotics, such as amoxicillin, clarithromycin, levofloxacin and metronidazole. In a study by Lee et al. [48], in vitro susceptibility tests were performed on the same strains of H. pylori (ATCC 43504) as in this study, and the MIC values of amoxicillin, clarithromycin and metronidazole were 0.029, 0.06 and 21.6 µg/mL, respectively. These referenced MIC values suggested that oroxindin (MIC: 50 µg/mL) may have less potency than amoxicillin and clarithromycin, but higher potency than metronidazole. For some antibiotic resistant clinical strains, these MIC values were much higher. Recent studies from different countries have reported that MIC values of amoxicillin, clarithromycin and metronidazole on clinical H. pylori strains were 256 µg/mL [49,50]. The resistance mechanisms of these antibiotics do not seem to be related to the three target enzymes involved in this study, urease, shikimate kinase and aspartate-semialdehyde dehydrogenase [51]. Hence, theoretically, oroxindin, verbascoside and rosmarinic acid should have similar potencies in both resistant and non-resistant strains. The MIC value of oroxindin (50 µg/mL) could be better than that of antibiotics in the resistant strains, and produce comparable or enhanced anti-H. pylori effects. Also, as the anti-H. pylori pharmacological mechanisms are different between the antibiotics and phytochemicals, the co-administration of them may improve their efficiency and reduce side effects by decreasing the dosage typically administered.
This study demonstrated an additive inhibitory effect between oroxindin and amoxicillin. This could be due to their different pharmacological mechanisms, in which amoxicillin is a beta-lactam antibiotic that binds to penicillin-binding proteins and inhibits the synthesis of peptidoglycan of H. pylori cell walls. Whereas, this in silico study suggests that oroxindin binds to H. pylori urease, which is responsible for converting stomach urea to ammonia and neutralise the gastric acid and protect H. pylori from the surrounding strong acids [52]. The suggested pharmacological mechanisms of verbascoside and rosmarinic were also different from that of amoxicillin, however, there were no synergistic or additive effects. We believe this is due to their large differences between their potency, indicated by their MIC values ( Table 2). The very potent amoxicillin would have killed most H. pylori before the phytochemicals produced a significant effect. However, in an antibiotic resistant strain, the difference between their potency will be reduced, and hence there will be a higher chance for producing synergistic or additive effects. Certainly, further studies on clinically obtained antibiotic resistance H. pylori strains are required to prove this theoretical idea.

Materials and Methods
This study performed molecular docking and drug-like properties analysis to select phytochemicals with desired pharmacokinetic properties that may interact with three anti-H. pylori targets, urease, shikimate kinase and aspartate-semialdehyde dehydrogenase. In vitro anti-H. pylori assays were performed to test the inhibitory abilities of these selected phytochemicals. We also investigated the synergistic anti-H. pylori effects between the phytochemicals and an antibiotic.

Molecular Docking
Docking has been used to identify inhibitors from phytochemicals by analysing the interactions between the target protein and ligands. Several molecular docking software programmes have been successfully applied in computational drug design (CADD). Here, the automated docking suite, GOLD v5.5 [53,54] was used to study the binding potential between the three target enzymes, urease, shikimate kinase and aspartate-semialdehyde dehydrogenase, with the selected pool of phytochemicals. According to the Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform (TCMSP) [55], 38 herbs can inhibit urease, and these herbs contain 5015 phytochemicals. All these phytochemicals were docked to the urease X-ray crystallographic structure with PDB code 1E9Y [56]. All ligands, ions and water were removed before the docking simulations. The ChemPLP scoring function [57], genetic algorithms with 100% search efficiency, no early termination, the slow option with high accuracy and the default parameters were used for all the docking simulations. Atoms within an area of 6 Å of the cognate ligands in the X-ray crystallographic structures were set as the binding sites. For shikimate kinase, 14 herbs were identified as inhibitors, and these herbs contained 1548 phytochemicals. The structure of shikimate kinase employed was PDB: 3N2E [9]. As there was no H. pylori X-ray crystallographic structures of aspartate-semialdehyde dehydrogenase, the homology structure built by SWISS-MODEL [58] with sequence identity and sequence similarity of 48.33% and 0.41, respectively, was employed for docking. A report with details of the sequence alignment can be obtained from https://swissmodel.expasy.org/repository/uniprot/O25801#none. The number of herbs and phytochemicals involved in the aspartate-semialdehyde dehydrogenase docking simulations were 32 and 4541, respectively. The docking procedures of shikimate kinase and aspartate-semialdehyde dehydrogenase were the same as that of the urease.

Validation of Molecular Docking Methods
The accuracy of molecular docking varies substantially between different docking algorithms, scoring functions and the type of protein-ligand interactions [59]. The docking software that we used in this study was GOLD v5.5 [53,54], which has been extensively tested across various proteins and ligands by both the software supplier (The Cambridge Crystallographic Data Centre, https: //www.ccdc.cam.ac.uk) and researchers [17]. GOLD has also been successfully used to perform docking experiments on the targeted enzymes in this study, urease, shikimate kinase and aspartate-semialdehyde dehydrogenase [9,11,19,20]. However, we still believe it is necessary to further evaluate its specific accuracy on the two targeted enzymes using receiver operating characteristic (ROC) analysis. Here, both the X-ray crystallography structures of urease (PDB: 1E9Y) and shikimate kinase (PDB: 3N2E) were docked separately to the 11,421 ligands of the Zinc In Man (ZIM) database [60]. A total of 24 experimentally approved urease inhibitors and 20 shikimate kinase inhibitors with IC 50 less than 100 µM were also docked with the urease and shikimate kinase, respectively. These inhibitors served as the 'positive' hits, whereas the 11,421 ZIM ligands were 'negative' hits. ROC analysis statistically revealed the ability of the docking methods to distinguish between the 'positive' and 'negative' ligands. All the experimentally approved inhibitors were identified from the BindingDB database [61], and our literature searches on the Cochrane Library, Embase, Medline and the China National Knowledge Infrastructure database. For aspartate-semialdehyde dehydrogenase, only two experimentally approved antibacterial inhibitors were found in the literature and databases [10], and this small number of inhibitors did not provide sufficient statistical power for the ROC analysis to validate the docking procedures.

Drug-Like Properties Analysis
Drug-like properties analyses of the phytochemicals were performed using the ACD/Percepta 14.0 software [62]. The analysis involved three categories of properties: ADME, physicochemical and drug safety. ADME predicts factors that affect absorption, distribution, metabolism and excretion. These factors include human intestinal absorption (HIA), passive permeability across Caco-2 cell monolayers and plasma protein binding [63]. Physicochemical properties, such as molecular weight, the number of H-bond donors/receptors, solubility, log P and predefined lead-like categories were predicted to evaluate the pharmacokinetic and pharmacodynamic properties of the phytochemicals. Drug safety properties evaluated the toxicity of the phytochemicals, including the probability of causing mutagenicity (positive Ames), cardiotoxicity (human ether-à-go-go-related gene, hERG) and drug-drug interactions (Cytochrome P450 regioselectivity) [64,65]. The performance of all these predictions were successfully analysed by the supplier company (ACDS/Labs), which is available at https://www.acdlabs.com/products/percepta/index.php.

Susceptibility Testing
The minimum inhibitory concentration (MIC) was used to evaluate the inhibitory abilities of the selected phytochemicals against H. pylori. MIC is the lowest concentration of the phytochemicals required to inhibit the visible growth of H. pylori. The strain of H. pylori used in this study was ATCC-43504, which were recovered according to the supplier product information sheet (https: //www.atcc.org/~{}/ps/43504.ashx), wherein the frozen strain was thawed in a water bath at 37°C and inoculated in solutions (100 µL) containing 6% sheep blood Colombian agar (Huankai Microbial, Guangdong, China) and 5% foetal bovine serum (Huankai Microbial, Guangdong, China) in a facultative anaerobic environment for 3 days. The H. pylori solution were added to cryopreservation media containing brain heart infusion broth and glycerine and stored at −80°C. To identify the H. pylori strain, a combination of the test was performed. The microscopy visualisation of colony morphology was used to assess the needle-like translucent appearance of H. pylori. Three biochemical assays, urease, oxidase and catalase tests (Huankai Microbial, Guangdong, China), were performed to confirm which H. pylori strains were present.
After identification, colonies of H. pylori were added to 0.85% normal saline to produce a solution with 0.5 (approximately 1.5 × 10 8 CFU/mL) turbidity under the microbial turbidimeter (DensiCHEK Plus, bioMerieux, USA). The solution was then diluted 50 times to make an approximately 3.0 × 10 6 CFU/mL H. pylori turbidity solution.
The inhibitory effects of the selected phytochemicals were obtained using microdilution methods, in which sterile Brucella broth (HopeBio, Qingdao, China) was used to dilute each of the most concentrated phytochemical solutions by six half-dilutions to produce solutions with a range of different concentrations. Two microliters of the prepared H. pylori McFarland Turbidity solution (approximately 3.0 × 10 6 CFU/mL) was then added to each of the phytochemical solutions and incubated at 37°C.
The MIC experiments were performed in sterile 96-well microliter plates (ChunBo Biologics, Haimen, China) at 37°C. The optical density (OD) values were collected from a spectrophotometer (Multiskan™ GO, Thermo Fisher Scientific) at an absorbance of 540 nm after 72 h. These experiments were performed with quality, negative and blank controls, which contained no H. pylori or phytochemicals, only H. pylori (no phytochemicals) and only phytochemicals (no H. pylori), respectively. These experiments were also performed with a commonly prescribed antibiotic (amoxicillin) for H. pylori eradication. Amoxicillin served as a parallel positive control, and the MIC value of amoxicillin was calculated for comparison with those of the phytochemicals. All experiments followed the guidelines from the Clinical and Laboratory Standards Institute [66].
The MIC 90 value is the minimum concentration of phytochemicals that inhibit 90% of H. pylori. The inhibition percentage was calculated using the following equation: Inhibition percentage (%) = OD value of the sample−OD value of blank control OD value of the negative control−OD value of quality control × 100%

Synergistic Testing
Synergistic testing was used to study whether the additive effect of two compounds or antibiotics was superior to that of the effect of the individual compounds. An additive effect may help to reduce the dose of each compound and may reduce adverse pharmacological effects. Here, the MIC values of the phytochemicals and antibiotics obtained from the susceptibility testing were used to calculate the concentrations required for the synergistic testing. Various combinations of two compounds or antibiotics at different concentrations were tested. Their concentrations were 1/8 MIC, 1/4 MIC, 1/2 MIC, 1 MIC and 2 MIC. The checkerboard methods [67] were used to calculate the fractional inhibitory concentration (FIC) values between the phytochemicals and amoxicillin against H. pylori. FIC values are measurements of synergistic, additive or antagonistic effects. An FIC value of less than 0.5 indicated synergistic action, between 0.5 and 1.0 indicated additive action, 1.1 to 4.0 meant indifferent and larger than 4.0 indicated antagonism [68,69]. All experiments used to calculate FIC were performed in the same manner as that of the MIC. The FIC values were calculated using the following equation:

. Statistical Analysis
A paired t-test was used to calculate the p-values for comparing the OD, MIC and FIC values of the sample groups and control groups at different concentrations. A p-value of less than 0.05 was considered to be statistically significant. The SPSS software (version 22.0, IBM Corp., Armonk, NY) was used.

Conclusions
This study performed validated in silico techniques to identify a urease, a shikimate kinase and an aspartate-semialdehyde dehydrogenase inhibitor from 5015, 1548 and 4541 phytochemicals, respectively. The three identified inhibitors with appropriate drug-like properties were oroxindin, verbascoside and rosmarinic acid. In our in vitro susceptibility testing, all three phytochemicals were shown to have anti-H. pylori effects, in which oroxindin had the highest potency. Their MIC values were higher than the current anti-H. pylori treatment, amoxicillin. Hence, their potential use as monotherapy of H. pylori treatment cannot be justified here. In our synergistic testing, oroxindin demonstrated its additive effects with amoxicillin. We believe further investigations on these phytochemicals as an adjuvant therapy with current antibiotics treatment on both resistant and non-resistant strains are well-intentioned.

Conflicts of Interest:
The authors declare no conflict of interest.