Abstract
Background
Recent studies have demonstrated that corticosteroid delivered by nasal irrigation is superior to nasal spray in the treatment of chronic rhinosinusitis patients who have undergone sinus surgery. However, the local cytotoxicity of both delivery methods has not been previously evaluated. In this study we aimed to evaluate the cytotoxicity of corticosteroid prepared nasal irrigation solution and commercially available corticosteroid nasal spray.
Methods
Primary human nasal epithelial air-liquid interface 3D cultures established from nasal brushes from patients with nasal diseases with polyps, and healthy nasal mucosa (n = 6 each) were used to assess the cytotoxicity of different drug concentrations. The following drugs were screened: budesonide rinse, mometasone rinse, mometasone spray, fluticasone spray, azelastine/fluticasone spray, xylometazoline drops and benzalkonium chloride. Nasal epithelial cell characterization at passage zero was evaluated with β- Tubulin and FOXJ1 immunostaining as well as fluorescence-activated cell sorting using epithelial cell marker (EpCAM-488). To assess the drug induced cytotoxicity, an Alamar Blue assay, transepithelial electrical resistance and optical microscopy were applied.
Results
Nasal irrigation using standard therapeutic concentrations of mometasone or budesonide demonstrated less cytotoxicity when compared to nasal spray of mometasone and fluticasone in stock concentration prescribed by the manufacturer. Nasal spray diluted by a factor of 10 demonstrated similar cytotoxicity to nasal irrigation using therapeutic concentration. This study reinforces the suspicion that it is not the drugs themselves but benzalkonium chloride that is the most involved in cytotoxicity.
Conclusion
Corticosteroid irrigation is less cytotoxic than corticosteroid spray to nasal epithelial cells.
Keywords: Sinusitis, Nasal epithelium, Drug delivery system, Corticosteroid, Drug toxicity
Introduction
Many studies have focused their attention on the mechanisms involving the delivery of medications, especially corticosteroids, in the nasal mucosa affected by rhinosinusitis [1, 2].
In comparison with commercially available nasal sprays, prepared corticosteroids irrigation solutions (delivered through high-volume nasal lavage) present a better homogeneous distribution of the drug and at greater depth of reach [1], especially in patients with surgically enlarged sinus drainage. However, if the high-volume nasal lavage is inadvertently ingested by the patient, it raises concerns regarding the systemic effects the corticosteroids may cause due to the total amount of absorbable medication [2]. Furthermore, little is known about the local effects of the prepared corticosteroids irrigation solutions and the commercially available nasal sprays on the nasal mucosa. Therefore, this study aims to compare the effects of nasal lavage with diluted corticosteroids versus commercially available nasal sprays containing corticosteroids.
Materials and methods
This study was divided into two groups, six individuals with nasal disease with polyps (Chronic Rinosinusitis with Nasal Polyps-CRSwNP) and six control individuals without nasal inflammatory disease. Middle meatus nasal brushes were collected from CRSwNP group during endoscopic sinus surgery. Whereas for control subjects, the nasal brushes were collected from the middle meatus area during septoplasty procedure. All experiments and sample collection were performed as per University of British Columbia Research Ethic board approval (H19-00249). Informed consent was obtained from each subject, under the Helsinki Declaration. The diagnosis of the disease was performed according to the European position paper on Rhinosinusitis and Nasal Polyps [3].
To compare the cytotoxicity between prepared corticosteroid rinse solutions used for high-volume nasal irrigation versus commercialized nasal sprays, the following drugs were screened: budesonide rinse, mometasone rinse, benzalkonium chloride for the former and mometasone spray, fluticasone spray, azelastine/fluticasone spray, and xylometazoline drops for the latter. The two groups were tested in primary nasal epithelium cell culture from middle meatus brushes of healthy and CRSwNP patients. The BCi-NS1.1 cell line was used as in vitro 2 D monolayer model to evaluate the drugs cytotoxicity in diluted concentration.
The cytotoxicity was assessed by optic microscopy using the Alamar Blue assay in both cultures (air-liquid interface and monolayer culture) and trans epithelial electrical resistance (TEER) in air-liquid interface (ALI) culture, before and after 72 h of cell culture treatment. To characterise the human nasal ephitelium cells (hNECs) in ALI culture, β-tubulin and FOX-J1 staining was used and flow cytometry stained with EpCAMm-488 and CD45-PerCP antibodies was performed.
Antibodies: Mouse anti-Forkhead box protein J1 (FOXJ1) (cat# AMAb91254) was purchased from Atlas Antibodies, Stockholm, Sweden. Mouse anti β-tubulin IV antibody (cat#, MA5-25559) was purchased from Thermo Fisher Scientific). Alexa Fluor 488 anti-human CD326 (EpCAM) Antibody (cat# 324210), PerCP Mouse IgG1, κ Isotype Ctrl Antibody (cat# 400148), Alexa Fluor® 488 Mouse IgG2b, κ Isotype Ctrl Antibody (cat# 400329), PerCP anti-human CD45 Antibody (cat# 368506) were purchased from BioLegend. LIVE/DEAD Fixable Violet Dead Cell Stain Kit (cat# L34955) was purchased from Thermo Fisher Scientific.
Drugs: budesonide rinse (Taro, budesonide suspension for inhalation, 0.5 mg/ml, DIN 02494272, LOT: PHPF0164), mometasone rinse (NeilMed sinus rinse, mometasone furoate 0.6 mg/capsule), mometasone spray (Apotex Inc, mometasone furoate a1ueous nasal spray, 50 µg, DIN: 02403587, LOT: TL6381), fluticasone spray (Teva, fluticasone propionate nasal spray, 50 µg, DIN: 02453738, LOT/barcode 6851017801), azelastine/fluticasone spray (Dymista 137 µg/50µg, Mylan, DIN: 02432889, LOT: 2020186), xylometazoline drops (Pharmasave, 01% w/v xylometazoline hydrochloride nasal solution, DIN: 02452863, LOT: 11325) and benzalkonium chloride (benzalkonium chloride solution, NF, 50%, LOT: 645918/F).
Cell culture
Patient-derived human nasal epithelial cells (hNEC) were isolated from the middle meatus by gently scraping the surface mucosa of the middle meatus with Puritan Sterile Histobrush (cat# P25-2199 from Harmony Lab & Safety Supplies). The tip of the histobrush was cut off and collected in the ice-cold expansion media (PneumaCult-Ex, cat#05008, STEMCELL Technologies, Vancouver, BC, Canada) and transferred to lab. Samples were processed right away by harvesting cells from the brushes, spinning down cell pellet in the expansion media and seeding in T75 tissue culture flasks (cat# 83.3911.302, Sarstdet). In two weeks, hNEC were the only ones that expanded from explants to ~ 80% confluency. Non-epithelial cells and immune cells were washed out with 48 h change of media. The cultured nasal epithelial cells were either used as monolayer cultures.
Air–liquid interface model
For air liquid interface (ALI) model, 1 × 10 5 hNEC were cultured on 0.4 μm pore size PET membrane 24-well transwell inserts (cat# 3470, Corning) in the PneumaCult Ex-Plus media (cat #0540, StemCell Technologies). Cells were cultured as submerged model with media both in the apical and basal compartments at 37 °C and 5% CO 2 until they get to ~ 90% confluency. At confluency, the apical cells were air-lifted, and basal media was replaced with PneumaCult ALI media (cat# 05001, STEMCELL Technologies, Vancouver, BC, Canada). The basal medium was changed every second day to facilitate pseudo stratification of the epithelium mimicking in vivo nasal epithelia.
Epithelium cell characterization
Fluorescence-activated cell Sorting (FACS) analysis
Freshly isolated human nasal epithelial cells were stained with live dead marker, followed by EpCAM-488 and CD45-PerCP to quantify epithelial cell purity. Single cell gating was done to analyze viable cells for EpCam positivity using Gallios instrument from Beckman Coulter (Mississauga, Canada).
Immunohistochemical (IHC) analysis
ALI cultures with or without drug treatments were processed as formalin-fixed paraffin-embedded (FFPE) blocks. Four µm thick section from each block was used for immunostaining against Fox-J1 and β-Tubulin IV proteins using the antibodies listed above and with the Leica autostainer.
Microscopy
hNEC ALI cultures after 28 days of differentiation were imaged using the EVOS M5000 Imaging System (Thermo Fisher Scientific, Waltham MA, USA). Using the transillumination function, the apical surface was focused on based on the presence of motile cilia. Representative images were taken at 100X magnification near the centre of the ALI culture.
Drug concentrations
Initially the budesonide rinse, mometasone rinse, mometasone spray, fluticasone spray, azelastine/fluticasone spray, xylometazoline drops and benzalkonium chloride cytotoxicity were tested in ALI cultures in the approximate concentration of budesonide 0.5 mg/240 ml (2.08 µg /ml), mometasone 2.4 mg/240 ml (10 µg/ml), mometasone 50 µg/puff (300 µg/ml), fluticasone 50 µg/puff (190 µg/ml), azelastine/fluticazone 125 µg/50µg per puff (azelastine 725 µg/ml, fluticasone 300 µg/ml), xylometazoline 0.1% and benzalkonium chloride (BZK, BKC or BAK) 0.016%, respectively, being the commercialized drugs tested in stock concentration.
Drug exposure
To compare the cytotoxicity between prepared rinse solution with corticosteroids used for high-volume nasal irrigation to commercialized nasal sprays containing corticosteroids, the following drugs: budesonide rinse, benzalkonium chloride and the commercialized drugs: mometasone spray, azelastine/fluticasone spray, were tested in BCi-NS1.1 epithelial cell line in an in vitro 2 D monolayer model at 1/2 and 1/10 dilute treatment dose. The cytotoxicity was assessed by optic microspic observation, and by Alamar Blue assay for cell viability.
Cytotoxicity
Alamar blue
Alamar blue (Resazurin) has been widely and easily used as a cost effective assay for proliferation and viability. Under cytotoxic conditions, blue and weakly-fluorescent resazurin gets reduced to pink and highly fluorescent resorufin. After 72 h hours of drug treatment, cell free condition media were collected, cells well washed with warm sterile phosphate buffered saline (PBS), and 10% solution of alamar blue was added to the basal compartment. Cultures were incubated 37 °C and 5% CO 2 for one hour. Samples of basal condition media were taken to quantify reduction of resazurin to resorufin. The basal media with alamar blue was quantified under a plate spectrophotometer with excitation at 550 nm and emission at 590 nm. Fluorescence intensity is proportional to aerobic respiration and quantifies cytotoxic/proliferation rate between treatments.
Trans-epithelial electrical resistance (TEER)
TEER is a measurement of barrier function and integrity and was measured before and after 72 h post-treatment using an Epithelial Tissue Voltohmmeter (EVOM, World Precision Instruments, Sarasota, FL, USA) as per manufacturer’s instructions. TEER is expressed as Ω.cm 2. PBS only served as a blank measurement.
Statistical analysis
The results were assessed for normal distribution using the Kolmogorov–Smirnov test. Dunnett´s test was applied to compare a few treatments with a single control. Whereas, the Wilcoxon test was applied to compare paired groups (before and after treatment). For all tests, a p-value < 0.05 was considered significant. All statistical tests were performed using smSTATA 14.0 (Stata, College Station, Texas, US).
Results
Epithelium cell characterization
The purity of the expanded nasal epithelial cells were confirmed by flow cytometry analysis where cells signaled the presence of Epcam-488 antibody to identify epithelium cells and not the presence of CD45-PerCP to identify immune cells, shown in Fig. 1B, and by immunostaining for Fox-J1 and β-Tubulin proteins, shown in Fig. 1A.
Fig. 1.
(A) representative images of primary hNECs as differentiated pseudostratified ALI cultures at 40X magnification. Four-micron thick tissue sections were used from FFPE blocks made from hNEC- ALI cultures for β- Tubulin staining (panel A) and FOX J1 staining (panel B). β- Tubulin positivity is demonstrated in the ciliary layer, supported by red staining in the nuclei for FOX-J1 in the panel B. ALI section showing only blue nuclear counter staining indicated isotype control of the matched ALI section. (B) representative image for epithelial cell purity via FACS analysis. Freshly isolated primary nasal epithelial cells were stained with Live Dead kit, Epcam-488 and CD45-PerCP antibodies. Single cell gating was used to analyze viable cells for their expression of Epcam positivity (97.55% of total 500 gated cells). CD45-PerCP was used as immune cell marker
Drug cytotoxicity
When comparing the nasal epithelium air-liquid interface 3D culture after incubating the drug prepared rinse solution for 72 h, the cultures treated with budesonide and mometasone did not differ from the control culture (untreated) in both healthy and polyp nasal epithelium, as shown in Fig. 2A and B (Alamar Blue) and 3 A, 3B (TEER). In contrast, the cultures treated with stock commercialized drugs differed from the control culture in healthy and polyp nasal epithelium, Fig. 2, p < 0.0001 and Fig. 3, p < 0.001.
Fig. 2.
Toxicity of the drugs: budesonide rinse, mometasone rinse, mometasone spray, fluticasone spray, azelastine/fluticasone spray and xylometazoline drops to the air-liquid interface (ALI) cultures grown from healthy nasal mucosa (A) and nasal polyp cells (B) tested by Alamar Blue assay. Resazurin fluorescence was decreased when ALI cultures were treated with stock commercialized drugs: mometasone spray, fluticasone spray, azelastine/fluticasone spray and xylometazoline drops demonstrating a increased cellular toxicity caused by these drugs
Fig. 3.
Change (Δ) in trans epithelial electrical resistance (TEER) in the air-liquid interface (ALI) cultures grown from control and nasal polypose. The commercialized drugs: mometasone spray, fluticasone spray, azelastine/fluticasone spray and xylometazoline drops, demonstrated a decrease in TEER
Monolayer culture– drug concentration assessment
Mometasone spray diluted 1/10 and azelastine/fluticasone spray diluted 1/10 treated culture showed similarity in microscopy evaluation with the untreated culture (control), Fig. 4A, while mometasone stock spray and diluted 1/2 and azelastine/fluticasone stock spray and diluted 1/2 treated culture showed cytotoxicity in microscopy evaluation when compared to untreated culture (control), Fig. 4A.
Fig. 4.
(A) optical microscopy from BCi-NS1 cultures. Mometasone spray treated culture with stock and 1/2 diluted concentrations demonstrated cytotoxicity, while 1/10 diluted concentration did not demonstrate cytotoxicity when compared to control; azelastine/fluticasone spray treated culture with stock and 1/2 diluted concentrations demonstrated cytotoxicity, while 1/10 diluted concentration did not demonstrate cytotoxicity when compared to control. (B) Alamar Blue assay in the Bci.NS1.1 monolayer cultures treated with budesonide rinse, mometasone spray, and benzalkonium chloride (BZK). Treatments were evaluated at stock concentration and 1/10 diluted drugs. Budesonide rinse treatment was not toxic to the cells, whereas mometasone spray and BZK treatments had a significant difference in cell viability between stock concentration and 1/10 dilution (**** p < 0.0001). Diluting the drug concentrations decreased cellular toxicity
Mometasone spray diluted 1/10 demonstrated no statistical difference in the fluorescence using Alamar Blue assay when compared to untreated culture, Fig. 4B. Mometasone spray and benzalkonium chloride diluted 1/10 demonstrated difference in the fluorescence using Alamar Blue assay when compared to mometasone stock spray and benzalkonium chloride treated culture, p < 0.0001, Fig. 4B.
Budesonide rinse treatment was not toxic to the cells showing no difference when compared to untreated culture, Fig. 4B.
Drug cytotoxicity in nasal epithelium cells using diluted 1/10 drugs
When comparing the nasal epithelium air-liquid interface 3D culture after 72 h incubation of drug prepared rinse solution versus 1/10 diluted commercialized drugs, no significant differences in cell viability (with respect with the control culture) where observed in healthy and polyp nasal epithelium, Fig. 5A and B (Alamar Blue). Furthermore, Δ TEER values after treatment for both groups were above baseline levels, as shown in Fig. 6A and B.
Fig. 5.
Toxicity of the drugs: rinse solutions (budesonide rinse, mometasone rinse), and 1/10 diluted commercialized drugs (mometasone nasal spray, fluticasone nasal spray, azelastine/fluticasone spray and xylometazoline drops) to the air-liquid interface (ALI) cultures grown from healthy nasal mucosa (A) and nasal polyp cells (B) tested by Alamar Blue assay. Resazurin fluorescence did not change when treated ALI cultures were compared to untreated ALI culture (control) demonstrating no cellular citotoxity caused by rinse drugs or 1/10 diluted commercialized drugs
Fig. 6.
Change (Δ) in trans epithelial electrical resistance (TEER) in the air-liquid interface (ALI) cultures grown from healthy nasal mucosa (A) and nasal polyp (B) with rinse solutions (budesonide rinse, mometasone rinse), and 1/10 diluted commercialized drugs (mometasone nasal spray, fluticasone nasal spray, azelastine/fluticasone spray and xylometazoline drops). TEER remained above the baseline after 72 h of treatment
Discussion
Much of the treatment of Acute or Chronic Rhinosinusitis uses topical nasal drugs exclusively or associated with systemic drugs, and it is of great importance to evaluate their impacts on Rhinosinusitis but also their side effects on the nasal mucosa.
The local effects on the nasal mucosa will depend on several factors such as: medication, drug concentration, time of use, means used to deliver the medication [4, 5], presence or absence of preservatives or drug stabilizers, the rhinosinusal region, the mode that the medication is applied in relation to the positioning of the head [6] and the presence or absence of sinus surgery to enlarge the sinus drainage ostia [6, 7].
High concentrations of drugs often become cytotoxic, even washing with water if the solution is not accompanied by salts that make it osmotically neutral (isotonic 0.9% saline) for the cell also becomes cytotoxic [8]. The concentration of corticosteroids has an inverse correlation with cell viability [9].
It is important to highlight that drugs associated with other medications, preservatives, or depending on the vehicle used, can add or enhance cytotoxicity. Benzalkonium chloride is a common preservative in commercialized corticosteroid nasal sprays, and is damaging to epithelium cells [10, 11].
As previously highlighted, the concentration of the drug is fundamental for the treatment but also for its cytotoxicity, nasal sprays present great heterogenicity in their distribution, presenting a higher concentration in the anterior mucosa of the nose [12], especially when the nasal spray is applied incorrectly. Depositing most of the drug in the septal region or lateral nasal wall [13], make these areas at greater risk for cytotoxicity and can lead to changes in epithelial quality and epistaxis [14].
The concentration of marketed corticosteroid nasal sprays is approximately 300 µg/ml for mometasone spray (50 µg/puff) and 195 µg/ml for fluticasone spray (50 µg/puff). While the prepared solution for corticosteroid rinse is approximately 10 µg/ml for mometasone and 2.08 µg/ml for fluticasone. Even with a 1/10 dilution of the nasal spray, this concentration was still above the concentration of the prepared irrigation solution.
In this study, we demonstrated that irrigation solutions prepared with corticosteroids are less cytotoxic to nasal epithelium cells than the commercialized nasal spray with corticosteroids. When corticosteroid spray solutions are diluted 10 times, they begin to show similar behaviour as the control and prepared corticosteroids irrigation solutions.
When benzalkonium chloride was evaluated alone in a monolayer culture, it was shown to be extremely cytotoxic, but reduced in a 1/10 dilution. Those results suggest that this component directly impacts the cytotoxicity of commercialized corticosteroids nasal sprays. In contrast, the 10 times diluted budesonide irrigation solution showed no difference in cytotoxicity when compared with the therapeutic dose.
The cytotoxicity of the solution in commercialized drops containing xylometazoline preserved in benzalkonium chloride showed similar toxicity as the nasal sprays preserved in benzalkonium chloride, suggesting that mainly the preservative substance is involved in the cytotoxicity. The fact that both the commercial azelastine/fluticasone and nasal sprays are preserved in benzalkonium chloride and both have statistically significant cytotoxicity effects reinforces the suspicion that the preservative is mainly involved in cytotoxicity rather than the drugs themselves.
Further studies are necessary to identify a safer preservative to avoid nasal mucosa cytotoxicity and associations between drug cytotoxicity and comorbidities that impair ciliary function or the tight junctions.
We concluded in this study that when nasal medication is delivered in its full concentration, commercially available corticosteroids nasal sprays present greater local cytotoxicity to the nasal epithelial cells when compared to corticosteroid rinse solutions for high-volume irrigation.
Author contributions
RP, design, experiments, writing and review. GKS, design, experiment and writing. TJFG, experiments, writing. DRD, design, experiments and review. AP, design and experiments. AT, design and review.
Funding
This study was supported by Capes PRINT/2024, Brazil and Mitacs Accelerate IT26548, Canada.
Declarations
Conflict of interest
The authors declare no conflict of interest.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Muenkaew Y, Tangbumrungtham N, Roongpuvapaht B, Tanjararak (2023) Comparison of sinus distribution between nasal irrigation and nasal spray using fluorescein-labelled in patients with chronic rhinosinusitis: a randomised clinical trial. Clin Otolaryngol 48(2):286–293 [DOI] [PubMed] [Google Scholar]
- 2.Smith KA, French G, Mechor B, Rudmik L (2016) Safety of long-term high-volume sinonasal budesonide irrigations for chronic rhinosinusitis. Int Forum Allergy Rhinol 6(3):228–232 [DOI] [PubMed] [Google Scholar]
- 3.Fokkens WJ, Lund VJ, Hopkins C et al (2020) European position paper on Rhinosinusitis and nasal polyps 2020. Rhinology 58(Suppl S29):1–464 [DOI] [PubMed] [Google Scholar]
- 4.Megow A, Bouras G, Ooi EH, Vreugde S, Psaltis A, Wormald PJ (2024) Double blind randomized controlled trial measuring the efficacy of nebulized steroid at half dose in comparison to high volume squeeze bottle steroid irrigation in management of patients with chronic rhinosinusitis. Int Forum Allergy Rhinol 14(8):1375–1377 [DOI] [PubMed] [Google Scholar]
- 5.Qi Y, Hong J, Wu D (2024) Necessary Factors for Efficient Frontal Sinus Irrigation After Endoscopic Sinus Surgery: A Systematic Review (2024) J Otolaryngol Head Neck Surg 53: 19160216241269375 [DOI] [PMC free article] [PubMed]
- 6.Snidvongs K, Chaowanapanja P, Aeumjaturapat S, Chusakul S, Praweswararat P (2008) Does nasal irrigation enter paranasal sinuses in chronic rhinosinusitis? Am J Rhinol 22(5):483–486 [DOI] [PubMed] [Google Scholar]
- 7.Grayson JW, Harvey RJ (2019) Topical corticosteroid irrigations in chronic rhinosinusitis. Int Forum Allergy Rhinol 9(S1):S9–S15 [DOI] [PubMed] [Google Scholar]
- 8.Jiao J, Yang J, Li J, Li Y, Zhang L (2020) Hypertonic saline and seawater solutions damage sinonasal epithelial cell air-liquid interface cultures. Int Forum Allergy Rhinol 10(1):59–68 [DOI] [PubMed] [Google Scholar]
- 9.Kuppermann BD, Zacharias LC, Kenney MC (2014) Steroid differentiation: the safety profile of various steroids on retinal cells in vitro and their implications for clinical use (an American Ophthalmological Society thesis). Trans Am Ophthalmol Soc 112:116–141 [PMC free article] [PubMed]
- 10.Mitra S, Tati V, Basu S, Shukla S (2024) Role of mesenchymal stem cell-derived conditioned medium in modulating the Benzalkonium Chloride-Induced Cytotoxic effects in cultured corneal epithelial cells in Vitro. Curr Eye Res 49(8):815–825 [DOI] [PubMed] [Google Scholar]
- 11.Park EJ, Jin SW, Kang MS et al (2022) Pulmonary inflammation and cellular responses following exposure to benzalkonium chloride: potential impact of disrupted pulmonary surfactant homeostasis. Toxicol Appl Pharmacol 440:115930 [DOI] [PubMed] [Google Scholar]
- 12.Harvey RJ, Goddard JC, Wise SK, Schlosser RJ (2008) Effects of endoscopic sinus surgery and delivery device on cadaver sinus irrigation. Otolaryngol Head Neck Surg 139(1):137–142 [DOI] [PubMed] [Google Scholar]
- 13.Verma RR, Verma R, Sinonasal (2023) Irrigation after endoscopic sinus surgery - past to Present and Future. Indian J Otolaryngol Head Neck Surg 75(3):1–8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wu EL, Harris WC, Babcock CM, Alexander BH, Riley CA, McCoul ED, Epistaxis, Risk (2019) Associated with Intranasal Corticosteroid sprays: a systematic review and Meta-analysis. Otolaryngol Head Neck Surg 161(1):18–27 [DOI] [PubMed] [Google Scholar]