Purpose To compare the effect of argon laser peripheral iridoplasty (ALPI) and conventional medical therapy in the immediate treatment of acute primary angle closure (APAC) using anterior segment optical coherence tomography (ASOCT).
Methods In this single tertiary centre, prospective comparative study, we randomised 30 consecutive patients with unilateral APAC into two groups: ALPI and medical treatment (n=15 each). Immediately before and 1 h after either intervention, ASOCT imaging was performed. Custom software was used to measure pupil diameter, anterior chamber depth, iris curvature (I-Curv), iris area (I-Area), and the angle opening distance (AOD750), trabecular iris space area (TISA750) and the iris thickness at 750 µm from the scleral spur. The main outcome measure was the change in anterior segment biometrical parameters.
Results The mean age of the patients was 62.8±7.7 years; 13 (43.3%) were male. APAC eyes treated with ALPI had a larger increase in AOD750 (p=0.002) and TISA750 (p=0.006); a smaller increase in I-Area (p=0.004) and a decrease in I-Curv (p=0.001) after treatment compared with those eyes which received medical therapy. An optimal model consisting of age, gender, pretreatment and post-treatment pupil diameter, treatment modality and pretreatment I-Curv explained 53.2% of the variance in AOD750 change after treatment, with the treatment modality accounting for 35.0% and I-Curv accounting for 12.4% of the variability.
Conclusions We observed a greater increase in angle width after ALPI compared with after medical treatment in eyes with APAC. Treatment modality and pretreatment I-Curv were the most significant predictors of angle width change after treatment.
Statistics from Altmetric.com
Acute primary angle closure (APAC) is an ophthalmic emergency and prompt treatment is necessary to reduce the rates of vision-threatening consequences.1 ,2 If left untreated, persistently high intraocular pressure (IOP) can lead to synechial angle damage and glaucomatous optic neuropathy.1 ,3 A long duration of raised IOP has been associated with retinal nerve fibre layer thinning,4 ,5 significant visual field defects6 ,7 and disc pallor.2 Conventional treatment of APAC involves reducing the IOP with systemic and topical medications, such as acetazolamide and β-blockers, followed by laser peripheral iridotomy to relieve the pupil block.8 ,9 However, medical therapy is ineffective in up to 10% of patients with APAC,9 and systemic medications may cause systemic side effects including blood dyscrasias, urinary retention and metabolic acidosis.10–14
In place of systemic IOP-lowering medications, the use of argon laser peripheral iridoplasty (ALPI) has been described in the treatment of APAC.15 ,16 ALPI has been shown to lower IOP more rapidly than conventional systemic medications in APAC eyes, though the reason for this increased efficacy has not been clearly elucidated.16 This procedure has also been used in the treatment of plateau iris syndrome17 and angle closure due to abnormal iris configurations or an enlarged lens.18–21 The presumed mechanism of ALPI involves the application of contraction burns to the peripheral iris, with consequent widening of the angle and increased drainage of aqueous via the trabecular meshwork.22 ,23 However, this proposed mechanism has not been objectively quantified or confirmed in APAC eyes.
The anterior segment optical coherence tomography (ASOCT) is a non-contact optical system that allows for rapid in vivo cross-sectional images of the anterior segment to be obtained using low-coherence interferometry.24 Furthermore, unlike gonioscopy, which is highly subjective, semiautomated and highly reproducible measurements of the anterior segment parameters can now be obtained with customised software.25–28 Unfortunately, patients with APAC usually present as an emergency after office hours, resulting in logistic and procedural challenges associated with performing anterior segment imaging during an APAC attack. Despite these difficulties, we provided novel data describing the anterior segment parameters of APAC eyes during an attack, before treatment was administered.29 In the current study, we subsequently sought to compare the changes in the anterior segment parameters of APAC eyes after two treatment modalities, ALPI and conventional medical therapy, which will provide an insight into the IOP-lowering mechanisms of each treatment modality.
This was a single tertiary centre, parallel group, prospective randomised comparative study of consecutive patients who presented with unilateral APAC to the Department of Ophthalmology at the National University Health System in Singapore. Informed consent was obtained from all participants before recruitment into the study. The patient was informed that participation in the study would mean being randomised to either medical treatment or ALPI. Only study personnel who have completed the Collaborative Institutional Training Initiative certification obtained informed consent from the patient. The study was conducted in accordance with the principles of the Declaration of Helsinki and the Good Clinical Practice guidelines.
Presence of at least two of these symptoms: headache, nausea and/or vomiting, ocular or periocular pain, blurring of vision
Presenting IOP exceeding 21 mm Hg (as measured by Goldmann applanation tonometry)
Presence of at least three of these signs: conjunctival injection, corneal epithelial oedema, shallow anterior chamber and mid-dilated unreactive pupil
Closed anterior chamber angles in four quadrants found on gonioscopy.
All patients had an oedematous cornea obscuring iris details, which rendered immediate laser peripheral iridotomy neither possible nor safe. We excluded patients with previous ocular surgery, and those with secondary angle closure, such as lens intumescence or subluxation, iris neovascularisation and a history of uveitis. Patients who had systemic contraindications to medical therapy (including renal impairment, sulfur allergy, asthma and heart failure), pre-existing corneal opacities obstructing laser access to more than one quadrant of the peripheral iris and single-eyed patients were also excluded. Patients who could not cooperate with ASOCT imaging or had poor quality ASOCT scans with indeterminate scleral spurs were not eligible for the study. Demographic data were recorded for each patient recruited, and the time interval between the onset of symptoms and presentation was noted.
At presentation and before therapeutic interventions, a fellowship-trained glaucoma specialist performed slit lamp examination of the anterior segment on all participants. Gonioscopy was performed under dark conditions with a four-mirror goniolens (Volk Optical, Mentor, Ohio, USA), using a narrow beam of light and avoiding the light being directed within the pupil. When the posterior trabecular meshwork was not visible during gonioscopy, the angle quadrant was considered closed.30 Before and 1 h after therapeutic interventions, three readings of IOP were measured with Goldmann applanation tonometry, and the average was calculated.
All study participants were randomised to receive either ALPI or conventional medical therapy. When the patient was recruited into the study and when informed consent was taken, the attending ophthalmologist and the patient were not aware of the subsequent treatment modality before randomisation. A computer-generated random allocation sequence at 1:1 ratio to one of the two treatment groups (with no blocks and restrictions) was implemented by concealing the treatment group within sealed envelopes just before treatment was administered. A research assistant who was masked to the study subjects performed this randomisation process. After recruitment into the study, co-investigator ophthalmologists assigned the participants to their treatment groups after opening the sealed envelope. While it was not possible to mask the patient or the ophthalmologist to the treatment modality, the outcome assessor (ie, the observer performing anterior segment measurements) was masked.
After randomisation and the allocation of the treatment group, each subject will undergo either ALPI or conventional medical therapy.
Argon laser peripheral iridoplasty
For patients randomised to ALPI, one drop of topical pilocarpine 4% was instilled into the APAC eye 15 min before laser was performed. A fellowship-trained ophthalmologist performed ALPI with an Abraham lens (Ocular Instruments, Bellevue, Washington, USA) under topical anaesthesia. The laser settings used were 200 ms duration, 100–150 µm spot size and 400–700 mW power.21 ,31 The laser beam was focused as peripherally on the iris as possible, without causing corneal burns. Localised iris contraction was the end point for laser treatment, and the laser energy was increased if there was no visible iris contraction. The laser energy was reduced if there was charring of the iris or the laser produced a ‘pop’ sound.16 The laser burns were placed in all four quadrants, with approximately 25–30 burns over 360°.
Conventional medical therapy
For patients randomised to conventional medical therapy, topical pilocarpine 4% four times daily, topical timolol maleate 0.5% twice daily and topical brimonidine tartrate 0.1% twice daily were instilled. Intravenous acetazolamide 500 mg was administered followed by 250 mg of oral acetazolamide prescribed four times daily; with oral potassium supplement 600 mg twice daily until IOP normalised.9 If the IOP was higher than 60 mm Hg at presentation, 200 mL of 20% intravenous mannitol will also be administered over 1 h.
Laser peripheral iridotomy
After immediate ALPI or conventional medical therapy, all patients underwent laser peripheral iridotomy as the definitive treatment, once the corneal oedema had reduced sufficiently and there was an adequate view of the iris.
ASOCT imaging and anterior segment measurements
Before and 1 h after therapeutic interventions, ASOCT imaging (Visante, Carl Zeiss Meditec, Dublin, California, USA) was performed by an experienced operator. Scans of the horizontal angles and centred on the pupil were obtained in a dark room (0 lux) with the standard anterior segment single-scan protocol, which produces 256 scans in 0.125 s. An image with the best quality was obtained by optimising the polarisation, adjusting the noise and image saturation, then choosing the image with the least artefacts for analysis.26 ,29
We used a previously validated technique for measurements of the anterior segment, which was performed by a single observer using customised software (Anterior Segment Analysis Program, National University Hospital, Singapore).27 ,29 The only observer input required was the accurate localisation of the scleral spurs, after which the anterior segment parameters were automatically calculated by the software (figure 1). The observer was masked to the treatment status (pretreatment vs post-treatment) and the treatment modality. The anterior chamber width was defined as the interscleral spur distance.32 The anterior chamber depth was the distance between the anterior lens surface and the corneal endothelium along the ocular axis.33 The anterior chamber area (ACA) was the area bordered by the anterior surface of the iris, the anterior surface of the lens within the pupil and the posterior surface of the cornea. The anterior chamber volume (ACV) was calculated by rotating ACA 360° around a vertical axis through the centre of the ACA.34 Angle opening distance at 750 µm from the scleral spur (AOD750) was defined as the perpendicular distance from the trabecular meshwork at 750 µm anterior to the scleral spur to the iris. Trabecular iris space area at 750 µm anterior to the scleral spur (TISA750) was the trapezoidal area bordered inferiorly by the anterior iris surface, superiorly by the inner corneoscleral wall, anteriorly by AOD750 and posteriorly by a perpendicular line to the plane of the inner corneoscleral wall drawn from the scleral spur to the opposing iris.26 Lens vault was the perpendicular distance from a horizontal line between the two scleral spurs to the anterior pole of the lens.35 Iris area (I-Area) was the cross-sectional area of the iris from the pupil to the scleral spur. Iris thickness at 750 µm from the scleral spur was measured. Iris curvature (I-Curv) was measured by drawing a line between the most central to the most peripheral points of the iris pigment epithelium, and calculating the perpendicular distance from this line to the posterior iris surface at the point of maximum convexity.25 Pupil diameter was defined as the length of a line drawn between the pupil edges of the iris.29 The intraclass correlation coefficient was >85% for all anterior segment measurements performed in our study using the Anterior Segment Analysis Program.
We performed statistical analyses using Stata Software V.12.1 (StataCorp, LP, College Station, Texas, USA). For descriptive statistics, we calculated the mean values and SDs for continuous data. We compared the demographic data between the two treatment groups (medical treatment and ALPI) and between included and excluded subjects, using the Mann-Whitney U test for continuous variables, and the Fisher's exact test for categorical variables. The Mann-Whitney U test was used to compare the IOP between APAC eyes in the two treatment groups, and the Wilcoxon signed-rank test was used to compare the IOP before treatment and 1 h after treatment. The adjusted mean values for differences in anterior segment measurements were calculated between 1 h post-treatment and pretreatment ASOCT images, after controlling for age, gender, the corresponding pretreatment anterior segment parameters and pupil diameter, as pupil diameter has been shown to significantly alter anterior segment characteristics. These differences in anterior segment measurements were compared between APAC eyes in the two treatment groups (medical treatment and ALPI) using linear regression models. A forward linear regression model, with the difference between 1 h post-treatment and pretreatment AOD750 as the dependent variable, was used to determine the most important pretreatment variable associated with a change in AOD750. We included pretreatment and post-treatment pupil diameters, age and gender as fixed variables in each iteration. The threshold to stop iterating was p<0.05. The effect size was calculated for each 1 mm increase in AOD750. We calculated the Pearson's correlation coefficient for the change in IOP and AOD750 1 h after treatment compared with before treatment. Statistical significance was set at p<0.05. Assuming that the anterior chamber angle was closed at presentation before treatment, with a sample size of 15 in each group with a 0.05 two-sided significance level, the statistical power was at least 85% for detecting a change in AOD750 of +0.12 mm after ALPI (ie, post-treatment AOD750 equivalent to primary angle closure suspects)27 and + 0.05 mm after medical therapy, with a SD of 0.06 mm or less.27 ,36
We recruited 30 patients who were randomised to receive either conventional medical therapy (n=15) or ALPI (n=15). The mean age of the patients (78.9%) included in this study was 62.8±7.7 years, and 13 patients (43.3%) were male. The patients were predominantly Chinese (26 Chinese, 2 Malay and 2 Indian). There were no significant differences in age, gender, race, duration of symptoms or presenting IOP between patients in the two treatment groups (medical treatment and ALPI), and the mean interval between the onset of APAC symptoms and presentation was 35 h (range 2–144 h). The mean IOP was 54.4±10.1 mm Hg before treatment and 32.1±14.1 mm Hg 1 h post treatment (p<0.001). The mean decrease in IOP was not significantly different between APAC eyes in the ALPI and medical treatment groups (27.5±10.8 mm Hg vs 22.6±13.1 mm Hg, p=0.206). All patients had successful treatment and cessation of APAC attack in both groups, with the corneal oedema resolving sufficiently to allow peripheral iridotomy without additional medical or laser treatment and the IOP was less than 21 mm Hg in all eyes 6 h after primary treatment. None of the patients developed major known or unintended side effects from either treatment.
The mean age and anterior segment parameters of APAC eyes before treatment are shown in table 1. There were no significant differences in anterior segment measurements between APAC eyes in the two treatment groups (medical treatment and ALPI) before therapeutic interventions.
Assessing the change in anterior segment measurements between APAC eyes 1 h after treatment and APAC eyes before treatment, the eyes which underwent ALPI had a larger increase in ACV (p=0.001), ACA (p=0.003), AOD750 (p=0.002) and TISA750 (p=0.006); a smaller increase in I-Area (p=0.004) and a decrease in I-Curv (p=0.001) compared with those which received medical treatment, after adjusting for age, gender, pupil diameter and the corresponding pretreatment anterior segment parameters (table 2 and figure 2).
In the forward logistic regression model for pretreatment parameters with the change in AOD750 as the dependent variable (table 3), after including age, gender and pretreatment and post-treatment pupil diameters in the model, the first iteration identified treatment modality as the most significant factor determining the change in AOD750 (p=0.002). In the second iteration, the next most significant factor determining the AOD750 change was the pretreatment I-Curv (p=0.028), after including age, gender, pretreatment and post-treatment pupil diameters and treatment modality in the model. In the third iteration, no other pretreatment parameters were significant determinants of the AOD750 change. The model consisting of age, gender, pretreatment and post-treatment pupil diameters, treatment modality and pretreatment I-Curv explained 53.2% of the variability in the AOD750 change, with the treatment modality accounting for 35.0% and the pretreatment I-Curv accounting for 12.4% of this variability. For every 1 mm increase in I-Curv, the per-mm increase in AOD750 was 0.42 (95% CI 0.05 to 0.79). Eyes that underwent ALPI had a larger per-mm increase in AOD750 of 0.12 (95% CI 0.05 to 0.20) compared with those receiving medical treatment. Comparing APAC eyes 1 h after treatment and APAC eyes before treatment, the difference in IOP and AOD750 were negatively correlated (Pearson's correlation coefficient=−0.44, p=0.014).
Recent advances in anterior segment imaging have led to an improved understanding of angle closure mechanisms, and the realisation that successful treatment of angle closure necessitates modification of the anatomical factors responsible.37–39 To our knowledge, this is the first study to assess the effect of ALPI and medical treatment on APAC eyes with anterior segment imaging, and yielded three interesting findings. First, we provide evidence that the drainage angle is wider after ALPI compared with after medical treatment. Second, treatment modality and pretreatment I-Curv were the most significant predictors of angle width change after treatment, accounting for 35% and 12% of the variability in AOD750 change, respectively. The use of ALPI as the primary treatment for APAC and a larger pretreatment I-Curv predicted a greater increase in AOD750 after therapeutic interventions. Third, we have also detected several significant differences in anterior segment parameters between APAC eyes treated with ALPI and those that received conventional medical therapy.
In our study, anterior segment imaging has provided novel data which objectively confirmed that ALPI resulted in a larger increase in angle width measurements (AOD750 and TISA750) compared with conventional medical therapy. Moreover, the treatment modality for APAC accounted for about a third of the variability in AOD750 change, and the choice of ALPI over medical treatment was the most significant predictor of greater angle widening after treatment. Among the ASOCT angle width parameters, AOD750 was identified as the most useful angle measurement for detecting gonioscopically narrow angles,40 and was also negatively correlated with IOP in our study. In a prospective randomised controlled trial conducted more than 10 years ago, which included 73 APAC eyes of 64 consecutive patients, Lam et al16 found that ALPI was more effective than conventional systemic medications in lowering the IOP within the first 2 h of treatment, when immediate laser peripheral iridotomy was neither possible nor safe due to corneal oedema. In our study, anterior segment imaging has demonstrated that ALPI leads to increased angle widening, explaining why ALPI lowers IOP more rapidly than medical therapy. This may also explain the efficacy of ALPI in eyes with severe APAC which are refractory to medical therapy.31 ,41 Conventional medical treatment and ALPI lower the IOP through two distinctly different mechanisms. The minimal angle widening after medical treatment suggests that IOP is lowered as a consequence of reduced aqueous secretion or osmotic shrinkage of the vitreous. In comparison, ALPI has a direct effect on the drainage angle itself by contraction burns on the iris which mechanically widen the angle and increase aqueous outflow.19 ,22 Lai et al42 observed that in 41 APAC eyes of 39 patients, after a mean follow-up period of 15 months, there was a reduced proportion of ALPI-treated APAC eyes with ≥90° peripheral anterior synechiae compared with those that received conventional medical therapy, though this attained statistical significance only when the analysis was by eyes, and not when the outcome was analysed by patients. A larger study with a longer follow-up duration would clarify whether the increased angle-widening from ALPI would result in reduced peripheral anterior synechiae formation in APAC eyes.
When analysing factors which affect the variability of angle width change after treatment, we found that pretreatment I-Curv was the second most significant predictor, with a larger I-Curv predicting a greater increase in AOD750. Interestingly, we previously showed that APAC eyes had a smaller I-Curv compared with fellow eyes before therapeutic interventions.29 This was contrary to expected results as pupil block is widely regarded as the primary mechanism responsible for APAC, and is associated with a convex forward bowing configuration of the iris,43 measured as an increased I-Curv. Similarly, one would expect that pupil block is more significant in eyes with severe APAC, which are less responsive to treatment. In line with this logical discourse, a larger pretreatment I-Curv should be associated with a smaller increase in angle width after treatment, which contradicts our findings. There are few possible explanations for this discrepancy. First, non-pupil block angle closure mechanisms may be significant in the course of APAC, determining the efficacy of therapeutic interventions in widening the angle during the acute phase. Second, the reduced I-Curv could indicate the presence of iris ischaemia,44 and is a possible marker of more severe APAC. Nevertheless, I-Curv was only a minor predictor of angle widening, accounting for 12% of the variability in AOD750 change. Pretreatment parameters explained only about half of the variability in AOD750 change in APAC eyes. Other factors, such as the duration of raised IOP and the dynamic properties of the choroid45 and iris,46 are also likely to be significant in determining the response to APAC treatment.
Our study found several other significant differences in iris measurements and anterior segment dimensions between APAC eyes that underwent ALPI and those that received medical treatment, which have not been previously reported. Compared with subtypes of asymptomatic angle closure, APAC eyes have the thickest iris and smallest anterior segment dimensions.36 In our study, ALPI resulted in a decrease in I-Curv, unlike conventional medical treatment which led to an increase in I-Curv. I-Area was increased after both treatment modalities, and could be the consequence of iris swelling following ischaemia, due to reperfusion in both eyes. The increase in I-Area was significantly greater after medical treatment compared with after ALPI, and this may be explained by the iris thinning caused by ALPI. The differing effects of these two treatment modalities on iris configuration and measurements may facilitate anterior chamber angle widening after ALPI. We also found that anterior segment dimensions (ACA and ACV) decreased after medical therapy, likely as a consequence of decreased aqueous secretion, but these measurements increased after ALPI. ALPI thins the iris, which may explain the increase in ACA and ACV after treatment. In a population-based study including 1067 Chinese subjects, Foo et al33 showed that ACA and ACV were strong determinants of angle width. Wang et al47 also reported that ACA/ACV ratio was the most prominent contributor to angle width variation for Chinese and Caucasian persons. These longitudinal changes in anterior segment measurements have not been assessed before and after therapeutic interventions in APAC eyes. Similar changes in anterior segment morphology are seen after laser peripheral iridotomy, which leads to an increase in AOD500, ACA and ACV, and a decrease in I-Curv.37 However, these observations require larger studies to confirm their significance in the pathogenesis and treatment response of APAC.
Despite our findings that ALPI resulted in a larger increase in angle width compared with conventional medical treatment, the latter remains the mainstay of APAC treatment, with resolution of APAC within 24 h in approximately 90% of patients.9 On the other hand, ALPI is associated with a potential risk of corneal decompensation48 and ischaemic pupil damage,49 though these were not reported in more recent studies on ALPI involving APAC eyes.15 ,42 The lack of availability of laser facilities also limits the application of this therapy to patients with APAC in developing countries. Our study, which was powered to detect a difference in angle width between the two treatment modalities, did not find a significant difference in the IOP responses between ALPI and medical treatment, likely a consequence of insufficient statistical power. In addition, patients who present early, as in the case of our study participants, would be expected to respond well to both treatment modalities. In the earlier study by Lam et al,16 patients in both treatment groups received topical IOP-lowering medications, but in our study, these were administered only to the patients randomised to conventional medical treatment. This may further explain why Lam et al found that ALPI lowered the IOP more rapidly than conventional medical treatment, while this was not evident in our study. Nevertheless, in tertiary centres with round-the-clock availability of laser facilities, the merits of ALPI should be considered, especially when faced with patients with APAC who present late. ALPI is effective in almost all cases of APAC, and is associated with a higher likelihood of a successful outcome than conventional medical treatment in patients with severe APAC.31 ,41 For patients with comorbidities, systemic medications may have potentially serious complications, such as metabolic acidosis and urinary retention,10–12 and ALPI would be a safer primary treatment option for APAC in such patients. Furthermore, ALPI prevents the recurrence of acute angle closure in patients who have angle closure mechanisms other than pupil block, such as plateau iris.50 ,51
Despite the logistical challenges associated with performing prompt anterior segment imaging before and after therapeutic interventions, while ensuring that patient safety was not compromised, our study was sufficiently powered to detect differences in anterior segment measurements between the two treatment modalities. However, we did not have sufficient statistical power to detect a difference in IOP between the two treatment groups. In our study, only the horizontal scans were analysed, as the visibility of the angle structures were generally superior for the nasal and temporal quadrants compared with the superior and inferior quadrants.52 This was compatible with most other studies which assessed the anterior segment using ASOCT.26 ,33 ,37 Although the axis of imaging was standardised in the APAC eyes, the locations of the iridoplasty burns were variable, hence it was not possible to consistently image either the iridoplasty burns or the area between the burns. This should be noted in the interpretation of the post-ALPI anterior segment measurements in our study. Nonetheless, we recognise that, with improvements in ASOCT technology, image acquisition and analysis could be improved in future studies. It was also not possible to distinguish between APAC secondary to pupil block and APAC secondary to plateau iris at presentation, as the diagnosis of plateau iris is made only in the presence of a patent laser iridotomy. Lastly, our study participants were predominantly Chinese Asians, and patients with APAC of other races may respond differently to treatment.
In conclusion, ASOCT imaging has provided novel insights into the anterior segment biometrical parameters of APAC eyes before and after therapeutic interventions. Compared with conventional medical treatment, ALPI resulted in a larger increase in angle width measurements. The choice of ALPI over medical treatment and a larger pretreatment I-Curv were the most significant predictors of angle widening after therapeutic interventions, explaining about half of the variability in AOD750 change. These findings may explain the increased efficacy of ALPI in severe APAC compared with medical treatment. Further studies are indicated to determine whether a wider anterior chamber angle after treatment translates into a better clinical outcome for APAC eyes, and to identify the role of other anatomical, dynamic and systemic factors in the pathogenesis and treatment response of APAC.
Contributors All authors met the ICJME criteria: substantial contributions to conception and design, acquisition of data, or analysis and interpretation of data; drafting the article or revising it critically for important intellectual content; and final approval of the version to be published.
Funding National Medical Research Council, Singapore (NIG10nov019).
Competing interests None declared.
Patient consent Obtained.
Ethics approval Domain Specific Review Board of the National Healthcare Group.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Additional data may be obtained from the corresponding author.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.