Association of eNOS Polymorphisms with Primary Angle-Closure Glaucoma

Abstract

Purpose.: Recently, several studies have investigated genetic associations between Cytochrome P450 (CYP1B1), Endothelial nitric oxide synthase (eNOS), and Neurotrophin-4 (NTF4) with primary angle-closure glaucoma (PACG) in various ethnic groups. We investigated the association of these candidate genes with PACG in samples from Australia and Nepal.

Methods.: A total of 235 patients with PACG (106 Nepalese and 129 Australian) and 492 controls (204 Nepalese and 288 Australian) was included. Tag single nucleotide polymorphisms (SNPs) were selected to cover the majority of common variation within the candidate genes and genotyped in DNA extracted from peripheral whole blood. Allele and haplotype analyses were conducted in PLINK. Bonferroni correction was applied for the total number of SNPs in this study (P = 0.05/15 = 0.003).

Results.: In the Australian cohort, one eNOS SNP rs3793342 showed significance association with PACG after Bonferroni correction (P value of 0.003, odds ratio [OR] 0.5, 95% confidence interval [CI] 0.3–0.8). After adjusting the results for sex and age, SNPs rs3793342 and rs7830 showed significance after Bonferroni correction (P value of 0.001 and 0.003, respectively). The eNOS haplotype of all 7 typed SNPs showed significant association with a global P value of 0.019, with the CGCAATC haplotype giving a specific P value of 0.008 and OR of 1.5 (95% CI 0.9–2.4). In the Nepalese cohort, SNPs in CYP1B1 and NTF4 genes showed borderline association with PACG, but did not survive Bonferroni correction.

Conclusions.: Our data support the involvement of common variations in eNOS with PACG pathogenesis. Differences were observed in the two populations studied, and additional replication studies in other populations are necessary to confirm these associations.

Introduction

Glaucoma is the leading cause of irreversible blindness worldwide, ¹ with primary open angle glaucoma (POAG) being the most prevalent subtype. However, primary angle-closure glaucoma (PACG) is responsible for almost half of all glaucoma blindness. ² Family history and ethnicity are important risk factors for PACG. The disease is more prevalent among Asians and Eskimos than in Caucasians and Africans. ³ In addition, first degree relatives of patients with PACG have a higher probability of narrow angles. ^4,5 The number of patients with PACG worldwide is expected to rise from approximately 16 million in 2010 to 21 million by the year 2020, ⁶ with the majority of bilaterally blind PACG patients expected to be of Asian ethnicity. ⁷  

Patients with PACG share certain anatomic biometric features, including short anterior chamber depth with narrowing in the iridocorneal drainage angle, increased lens thickness, and anterior apposition of the lens, short axial length, and hyperopic (farsightedness) refractive error. ⁸  

The leading cause of PACG is from obstruction of the trabecular meshwork, which leads to the accumulation of aqueous humor and subsequent increase in intraocular pressure. This, in turn, causes progressive destruction of the optic nerve with corresponding loss of the peripheral visual field. ⁶  

PACG is a complex heterogeneous disease. Recently, Vithana et al. conducted a large two-staged genome-wide association study and reported three susceptibility loci: rs11024102 in PLEKHA7, rs3753841 in COL11A1, and rs1015213 located between PCMTD1 and ST18. ⁹ Before this, evaluation of PACG genetics has been conducted through evaluation of candidate genes chosen for biologic plausibility or association with similar phenotypes. Several genes have been reported previously to be associated with other subtypes of glaucoma. For example Cytochrome P450 (CYP1B1) is well known to cause primary congenital glaucoma (PCG), ^10,11 while Endothelial nitric oxidase synthase (eNOS) ^12,13 and Neurotrophin-4 (NTF4) have been implicated in POAG. ¹⁴ As all three types of glaucoma (POAG, PCG, and PACG) are characterized by destruction of the optic nerve and progressive increase in the cup-to-disc ratio, generally with elevated intraocular pressure, we hypothesized that variations within these POAG and PCG genes also may be associated with PACG. 

CYP1B1 is the most common gene known to be involved with the pathogenesis of PCG. ^15,16 It accounts for approximately 20% of PCG cases in Australia. ¹⁷ Sarfarazi et al. hypothesized that CYP1B1 (OMIM 601771) is involved in the development of the anterior chamber angle of the eye, making it a gene of interest for PACG. ¹⁰ Chakrabarti et al. sequenced the coding region of the gene, and found association of mutations with PACG in an Indian cohort consisting of 90 cases and 200 controls. ¹⁸ A smaller study of 29 PACG patients from the Middle East did not report any known or novel polymorphisms in CYP1B1, using direct sequencing. ¹⁹  

Nitric oxide (NO) is synthesized in the vascular endothelium by endothelial nitric oxide synthase 3 (also known as eNOS), from the substrate L-arginine. ²⁰ eNOS (OMIM 163729) overexpression is thought to be neuroprotective by causing vasodilation and increased blood flow in human eye tissues. ²¹ Overexpression of eNOS also was reported to lower intraocular pressure in the mouse eye by increasing the pressure-dependent drainage. ²² Other factors, such as asymmetric dimethylarginine (ADMA) and symmetric dimethylarginine (SDMA), have an inhibitory role in the production of NO. Our laboratory found that serum levels of ADMA and SDMA are significantly elevated in patients with advanced POAG, suggesting dysregulation of the NO system in this disease. ²³  

Furthermore, NO enhances the activity of Matrix metalloprotinase-9 (MMP9) ²⁴ which also has been reported to be associated with PACG. ^25,26 Alteration in MMP9 activity during eye development may lead to hyperopic refractive error, which is a risk factor for PACG. ^26,27 It has been proposed that alteration of eNOS expression causes impairment of blood flow and subsequent development of angle-closure. The 27-base pair (bp) variable number of tandem repeat (VNTR) polymorphism in intron 4 of eNOS is believed to alter the production of NO and cause vascular deregulation. ²⁸ This variation was associated with PACG in Pakistani cohorts. ²⁹ A study of Han Chinese used tag single nucleotide polymorphisms (SNPs) to identify association of common variants in the eNOS gene in 88 patients with PACG, but no associations were observed. ³⁰  

Heterozygous mutations in NTF4 (OMIM 162662) recently were reported to be responsible for 1.7% of POAG cases of European descent. ¹⁴ Neurotrophin has a vital role in the neuronal cell development, survival, and differentiation, and it was suggested that the pathway also may prevent neuronal damage in retinal ganglion cells. ¹⁴ Two studies have looked for association with PACG in Indian ³¹ and Caucasian European ³² populations, respectively, by sequencing the coding exons, but failed to detect association. 

Thus, although these three genes are plausible candidates, results are inconsistent between published studies and even less so across ethnicities. Thus, in our study, we aimed to investigate the association between tag SNPs in these three genes, and PACG in Australian and Nepalese cohorts to determine if common variation in these genes could contribute to disease in these populations. 

Methods

Australian Caucasian participants were recruited from ophthalmology clinics in Australia through the Australian and New Zealand Registry of Advanced Glaucoma. ³³ Approval was obtained from the human research ethics committee of the Southern Adelaide Health Service and Flinders University. All participants were self-reported Caucasian. The Nepalese cohort was recruited from the Nepal Glaucoma Eye Clinic, Tilganga Institute of Ophthalmology, Kathmandu, Nepal, by one of us (SST). The study was approved by the Institutional Review Committee of the Tilganga Institute of Ophthalmology (TIO). All participants were from Nepal and detailed ethnic group information was collected. ^34,35 This study has been conducted in accordance with the tenets of the Declaration of Helsinki and its subsequent revisions. Informed consent was obtained from each individual. 

Following the International Society of Geographical and Epidemiological Ophthalmology (ISGEO), ³⁶ diagnosis of PACG was based on the presence of glaucomatous optic neuropathy with cup-to-disc ratio ≥ 0.7, IOP more than 21 mm Hg, peripheral visual loss, and presence of at least 180° of closed angle in which the trabecular meshwork is not visible on gonioscopy. In our study, 129 Australian and 106 Nepalese affected participants were recruited, and they were all identified with PACG. 

Controls were required to have none of the above characteristics, with no known family history of glaucoma. Participants with pseudophakia or secondary angle-closure glaucoma caused by events, such as uveitis, trauma, or lens subluxation, were excluded. The control groups consisted of 288 Australian and 204 Nepalese individuals. The Australian control cohort was ascertained from nursing home facilities in Adelaide, South Australia. Nepalese controls were participants in a population-based study of Kathmandu, Nepal, in which the individuals were chosen specifically to be matched for age, sex, and ethnic group to the Nepalese cases. ³⁴ Each participant underwent a complete eye examination, including slit-lamp examination of the anterior chamber, gonioscopy, best corrected visual acuity, measurement of intraocular pressure, fundus examination with special attention to optic disc parameters, and visual field assessment. Refraction was done using a streak retinoscope (Beta 200; Heine Optotechnik, Herrsching, Germany), which was followed by a subjective refraction. ³⁴  

Genomic DNA was extracted from peripheral whole blood using the QiaAmp Blood Midi (Nepalese samples) or Maxi (Australian samples) Kit (Qiagen, Valencia, CA). 

A total of 15 Tag SNPs was selected using the tagger program implemented in Haploview 4.2 (available in the public domain at http://www.broadinstitute.org/scientific-community/science/programs/medical-and-populationgenetics/haploview/haploview) to cover the majority of known genetic variations in and around the candidate genes (CYP1B1 5 SNPs, eNOS 7 SNPs, and NTF4 3 SNPs). For the Nepalese cohort, SNPs were selected from the HapMap (available in the public domain at http://hapmap.ncbi.nlm.nih.gov/) Han Chinese in Beijing, China (CHB) sample as the most closely related population available at the time of the study. For SNP selection in the Australian cohort we used CEU: CEPH (Utah residents with ancestry from northern and western Europe). Tag SNPs were chosen using pairwise tagging, to have an r ² > 0.8 with SNPs displaying a minor allele frequency of >5% in the relevant HapMap II population. 

Genotyping was conducted using the iPLEX Gold chemistry (Sequenom, Inc., San Diego, CA) on an Autoflex MassARRAY system (Sequenom, Inc.) at the Australian Genome Research Facility (AGRF), Brisbane. All analyses were conducted using PLINK. ³⁷ SNPs were assessed for compliance with Hardy-Weinberg equilibrium using the χ² test. Genetic association was assessed under an allelic model. Analysis was done with respect to minor allele of the tag SNPs. P values were adjusted for sex and age using logistic regression. A Bonferroni correction was applied to each P value according to the number of SNPs typed in this study (corrected P value of 0.05/15 = 0.0033). Haplotype analyses were conducted in PLINK based on the observed linkage disequilibrium blocks, as visualized using the “solid spine” definition in Haploview. ³⁸  

We further analyzed the association between eNOS and MMP9 with PACG in the Australian cohort by comparing the combined risk alleles with the protective one using the χ² test. The data for MMP9 were those published previously on a subset of the Australian cohort. ²⁵ We selected the significant SNP from each gene eNOS rs3793342 and MMP9 rs17576. 

The power of this study at α = 0.05 was assessed using the genetic power calculator. ³⁹ The prevalence is similar in Australian 0.4% ⁴⁰ and Nepalese 0.43% ⁴¹ cohorts. Assuming complete linkage disequilibrium between the disease causing variant and the marker, we will have a power of 89% to detect a genotypic relative risk of 1.3 with a risk allele frequency of 0.2 under an additive model. 

Results

We enrolled 417 Australian and 310 Nepalese participants in this study. All cases in the Nepalese cohort presented with PACG, of which 53 cases were reported to have had an acute attack. In the Australian cohort, 129 cases were identified with PACG (35 with previous history of acute attack). Table 1 displays the demographic characteristics and clinical data of the cases and controls for each cohort. No SNP deviated from Hardy-Weinberg equilibrium in either cohort (P > 0.05). The physical locations of the tag SNPs are presented in the Figure. 

Figure

 

Gene ideograms depicting the location of tag SNPs genotyped for each candidate gene. Exons are indicated by solid boxes and joined by introns indicated by lines. The direction of transcription is indicated by arrows. Translated regions of exons are colored dark and untranslated regions are light. * and † indicate the tag SNP selected to represent variation in the Australian and Nepalese cohorts, respectively. Unmarked SNPs are tagging SNPs in both populations. Figure adapted with permission from NCBI website (available in the public domain at http://www.ncbi.nlm.nih.gov/gene).(A) CYP1B1, (B) eNOS, and (C) NTF4.

Table 1. 

 

Characteristics of the Nepalese and Australian Cohorts

Variables	Australian			Nepalese
	Case	Control	P Value	Case	Control	P Value
N	129	288	–	106	204	–
Sex, % female	62%	53%	0.20	76%	75%	0.85
Mean age, y (SD)	72 (11.7)	69 (11.2)	0.01	57.3 (12.30)	60.3 (13.71)	0.07
Mean SE, diopters (SD)	2.2 (2.8)	0.12 (0.37)	<0.01	−0.30 (1.64)	0.10 (0.31)	0.16
IOP, mm Hg (SD)	24.5 (14)	14.6 (3.4)	<0.01	21.3 (17.7)	12.8 (2.3)	<0.01
Cup/disc ratio (SD)	0.5 (0.25)	0.2 (0.25)	<0.01	0.8 (0.11)	0.2 (0.12)	<0.01

 SE, spherical equivalent.

The minor allele frequencies and allelic association P values of typed SNPs in the Australian cohort are presented in Table 2. Three SNPs in eNOS showed significant association with PACG: rs3793342, T allele, with a P value of 0.003 (odds ratio [OR] 0.5, 95% confidence interval [CI] 0.3–0.8); rs3918188, A allele, with a P value of 0.014 (OR 1.5, 95% CI 1.1–1.9); and rs7830, A allele, with a P value = 0.007 (OR 0.7, 95% CI 0.5–1.0). Only rs3793342 survived correction for multiple testing of 15 SNP (P value of 0.001). However, after justifying the Australian cohort for age and sex, both rs3793342 and rs7830 were significant with a P value of 0.001 and 0.003, respectively. One haplotype in the eNOS gene also showed association with PACG in the Australian cohort, with a global P value of 0.019. The CGCAATC haplotype conferred risk, with significant P value of 0.009 (OR 1.5, 95% CI 0.9–2.4). This haplotype contained the risk alleles of all the three nominally associated SNPs (Table 3). 

Table 2. 

 

Allele Frequencies (%) of the SNPs in Australian Cohort and Unadjusted P Value for Association under the Allelic Model with OR (95% CI)

Gene	Chr	SNP Position	SNP	Minor Allele	MAF	P Value	OR (95% CI)	P*
CYP1B1	2	1	rs2855658	A	0.42	0.289	0.9 (0.7–1.2)	0.337
		2	rs10916	G	0.22	0.821	1.0 (0.7–1.4)	0.762
		3	rs162562	C	0.22	0.821	1.0 (0.7–1.4)	0.762
		4	rs162561	A	0.14	0.343	1.0 (0.7–1.4)	0.437
		5	rs2551188	T	0.30	0.675	0.9 (0.7–1.2)	0.619
eNOS	7	1	rs3793342	T	0.08	0.003†	0.5 (0.3–0.8)	0.001†
		2	rs1799983	T	0.32	0.776	1.1 (0.8–1.4)	0.671
		3	rs3918227	A	0.08	0.745	1.2 (0.7–2.0)	0.743
		4	rs3918186	T	0.06	0.492	0.6 (0.3–1.2)	0.268
		5	rs3918188	A	0.42	0.014†	1.5 (1.1–1.9)	0.033†
		6	rs1808593	G	0.27	0.186	1.1 (0.8–1.6)	0.164
		7	rs7830	A	0.29	0.007†	0.7 (0.5–1.0)	0.003†
NTF4	19	1	rs12973356	G	0.10	0.185	0.8 (0.5–1.2)	0.119
		2	rs11669977	G	0.36	0.739	0.9 (0.6–1.2)	0.961
		3	rs4802546	T	0.17	0.554	1.1 (0.8–1.7)	0.714

 Chr, chromosome; MAF, minor allele frequency.

*   P value adjusted for sex and age.

†  Significant at the P < 0.05 level.

Table 3. 

 

Haplotypes of eNOS Gene in the Australian Population (>1% Frequency) and its Association with PACG

Haplotype							Cases	Controls	OR (95% CI)	P Value
1	2	3	4	5	6	7
C	G	C	A	A	T	A	0.18	0.19	0.9 (0.6–1.5)	0.844
T	G	C	T	C	T	A	0.04	0.07	0.5 (0.2–1.2)	0.093
C	G	C	A	A	T	C	0.25	0.15	1.5 (0.9–2.4)	0.009
T	G	C	A	C	T	C	0.02	0.05	0.4 (0.1–1.4)	0.083
C	T	A	A	C	T	C	0.08	0.07	1.1 (0.5–2.3)	0.692
C	G	C	A	C	T	A	0.10	0.13	0.8 (0.5–1.5)	0.327
T	G	C	A	C	T	A	0.02	0.05	0.4 (0.1–1.2)	0.055
C	T	C	A	C	G	C	0.22	0.18	1.4 (0.9–2.3)	0.228
C	G	C	A	C	T	C	0.09	0.11	0.9 (0.4–1.9)	0.688

*   P value adjusted for sex and age.

†  Significant at the P < 0.05 level. The order of the SNPs follow those in Table 2.

When we further analyzed the association between eNOS and MMP-9 with PACG in the Australian cohort, a significant difference in the rate of PACG was found between individuals carrying the risk alleles of SNPs rs17576 in MMP-9 and rs3793342 in eNOS (105 individuals) with those who do carry the protective allele only (264 individuals) with P value of 0.048 (OR 2.8). 

No statistically significant association was observed across the CYP1B1 or NTF4 loci. 

The allele frequencies and association P values of typed SNPs in the Nepalese cohort are presented in Table 4. Two SNP from the CYP1B1 gene were nominally significant: rs10916, G allele, with an OR 2.1 (95% CI 1.1–4.0, P = 0.02); and rs162561, A allele, with an OR 2.2 (95% CI 1.1–4.3, P = 0.01). Both SNPs remained significant after adjustment for sex and age; however, did not survive Bonferroni correction. Similarly, in the NTF4 gene one SNP, rs11669977 T allele, showed a P value of 0.04 with OR of 1.5 (95% CI 1.0–2.4), but did not survive correction for the 15 SNPs typed. No significant associations were identified with eNOS in this cohort. 

Table 4. 

 

Allele Frequencies (%) of the SNPs in Nepalese Cohort and Unadjusted P Value for Association under the Allelic Model, with OR (95% CI)

Gene	Chr	SNP Position	SNP	Minor Allele	MAF	P Value	OR (95% CI)	P*
CYP1B1	2	1	rs2855658	A	0.19	0.46	1.1 (0.7–1.8)	0.31
		2	rs10916	G	0.09	0.02†	2.1 (1.1–4.0)†	0.02†
		3	rs162562	C	0.13	0.09	1.5 (0.9–2.6)	0.07
		4	rs162561	A	0.1	0.01†	2.2 (1.1–4.3)†	0.02†
		5	rs2551188	T	0.38	0.54	1.1 (0.7–1.5)	0.62
eNOS	7	1	rs3793342	T	0.1	0.99	1.0 (0.5–1.7)	0.95
		2	rs1799983	T	0.19	0.73	1.0 (0.7–1.6)	0.76
		3	rs3918227	A	0.04	0.17	0.5 (0.2–1.2)	0.22
		4	rs3918186	T	0.11	0.92	1.0 (0.6–1.7)	0.9
		5	rs3918188	A	0.33	0.87	0.9 (0.6–1.3)	0.99
		6	rs1808593	G	0.28	0.42	1.1 (0.8–1.7)	0.46
		7	rs7830	A	0.38	0.39	0.8 (0.6–1.2)	0.41
NTF4	19	1	rs12973356	G	0.49	0.67	1.1 (0.7–1.5)	0.77
		2	rs11669977	G	0.22	0.04†	1.5 (1.0–2.4)†	0.17
		3	rs4802546	T	0.14	0.46	1.1 (0.8–1.5)	0.36

*   P value adjusted for sex and age.

†  Significant at the P < 0.05 level.

In the Nepalese cohort, the global P value for the haplotypes of all five SNP markers tested in CYP1B1 gene was not significant (0.11). However, one haplotype, AGCAC, showed a nominally significant association with PACG, with a P value of 0.02 (OR 2.2, 95% CI 0.7–6.7, Table 5), but did not survive correction for multiple testing of five haplotypes (corrected P value of 0.1). The remaining genes did not show any significant haplotypic association in this cohort (data not shown). 

Table 5. 

 

Haplotypes of CYP1B1 and NTF4 Genes in the Nepalese Population (>1% Frequency) and its Association with PACG

Gene	Haplotype					Frequency Cases	Frequency Controls	P Value
	1	2	3	4	5
CYP1B1	A	T	C	C	T	0.03	0.04	0.76
	G	T	A	C	T	0.34	0.31	0.53
	A*	G*	C*	A*	C*	0.10*	0.05*	0.02*
	A	T	A	C	C	0.05	0.08	0.26
	G	T	A	C	C	0.48	0.52	0.24
NTF4	C	A	C			0.30	0.33	0.57
	G	G	C			0.22	0.16	0.18
	C	G	T			0.22	0.22	0.94
	C	G	C			0.26	0.29	0.38

*  Significant at the P < 0.05 level. The order of SNPs follows that of Table 4.

Discussion

PACG is a complex disease, believed to arise from interactions between genetic variants and environmental effects. Here, we chose three genes targeting different function in the pathogenesis of glaucoma; development of anterior chamber (CYP1B1), retinal ganglion cell development, and survival (NTF4), and regulation of intraocular pressure (eNOS). All three genes have been implicated previously in POAG or PCG, and have been proposed as likely candidates for PACG. 

CYP1B1 is well known for its association with PCG. ^11,42,43 It is expressed in tissues of the anterior chamber of the eye, such as the ciliary body, iris, and trabecular meshwork. ^29,30 CYP1B1 was hypothesized to take part in the normal development and function of the eye, and is involved in the development of the anterior chamber angle. ¹⁰ Association was found with PACG in an Indian population, ³⁰ but not in a Middle Eastern cohort using sequencing methodology. ¹⁹ Here, we genotyped tag SNPs to look for common variations in and around the gene. Our study showed nominal association in the Nepalese cohort under single SNP and haplotypic analyses; however, this association was not considered significant given the number of tests conducted. Thus, a larger cohort will be required to confirm this putative association. 

When ocular tissues are subjected to stress, NO is released from the ocular vascular endothelium, causing an increase in the ocular blood flow and oxygen delivery to the retina. ⁴⁴ NO is produced from the eNOS or NOS3 enzyme. NO synthesis sites are located abundantly in the ciliary muscle and the outflow pathway of normal human eye. It has been hypothesized to be involved in IOP regulation, either directly through affecting the outflow resistance at the level of the trabecular meshwork, or indirectly through affecting the tone of the ciliary muscle. ⁴⁵ In addition to a role in IOP modulation, eNOS also affects the blood flow to the optic nerve via altering the dilation of the ocular vasculature. ⁴⁶ This could be important in determining the nerve's response to stress, and dysfunction could lead to retinal ganglion cell death. eNOS was shown to be associated significantly with PACG in our Australian cohort at the allelic and haplotypic levels, suggesting that NO regulation has a pathogenic role in PACG. 

NO enhances expression of MMP9. ²⁴ As we previously have reported an association of the MMP9 gene with PACG in this Australian cohort, ²⁵ we evaluated the data at both genes combined and showed that patients carrying PACG risk alleles at MMP9 and eNOS have double the risk of the disease developing compared to carrying no risk alleles. As the genetic architecture of PACG is unravelled, more detailed genetic risk matrices could be developed, which will predict better which patients with primary angle-closure suspect are likely to have progression to PACG and, thus, require close monitoring. 

NTF4 variants were reported to have a minor contribution in the pathogenesis of POAG. ⁴⁷ Variations in NTF4 were not associated with PACG in an Indian population. ³¹ Our study reported that the most prevalent variant, A88V, was present in controls (4.91%) at a higher frequency than in cases (2.85%), ⁴⁸ which is opposite to the findings in a large cohort of POAG. ¹⁴ Neither of our cohorts showed robust association at this locus. Thus, NTF4 is unlikely to be a risk locus for PACG. 

In conclusion, we found no evidence for association of NTF4 or CYP1B1 with PACG in either the Australian or Nepalese cohort. SNPs and haplotypes in the eNOS gene were associated with PACG in the Australian cohort. The lack of association of this gene in the Nepalese may be due to a true negative finding, or insufficient power in this cohort to detect an effect, and larger cohorts will be required to determine this. The NO pathway long has been implicated in glaucoma. Our findings may indicate common molecular pathways leading to optic nerve susceptibility to glaucoma relevant to POAG and PACG. 

Footnotes

 Supported by a grant from the Flinders Medical Centre Foundation, and by a National Health and Medical Research Council (NHMRC) of Australia Career Development Award (KPB). JEC is an NHMRC Practitioner Fellow.

Footnotes

 Disclosure: M.S. Awadalla, None; S.S. Thapa, None; A.W. Hewitt, None; J.E. Craig, None; K.P. Burdon, None