| HOME | HELP | CONTACT US | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Endocrinology |
aCancer Genetics, Kolling Institute of Medical Research, University of Sydney, bDepartment of Surgery, Bankstown Hospital and University of New South Wales, cDepartment of Endocrinology, and dDepartment of Endocrine and Oncology Surgery, Royal North Shore Hospital, St. Leonards, Australia and University of Sydney
Key Words. Adrenocortical carcinoma • Adrenocortical adenoma • Adrenocortical tumors • Molecular marker
Correspondence: Stan Sidhu, MBBS, PhD, FRACS, Department of Endocrine and Oncology Surgery, Royal North Shore Hospital, St. Leonards, NSW 2065, Australia. Telephone: 61-2-94371731; Fax: 61-2-99268523; e-mail: stansidhu{at}nebsc.com.au
Received December 12, 2007; accepted for publication March 24, 2008.
Disclosure: No potential conflicts of interest were reported by the authors.
| ABSTRACT |
|---|
|
|
|---|
| INTRODUCTION |
|---|
|
|
|---|
2 is typically associated with an ACA while a score
3 is indicative of an ACC [4–6]. In a study of 24 malignant and 25 benign ACTs, a Weiss score
3 had a sensitivity of 100% and a specificity of 96% in predicting malignancy. The authors noted, however, that the interobserver agreement for certain criteria, such as architecture, sinusoid invasion, nuclear grade, and venous invasion, was not reliable [4]. Furthermore, even when applied correctly, the Weiss system is far from infallible, and individual ACTs may go on to behave in a malignant manner despite initially receiving a Weiss score
2 [7–9]. Another shortcoming of this classification system is its restriction to providing descriptive information only on the tumor's morphology; consequently, it does not offer any molecular targets for therapy development. Surgery is the mainstay of treatment for ACCs. Open adrenalectomy is recommended for excision of malignant primary adrenal lesions [10–12] because there have been some case reports that laparoscopic adrenalectomy for ACCs actually increases the risk for peritoneal dissemination and metastasis [13–15]. Despite radical surgery with curative intent, for patients with localized ACCs, the majority will develop metastases within 6–24 months of resection [16].
About one fifth of the patients with ACC present with advanced disease that is not cured by surgery [17, 18]. The treatment of choice for these patients is mitotane (o,p'-dichlorodiphenyldichloroethane). Up to one third of patients have at least a partial response to mitotane [19]. The largest series of mitotane use after surgery compared with surgery alone, in 177 patients with ACCs from eight centers in Italy and 47 centers in Germany, was published recently. All patients had radical resection with a follow-up of up to 10 years. Forty-seven of the 177 patients had mitotane after surgery, while the remainder of the patients had surgery alone. Mitotane treatment was associated with significantly longer recurrence-free survival and overall survival times compared with the control [16].
Radiotherapy to the tumor bed has also been used as adjuvant treatment after radical resection of ACC. Reports with small numbers of patients have described response rates of up to 42%. Fassnacht et al. [20] compared a group of 14 patients who received radiotherapy to the tumor bed with a matched control group of 14 patients. The local recurrence rate was significantly lower in the radiotherapy group, at 14%, compared with 79% in the control group. The disease-free and overall survival times, however, were not significantly different between the two groups [20].
ACCs have a 10%–40% response rate to various chemotherapeutic agents. Chemotherapeutic drugs that have been reported to have some effect against ACCs include etoposide, doxorubicin, and cisplatin. ACCs tend to express the multidrug resistance gene MDR-1, which results in the production of P-glycoprotein, which is involved in the removal of the drug from cancer cells. Because of this multidrug resistance gene, single-agent chemotherapy is not favored for ACCs [21].
Advancement in the understanding of the pathophysiology of ACCs is essential for the development of more sensitive means of diagnosis and treatment, resulting in better clinical outcome. The aim of this review is to summarize the current knowledge of the molecular mechanisms involved in adrenocortical tumorigenesis.
| MOLECULAR ASPECTS OF ACCS |
|---|
|
|
|---|
|
Li-Fraumeni Syndrome
Li-Fraumeni Syndrome (LFS; OMIM 151623) is an autosomal dominant familial disease characterized by the early onset of tumors and multiple tumors in affected individuals. Families with this syndrome also have multiple affected family members. The most common tumor types that occur in LFS are soft tissue sarcomas, osteosarcomas, breast cancer, brain tumors, leukemia, and ACCs [22, 23]. ACCs have been reported to occur in 3%–4% of patients with LFS, often under the age of 20 [23]. Seventy percent of LFS cases are a result of a germline mutation in the TP53 gene [24]. A second variant is caused by a heterozygous germline mutation in the hCHK2 gene [25], while a four centi-Morgan region on 1q23 has been implicated in a third variant [26].
Beckwith-Wiedemann Syndrome
Beckwith-Wiedemann syndrome (BWS; OMIM 130650) is a congenital overgrowth syndrome characterized by exomphalos, macroglossia, and gigantism in the neonate as well as the development of childhood tumors. These tumors, which include ACC, nephroblastoma, hepatoblastoma, and rhabdomyosarcoma, occur in 5% of patients [27]. About 15% of cases with BWS are familial, with the remainder being sporadic. BWS is linked to the 11p15 chromosomal locus. This region is subject to parental imprinting, a genetic phenomenon in which specific genes are expressed solely either from the maternal or paternal allele. Genes located at 11p15 and implicated in the pathogenesis of BWS are the insulin-like growth factor 2 (IGF2), H19, and cyclin-dependent kinase inhibitor 1C (CDKN1C also known as p57kip2) genes. IGF2 is maternally imprinted, while H19 and p57kip2 are both paternally imprinted. Paternal uniparental isodisomy (duplication of the paternal allele and loss of the maternal allele) of 11p15 [28], germline mutations of p57kip2, or methylation of H19 and potassium channel, voltage-gated, KQT-like subfamily, member 1 (KCNQ1) have been implicated in the pathogenesis of BWS [27]. p57kip2 encodes a cyclin-dependent kinase (CDK) inhibitor that belongs to the CIP/KIP family of cell-cycle regulators. Overexpression of this gene arrests cells in the G1 phase of the cell cycle. H19 is transcribed to RNA but not translated to protein. It is thought to be involved in the regulation of IGF2 [27].
Carney Complex
Carney complex (CNC; OMIM 160980) is a dominantly inherited syndrome characterized by cardiac, endocrine, cutaneous, and neural myxomatous tumors, as well as pigmented lesions of the skin and mucosa [29–31]. Primary pigmented nodular adrenocortical disease (PPNAD), a main feature of CNC, is a rare cause of adrenocorticotropic hormone (ACTH)-independent Cushing's syndrome, usually in children and young adults [32]. There are two types of CNC—type 1 is caused by mutation of the protein kinase, cAMP-dependent, regulatory, type 1, alpha (PRKAR1A) gene, located on 17q23-q24 [33], while type 2 has been attributed to the 2p16 chromosomal locus [34].
Multiple Endocrine Neoplasia 1
Multiple endocrine neoplasia 1 (MEN1; OMIM 131100) is an autosomal dominant syndrome characterized by the occurrence of parathyroid, pancreatic islet cell, and anterior pituitary tumors [35]. In 55% of individuals diagnosed with MEN1, ACTs (typically ACAs) have also been reported [36–39]. ACCs have only rarely been reported with MEN1 [37–39]. The MEN1 gene, located on 11q13, encodes the menin protein. Although the function of menin is currently unknown, mutations have resulted in the loss of its function, suggesting that menin has tumor suppressor activities [40, 41]. Because menin is located in the nucleus, it is thought to play a role in the cell cycle, regulation of transcription, or DNA replication [35].
Congenital Adrenal Hyperplasia
Congenital adrenal hyperplasia (CAH) is an autosomal recessive disorder resulting from an enzyme deficiency in the cortisol synthesis pathway. Typically, the enzyme 21-hydroxylase, which is encoded by cytochrome P450, family 21, subfamily B (CYP21B; OMIM 201910), is deficient. A lack of this enzyme leads to compensatory stimulation of the adrenal cortex by corticotrophin-releasing hormone and ACTH with consequent adrenal hyperplasia and overproduction of cortisol precursors, engendering higher levels of androgens. Deficiencies of 11β-hydoxylase, 17
-hydroxylase, and 3β-hydroxysteroid dehydrogenase are less commonly the cause of CAH. Clinically, CAH is divided into a classic (severe) salt wasting or simple virilizing form and a nonclassic (mild) form [42–44]. Assessment for adrenal lesions with abdominal CT in CAH patients and heterozygous carriers of CAH found that all CAH patients had adrenal hyperplasia as well as twice the rate of adrenal tumors, at 82%, compared with CAH carriers, at 45%. One in 11 patients with CAH and one in 20 carriers of CAH had adrenal tumors >5 cm [45]. Rarely, ACCs have been described in CAH [46, 47].
| GENETICS OF SPORADIC ACTS |
|---|
|
|
|---|
TP53 mutation is thought to be a late event in the evolution of malignant transformation in sporadic ACTs. Mutations in exons 5–8 of TP53 have been found in 20%–27% of sporadic ACCs and 0%–6% of sporadic ACAs [51, 52]. Sequencing of exons 2–11 of TP53 in one study found mutations in 25% of sporadic ACCs, all of which occurred in exons 5–8. Patients with TP53 mutation showed a trend towards a shorter survival duration (p = .098) [53]. Frequent somatic mutations in exon 4 of TP53, however, have been reported in 60% of ACAs and 50% of pheochromocytomas in Taiwanese patients [54], but these findings were not confirmed in a study on white patients [55], suggesting that perhaps different ethnic backgrounds may result in different TP53 mutations.
The majority of TP53 mutations occur in its DNA-binding domain. Less commonly, mutations in the oligomerization domain can occur [56]. Numerous missense mutations have been described for TP53 [57]. In ACCs, the R337H mutation in the oligomerization domain of the p53 protein results in the substitution of histidine for arginine at codon 337. This mutation is commonly found as a germline mutation in children with ACCs in southern Brazil [58]. It has been estimated that one in 10 carriers of this mutation develops ACC. In these cases of ACC, there is loss of heterozygosity (LOH) of the normal wild-type TP53 allele with retention of the R337H mutated allele [59]. Other missense mutations as well as nonsense mutations and deletions of TP53 have also been described [51–53, 60, 61].
The IGF2, p57kip2 (CDKN1C), and H19 Genes
Rearrangements, LOH (loss of one of two alleles of a gene), and abnormal imprinting of the 11p15.5 locus, resulting in low p57kip2 and H19 and elevated IGF2 mRNA expression levels, have been reported in sporadic ACCs [62–64].
The IGF system is comprised of two peptide ligands (IGF1 and IGF2), two IGF receptors (IGF1R and IGF2R/ mannose-6-phosphate receptor), and six high-affinity binding proteins (IGF binding proteins 1–6) [65]. In the adrenal gland, both IGF1 and IGF2 have growth-promoting as well as differentiating functions. They induce steroidogenesis in adrenocortical cells both in vitro and in vivo. IGF2 at high levels (50 times that of insulin) can also exert an insulin effect, leading to hypoglycemia.
In sporadic ACCs, IGF1 has not been shown to be overexpressed [66]. IGF2, however, has frequently been reported to be overexpressed in ACCs compared with ACAs or normal adrenal cortices [67–69]. Higher IGF2 expression levels are associated with a more malignant phenotype [66], and overexpression of IGF2 is associated with a higher risk for ACC recurrence [63]. Furthermore, LOH of the 11p15 locus has been demonstrated more frequently in ACCs than in ACAs—in 67% of ACCs versus 13% of ACAs. It is suggested that this LOH event leads to overexpression of IGF2 because the maternal allele is lost while the paternal allele is duplicated, leading to a double dose of the expressed allele [62].
Studies with IGF2–transgenic mice have shown that the weights of the adrenal glands of these animals are significantly higher than those of controls, a result of hyperplasia of the zona fasciculata. Despite adrenal hyperplasia, over an 18-month period, these transgenic mice did not develop tumors in their adrenal glands, indicating that overexpression of IGF2 alone is insufficient to cause ACT formation and that other factors are required for tumorigenesis [70].
The p57kip2 gene is located within the 11p15 region and is paternally imprinted. It encodes a CDK inhibitor, which binds to cyclin–CDK complexes and inactivates their catalytic domain. It therefore functions as a negative regulator of cell cycle progression. By northern blot, p57kip2 mRNA was easily detected in normal adrenals and in all tumors with normal expression of IGF2, but was absent or low in tumors with overexpression of IGF2 and in the NCI-H295R adrenocortical cell line [71].
Somatic mutations of the p57kip2 gene are rare and do not account for the lower levels of p57kip2 mRNA and protein expression in ACCs. Instead, other mechanisms, such as LOH and abnormalities of imprinting or methylation, could be responsible for the lower mRNA and protein levels with this gene [61].
MEN1 Gene
Because LOH of 11q13 occurs in about 20% of sporadic ACTs, and adrenal tumors occur in up to 40% of patients from MEN1 kindreds, MEN1 was considered to be a prime candidate gene in the pathogenesis of these lesions. LOH of 11q13 is frequently found in ACCs but not in ACAs. Because the MEN1 mRNA expression by quantitative polymerase chain reaction and northern blot has been found to be similar in ACCs, ACAs, and normal adrenal cortices, and because no mutations within the MEN1 coding region were found in 33 sporadic ACTs and cell lines, it is unlikely that the MEN1 gene plays a prominent part in the pathogenesis of sporadic ACTs [72–74].
PRKAR1A Gene
PRKAR1A is the main mediator of cAMP signaling [75]. One study found LOH of 17q22–24, the locus for PRKAR1A, in 23% of ACAs and 53% of ACCs. Direct sequencing of the PRKAR1A gene revealed inactivating mutations in 10% of ACAs, with corresponding lower mRNA and protein levels in these tumors. These tumors were also smaller in size and exhibited paradoxical cortisol responses to dexamethasone, all features that are found in PPNAD. No mutations were found in ACCs. This is consistent with PPNAD, in which there has not been a case of ACC reported to date [76].
| THE GUANINE NUCLEOTIDE-BINDING PROTEIN, ALPHA-STIMULATING ACTIVITY POLYPEPTIDE GENE |
|---|
|
|
|---|
). In abnormal Gs
, there is a substitution of arginine 201 with histidine or cysteine, with subsequent lower GTPase activity, resulting in constitutive adenylate cylase activation and consequent cAMP signaling [77–79]. MAS is a sporadic disease that predominantly affects the skeleton, skin, and endocrine system. Classic manifestations include a triad of polyostotic fibrous dysplasia, large irregular café-au-lait spots, and endocrine dysfunction, including precocious puberty, hyperthyroidism, gigantism, and Cushing's syndrome. Because the genes involved in MAS and CNC both act on the cAMP pathway, the endocrine features of the two syndromes are very similar. GNAS has been reported to rarely be mutated in sporadic ACAs. Mutation of codon 201 of GNAS has been reported in ACAs and tumors of patients with ACTH-independent macronodular adrenocortical hyperplasia (AIMAH) [80, 81]. There have been no reports, however, of GNAS mutations in sporadic ACCs.
| SIGNALING PATHWAYS IN ACTS |
|---|
|
|
|---|
The ACTH–cAMP–PKA Pathway
The binding of ACTH to its receptor, a member of the G protein-coupled receptor family, results in the dissociation of the heterotrimeric Gs, causing the separation of the
subunit from the β and
subunits and stimulation of adenylate cyclase, which in turn leads to the production of cAMP from ATP. cAMP then binds to the regulatory subunits of PKA, releasing the catalytic subunits, which results in phosphorylation of proteins in the cytoplasm and nucleus, and subsequent signal transduction [82] (Fig. 1).
|
The Wnt Pathway
The Wnt family is comprised of a group of highly conserved growth factors with similar amino acid sequences, which play roles in developmental and homeostatic processes. The central event in the canonical Wnt signaling pathway is the accumulation of β-catenin in the cytoplasm with subsequent translocation into the nucleus. Wnt binds to its receptor complex, which is composed of members of the frizzled family and low-density lipoprotein receptor-related protein. This results in the inhibition of the axin–adenomatous polyposis coli (APC)–glycogen synthase kinase 3β (GSK-3) complex, leading to a block in β-catenin phosphorylation by GSK-3 and accumulation of β-catenin in the cytoplasm. β-catenin then translocates into the nucleus where it interacts with the T cell–specific transcription factor/lymphoid enhancer-binding factor 1 family of transcription factors to regulate transcription of Wnt target genes (Fig. 2A). In the absence of Wnt stimulation of its receptor, GSK-3 phosphorylates β-catenin, resulting in its ubiquitylation and degradation by proteosomes [90] (Fig. 2B).
|
| CLONAL ANALYSIS OF ACTS |
|---|
|
|
|---|
Clonal analysis of tumors determines the X inactivation patterns in females heterozygous for X-linked polymorphisms. It is based on the rationale that only a single X chromosome is active in each somatic cell of a female. As a result, either the maternal or paternal X chromosome is randomly inactivated and this is transmitted in a highly stable fashion to the progeny cell [97]. Because X chromosome inactivation is random, it is expected that there are equal proportions of cells with the maternal and paternal X chromosomes. The presence of only either the maternal or the paternal X chromosome in all the cells indicates that the tumor is monoclonal.
Three clonal composition studies have shown that 60%–100% of ACCs are monoclonal while 77.4%–100% of adrenal hyperplasias and 12.5%–43% of ACAs are polyclonal [98–100]. Interestingly, one study examined both adrenal glands of a patient with AIMAH—a diffuse hyperplastic area and a <1-cm nodule from the right adrenal gland and a 3.5-cm nodule from the left adrenal gland—and found polyclonality in the two right lesions and monoclonality in the left lesion, suggestive of a transition from hyperplasia to autonomous adenoma-like growth in larger nodules [99]. All three studies concur that ACCs are more often monoclonal, adrenal hyperplasia is more often polyclonal, and ACAs can be either monoclonal or polyclonal. Polyclonality favors the idea that the tumor developed from a group of cells under the common stimulus of a growth factor, while monoclonality suggests that it developed from a single genetically aberrant cell. The presence of monoclonal and polyclonal ACAs could be a result of either different pathological mechanisms or different stages of a common multistep process. It has been suggested that progression to a monoclonal tumor could occur as a result of a first event that initiates the growth of a polyclonal or partially monoclonal tumor with the maintenance of a normal steroid secretory pattern, while a second event would confer a growth advantage in a selected clone of cells, with a concomitant loss of differentiated functions and an aberrant steroid secretory pattern [98].
| COMPARATIVE GENOMIC HYBRIDIZATION ANALYSIS |
|---|
|
|
|---|
| LOH ANALYSIS |
|---|
|
|
|---|
|
| CURRENT MOLECULAR MARKERS IN CLINICAL PRACTICE |
|---|
|
|
|---|
, human epidermal growth factor receptor 2/neu, E-cadherin, and the retinoblastoma gene product [108–112]. Many of these molecular markers, however, lack specificity to achieve discrimination between ACCs and ACAs. The utility of IGF2 and MIB1 (a mouse monoclonal antibody that recognizes a formalin-fixation resistant epitope on the cell proliferation-associated antigen Ki-67) IHC in discriminating between ACCs and ACAs was assessed in one study. Tumors were classified as ACAs or ACCs based on Weiss [5], Hough [113], and van Slooten [114] scores. For IGF2 IHC, 21 of 22 ACAs were negative and 13 of 17 ACCs were positive, giving a specificity of 95.5% and a sensitivity of 76.5%. For MIB1, 21 of 22 ACAs were negative while 14 of 16 ACCs were positive, yielding a specificity of 95.5% and a sensitivity of 87.5%. Combining IGF2 and MIB1 IHC yielded a sensitivity of 100% and a specificity of 95.5% in differentiating ACCs from ACAs [110].
Transcription factors have also been used as possible molecular markers that can differentiate ACCs from ACAs. A member of the nuclear receptor family of transcription factors, steroidogenic factor 1 (SF1) maps to 9q33.3. It has a key role in the development and function of the adrenal cortex [115]. A study on SF1 knockout mice demonstrated that these mice died on postnatal day 8 with severe adrenocortical insufficiency resulting from an absence of the adrenal glands [116]. SF1 heterozygous mice have also been found to develop adrenal insufficiency [117]. In CGH studies of 11 pediatric ACTs, James et al. [118] found that 9q34 showed amplification in 10 of the 11 ACTs. This finding was confirmed by Figueiredo et al. [119], who went on to use fluorescence in situ hybridization to confirm that there was a higher SF1 copy number in these tumors. Another study from this group also showed a higher SF1 copy number in eight of 10 pediatric ACTs. SF1 protein levels, however, were noted to be higher in all ACTs than in the normal adrenal cortex [120]. IHC with SF1 has not been shown to differentiate between ACCs and ACAs [121], but it is useful in distinguishing between primary ACC and metastasis from other sites [122].
GATA6 is from the GATA family of transcription factors, which is characterized by binding to the DNA consensus sequence (A/T)GATA(A/G). GATA6 plays a role in cellular maturation and differentiation [123]. GATA6 protein expression has been found to be significantly lower in ACCs than in ACAs on IHC. Accordingly, ACTs with Weiss scores of 4–9 had a significantly lower GATA6 level than ACTs with Weiss scores of 1–3 [121].
Vascular endothelial growth factor (VEGF) plays a pivotal role in the regulation of both normal and tumor angiogenesis [124]. Angiogenesis is critical for tumor growth and metastasis [125]. VEGF has been found to be increased in the majority of cancers and is associated with a poorer outcome [126–129]. One study assessed VEGF expression in 18 ACAs (Weiss score, 0), 12 transitional tumors (Weiss score, 1–3), and 13 ACCs (Weiss score >3) by enzyme-linked immunosorbent assay, and found that VEGF levels were significantly lower in ACAs and transitional tumors than in ACCs; these VEGF levels were unrelated to tumor weight. The VEGF levels for transitional tumors or ACCs that recurred were also higher than for those that did not recur [130]. Serum VEGF levels have also been assessed in patients with ACCs versus patients with ACAs, and they were not found to be significantly different between the two groups [131]. VEGF, however, as a molecular marker for ACC, has not been integrated into clinical practice.
| THE HUNT FOR NEW MOLECULAR MARKERS |
|---|
|
|
|---|
Microarray gene-expression analysis is a high-throughput technique that allows the simultaneous analysis of the expression of thousands of genes in a tissue. By comparing the gene-expression profiles of two different groups, such as ACCs and ACAs, it is possible to identify genes that are significantly upregulated or downregulated in one group relative to the other. The assumption is that these genes are in some way involved in the pathogenesis of these tumors. Overexpressed genes specific to ACCs have the potential to become therapeutic targets. At the time of writing this review, five microarray gene-expression profiling studies comparing ACCs with ACAs had been conducted, four specifically examining adult tumors and one examining pediatric tumors. A summary of the top genes identified to be significantly differentially expressed between ACCs and ACAs are listed in Table 3.
|
Several studies have noted the prognostic significance of molecular markers in ACCs [63, 68, 132]. In one study, an IGF2 gene–related cluster was identified that could select the subgroup of patients with ACCs who were at a high risk for recurrence and who would therefore benefit from adjuvant therapy [68]. This IGF2 gene–related cluster contained eight genes. Ninety percent of tumors with low expression of the IGF2 gene–related cluster were ACAs, while 75% of tumors with high expression of these genes were ACCs. In contrast a 14-gene steroidogenesis cluster could identify ACAs with high accuracy. Analyzing a subgroup of 40 tumors with follow-up data showed that either the IGF2 or steroidogenesis cluster of genes alone was not as effective as the Weiss score in terms of predicting malignancy and postoperative recurrence [68]. Another study found that LOH of the 17p13 locus in a cohort of 96 localized ACTs was a strong predictor of a shorter disease-free survival time, with a relative risk of 21.5 by multivariate analysis [63]. Volante et al. [132] studied the protein expression of matrix metalloproteinase type 2 (MMP2), also known as gelatinase A, by IHC in a cohort of 50 ACCs and 50 ACAs. They showed that MMP2 was detected in one of 50 (2%) ACAs and 37 of 50 (74%) ACCs (p < .001). MMP2 protein expression in ACCs was focal in two thirds of cases and diffuse in the remainder. It was also noted that more diffuse expression of MMP2 in ACCs was associated with shorter overall and disease-free survival times [132]. Interestingly, Kjellman et al. [133] assessed MMP2 mRNA in 16 ACCs and 14 ACAs by an mRNA in situ hybridization technique and found that 13 of 16 (81%) ACCs and one of 14 (7%) ACAs expressed MMP2 mRNA. However, the MMP2 mRNA was actually found in surrounding stromal tissue and not in the neoplastic cell itself [133]. Serum levels of MMP2 have not been found to be useful in predicting either ACC or ACA [131].
| CONCLUSION |
|---|
|
|
|---|
| AUTHOR CONTRIBUTIONS |
|---|
|
|
|---|
Collection/assembly of data: Patsy S. H. Soon
Data analysis and interpretation: Patsy S. H. Soon
Manuscript writing: Patsy S. H. Soon, Kerrie L. McDonald, Bruce G. Robinson, Stan B. Sidhu
Final approval of manuscript: Patsy S. H. Soon, Kerrie L. McDonald, Bruce G. Robinson, Stan B. Sidhu
| ACKNOWLEDGMENTS |
|---|
|
|
|---|
| REFERENCES |
|---|
|
|
|---|