Variations in the AURKA Gene: Biomarkers for the Development and Progression of Hepatocellular Carcinoma.

Hepatocellular carcinoma (HCC) is a liver malignancy and a major cause of cancer mortality worldwide. AURKA (aurora kinase A) is a mitotic serine/threonine kinase that functions as an oncogene and plays a critical role in hepatocarcinogenesis. We report on the association between 4 single nucleotide polymorphisms (SNPs) of the AURKA gene (rs1047972, rs2273535, rs2064836, and rs6024836) and HCC susceptibility as well as clinical outcomes in 312 patients with HCC and in 624 cancer-free controls. We found that carriers of the TT allele of the variant rs1047972 were at greater risk of HCC compared with wild-type (CC) carriers. Moreover, carriers of at least one A allele in rs2273535 were less likely to progress to stage III/IV disease, develop large tumors or be classified into Child-Pugh class B or C. Individuals with at least one G allele at AURKA SNP rs2064863 were at lower risk of developing large tumors or progressing to Child-Pugh grade B or C. Our results indicate that genetic variations in the AURKA gene may serve as an important predictor of early-stage HCC and be a reliable biomarker for the development of HCC.

Introduction

Hepatocellular carcinoma (HCC) is the fifth most common cancer among men worldwide and the ninth in women, and a major cause of cancer-related mortality 1. HCC is associated with a low 5-year survival rate and an increasing mortality rate 2, 3. In Taiwan, HCC is the second major cause of cancer-associated deaths 4, 5.

Genetic variation plays a key role in HCC susceptibility and development of the disease. The majority of people who are exposed to the well-known infectious, lifestyle or environmental risk factors (i.e., hepatitis B or C virus infection, alcohol abuse or nonalcoholic fatty liver disease caused by obesity, type 2 diabetes and insulin resistance) do not develop HCC, which suggests that individual susceptibility modulates the tumor process 4. Genotype distribution frequency data can be used to map single nucleotide polymorphism (SNP) diversity in a population and to examine the risk and development of specific diseases 6. Emerging reports indicate an association between SNPs in certain genes and the susceptibility and clinicopathological status of HCC. For instance, individuals carrying specific interleukin-18 (IL-18) or high-mobility group box protein 1 (HMGB1) SNPs are at higher risk of HCC than wild-type carriers 7, 8.

AURKA, also known as aurora kinase A, is a mitotic serine/threonine kinase that plays a critical role in centrosome duplication and separation, spindle assembly, maturation, chromosomal alignment, spindle assembly checkpoint, and cytokinesis 9. Increased expression of AURKA may cause to chromosomal instability and transformation as well as centrosome amplification in mammalian cells 10. AURKA overexpression has been observed in many human tumors 11-13, particularly in HCC 14. It has also been reported that AURKA promotes the oncogenic effects of c-Myc, which is frequently amplified and overexpressed in many human cancers including HCC 15. Genetic polymorphisms of AURKA have been indicated in several different cancer types (oral cancers, breast and ovarian cancers) 13, 16, 17. It has been suggested that carriers of the AURKA 31Phe allele are less susceptible to hepatitis B virus (HBV)-related HCC when compared with noncarriers 18. Scant research has examined the association between AURKA SNPs, HCC risk and prognosis. We therefore conducted a case-control study to evaluate the role of four AURKA SNPs on HCC susceptibility and clinicopathological features in a cohort of Chinese Han individuals.

Materials and Methods

Participants

We enrolled 312 patients (cases) presenting with HCC to Chung Shan Medical University Hospital, Taiwan, between 2007 and 2015. A total of 624 anonymised healthy controls (HCs) were randomly selected from the Taiwan Biobank Project. All study participants were of Chinese Han ethnicity. HCC patients were staged according to the 2002 American Joint Committee on Cancer (AJCC) TNM staging system, which incorporates tumor morphology, number of lymph nodes affected, and metastases 19. Before entering the study, each participant provided informed written consent and completed a structured questionnaire about sociodemographic status, cigarette and alcohol use. Liver cirrhosis was diagnosed by biopsy, appropriate sagittal CT or MRI scans, or biochemical evidence of liver parenchymal damage with endoscopic esophageal or gastric varices. The study was approved by the Institutional Review Board of Chung Shan Medical University Hospital prior to commencement.

Determination of genotypes

Total genomic DNA was isolated from whole blood specimens using QIAamp DNA blood mini kits (Qiagen, Valencia, CA), as per the manufacturer's instructions. This DNA was dissolved in TE buffer (10 mM Tris pH 7.8, 1 mM EDTA] and stored at -20°C until it was subjected to quantitative polymerase chain reaction (PCR) analysis. Four AURKA SNPs (rs1047972, rs2273535, rs2064836, and rs6024836) with minor allele frequencies >5% in the HapMap population were selected. Moreover, these SNPs have previously been found to associate with the development of cancer 13, 20, 21. The AURKA SNPs were examined by the commercially available TaqMan SNP genotyping assay (Applied Biosystems, Warrington, UK), according to the manufacturer's protocols 22, 23.

Statistical analysis

The genotype distribution of each SNP was analyzed for Hardy-Weinberg equilibrium and confirmed by Chi-square analysis. Demographic characteristics were compared between patients and controls using the Mann-Whitney U-test and Fisher's exact test. Associations between genotypes, HCC risk and clinicopathological characteristics were estimated using adjusted odds ratios (AORs) and 95% confidence intervals (CIs) obtained from age- and gender-adjusted multiple logistic regression models. A p value of <0.05 was considered statistically significant. Data were analyzed using SAS statistical software (Version 9.1, 2005; SAS Institute Inc., Cary, NC).

Results

Demographic characteristics did not differ significantly between the 312 patients with HCC and 624 cancer-free healthy controls (HCs) (Table 1). Significantly fewer (p < 0.001) controls compared with patients reported that they consumed alcohol, but cigarette smoking status did not differ between the two groups (p = 0.604) (Table 1). Compared with controls, significantly higher proportions of HCC patients were positive for HBsAg (11.1% vs 43.9%; p < 0.001) and anti-HCV antibodies (4.5% vs 47.4%; p < 0.001) (Table 1). 213 patients (68.3%) had stage I/II HCC and 99 (31.75%) had stage III/IV disease (Table 1).

Table 1

Demographic characteristics of 624 healthy controls and 312 patients with HCC.

Variable	Controls (N=624)	Patients (N=312)	p value
Age (yrs)	Mean ± S.D.	Mean ± S.D.
	59.53 ± 7.53	60.41 ± 9.44	p = 0.155
Gender
Male	452 (72.4%)	226 (72.4%)
Female	172 (27.6%)	86 (27.6%)	p = 1.000
Cigarette smoking
No	379 (60.7%)	184 (59.0%)
Yes	245 (39.3%)	128 (41.0%)	p = 0.604
Alcohol drinking
No	537 (86.1%)	194 (62.2%)
Yes	87 (13.9%)	118 (37.8%)	p < 0.001*
HBsAg
Negative	555 (88.9%)	175 (56.1%)
Positive	69 (11.1%)	137 (43.9%)	p < 0.001*
Anti-HCV
Negative	596 (95.5%)	164 (52.6%)
Positive	28 (4.5%)	148 (47.4%)	p < 0.001*
Stage
I+II		213 (68.3%)
III+IV		99 (31.7%)
Tumor T status
T1+T2		215 (68.9%)
T3+T4		97 (31.1%)
Lymph node status
N0		302 (96.8%)
N1+N2+N3		10 (3.2%)
Metastasis
M0		297 (95.2%)
M1		15 (4.8%)
Child-Pugh grade
A		242 (77.6%)
B or C		70 (22.4%)
Liver cirrhosis
Negative		52 (16.7%)
Positive		260 (83.3%)

Mann-Whitney U test or Fisher's exact test was used between healthy controls and patients with HCC. * p value < 0.05 as statistically significant.

The distribution of the AURKA genotypes between the HCC patients and HCs is shown in Table 2. In the HCs, all genotypic frequencies were in Hardy-Weinberg equilibrium (p > 0.05). In both patients and controls, most of those with the rs1047972 SNP were homozygous for the C/C genotype, most of those with the rs2273535 SNP were homozygous for the T/T genotype, most of those with the rs2064836 SNP were homozygous for T/T, and most of those with the rs6024836 SNP were homozygous for A/A (Table 2). After adjusting for potential confounders, subjects with T/T homozygotes of the AURKA rs1047972 polymorphism had a 2.678-fold (95% CI: 1.012-7.092; p < 0.05) higher risk of developing HCC compared to those with C/C homozygotes. However, no significant differences in the incidences of HCC patients with the rs2273535, rs2064836, and rs6024836 polymorphisms compared to HCs.

Table 2

Genotyping and allele frequency of AURKA single nucleotide polymorphisms (SNPs) in HCC patients and healthy controls.

Variable	Controls (N=624 (%)	Patients (N=312 (%)	OR (95% CI)a
rs1047972
CC	485 (77.7%)	235 (75.3%)	1.000 (reference)
TC	131 (21.0%)	67 (21.5%)	1.094 (0.751-1.594)
TT	8 (1.3%)	10 (3.2%)	2.678 (1.012-7.092)b
TC+TT	139 (22.3%)	77 (24.7%)	1.203 (0.841-1.720)
rs2273535
TT	310 (49.7%)	152 (48.7%)	1.000 (reference)
AT	257 (41.2%)	124 (39.8%)	1.004 (0.716-1.407)
AA	57 (9.1%)	36 (11.5%)	1.263 (0.768-2.078)
AT+AA	314 (50.3)	160 (51.3%)	1.061 (0.775-1.451)
rs2064863
TT	444 (71.1%)	217 (69.6%)	1.000 (reference)
GT	162 (26.0%)	88 (28.2%)	1.143 (0.804-1.624)
GG	18 (2.9%)	7 (2.2%)	0.600 (0.221-1.628)
GT+GG	180 (28.9%)	95 (30.4%)	1.073 (0.764-1.506)
rs6024836
AA	284 (45.5%)	147 (47.1%)	1.000 (reference)
AG	268 (43.0%)	130 (41.7%)	1.052 (0.755-1.466)
GG	72 (11.5%)	35 (11.2%)	0.949 (0.569-1.582)
AG+GG	340 (54.5%)	165 (52.9%)	1.029 (0.752-1.407)

a adjusted for the effects of age and gender.

b p = 0.047.

Next, we compared the distributions of the clinical aspects and AURKA genotypes in HCC patients. Compared with patients with the T/T genotype, those with at least one polymorphic allele at the rs2273535 SNP (A/T or A/A genotype) were less prone to developing stage III/IV disease (p = 0.033), large tumors (p = 0.033) and Child-Pugh B or C grade (p = 0.033), but were more likely to develop liver cirrhosis (p = 0.045) (Table 3). Moreover, carriers of the G/T+G/G genotype of rs2064863 had a lower risk than T/T carriers of developing large tumors (p = 0.047) and Child-Pugh grade B or C (p = 0.033), but were more likely to have HCV infection (p = 0.039) (Table 4).

Table 3

Odds ratios (ORs) and 95% confidence intervals (CIs) of clinical status and AURKA rs2273535 genotype frequencies in 312 HCC patients.

Variable	Genotypic frequencies
	TT (N=152)	AT+AA (N=160)	OR (95% CI)	p value
Clinical Stage
Stage I/II	95 (62.5%)	118 (73.7%)	1.00	P=0.033*
Stage III/IV	57 (37.5%)	42 (26.3%)	0.593 (0.367-0.960)
Tumor size
≤ T2	96 (63.2%)	119 (74.4%)	1.00	P=0.033*
> T2	56 (36.8%)	41 (25.6%)	0.591 (0.364-0.959)
Lymph node metastasis
No	146 (96.1%)	156 (97.5%)	1.00	P=0.472
Yes	6 (3.9%)	4 (2.5%)	0.624 (0.173-2.256)
Distant metastasis
No	144 (94.7%)	153 (95.6%)	1.00	P=0.714
Yes	8 (5.3%)	7 (4.4%)	0.824 (0.291-2.329)
Vascular invasion
No	124 (81.6%)	134 (83.8%)	1.00	P=0.613
Yes	28 (18.4%)	26 (16.2%)	0.859 (0.478-1.546)
Child-Pugh grade
A	110 (72.4%)	132 (82.5%)	1.00	P=0.033*
B or C	42 (27.6%)	28 (17.5%)	0.556 (0.323-0.954)
HBsAg
Negative	79 (52.0%)	96 (60.0%)	1.00	P=0.496
Positive	73 (48.0%)	64 (40.0%)	0.898 (0.659-1.224)
Anti-HCV
Negative	86 (56.6%)	78 (48.8%)	1.00	P=0.099
Positive	66 (43.4%)	82 (51.2%)	1.320 (0.949-1.836)
Liver cirrhosis
Negative	32 (21.1%)	20 (12.5%)	1.00	P=0.045*
Positive	120 (78.9%)	140 (87.5%)	1.867 (1.015-3.434)

The ORs with analyzed by their 95% CIs were estimated by logistic regression models.

> T2: multiple tumor more than 5 cm or tumor involving a major branch of the portal or hepatic vein(s)

* p value < 0.05 as statistically significant.

Table 4

Odds ratio (OR) and 95% confidence interval (CI) of clinical status and AURKA rs2064863 genotypic frequencies in 312 HCC patients.

Variable	Genotypic frequencies
	TT (N=217)	GT+GG (N=95)	OR (95% CI)	p value
Clinical Stage
Stage I/II	141 (65.0%)	72 (75.8%)	1.00	P=0.061
Stage III/IV	76 (35.0%)	23 (24.2%)	0.593 (0.343-1.023)
Tumor size
≤ T2	142 (65.4%)	73 (76.8%)	1.00	P=0.047*
> T2	75 (34.6%)	22 (23.2%)	0.571 (0.328-0.992)
Lymph node metastasis
No	209 (96.3%)	93 (97.9%)	1.00	P=0.471
Yes	8 (3.7%)	2 (2.1%)	0.562 (0.117-2.697)
Distant metastasis
No	205 (94.5%)	92 (96.8%)	1.00	P=0.374
Yes	12 (5.5%)	3 (3.2%)	0.557 (0.154-2.021)
Vascular invasion
No	176 (81.1%)	82 (86.3%)	1.00	P=0.265
Yes	41 (18.9%)	13 (13.7%)	0.681 (0.346-1.339)
Child-Pugh grade
A	161 (74.2%)	81 (85.3%)	1.00	P=0.033*
B or C	56 (25.8%)	14 (14.7%)	0.497 (0.261-0.946)
HBsAg
Negative	117 (53.9%)	58 (61.1%)	1.00	P=0.193
Positive	100 (46.1%)	37 (38.9%)	0.793 (0.559-1.125)
Anti-HCV
Negative	119 (54.8%)	45 (47.4%)	1.00	P=0.039*
Positive	98 (45.2%)	50 (52.6%)	1.441 (1.019-2.038)
Liver cirrhosis
Negative	38 (17.5%)	14 (14.7%)	1.00	P=0.546
Positive	179 (82.5%)	81 (85.3%)	1.228 (0.631-2.392)

The ORs with analyzed by their 95% CIs were estimated by logistic regression models.

> T2: multiple tumor more than 5 cm or tumor involving a major branch of the portal or hepatic vein(s)

* p value < 0.05 as statistically significant.

When we investigated associations between AURKA gene polymorphisms and serum levels of alpha-fetoprotein (AFP), aspartate transaminase (AST) and alanine transaminase (ALT) in HCC patients 24, we found significantly lower AFP levels in those carrying the rs1047972 T/C or T/T genotypes (p = 0.037; Table 5).

Table 5

Association of AURKA genotype frequencies with laboratory findings in liver tests from HCC patients.

Characteristic	α-Fetoprotein a (ng/mL)	AST (IU/L)	ALT (IU/L)	AST/ALT ratio
rs1047972
CC	1226.8 ± 365.6	58.61 ± 5.88	54.82 ± 5.48	1.23 ± 0.03
TC+TT	396.8 ± 153.0	47.51 ± 3.91	49.10 ± 4.84	1.19 ± 0.04
p value	0.037*	0.116	0.434	0.470
rs2273535
TT	1480.6 ± 523.4	63.39 ± 8.66	57.91 ± 8.00	1.23 ± 0.03
AT+AA	601.2 ± 230.3	48.89 ± 3.41	49.20 ± 3.63	1.21 ± 0.05
p value	0.125	0.120	0.322	0.654
rs2064863
TT	1144.4 ± 369.3	57.82 ± 6.18	54.39 ± 5.77	1.22 ± 0.02
GT+GG	772.8 ± 380.2	51.79 ± 5.12	51.36 ± 5.25	1.21 ± 0.07
p value	0.483	0.453	0.698	0.889
rs6024836
AA	1365.9 ± 493.4	65.00 ± 9.21	59.93 ± 8.49	1.26 ± 0.05
AG+GG	753.0 ± 314.7	48.41 ± 3.35	48.01 ± 3.57	1.19 ± 0.02
p value	0.295	0.091	0.196	0.193

Mann-Whitney U test was used between two groups.

a Mean ± S.E.

* p value < 0.05 as statistically significant.

Discussion

AURKA, a centrosome-associated serine/threonine kinase, has demonstrated higher expression in various human cancers including colorectal cancer, breast cancer, head and neck squamous cell carcinoma, as well as HCC 11-13, 25. It is postulated that this increase in expression might result in high chromosome instability in cancer and encourage susceptibility to malignant transformation 26, processes that may arise from the acquisition of the chromosome 20q amplicon, which promotes the adenoma to carcinoma progression 27. In addition, overexpression of AURKA has been found to enhance tumor proliferation, differentiation, and metastasis 28-30. AURKA also promotes cancer metastasis and cancer stem cells in HCC 31. Inhibition of AURKA promotes autophagy and cell cycle arrest, and induces chemosensitivity in HCC 32. These results suggest that knockdown AURKA might be a valuable therapeutic strategy for HCC. However, we dose not recruited the survival results of HCC patients. Future research could evaluate the association of AURKA polymorphisms with survival of HCC patients. In addition, it would be advisable to collect data on a larger number of patients for analysis of the functions of AURKA polymorphisms in HCC.

Since HCC is one of the most common and lethal tumors worldwide, preventing its occurrence and lowering its mortality rate is an important challenge. Infection with HBV or HCV, a history of liver cirrhosis, family history of HCC, and alcohol consumption are the dominant etiological factors for HCC in Taiwan 33. In this study, there is no difference between the ratios of cigarette smokers/nonsmokers in controls (60.7:39.3) and HCC patients (59:41), whereas a higher proportion of HCC patients consumed alcohol (37.8%) compared with controls (13.9%). This suggests that alcohol consumption is a risk factor for HCC development. Chronic alcohol consumption promotes hepatobiliary tumors by increasing microRNA-122-controlled HIF-1α activity and stemness 34. In a pig model, moderate alcohol consumption changed autophagy- and apoptosis-regulated pathways 35. Exposure alcohol frequently changed genes at fragile sites, and promoted AURKA functioning. An increasing body of evidence shows that alcohol consumption is a risk factor for HCC 36, 37. Our data is consistent with this finding, as those HCC patients who consumed alcohol were at higher risk of worsening disease.

The risk of breast cancer is high in individuals with the AURKA rs2273535 polymorphism 38, while the AURKA 91A (rs2273535) polymorphism is associated with a high risk of oral cancer 39. In Caucasians, the AURKA rs1047972 polymorphism is associated with a decreased risk of breast cancer 40. In this study, we did not find that the AURKA rs2273535 polymorphism was associated with HCC risk. However, our data does indicate that the AURKA rs1047972 polymorphism increases the risk of developing HCC. These findings suggest that different AURKA polymorphisms play different roles in cancer development.

This study found that HCC patients with the AURKA rs2273535 polymorphism had a lower risk of developing stage III/IV disease, large tumors, and Child-Pugh grade B or C. Similarly, the AURKA rs2064863 polymorphism was also associated with a lower risk of developing large tumors and Child-Pugh grade B or C. It is established that overexpression of the AURKA gene is implicated in the development of colorectal adenoma to colorectal cancer 26. In addition, AURKA upregulation promotes high chromosome instability in cancerous tissue and induces increased susceptibility to tumor transformation 26. However, more research is required to determine whether an association exists among advanced-stage disease, AURKA expression levels, and AURKA genotype, and clarification is needed in regard to the effects of the AURKA genotype on HCC risk.

In conclusion, the current study suggests a potentially clinically significant finding showing that several variants of the AURKA gene are associated with the clinical status and susceptibility of HCC. We found that individuals carrying the T/T allele of the AURKA SNP rs1047972 were at higher risk of HCC than wild-type (C/C) carriers. Genetic variations in the gene encoding AURKA may be a significant predictor of early HCC occurrence and a reliable biomarker for disease progression.