Imaging, Diagnosis, Prognosis

ASPM Is a Novel Marker for Vascular Invasion, Early Recurrence, and Poor Prognosis of Hepatocellular Carcinoma Shih-Yeh Lin,1Hung-Wei Pan,1Shu-Hsiang Liu,1Yung-Ming Jeng,3 Fu-Chang Hu,2,4 Shian-Yang Peng,5 Po-Lin Lai,3 and Hey-Chi Hsu1, 3

Abstract Purpose: Abnormal spindle-like microcephaly associated (ASPM) plays an important role in neurogenesis and cell proliferation. This study is to elucidate its role in hepatocelllular carcinoma (HCC), particularly early tumor recurrence (ETR) and prognosis. Experimental Design:We used reverse transcription-PCR assays to measure theASPM mRNA levels in 247 HCC and correlated with clinicopathologic and molecular features. Results: ASPM mRNA levels were high in fetal tissues but very low in most adult tissues.ASPM mRNAwas overexpressedin162 HCC (66%)but notinbenignliver tumors.ASPM overexpression correlated with high a-fetoprotein (P =110 -8), high-grade (grade II-IV) HCC (P =2 10 -6), high-stage (stage IIIA-IV) HCC (P =110 -8), and importantly ETR (P =110 -8). ETR is the most criticalunfavorableclinicalprognostic factor. Among the various independent histo- pathologic (tumor size, tumor grade and tumor stage) and molecular factors (p53 mutation, high a-fetoprotein, and ASPM overexpression), tumor stage was the most crucialhistologic factor (odds ratio, 14.7; 95% confidence interval, 6.65-33.0; P =110 -8), whereas ASPM overexpres- sion (odds ratio, 6.49; P =110 -8) is the most important molecular factor associated with ETR. ASPM overexpression was associated with vascular invasion and ETR in both p53-mutated (all P values = 1 10 -8) and non-p53-mutated HCC (P =110 -8 and 0.00088, respectively). Hence, patients with APSM-overexpressing HCC had lower 5-year survival (P = 0.000001)in both p53-mutated (P = 0.00008) and non-p53-mutated HCC (P = 0.0027). Inlow-stage (stage II) HCC, ASPM overexpression also correlated withhigher ETR (P =0.008). Conclusion: ASPM overexpression is a molecular marker predicting enhanced invasive/meta- static potentialof HCC, higher risk of ETR regardless of p53 mutation status and tumor stage, and hence poor prognosis.

Abnormal spindle-like microcephaly associated (ASPM) gene is the mutation of Drosophila asp causes abnormal spindles, frequent human orthologue of the Drosophila abnormal spindle (asp) and polyploid cells, and cytokinesis failure (6), leading to arrest of the most commonly mutated gene of autosomal recessive neuroblasts in metaphase (7, 8) and larval-pupal lethality primary microcephaly (1–5). The homozygous semilethal (6, 9). In human, the defective neurogenesis caused by homozygous mutation of ASPM leads to microcephaly and mental retardation (4, 10). The ASPM gene encodes a large protein of 3,477 amino acids with a NH2-terminal microtu- Authors’ Affiliations: 1Graduate Institute of Pathology, College of Medicine; bule-binding domain, two calponin homology domains, 74 2College of Public Health, National Taiwan University;3Department of Pathology 4 repeated calmodulin-binding isoleucine-glutamine domains, and NationalCenter of Excellence for GeneralClinicalTrialand Research, National and a COOH-terminal region (4, 11, 12). The NH -terminal Taiwan University Hospital; and 5Department of GeneralEducation, NationalTaipei 2 College of Nursing,Taipei,Taiwan, Republic of China microtubule-binding domain is highly conserved in eukaryotes, Received 12/28/07; revised 4/8/08; accepted 4/11/08. suggesting that ASPM may possess the fundamental function in Grant support: National Health Research Institute, Department of Health of the cytokinesis. However, the multidomain feature suggests that Republic of China,Taiwan grant NHRI-EX94-9427NI, and National Science Council ASPM have more diverse function than the Drosophila of the Republic of China,Taiwan grant NSC94-3112-B-002-018. The costs of publication of this article were defrayed in part by the payment of page counterpart. charges. This article must therefore be hereby marked advertisement in accordance Drosophila Asp protein associates with centrosomes and is with 18 U.S.C. Section 1734 solely to indicate this fact. involved in organizing microtubules at the spindle poles in Note: Supplementary data for this article are available at Clinical Cancer Research mitosis and meiosis (6, 8, 9, 11, 13–16). Centrosome defects Online (http://clincancerres.aacrjournals.org/). S-Y.Lin and H-W. Pan contributed equally to this work. cause chromosome misaggregation and aneuploidy, leading to Current address for S-H. Liu: Department of MedicalResearch, Mackay Memorial genetic instability, a major driving force for malignant transfor- Hospital,Danshui,TaipeiCounty,Taiwan,RepublicofChina. mation and tumor progression (17–22). Recently, ASPM is Requests for reprints: Hey-Chi Hsu, Department of Pathology, National Taiwan shown to express in nearly all transformed human cell lines and University Hospital, Taipei, Taiwan, Republic of China. Phone: 886-2-23123456, ext. 5455; Fax: 886-2-23410876; E-mail: [email protected]. in multiple fetal tissues (23, 24). These findings suggest ASPM F 2008 American Association for Cancer Research. play an important role in cell cycle progression and cell doi:10.1158/1078-0432.CCR-07-5262 proliferation in embryonic development and tumorigenesis.

Clin Cancer Res 2008;14(15) August 1, 2008 4814 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. ASPM and Hepatocellular Carcinoma

Using differential display analysis of gene expression profile Statistical analysis. The statistical significance of differences of of hepatocelllular carcinoma (HCC), we found frequent over- selected clinicopathologic features between those with and without m2 expression of ASPM, which is located at chromosome 1q31, a ASPM mRNA overexpression was assessed by tests. For 5-year region with frequent gain in HCC (25, 26). In this study, we survival analysis, person-months were calculated from the date of resection to the date of death or 60 months for those who survived for showed that ASPM was often overexpressed in human HCC >5 years. P values < 0.05 or a 95% confidence interval (95% CI), not and associated with tumor progression, early tumor recurrence including 1, were considered statistically significant. Multivariate (ETR), and poor prognosis. analyses of grade and stage were conducted by fitting multiple logistic regression models (42) and then times to ETR and death were analyzed by fitting multiple Cox’s proportional hazards models (43). Basic Patients and Methods model-fitting techniques for (a) variable selection, (b) goodness-of-fit assessment, and (c) regression diagnostics (including residual analysis, Differential display, reverse transcription-PCR, and definition of ASPM influence analysis, and check of multicollinearity) were used in our overexpression. Differential display was done using anchor primer regression analyses to assure the quality of analysis results (42, 43). HT11C (5¶-AAGCTTTTTTTTTTTC-3¶) and arbitrary primer HAP7 (5¶- Two-tailed P < 0.05 was considered statistically significant. AAGCTTAACGAGG-3¶) from the RNAimage kit (GenHunter) as described (27–31). Reverse transcription-PCR assays in the linear range were used for Results the measurements of ASPM mRNA levels using ribosomal protein S26 ASPM (RPS26; 32) or porphobilinogen deaminase (PBGD) mRNA for internal mRNA expression in fetal and adult tissues and various controls (33) as described (29–31, 34). Primers for ASPM spanning solid types of human cancer. By the differential display analysis, intron 20 to avoid DNA contamination are ASPM-forward (5¶- we identified a 240-bp cDNA, which was often expressed in TAAAAAGACATCGAGCTGCTT-3¶) and ASPM-reverse (5¶-CCTCTCCA- HCC and had sequence identical to human ASPM (Genbank TAATGCCTGAATT-3¶). The primers for RPS26 are RPS26-forward NM_018136; Fig. 1A). Using reverse transcription-PCR at the (5¶-CCGTGCCTCCAAGATGACAAAG-3¶) and RPS26-reverse (5¶- linear range, ASPM mRNA was expressed in relative abundance TGTCTGGTAACGGCAATGCGGCT-3¶) and the primers for PBGD are in most fetal tissues, including the liver (Fig. 1B), and all the ¶ ¶ ¶ ¶ ¶ PBGD-5 (5 -TGTCTGGTAACGGCAATGCGGCT-3 ) and PBGD-3 (5 - cell lines examined (Fig. 1C), but the level was very low in most GGTCCACTTCATTCTTCTCCAG-3¶). Relative ASPM mRNA level was adult tissues (Fig. 1B). determined as the ratio of ASPM to PBGD as described (29–31, 34, 35) and scored as high (ratio z 1.0; 128 cases), medium (0.5 < In clinical tissue samples, ASPM mRNA overexpression was ratio < 1.0; 18 cases), low (ratio < 0.5; 39 cases), trace (ratio < 0.2; found in 162 (66%) of 247 surgically removed, unifocal, 44 cases), and negative (24 cases). In 227 nontumorous livers primary HCC (Fig. 2) but in none of hepatocellular adenomas examined, none had a ratio exceeding 0.5. Hence, a ratio z 0.5 was (4 cases), focal nodular hyperplasia (10 cases), and non- regarded as ASPM overexpression. tumorous livers (227 cases). Patients, histologic study, ETR, follow-up observation, and p53 Correlation of ASPM overexpression with HCC progres- mutation. Between 1983 and 1997, 247 surgically resected, unifocal, sion. To better understand the significance of ASPM in HCC, primary HCC pathologically assessed at the National Taiwan University we correlated its mRNA expression with the major clinicopath- Hospital, as described (36–39), formed the basis of this study. This ologic features. As shown in Table 1, ASPM overexpression was study was executed according to the regulations of the Ethical strongly associated with high serum AFP level (>200 ng/mL; Committee of the National Taiwan University Hospital. These patients, -8 ages 14 to 88 years (mean, 55.5 years) and with adequate liver function P =1 10 ). Importantly, ASPM overexpression closely reserve, had survived for 2 months after hepatectomy, and none correlated with high-grade HCC [grade II-IV; odds ratio (OR), -7 received transhepatic arterial embolization or chemotherapy before 5.1; 95% CI, 2.6-10; P =1 10 ] and high-stage HCC (stages -8 surgery. The tumor was V3 cm in 56cases and >5 cm in 137 cases. The IIIA-IV versus II; OR, 30.6; 95% CI, 13.1-73.6; P =1 10 ). tumor grade was classified into grade I to IV. The tumor staging was ASPM overexpression also occurred more often in bigger classified into five groups, which closely correlated with survival in 781 tumors (>5 cm) but less distinct (P = 0.014). unifocal surgical HCC patients, as described (35). Stage I and II HCC ASPM overexpression as an important predictive marker for had no vascular invasion. In contrast, stage IIIA to IV HCC had vascular ETR and poor prognosis. ETR is the most critical, early clinical invasion, with various extents of microscopic intrahepatic spread in factor predictive of poor prognosis of HCC after hepatectomy stage IIIB and IV HCC. (30, 35). ETR occurred two times higher in HCC with ASPM During the follow-up observation up to 236months until January 2008, 245 (99%) cases had been followed for >5 years (90 cases) or overexpression than in HCC without the overexpression (P =1 -8 until death (155 cases). Intrahepatic tumor recurrence and/or distant 10 ; Table 2). Early metastasis was also found more often in metastasis detected within 12 months after tumor resection were HCC with ASPM overexpression [28% (30 of 107) versus 11% defined as ETR, which was detected in 107 (47%) of 226eligible cases. (8 of 73); P = 0.006]. With the close correlation with high The diagnosis of recurrence and/or metastasis was based on serum tumor stage and ETR, the two most crucial histopathologic and a-fetoprotein (AFP) elevation, imaging findings (ultrasonography, clinical unfavorable prognostic factors, HCC with ASPM computer tomography, angiography, bone scan, and magnetic reso- overexpression had lower 5-year survival than those without nance imaging), and histology as described (30, 34, 35). The details of the overexpression (P = 0.000001; Fig. 3). We then examined category in patients, histologic study, and ETR were described in the postoperative treatment modalities and metastasis, which Supplementary Material. In cases with ETR, patients with only early could affect significantly the patient’s outcome, and found no intrahepatic recurrence had much better survival. Mutations of p53 tumor suppressor gene, spanning exons 2 to 11, significant difference in both groups of patients. Surgical tumor were detected by DNA sequencing as described (38, 40, 41). resection and/or transarterial chemoembolization of the intra- To validate the clinicopathologic significance of ASPM expression, hepatic recurrent and distant metastatic lesions were done in patients were randomly assigned to two groups, the learning set (124 50 (56%) of 90 HCC with ETR and ASPM overexpression and cases) and test set (123 cases). in 13 (76%) of 17 HCC with ETR but without ASPM

www.aacrjournals.org 4815 Clin Cancer Res 2008;14(15) August 1, 2008 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

Fig. 1. A, differentialdisplayanalysis. Among 10 pairs of hepatocellular carcinoma (T) and nontumorous liver (L) samples, a significant differential band f240 bp (arrow) was overexpressed in 6 tumors. B, ASPM mRNA expression in adult and fetal tissues. Using reverse transcription-PCR measurements, ASPM mRNA was expressed in high level in multiple fetus tissues but not in adult tissues. brain-c, brain cortex; brain-m, brain white matter. C, ASPM mRNA expression in multiple cancer and noncancer cell lines.ASPM,28cycles; RPS26, RPS26 mRNA was amplified at 22 cycles as internalcontrol.

overexpression (P = 0.108). Surgical resection was done for evidence of vascular invasion [28% (12 of 31) versus 11% early metastatic tumor in 14% (4 of 29) of HCC with ASPM (10 of 68); P = 0.008]. overexpression and in 25% (2 of 8) without ASPM over- Interaction of ASPM overexpression with p53mutation and expression (P = 0.591). high AFP in relation to ETR and prognosis. The p53 gene is the In addition to the close association with vascular invasion most commonly mutated gene in HCC and associated with (stage IIIA-IV) and ETR, ASPM overexpression was also more aggressive tumor (38, 40). High serum AFP is the most associated with higher ETR in stage II HCC, which had no widely used diagnostic marker and associated with more

Fig. 2. Expression of ASPM mRNA in hepatocellular carcinoma. The ASPM/PBGD ratios were listed below. Overexpression of ASPM mRNA (ASPM/PBGD ratioz 0.5; asterisk) was detected in 15 of 31representative HCC but in none of the 13 nontumorous livers.Tumor stage and tumor size are depicted above. ASPM and internal control PBGD were both amplified at 28 cycles of reverse transcription-PCR.

Clin Cancer Res 2008;14(15) August 1, 2008 4816 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. ASPM and Hepatocellular Carcinoma

Table 1. Correlation of ASPM mRNA expression Discussion with clinicopathologic features and aberrant gene ASPM plays an important role in the neuronogenesis and its expression in hepatocellular carcinoma by mutation is the major cause of primary microencephaly (1–5, univariate logistic regression analyses 44). Recent studies have shown that ASPM is located to the Variable ASPM overexpression centrosomes, spindle poles, and midbody (23, 45) and plays an important role in cytokinesis and cell proliferation in the Total n (%) OR (95% CI) P transformed human cell lines, fetal tissues, and human cancer Age (y) cells (23). Despite these elegant observations, the significance V56 116 83 (60) 1.0 of ASPM in human cancer has not been fully investigated. In >56 131 79 (603) 1.66 (0.94-2.93) 0.0063 Gender this study, we showed ASPM mRNA overexpression in 66% of Male 194 121 (62) 1.0 247 surgical removed, unifocal primary HCC and in 54% of Female 53 41 (77) 2.06 (0.97-4.44) 0.042 167 other types of human cancer, including hepatoblastoma, Serum HBsAg cholangiocarcinoma, and carcinomas of other anatomic sites - 88 47 (53) 1.0 (data not shown), consistent with the observations reported by + 159 115 (72) 2.28 (1.28-4.08) 0.0027 AFP (ng/mL) other investigators (23). Besides, ASPM mRNA was expressed V200 131 63 (48) 1.0 abundantly in most fetal tissues, including the liver. In contrast, >200 116 99 (85) 6.29 (3.25-12.26) 1 10-8 ASPM mRNA level was very low or undetectable in many adult Liver cirrhosis tissues, including liver, and benign hepatic tumor conditions, Yes 95 60 (63) 1.0 No 152 102 (67) 1.19 (0.67-2.11) 0.525 such as hepatocellular adenoma and focal nodular hyperplasia. Tumor size (cm) V5 110 63 (57) 1.0 >5 137 99 (72) 1.94 (1.10-3.43) 0.014 Tumor grade I 59 20 (34) 1.0 Table 2. Correlation of clinicopathologic variables II 99 60 (61) 3.0 (1.45-6.23) 0.0012 -8 and ASPM expression with ETR of resected III-IV 89 76 (85) 11.4 (4.8-27.6) 1 10 Tumor stage hepatocellular carcinoma I 6 1 (17) —* II 104 33 (32) 1.0 Variable* ETR -8 IIIA-IV 137 128(93) 30.6 (13.1-73.6) 1 10 Total n (%) OR (95% CI) P ETRc No 11853 (45) 1.0 Clinical features Yes 107 90 (84) 6.49 (3.31-12.9) 1 10-8 Age (y) p53 mutation >56 115 49 (43) 1.0 No 114 62 (54) 1.0 V56 110 58(53) 1.5 (0.86-2.63) 0.129 Yes 98 80 (82) 3.73 (1.90-7.37) 0.000026 Sex Male 177 79 (45) 1.0 Female 49 28(57) 1.65 (0.83-3.29) 0.121 *Stage I HCC was not used for comparison because of the small Cirrhosis number of cases, low frequency of ASPM overexpression, and No 141 68(48) 1.0 unique lack of ETR. Yes 85 39 (46) 0.91 (0.51-1.62) 0.732 cETR within 12 mo after hepatectomy. HBsAg Negative 83 35 (42) 1.0 Positive 143 72 (50) 1.39 (0.78-2.49) 0.235 aggressive HCC, ETR, and poor prognosis (35). In this study, Histopathologic variables Tumor size (cm) we showed that ASPM overexpression correlated with high AFP V5 102 31 (30) 1.0 -8 (P =1 10 ) and p53 mutation (P = 0.000026; Table 1). We >5 123 76 (62) 3.7 (2.05-6.73) 3 10-6 then analyzed their roles and the major clinicopathologic Tumor grade factors in relation to ETR. As shown in Table 2, tumor size, I 5814 (24) 1.0 tumor grade, and tumor stage were important histopathologic II-IV 167 93 (56) 3.95 (1.92-8.22) <0.00004 Tumor stage -8 risk factors for ETR, particularly high tumor stage (P =1 10 ), I 6 0 (0) — whereas ASPM overexpression, high AFP, and p53 mutation II 99 22 (22) 1.0 were the molecular risk factors, particularly ASPM overexpres- IIIA-IV 120 85 (71) 8.5 (4.4-16.6) 1 10-8 sion (P =1 10-8). By multivariate regression analysis, ASPM Molecular markers p53 mutation overexpression directly correlated with ETR after considering No 103 35 (34) 1.0 tumor size, tumor stage, and p53 mutation (Table 3). The Yes 90 55 (61) 3.05 (1.62-5.75) 0.00016 multivariate regression analysis also showed that ASPM over- AFP (ng/mL) expression directly correlated with high tumor stage after V200 121 40 (33) 1.0 -6 considering tumor size (Table 3). Regardless of p53 mutation, >200 104 67 (64) 3.67 (2.04-6.63) 3 10 ASPM "* ASPM overexpression was associated with vascular invasion No 82 17 (22) 1.0 -8 (P values = 1 10 ), ETR (P values < 0.001) (Supplementary Yes 143 90 (63) 6.49 (3.31-12.9) 1 10-8 Table S1), and hence lower 5-year survival rate in both non- p53-mutated and p53-mutated HCC (P = 0.0027 and 0.00008, *", mRNA overexpression. respectively; Fig. 3).

www.aacrjournals.org 4817 Clin Cancer Res 2008;14(15) August 1, 2008 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

These findings suggest that ASPM is actively involved in cell improvement of earlier diagnosis and better management, the proliferation in embryonic development and the tumorigenesis outcome of HCC after tumor resection remains unsatisfactory of HCC and various other types of human cancer, whereas because of the high tumor recurrence rate (30, 46–48). ETR is ASPM down-regulation is associated with tissue maturation the most crucial unfavorable, clinical prognostic factor for and cell differentiation. surgical HCC patients, and <20% of the patients with ETR To elucidate the role of ASPM in HCC, we correlated its could survive >5 years after tumor resection, whereas >50% of mRNA expression with the major clinical and histopathologic patients without ETR did (30). Hence, ETR can be regarded as a variables associated with tumor progression. We showed that crucial clinical event before death, and markers for the ASPM overexpression correlated with high serum AFP level prediction of ETR are needed. Although the list of molecular (>200 ng/mL; P =1 10-8) and poorly differentiated (grade II- factors associated with ETR expands rapidly (30, 34, 35, IV) tumor (P =1 10-7) but less significantly with bigger 49–51), the number of markers useful as predictors of ETR tumor (>5 cm; P = 0.014). These findings suggest that ASPM remains limited and their interaction has not been well overexpression in HCC facilitates tumor cell proliferation, investigated. Here, we showed that HCC with ASPM over- leading to bigger tumor and less differentiation, which are expression had ETR f2-fold higher than HCC without the associated with high AFP level (35). Importantly, ASPM overexpression (OR, 6.49; 95% CI, 3.31-12.9; P =1 10-8). In overexpression was associated with high-stage (stage IIIA, IIIB, addition to the close association with vascular invasion (stage and IV) HCC that had vascular invasion and various extents of IIIA-IV) and ETR, ASPM overexpression was also associated intrahepatic metastasis (OR, 30.6; 95% CI, 13.1-73.6; P =1 with higher ETR in stage II HCC (low stage), which had no 10-8). These findings suggest that HCC with ASPM over- microscopic evidence of vascular invasion (P = 0.008). These expression harbor enhanced invasion/metastasis potential. findings suggest that ASPM overexpression serves as a useful HCC is a dreadful disease difficult to treat; surgical resection predictive marker to identify surgical HCC at high risk of ETR provides an opportunity for cure. Despite the significant regardless of p53 mutation status and tumor stage.

Fig. 3. A, cumulative survival curves for HCC in relation toASPM overexpression ("). HCC with ASPM mRNA overexpression (Ye s ) had significantly lower 5-yr survival rate than HCC without ASPM overexpression (No ; P = 0.000001, log-rank test). B, cumulative survival for HCC without p53 mutation. The 5-yr survivalrate of HCC with ASPM overexpression was lower than HCC without ASPM overexpression (P = 0.0055, log-rank test). C, cumulative survival for HCC with p53 mutation. The 5-yr survivalrate of HCC with ASPM overexpression was lower than HCC without ASPM overexpression (P = 0.0009, log-rank test).

Clin Cancer Res 2008;14(15) August 1, 2008 4818 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. ASPM and Hepatocellular Carcinoma

Table 3. Multivariate analyses of risk factors associated with ETR, tumor stage, and survival of patients with unifocal hepatocellular carcinoma

Covariate Variable estimate SE Wald C2 P Odds/hazard ratio

ETR (yes)* Intercept -2.2895 0.3879 34.8380 <0.0001 — Size 0.1147 0.0392 8.5624 0.0034 1.122 Stage (III-IV) 1.3713 0.4024 11.6153 0.0007 3.940 ASPM 0.9186 0.4181 4.8259 0.0280 2.506 High stage (stage III-IV; vascular invasion; yes)c Intercept -5.7916 1.045830.6664 <0.0001 — Size 0.2100 0.0809 6.7391 0.0094 1.234 ASPM 2.6917 0.7291 13.6304 0.0002 14.757 Survival time (death)b Size 0.05917 0.01987 8.8623 0.0029 1.061 High stage 0.17570 0.064787.3564 0.0067 1.192 ETR 1.82381 0.22868 63.6048 <0.0001 6.195

*Logistic regression model (n = 226), percentage of concordant pairs = 80.2%, percentage of discordant pairs = 18.7%, Hosmer and Lemeshow goodness-of-fit test P = 0.8374 > 0.05 (df = 7). cLogistic regression model (n = 141), percentage of concordant pairs = 94.9%, percentage of discordant pairs = 4.9%, Hosmer and Lemeshow goodness-of-fit test P = 0.7419 > 0.05 (df =8). bCox’s proportional hazards model (n = 247).

Using univariate analysis, tumor size, tumor grade, and important roles in cell cycle progression, and p53 mutation is tumor stage were important histopathologic risk factors for the most commonly mutated gene in HCC in association with ETR, particularly high tumor stage (P =1 10-8), whereas more aggressive tumor (38, 40). We then analyzed the ASPM overexpression, high AFP, and p53 mutation were the interaction between ASPM overexpression and p53 mutation, molecular risk factors, particularly ASPM overexpression the latter would be accompanied by the loss of G1 (52) and -8 (P =1 10 ). The multivariate regression analysis showed G2-M checkpoints (53), in HCC progression. In both HCC with that ASPM overexpression directly correlated with high tumor and without p53 mutation, ASPM overexpression was associ- stage after considering tumor size and ASPM overexpression ated with higher vascular invasion (P values = 1 10-8), ETR directly correlated with ETR after considering tumor size, (P values < 0.001 and P =1 10-8, respectively), and lower tumor stage, and p53 mutation. These findings indicate 5-year survival rate (P = 0.0027 and 0.0008, respectively). These that ASPM overexpression is an independent risk factor of findings indicate that ASPM overexpression is an important high-stage HCC, and conditioning on high stage and tumor factor associated with HCC progression. However, it was also size, ASPM overexpression is also an independent risk factor noted that HCC with p53 mutation and ASPM overexpression of ETR. had the highest frequencies of vascular invasion and ETR and With the close association with high tumor stage and ETR, the the lowest 5-year survival. These findings suggest that the loss two most crucial histopathologic and clinical factors associated of p53-mediated checkpoint contributes cooperatively with the with poor prognosis, HCC with ASPM overexpression had a ASPM overexpression toward more advanced disease, with significantly lower 5-year survival than HCC without the higher ETR and poorer prognosis. overexpression (P = 0.000001). Together, our findings indicate that ASPM overexpression contribute to higher tumor stage Disclosure of Potential Conflicts of Interest and frequent ETR, leading to poor prognosis, and a combined analysis of multiple molecular factors provides a more accurate No potentialconflicts of interest were disclosed. assessment of risk for ETR and prognosis of HCC. In addition to the association with centrosomes and Acknowledgments involvement in cell division (23, 45), we showed that ASPM was cell cycle regulated (data not shown). The p53 plays We thank Chia-Yen Su for excellent technical assistance.

References 1. Bundey S. Abnormalmentaldevelopment. In: Rimoin human microcephaly. Curr Opin Neurol 2001;14: 6. Wakefield JG, Bonaccorsi S, Gatti M. The Drosophila DL, Connor JM, Pyeritz RE, editors. Emery and 151 ^ 6. protein asp is involved in microtubule organization Rimoin’s principles and practice of medical genetics. 4. BondJ, Roberts E, Mochida GH, et al. ASPMis a ma- during spindle formation and cytokinesis. J Cell Biol 3rd ed. New York: Churchill Livingstone; 1997. p. jor determinant of cerebralcorticalsize. Nat Genet 2001;153:637^ 48. 730 ^ 1. 2002;32:316^ 20. 7. Gonzalez C, Molina I, Casal J, Ripoll P. Gross genetic 2. Aicardi J. Malformations of the central nervous sys- 5. Kumar A, Blanton SH, Babu M, Markandaya M, dissection and interaction of the chromosomalregion tem. Diseases of the nervous system in childhood. Girimaji SC. Genetic analysis of primary microcephaly 95E;96F of Drosophila melanogaster.Genetics1989; 2nd ed. London: Mac Keith; 1998. p. 90 ^ 1. in Indian families: novel ASPM mutations. Clin Genet 12 3 :371 ^ 7. 3. Mochida GH, Walsh CA. Molecular genetics of 2004;66:341 ^8. 8. Gonzalez C, Saunders RD, Casal J, et al. Mutations at

www.aacrjournals.org 4819 Clin Cancer Res 2008;14(15) August 1, 2008 Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. Imaging, Diagnosis, Prognosis

the asp locus of Drosophila leadtomultiplefree25. Guan XY, FangY, Sham JS, et al. Recurrent chromo- carcinoma with multiple nodules. Hepatology 1991; centrosomes in syncytialembryos, but restrict centro- some alterations in hepatocellular carcinoma detected 13 :92 3 ^ 8 . some duplication in larval neuroblasts. J Cell Sci 1990; by comparative genomic hybridization. Genes Chro- 38. Hsu HC, Tseng HJ, Lai PL, Lee PH, Peng SY. Ex- 96:605^16. mosomes Cancer 2000;29:110^ 6. pression of p53 gene in 184 unifocal hepatocellular 9. Ripoll P, Pimpinelli S,Valdivia MM, Avila J. A cell divi- 26. Leung TH, Wong N, Lai PB, et al. Identification of carcinomas: association with tumor growth and inva- sion mutant of Drosophila with a functionally abnor- four distinct regions of allelic imbalances on chromo- siveness. Cancer Res 1993;53:4691 ^ 4. malspindle. Cell1985;41:907 ^ 12. some1by the combined comparative genomic hybrid- 39. Hsu HC, JengYM, MaoTL, Chu JS, Lai PL, Peng SY. 10. Bond J, Scott S, Hampshire DJ, et al. Protein-trun- ization and microsatellite analysis on hepatocellular h-Catenin mutations are associated with a subset of cating mutations in ASPM cause variable reduction in carcinoma. Mod Pathol2002;15:1213^ 20. low-stage hepatocellular carcinoma negative for hep- brain size. Am J Hum Genet 2003;73:1170 ^ 7. 27. Hsu HC, Cheng W, Lai PL. Cloning and expression atitis B virus and with favorable prognosis. Am J 11. Saunders RD, Avides MC, Howard T, Gonzalez C, of a developmentally regulated transcript MXR7 in he- Pathol2000;157:763 ^ 70. Glover DM. The Drosophila gene abnormalspindle patocellular carcinoma: biological significance and 40. HsuHC,PengSY,LaiPL,ChuJS,LeePH.Muta- encodes a novelmicrotubule-associated protein that temporospatialdistribution. Cancer Res 1997;57: tions of p53 gene in hepatocellular carcinoma (HCC) associates with the polar regions of the mitotic spin- 5179^ 84. correlate with tumor progression and patient progno- dle. J Cell Biol 1997;137:881 ^ 90. 28. Huang LR, Hsu HC. Cloning and expression of sis. Int J Oncol1994;4:1341 ^ 7. 12. Craig R, Norbury C. The novelmurine calmodulin- CD24 gene in human hepatocellular carcinoma: a po- 41. Hsu HC, Peng SY, Lai PL, et al. Allelotype and loss binding protein Sha1disrupts mitotic spindle and repli- tentialearlytumor marker gene correlateswith p53 of heterozygosity of p53 in primary and recurrent he- cation checkpoint functions in fission yeast. J Cell Sci mutation and tumor differentiation. Cancer Res 1995; patocellular carcinomas. A study of150 patients. Can- 19 9 8 ; 111: 3 6 0 9 ^ 19 . 55:4717^ 21. cer 1994;73:42 ^ 7. 13. do Carmo Avides M, Glover DM. Abnormal spindle 29. Liu SH, Lin CY, Peng SY, et al. Down-regulation of 42. Hosmer DW, Lemeshow S. Applied logistic regres- protein, Asp, and the integrity of mitotic centrosomal Annexin A10 in hepatocellular carcinoma is associated sion. 2nd ed. New York (NY): John Wiley & Sons; microtubule organizing centers. Science 1999;283: with vascular invasion, early recurrence, and poor 2000. 17 3 3 ^ 5 . prognosis in synergy with p53 mutation. Am J Pathol 43. Hosmer DW, Lemeshow S. Applied survival analy- 14. do Carmo Avides M,Tavares A, Glover DM. Polo ki- 2002;160:1831 ^ 7. sis: regression modeling of time to event data. New nase and Asp are needed to promote the mitotic orga- 30. Pan HW, Ou YH, Peng SY, et al. Overexpression of Yo r k (NY ): Jo hn W ile y & S o n s ; 19 9 9 . nizing activity of centrosomes. Nat CellBiol2001;3: osteopontin is associated with intrahepatic metasta- 44. Kumar A, Markandaya M, Girimaji SC. Primary 421 ^ 4. sis, early recurrence, and poorer prognosis of surgical- microcephaly: microcephalin and ASPM determine 15. Andersen JS, Wilkinson CJ, MayorT, Mortensen P, ly resected hepatocellular carcinoma. Cancer 2003; the size of the human brain. J Biosci 2002;27: Nigg EA, Mann M. Proteomic characterization of the 98:119^27. 629 ^ 32. 31. Pan HW, Chou HYE, Liu SH, Peng SY, Liu CL, Hsu human centrosome by protein correlation profiling. 45. Paramasivam M, Chang YJ, LoTurco JJ. ASPM and Nature 2003;426:570 ^ 4. HC. Role of L2DTL, cell cycle-regulated nuclear and citron kinase co-localize to the midbody ring during centrosome protein, in aggressive hepatocellular car- 16. Ponting CP. A noveldomain suggests a ciliary func- cytokinesis. Cell Cycle 2007;6:1605 ^12. tion for ASPM, a brain size determining gene. Bioinfor- cinoma. Cell Cycle 2006;5:2676 ^ 87. 46. Nagasue N, Uchida M, Makino Y, et al. Incidence matics 2006;22:1031 ^ 5. 32. Vincent S, Marty L, Fort P. S26 ribosomalprotein and factors associated with intrahepatic recurrence 17. Lengauer C, Kinzler KW, Vogelstein B. Genetic RNA: an invariant controlfor gene regulation experi- following resection of hepatocellular carcinoma. Gas- ments in eucaryotic cells and tissues. Nucleic Acids instabilities in human cancers. Nature 1998;396: troenterology 1993;105:488 ^ 94. 643^9. Res 1993;21:1498. 47. Hu RH, Lee PH,Yu SC, et al. Surgical resection for 18. Cahill DP, Kinzler KW,Vogelstein B, Lengauer C. Ge- 33. Grandchamp B, De VerneuilH, Beaumont C, netic instability and Darwinian selection in tumours. Chretien S, Walter O, Nordmann Y. Tissue-specific recurrent hepatocellular carcinoma: prognosis and analysis of risk factors. Surgery 1996;120:23 ^9. Trends Cell Biol 1999;9:M57 ^ 60. expression of porphobilinogen deaminase. Two isoen- 19. Pihan GA, Purohit A, Wallace J, et al. Centrosome zymes from a single gene. Eur J Biochem 1987;162: 48. PoonRT,FanST,NgIO,LoCM,LiuCL,WongJ. defects and genetic instability in malignant tumors. 10 5 ^ 10. Different risk factors and prognosis for early and late Cancer Res 1998;58:3974 ^ 85. 34. Peng SY, Ou YH, Chen WJ, et al. Aberrant intrahepatic recurrence after resection of hepatocellu- lar carcinoma. Cancer 2000;89:500 ^ 7. 20. Pihan GA, Doxsey SJ. The mitotic machinery as a expressions of Annexin A10 short isoform, osteo- source of genetic instability in cancer. Semin Cancer pontin and a-fetoprotein at chromosome 4q co- 49. Tang ZY,Ye SL, Liu YK, et al. A decade’s studies on Biol1999;9:289^302. operatively contribute to progression and poor metastasis of hepatocellular carcinoma. J Cancer Res 21. Carroll PE, Okuda M, Horn HF, et al. Centrosome prognosis of hepatocellular carcinoma. Int J Oncol Clin Oncol 2004;130:187^96. hyperamplification in human cancer: chromosome in- 2005;26:1053 ^ 61. 50. Matoba K, Iizuka N, GondoT, et al. Tumor HLA-DR stability induced by p53 mutation and/or Mdm2 over- 35. PengSY,ChenWJ,LaiPL,JengYM,SheuJC,Hsu expression linked to early intrahepatic recurrence of expression. Oncogene 1999;18:1935^ 44. HC. High a-fetoprotein level correlates with high hepatocellular carcinoma. Int J Cancer 2005;115: 22. Brinkley BR. Managing the centrosome numbers stage, early recurrence and poor prognosis of hepato- 231 ^ 40. game: from chaos to stability in cancer cell division. cellular carcinoma: significance of hepatitis virus in- 51. Uenishi T, Kubo S, Yamamoto T, et al. Cytokeratin Trends CellBiol2001;11:18^ 21. fection, age, p53 and h-catenin mutations. Int J 19 expression in hepatocellular carcinoma predicts 23. Kouprina N, Pavlicek A, Collins NK, et al. The micro- Cancer 2004;112:44 ^ 50. early postoperative recurrence. Cancer Sci 2003;94: cephaly ASPM gene is expressed in proliferating tis- 36. HsuHC,WuTT,WuMZ,SheuJC,LeeCS,ChenDS. 851^7. sues and encodes for a mitotic spindle protein. Hum Tumor invasiveness and prognosis in resected hepato- 52. Eastman A. Cell cycle checkpoints and their impact MolGenet 2005;14:2155 ^ 65. cellular carcinoma. Clinical and pathogenetic implica- on anticancer therapeutic strategies. J Cell Biochem 24. Rhoads A, Kenguele H. Expression of IQ-motif tions. Cancer 1988;61:2095 ^ 9. 2004;91:223 ^ 31. genes in human cells and ASPM domain structure. 37. Hsu HC, ChiouTJ, Chen JY, Lee CS, Lee PH, Peng 53. Stark GR, Taylor WR. Analyzing the G2-M check- Ethn Dis 2005;15:S5^ 88 ^ 91. SY. Clonality and clonal evolution of hepatocellular point. Methods MolBiol2004;280:51 ^ 82.

Clin Cancer Res 2008;14(15) August 1, 2008 4820 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research. ASPM Is a Novel Marker for Vascular Invasion, Early Recurrence, and Poor Prognosis of Hepatocellular Carcinoma

Shih-Yeh Lin, Hung-Wei Pan, Shu-Hsiang Liu, et al.

Clin Cancer Res 2008;14:4814-4820.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/14/15/4814

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2008/08/25/14.15.4814.DC1

Cited articles This article cites 49 articles, 9 of which you can access for free at: http://clincancerres.aacrjournals.org/content/14/15/4814.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/14/15/4814.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/14/15/4814. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from clincancerres.aacrjournals.org on September 26, 2021. © 2008 American Association for Cancer Research.