This article is mentioned in:

Introduction

Hepatocellular carcinoma (HCC) is one of the most prevalent cancers and the fifth leading cause of cancer-related deaths worldwide (1,2). Although partial hepatectomy is considered as one of the first line treatments for patients with early stage HCC, the outcome is far from satisfactory due to a high recurrence rate, which is up to 60% at postoperative five years (3–7). In clinical practice, it is difficult to predict the prognosis of early stage HCC patients due to similarities in clinicopathologic characteristics. Although great efforts have been made in prognostic biomarker investigation, the optimal candidate with clinical applicability is still lacking but urgently required (7–12).

Notch1 is a member of the Notch family and its association with human malignancy was first established in T-cell acute lymphoblastic leukemia (13–16). It has been reported that aberrant expression of Notch1 may contribute to carcinogenesis in additional types of malignancies (17–19). Recent studies also indicate that Notch1 may play an important role in hepatic carcinogenesis (20–22). Qi et al reported that upregulated Notch1 in the SMMC-7721 cell line induced cell cycle arrest and inhibited cell growth (20). They also found that Notch1 sensitized tumor necrosis factor-related apoptosis inducing ligand (TRAIL)-induced apoptosis in HCC cell lines (21). A previous study revealed that HBx overexpression in liver cancer cells decreased Notch1 signaling activity, promoting cell proliferation and inducing cell cycle progression, while blunting senescence-like growth arrest in vitro and in vivo (22). These studies have suggested an important role of Notch1 as a tumor suppressor in the carcinogenesis of liver cancer. However, the relationship between Notch1 expression and disease outcome has not yet been evaluated.

We examined the differential expression of Notch1 in paired tumor tissue (TT) and non-tumorous tissue (NTT). The association of Notch1 expression with outcome of early stage (BCLC stage 0 and A) HCC patients, as well as their ≤5 cm subgroups (stage 0 and ≤5 cm stage A) was analyzed. To the best of our knowledge, this is the first study exploring the clinical significance of Notch1 in HCC.

Materials and methods

Patients

We determined the differential expression of Notch1 in paired TT and NTT in 40 HCC patients, who were randomly selected from the Eastern Hepatobiliary Surgery Hospital (EHBH) between January and February 2004. All resected fresh TT and NTT samples were subjected to RT-PCR and western blot analysis.

We then investigated the relationship between Notch1 expression in TT and surgical outcome. The inclusion criteria required patients to have BCLC stage 0 or A HCCs with Child-Pugh A liver function, to have received curative hepatectomy and no evidence of distant metastasis, image visible ascites or severe varices, and to have no history of preoperative anticancer therapy. The definition of curative hepatectomy was described in our previous studies (23). Patients were excluded from the present study if they had a history of another malignancy or died from severe postoperative complications. According to the criteria, 206 patients undergoing hepatectomy between October 1997 and September 2002 were randomly retrieved from a database of the EHBH which was recognized as the training cohort. Formalin-fixed, paraffin-embedded specimens, clinicopathologic and follow-up data were obtained for immunohistochemistry (IHC) staining and prognostic analyses.

For validating the results from the training cohort, we prospectively collected 185 consecutive patients from the EHBH between March and September 2004, as validation cohort 1. Another independent cohort of 129 consecutive patients from the First Affiliated hospital of Fujian Medical university between September 2001 and March 2009, was retrospectively recruited as validation cohort 2. Identical inclusion and exclusion criteria were used in the validation cohorts of patients as compared with the patients in the training cohort.

The clinical staging was determined by the Barcelona-Clinic-Liver-Cancer (BCLC) and Tumor-Node-Metastasis (TNM) classification systems (7,8,24). According to AASLD guidelines, the BCLC stage 0 and A was defined as early stage of HCC in the present study (7,8). The present study protocol was approved by the Institutional Review Board of the Eastern Hepatobiliary Surgery Hospital and the Institutional Review Board of the First Affiliated hospital of Fujian Medical University. Written informed consent for each patient was obtained before surgery.

RNA analysis

Total RNA from 40 paired TT and NTT was isolated using TRIzol (Life Technologies, Inc., Rockville, MD, USA) according to the manufacturer's instructions. Reverse transcription was performed on 1 µg of total RNA from each sample using oligo(dT)18 primers and 200 units of SuperScript II (Life Technologies) for extension.

The primers used in semi-quantitative PCR and real-time PCR were as follows: Notch1 sense, 5′-CCGCAGTTGTGCTCCTGAA-3′ and Notch1 antisense, 5′-ACCTTGGCGGTCTCGTAGCT-3′; GAPDH sense, 5′-GTTGGAGGTCGGAGTCAACGGA-3′ and antisense, 5′-GAGGGATCTCGCTCCTGGAGGA-3′. GAPDH was used as an internal quantitative control. Semi-quantitative PCR amplification was performed with 1.25 units Ex Taq polymerase (Takara, Dalian, China). All of the PCR products were resolved on a 2% agarose gel containing ethidium bromide. Real-time PCR was performed using the SYBR-Green detection of PCR products in real time with the LightCycler (Roche Diagnostics, Meylan, France). The analyses were carried out according to the manufacturer's instructions. The relative expression level of Notch1 was obtained as a ratio normalized to GAPDH expression level. A no-template negative control was included in each experiment. All experiments were repeated three times, and the results are presented as the mean value.

Western blot analysis

The tissues were lysed in T-PER Tissue Protein Extraction Reagent (Pierce, Rockford, IL, USA) containing proteinase inhibitors (Calbiochem, San Diego, CA, USA). The extracts were collected and centrifuged at 12,000 × g for 5 min. The protein concentrations were determined using the BCA protein assay (Pierce), according to the manufacturer's instructions. Total proteins (20 µg) from whole lysates were boiled for 5 min in 1X SDS buffer, resolved by 8% SDS-PAge and transferred to nitrocellulose membranes. Membranes were blocked with 0.1 M Tris (pH 7.5), 0.9% NaCl and 0.05% Tween-20 (TBST) containing 10% non-fat milk powder and then incubated with the appropriate primary antibody (1:200; Notch1 and β-actin; Santa Cruz Biotechnology, Santa Cruz, CA, USA), followed by incubation with anti-goat (rabbit) horseradish peroxidase-conjugated antibody (1:5,000; Santa Cruz Biotechnology). Finally, the proteins were detected using the Western Blotting Luminol reagent (Santa Cruz Biotechnology).

Tissue microarray and immunohistochemical staining

Tissue microarrays (TMAs) were constructed as previously described (23). The first antibody was purchased from Santa Cruz Biotechnology (1:50). Immunohistochemical staining was performed using the Dako Envision Plus System (Dako, Carpinteria, CA, USA) according to the manufacturer's instructions. Appropriate negative and positive controls were used. The tissue was evaluated as positive for Notch1 staining when there were >10% of tumor cells demonstrating cytoplasmic and/or nuclear immunoreaction deposits. The sections were scored with a four-tier scale: 0, negative (0–10%), 1, weak signal (>10–20%), 2, intermediate signal (>20–50%) and 3, strong signal (>50%). 0 and 1 were defined as loss of Notch1, while 2 and 3 were defined as normal Notch1. All sections were independently scored by two observers who did not have any prior knowledge of the clinicopathological data. The concordance between scores from different sections of the same tumor was >90%. All discrepancies in scoring were reviewed and a consensus was reached.

Follow-up

Patients were examined every 2–3 months during the first two years and every 3–6 months from the third year after surgery. The standard examination of each visit and diagnosis of recurrence were described in our previous study (23). Patients with HCC recurrence received further treatments according to the tumor location, number of recurrent lesions, evidence of portal hypertension and hepatic compensatory function.

Time to recurrence (TTR) and overall survival (OS) were considered as primary endpoints of the present study. TTR was calculated from the date of resection to the date when tumor recurrence was diagnosed. OS was defined as the interval between surgery and death or last observation for surviving patients (25).

Statistical analysis

Analysis was performed with SPSS, version 10.0 for Windows (SPSS, Inc.); the χ2 or Fisher's exact tests were used to compare qualitative variables, while continuous variables were compared using the Student's t-test or Mann-Whitney test for variables with an abnormal distribution. Receiver operating characteristic curve analysis was used to determine the optimal cut-offs of continuous variables. Expression of Notch1 in TT and paired NTT was compared by the Wilcoxon test. Survival curves were calculated by the Kaplan-Meier method and compared using the log-rank test. The Cox proportional hazards model was used to determine the independent factors based on the variables selected by the univariate analysis. P<0.05 was considered to indicate a statistically significant result.

Results

Characteristics of the patients

Table I summarizes the clinicopathological characteristics of all the patients. The 1-, 3- and 5-year recurrence and survival rates were: 27, 59 and 74%, and 82, 49 and 36% in the training cohort; 33, 63 and 64%, and 78, 52 and 49% in the validation cohort 1; 31, 54 and 57%, and 84, 58 and 45% in the validation cohort 2.

Table I Characteristics of the patients in the three cohorts.

Table I

Characteristics of the patients in the three cohorts.

Variables	Training cohort (n=206)	Validation cohort 1 (n=185)	P-valueb	Validation cohort 2 (n=129)	P-valuec
Age, in years, median (range)	50 (15–74)	47 (21–77)	0.149a	53 (14–78)	0.029a
Female, n (%)	23 (11.2)	29 (15.7)	0.190	23 (17.8)	0.085
HBsAg, n (%)	168 (81.6)	163 (88.1)	0.073	108 (83.7)	0.612
HBeAg, n (%)	18 (8.7)	36 (19.5)	0.002	22 (17.1)	0.022
Liver cirrhosis, n (%)	194 (94.2)	121 (65.4)	<0.001	99 (76.7)	<0.001
Multiple tumors, n (%)	52 (25.2)	33 (17.8)	0.076	49 (38.0)	0.013
Complete encapsulation, n (%)	60 (29.1)	43 (23.2)	0.187	30 (23.3)	0.238
Microvascular invasion, n (%)	141 (68.4)	99 (53.5)	0.002	56 (43.4)	<0.001
AFP ≥20 µg/l, n (%)	142 (68.9)	128 (69.2)	0.956	83 (64.3)	0.384
Diameter ≤5 cm, n (%)	102 (49.5)	75 (40.5)	0.075	80 (62.0)	0.025
Differentiation, n (%)
I–II	35 (17.0)	42 (22.7)	0.156	39 (30.2)	0.004
III–IV	171 (83.0)	143 (77.3)		90 (69.8)
Transfusion, n (%)	70 (34.0)	38 (20.5)	0.003	19 (14.7)	<0.001
Laboratory values, median (range)
Total bilirubin, µmol/l	12.2 (3.0–48.9)	13.6 (3.7–64.6)	0.011a	15.5 (3.4–147.7)	<0.001a
Albumin, g/l	42.1 (31.5–54.3)	40.2 (28.7–50.6)	<0.001a	41.0 (31.4–51.4)	<0.001a
Platelets, 109/l	120.5 (20–342)	146.0 (24–468)	<0.001a	Not collected
Prothrombin time, sec	12.9 (9.7–21.1)	13.5 (9.8–18.5)	<0.001a	Not collected
GGT, U/l	56.7 (10.0–391.0)	75.0 (11.0–576.0)	0.001a	Not collected
ALP, U/l	133.0 (35–553)	94.0 (38–377)	<0.001a	Not collected
ALT, U/l	47.5 (17.0–699.0)	49.7 (8.2–104.0)	0.356a	45.0 (7.0–182.0)	0.292a

{ label (or @symbol) needed for fn[@id='tfn1-or-34-06-3174'] } HBsAg, hepatitis B surface antigen; HBeAg, hepatitis B e antigen; AFP, α-fetoprotein; GGT, γ-glutamyl transferase; ALP, alkaline phosphatase; ALT, alanine aminotransferase.

a Mann-Whitney test;

b P-value, training cohort vs. validation cohort 1;

c P-value, training cohort vs. validation cohort 2.

Notch1 expression in TT and matched NTT

Compared with NTT, TT showed loss of Notch1 mRNA and protein by real-time RT-PCR, semi-quantitative RT-PCR and western blotting (Fig. 1A–C). Similarly, TT displayed a relatively weaker immunostaining of Notch1 when compared with NTT (Fig. 1D–G). As shown in Table II, the expanded IHC investigation suggested that Notch1 expression in TT was significantly lower than that in NTT in the training cohort (P<0.001), which was verified by two validation cohorts (P<0.001 for both).

Figure 1 Expression status of Notch1. Loss of Notch1 in TT relative to matched NTT demonstrated by real-time quantitative RT-PCR analysis (A), RT-PCR (B), western blotting (C) and immunohistochemistry (D–G). (D and E) and (F and G) are paired TT and NTT of two patients randomly selected. Positive cells are stained brown. TT, tumor tissue; NTT, non-tumorous tissue.

Table II Expression of Notch1 in TT and NTT.

Table II

Expression of Notch1 in TT and NTT.

	Density in TT vs. NTT			P-valuea
	Normal (n)	Loss (n)	Equal (n)
Training cohort (n=206)	24	168	14	<0.001
	9	84	9
Validation cohort 1 (n=185)	22	152	11	<0.001
	9	62	4	<0.001
Validation cohort 2 (n=129)	18	99	12	<0.001
	7	64	9	<0.001

{ label (or @symbol) needed for fn[@id='tfn5-or-34-06-3174'] } TT, tumor tissue; NTT, non-tumorous tissue.

a Paired Wilcoxon test.

Correlation between Notch1 status and surgical prognosis in the training cohort

Loss of Notch1 was statistically associated with tumor diameter (P=0.036) and microvascular invasion (MVI; P=0.007) (Table III). Compared with those with loss of Notch1 in TT, patients with normal Notch1 had a prolonged median TTR (38.5±6.1 vs. 16.0±3.2 months, P<0.001) and an improved 1-, 3- and 5-year survival rates (91, 64 and 49% vs. 73, 31 and 22%, P<0.001) (Fig. 2A and B).

Figure 2 Kaplan-Meier analysis of the correlation between Notch1 status and prognosis of early stage HCC patients. Time to recurrence and survival curves of patients in the training cohort (A and B), validation cohort 1 (C and D) and validation cohort 2 (E and F).

Table III Notch1 expression and clinicopathological features.

Table III

Notch1 expression and clinicopathological features.

Variables	Training cohort (n=206)
	Loss	Normal	P-value
Gender
Female	7	16	0.081
Male	91	92
Age (years)
Median	50	49	0.458a
Range	15–72	32–74
AFP, µg/l
<20	27	37	0.299
≥20	71	71
HBsAg
Positive	79	89	0.740
Negative	19	19
HBeAg
Positive	6	12	0.205
Negative	92	96
Liver cirrhosis
Yes	91	103	0.442
No	7	5
Differentiation
I-II	20	15	0.213
III-IV	78	93
Diameter, cm
≤5	41	61	0.036
>5	57	47
Encapsulation
Complete	24	36	0.163
No	74	72
MVI
Yes	76	65	0.007
No	22	43
Tumor no.
Single	68	86	0.091
Multiple	30	22
TNM stage
I	16	38
II	62	56	0.007
III	20	14
ALT, U/l
Median	52.2	42.1	0.122a
Range	18.0–250.0	17.0–699.0
TBL, µmol/l
Median	12.9	11.4	0.115a
Range	4.0–35.4	3.0–48.9
ALB, g/l
Median	42.5	42.0	0.766a
Range	31.5–54.3	32.4–52.2
PLT, 109/l
Median	122	117	0.439a
Range	20–342	23–302
PT, sec
Median	12.8	13.0	0.439a
Range	10.0–17.0	9.7–21.1
GGT, U/l
Median	65.7	50.5	0.033a
Range	10.0–333.0	13.0–391.0
ALP, U/l
Median	146	126	0.031a
Range	38–553	35–313

{ label (or @symbol) needed for fn[@id='tfn6-or-34-06-3174'] } AFP, α-fetoprotein; HBsAg, hepatitis B surface antigen; HBeAg, hepatitis B e antigen; MVI, microvascular invasion; ALT, alanine aminotransferase; TBL, total bilirubin; ALB, human serum albumin; PLT, platelets; PT, prothrombin time; GGT, γ-glutamyl transferase; ALP, alkaline phosphatase; NA, not adopted.

a Mann-Whitney test;

b Fisher's exact test.

Tumor size ≤5 cm in diameter is frequently used in clinical staging systems and therapeutic criteria (24,26–29). We further investigated the patients with tumors ≤5 cm in diameter (n=102). We also found that patients with normal Notch1 in TT had longer median TTR and higher survival rates (TTR: 53.0±6.1 vs. 21.7±3.5 months, P=0.004 (Table IV); 1-, 3- and 5-year survival rates: 93, 71 and 57% vs. 76, 39 and 24%, P<0.001) (Fig. 3A and B).

Figure 3 Kaplan-Meier analysis of the correlation between Notch1 expression and prognosis of early stage HCC patients (≤5 cm). Time to recurrence and survival curves of patients in the training cohort (A and B), validation cohort 1 (C and D) and validation cohort 2 (E and F).

Table IV Univariate analysis for prognosis in the training cohort.

Table IV

Univariate analysis for prognosis in the training cohort.

Variables	n	Recurrence (months)		Survival (months)
		Median time to event	P-value	Median time to event	P-value
Training cohort
Diameter, cm
≤5	102	41.5±8.4	<0.001	53.7±6.0	0.001
>5	104	21.7±1.9		27.9±3.7
Encapsulation
Complete	60	46.7±9.2	0.011	62.1±7.4	0.001
No	146	22.6±2.3		30.1±3.8
Transfusion
Yes	70	21.3±3.1	0.074	28.8±5.5	0.029
No	136	31.3±3.9		44.3±7.5
Tumor no.
Single	154	32.1±3.6	0.001	51.3±7.2	0.001
Multiple	52	14.0±2.7		24.0±4.0
AFP, µg/l
<20	64	41.7±10.0	0.026	58.2±5.9	0.051
≥20	142	21.9±3.0		30.1±3.6
MVI
Yes	141	23.4±3.3	0.031	32.9±3.2	0.095
No	65	38.4±9.1		55.3±6.1
Notch1 in TT
Loss	98	16.0±3.2	<0.001	22.6±3.1	<0.001
Normal	108	38.5±6.1		56.6±6.6
≤5 cm subgroup
HBsAg
Positive	83	32.2±4.8	0.008	46.5±8.7	0.128
Negative	19	86.7±10.1		105.1±35.1
Tumor no.
Single	79	46.7±7.0	0.125	63.0±8.3	0.002
Multiple	23	8.0±3.8		24.0±9.4
Encapsulation
Complete	36	53.0±9.1	0.243	105.1±38.3	0.032
No	66	36.2±4.0		44.3±11.2
Notch1 in TT
Loss	41	21.7±3.5	0.004	25.8±6.9	<0.001
Normal	61	53.0±6.1		76.3±21.5

[i] AFP, α-fetoprotein; MVI, microvascular invasion; HBsAg, hepatitis B surface antigen; TT, tumor tissue.

The factors, listed in Table I, showing significance by univariate analysis for prognosis were adopted to multivariate analysis (Table IV). Notch1 status was an independent factor for both recurrence and survival in the cohort, with highest hazard ratio (HR) values among all independent factors (HR, 1.901; 95% CI, 1.366–2.646; P<0.001 for recurrence; HR, 2.038; 95% CI, 1.468–2.829; P<0.001 for survival). It was also an independent factor for both recurrence and survival in ≤5 cm subgroup, and had a highest HR value for survival (HR, 2.337; 95% CI, 1.431–3.818; P=0.001) (Table V).

Table V Multivariate analysis and hazard ratios for recurrence and survival.

Table V

Multivariate analysis and hazard ratios for recurrence and survival.

	Variables	Time to recurrence		Overall survival
		HR (95% CI)	P-value	HR (95% CI)	P-value
Cohorts
Training cohort	Diameter: >5 vs. ≤5 cm	1.674 (1.204–2.328)	0.002	1.487 (1.071–2.065)	0.018
	Tumor no.: multiple vs. single	1.709 (1.195–2.443)	0.003	1.629 (1.140–2.327)	0.007
	Encapsulation: no vs. complete	1.644 (1.144–2.363)	0.007	1.886 (1.292–2.751)	0.001
	Notch1 in TT: loss vs. normal	1.901 (1.366–2.646)	<0.001	2.038 (1.468–2.829)	<0.001
Validation cohort 1	Diameter: >5 vs. ≤5 cm	2.418 (1.580–3.701)	<0.001	2.932 (1.742–4.934)	<0.001
	Transfusion: yes vs. no	2.147 (1.396–3.301)	<0.001	1.965 (1.245–3.103)	0.004
	Encapsulation: no vs. complete	2.200 (1.314–3.684)	0.003	3.020 (1.551–5.878)	0.001
	Notch1 in TT: loss vs. normal	2.056 (1.409–3.000)	<0.001	2.381 (1.551–3.656)	<0.001
Validation cohort 2	Diameter: >5 vs. ≤5 cm	4.269 (2.495–7.303)	<0.001	4.111 (2.379–7.106)	<0.001
	Tumor no.: multiple vs. single	2.003 (1.206–3.325)	0.007	2.161 (1.262–3.703)	0.005
	Notch1 in TT: loss vs. normal	4.341 (2.517–7.487)	<0.001	4.721 (2.680–8.315)	<0.001
≤5 cm subgroups
Training cohort	HBsAg: positive vs. negative	2.815 (1.332–5.946)	0.007		nA
	Tumor no.: multiple vs. single		NA	2.134 (1.243–3.664)	0.006
	Notch1 in TT: loss vs. normal	2.154 (1.317–3.524)	0.002	2.337 (1.431–3.818)	0.001
Validation cohort 1	MVI: yes vs. no	2.163 (1.080–4.331)	0.030	2.721 (1.093–6.776)	0.032
	Notch1 in TT: loss vs. normal	2.514 (1.251–5.049)	0.010	4.621 (1.814–11.773)	0.001
Validation cohort 2	Tumor no.: multiple vs. single	3.269 (1.556–6.866)	0.002		nA
	Notch1 in TT: loss vs. normal	2.599 (1.236–5.466)	0.012	3.496 (1.469–8.322)	0.005

[i] HBsAg, hepatitis B surface antigen; MVI, microvascular invasion; NA, not adopted; TT, tumor tissue; CI, confidence interval.

Validation of the correlation between Notch1 status and surgical prognosis

In the validation cohort 1, normal Notch1 in TT was closely associated with longer median TTR (39.0±2.3 vs. 14.2±2.2 months, P<0.001 (Table VI) and higher survival rates (1- and 3-year, 88 and 68% vs. 66 and 34%, P<0.001) (Fig. 2C and D). In the ≤5 cm subgroup (n=75), normal Notch1 again predicted a prolonged TTR (49.3±2.5 vs. 26.7±7.2 months, P= 0.006) and improved survival rates (1- and 3-year, 100 and 86% vs. 96 and 50%, P<0.001) (Fig. 3C and D).

Table VI Univariate analysis for prognosis in the validation cohort 1.

Table VI

Univariate analysis for prognosis in the validation cohort 1.

Variables	No.	Recurrence (months)		Survival (months)
		Median time to event	P-value	Median time to event	P-value
Validation cohort 1
AFP, µg/l
<20	57	38.8±3.1	0.009	45.0±2.8	0.015
≥20	128	21.0±4.2		33.3±8.5
Encapsulation
Complete	43	45.5±3.0	<0.001	52.7±2.5	<0.001
No	142	18.3±3.0		24.2±5.6
MVI
Yes	99	16.2±4.1	0.001	24.2±7.3	0.001
No	86	37.7±2.6		44.4±2.4
Tumor no.
Single	152	27.1±2.6	0.007	41.2±1.9	0.058
Multiple	33	11.4±3.5		14.8±2.9
Diameter, cm
≤5	75	44.8±2.3	<0.001	51.9±2.0	<0.001
>5	110	12.2±1.8		15.9±1.4
Notch1 in TT
Loss	86	14.2±2.2	<0.001	17.8±2.0	<0.001
Normal	99	39.0±2.3		46.5±2.2
Transfusion
Yes	38	10.4±2.4	<0.001	14.9±2.4	<0.001
No	147	28.6±3.0		41.7±1.9
≤5 cm subgroup
MVI
Yes	28	28.3±7.2	0.020	44.1±3.5	0.023
No	47	49.1±2.6		55.4±2.0
Notch1 in TT
Loss	24	26.7±7.2	0.006	39.4±3.6	<0.001
Normal	51	49.3±2.5		56.5±1.8

[i] AFP, α-fetoprotein; MVI, microvascular invasion; TT, tumor tissue.

Normal Notch1 in TT also predicted a better prognosis in the validation cohort 2 (TTR, 59.6±4.7 vs. 11.9±2.4 months, P<0.001; 1-, 3- and 5-year survival rates, 97, 75 and 61% vs. 68, 35 and 22%, P<0.001, Fig. 2E and F) and the ≤5 cm subgroup (n=80, TTR, 65.9±5.3 vs. 30.6±11.4 months, P=0.021; 1-, 3- and 5-year survival rates, 98, 85 and 71% vs. 86, 54 and 45%, P=0.003) (Fig. 3E and F).

The univariate analysis (Tables VI and VII) and multivariate analysis (Table V) indicated again that the Notch1 status was an independent factor for recurrence and survival in the validation cohorts and their ≤5 cm subgroups. The HR value of Notch1 status ranked in the forefront among all independent factors for both recurrence and survival in the validation cohort 1 (HR, 2.056; 95% CI, 1.409–3.000; P<0.001 for recurrence; HR, 2.381; 95% CI, 1.551–3.656; P<0.001 for survival), and was the highest for both prognostic indexes in the validation cohort 2 (HR, 4.341; 95% CI, 2.517–7.487; P<0.001 for recurrence; HR, 4.721; 95% CI, 2.680–8.315; P<0.001 for survival). The similar results were obtained in the ≤5 cm subgroups (Table V). The prognostic value of Notch1 in the training cohort was fully confirmed by the validation studies.

Table VII Univariate analysis for prognosis in the validation cohort 2.

Table VII

Univariate analysis for prognosis in the validation cohort 2.

Variables	No.	Recurrence (months)		Survival (months)
		Median time to event	P-value	Median time to event	P-value
Validation cohort 2
MVI
Yes	56	18.6±5.5	0.044	29.1±7.6	0.006
No	73	53.6±5.2		63.4±5.0
Tumor no.
Single	80	58.8±4.8	<0.001	63.7±4.5	0.001
Multiple	49	15.3±2.8		25.9±5.7
Diameter, cm
≤5	80	58.7±4.5	<0.001	66.0±4.2	<0.001
>5	49	9.9±3.6		25.9±6.4
Notch1 in TT
Loss	57	11.9±2.4	<0.001	21.1±4.2	<0.001
Normal	72	59.6±4.7		68.8±4.2
Transfusion
Yes	19	9.9±4.4	<0.001	15.6±6.5	<0.001
No	110	52.9±4.1		60.8±3.9
≤5 cm subgroup
Tumor no.
Single	53	68.4±5.1	0.002	69.6±5.0	0.224
Multiple	27	19.9±5.2		54.5±6.8
Notch1 in TT
Loss	34	30.6±11.4	0.021	39.0±12.6	0.003
Normal	46	65.9±5.3		74.4±4.6

Correlation between Notch1 status and early tumor recurrence

It was reported that the recurrence of HCC within and after postoperative two years has different molecular background (12). In the training cohort, patients with loss of Notch1 had an increased incidence of early recurrence, compared with those with high Notch1 expression (61.2 vs. 26.9%, P<0.001). Additionally, Notch1 status was also an independent factor for early recurrence with the highest HR value (HR, 3.228; 95% CI, 2.059–5.062; P<0.001). These results could be verified by the analysis of the validation cohorts (Table VIII).

Table VIII Multivariate analysis and hazard ratios for early tumor recurrence.

Table VIII

Multivariate analysis and hazard ratios for early tumor recurrence.

Cohorts	Variables	HR (95% CI)	P-value
Cohorts
Training cohort	Diameter: >5 vs. ≤5 cm	1.553 (1.007–2.394)	0.046
	Tumor no.: multiple vs. single	2.521 (1.637–3.882)	<0.001
	Encapsulation: no vs. complete	1.940 (1.163–3.238)	0.011
	Notch1 in TT: loss vs. normal	3.228 (2.059–5.062)	<0.001
Validation cohort 1	Diameter: >5 vs. ≤5 cm	2.828 (1.658–4.821)	<0.001
	Transfusion: yes vs. no	2.292 (1.424–3.689)	0.001
	Encapsulation: no vs. complete	3.736 (1.781–7.837)	<0.001
	Notch1 in TT: loss vs. normal	2.446 (1.570–3.810)	<0.001
Validation cohort 2	Diameter: >5 vs. ≤5 cm	4.707 (2.637–8.402)	<0.001
	Tumor no.: multiple vs. single	1.791 (1.041–3.083)	0.035
	Notch1 in TT: loss vs. normal	5.071 (2.823–9.110)	<0.001
≤5 cm subgroups
Training cohort	HBsAg: positive vs. negative	4.952 (1.184–20.713)	0.007
	Notch1 in TT: loss vs. normal	3.639 (1.812–7.308)	0.002
Validation cohort 1	MVI: yes vs. no	3.215 (1.244–8.309)	0.016
	Notch1 in TT: loss vs. normal	4.032 (1.557–10.436)	0.004
Validation cohort 2	Tumor no.: multiple vs. single	3.544 (1.562–8.042)	0.002
	Notch1 in TT: loss vs. normal	3.122 (1.371–7.110)	0.007

[i] HBsAg, hepatitis B surface antigen; MVI, microvascular invasion; TT, tumor tissue; CI, confidence interval.

The similar results were obtained in the ≤5 cm subgroups (2-year recurrence rate, 51.2 vs. 21.3%, P=0.002; 45.8 vs. 13.7%, P=0.002; 41.2 vs. 21.7%, P=0.014 in the subgroup of the training cohort, validation cohort 1 and 2, respectively). Notch1 status was also an independent factor for early recurrence (Table VIII).

Discussion

The present study, for the first time, investigated the relationship between Notch1 status and outcome of early stage HCC patients undergoing hepatectomy. We found that loss of Notch1 was often observed in HCC, and early HCC with loss Notch1 was more likely to exhibit a malignant phenotype and presented a worse surgical prognosis.

Previous studies indicated that Notch1 is involved in carcinogenesis in some types of malignancies (17,30). It was reported that the function of Notch1 in malignancy depended on the different target gene(s) or downstream pathway(s) it turned on/off (31,32). Weng et al found that c-myc as a developmentally regulated direct downstream target of Notch1 contributed to the growth of T acute lymphoblastic leukemia/lymphoma (31), in contrast, in another study which described Notch1 binding to P21, a decrease in keratinocyte proliferation and a delay in terminal differentiation was observed (32). The function of Notch1 was also affected by the interplay between Notch1 and other signaling pathways such as Wnt and Ras (33). The authors reported that in Notch1-deficient mice bearing basal cell carcinomas, Wnt signaling seemed to be abnormal as they showed an increase in both levels of β-catenin and activity of LEF-1. Notch1-deficient mice were also more susceptible to skin tumor development in the context of Ras activation or carcinogen exposure (33). These studies suggested that activation of the Notch1 pathway may display an inhibitory function in carcinogenesis of the malignancy.

Notch1 may also play an important role in hepatic carcinogenesis, and activation of Notch1 signaling suppressed HCC cell proliferation (20–22,34). The function of Notch1 in HCC seems to be closely associated with HBV infection. A recent study revealed that HBx overexpression in the Huh7 cell line decreased the endogenous protein level of the intracellular domain of Notch1, and mRNA levels of its downstream target genes through suppressing presenilin1 transcription. This process enhanced cell proliferation, induced G1-S cell cycle progression and blunted cellular senescence in vitro and in vivo (22). Indeed, in the present study, the majority of patients (513/520, 98.7%) undergoing the clinical observation had a background of HBV infection. Our results revealed that Notch1 expression was significantly decreased in HCC and that lower expression of this molecule was associated with certain invasive pathological features and was an independent and powerful risk factor for poor prognosis (Table V). These observations suggest that normal Notch1 in TT is a protective factor, a notion that is supported by previous findings that Notch1 acts as a tumor suppressor gene in HCC (20–22).

Hepatectomy is recommended for HCC patients at early stage such as BCLC stage 0 and A (7,8,35). However, some early stage HCC patients also have a poor prognosis. It has become increasingly more important that biological prognostic predictors for early HCC should be found, as the number of patients with early HCC who are receiving surgical treatment is rising (7,8,25). There were several molecular markers which were also found to be associated with the prognosis of ≤5 cm HCC (36,37), however, an ideal one that has strong correlations with clinical outcomes and that is easy to measure is still lacking. Our results indicated that Notch1 status in TT is closely associated with tumor recurrence and survival in patients at BCLC stage 0/A. Patients with normal Notch1 have a longer TTR and survival compared with those with low Notch1 expression, a observation made in a retrospective study and validated by a prospective cohort from the same medical center, and also verified in an independent cohort from another center. Furthermore, the HR value of Notch1 status for recurrence and survival from multivariate analyses was either higher or highest among all independent factors.

Considering that HCC patients at BCLC stage 0 and ≤5 cm stage A are usually thought to be optimal candidates for surgical resection (24,26,29), we therefore evaluated the prognostic values of this molecule in these patients. The Notch1 status was also statistically associated with TTR and survival in three subgroups, and was unexclusively an independent factor for both prognostic indexes, with higher or highest HR values for prognosis (Table V). In addition, the prognostic role of the Notch1 status in the ≤5 cm subgroup seemed to be more important than that in the whole cohort, as fewer independent prognostic factors were found in all three ≤5 cm subgroups. Therefore, the earlier the HCC stage the more difficult the prognostic prediction; thus, the predictive value of Notch1 status in early HCC should be fully assessed.

The present study also suggests that tumors with loss of Notch1 tend to undergo early relapse after surgery, as the 2-year recurrence rate was higher in the patients with loss of Notch1 relative to those with normal Notch1 in all three cohorts and their ≤5 cm subgroups (Figs. 2 and 3). It has been reported that HCC recurrence within two years after surgery is closely associated with molecular features of the primary tumor, but that the recurrence after postoperative two years appears to result from new primary tumors arising in a damaged liver (12). Our results suggest that loss of Notch1, which was associated with invasive features such as larger tumor diameter and higher incidence of vascular invasion, may contribute to early tumor recurrence. While there is a lack of mechanistic understanding regarding these issues in HCC, studies on breast cancer indicate that aberrant expression of this molecule can promote tumor growth and metastasis, which was characterized by inhibition of anoikis and induction of epithelial-to-mesenchymal transition (38). Our results thus suggest that Notch1 status may have meaningful prognostic discrimination for early postoperative recurrence of HCC after hepatectomy.

In conclusion, our results suggest that loss of Notch1 in HCC may be an informative warning sign that these early HCC patients should receive more intensive monitoring and appropriate adjuvant therapies. Further studies are needed to establish whether Notch1 has full potential as a new therapeutic target in liver cancer.

Acknowledgments

We thank Dr Jian Lu for statistical assistance, Dr Li Gao for the pathological examinations, Dr Anita S.F. Chong and Dr Yi-Zheng Wang for critical reading of the manuscript. The present study was supported by a grant from the State Key Project on Infectious Diseases of China (no. 2008ZX10002-025,019,2012ZX10002-016 to F.S.), grants from national natural Science Foundation of China (no. 30772141 to F.S and no. 81071867 to W.K.), a grant from the Science Fund for Creative Research Groups, and a grant of the Shanghai hospital Development Program (SHDC12010121 to F.S.).

Abbreviations:

References