Proliferative ductular reactions correlate with hepatic progenitor cell and predict recurrence in HCC patients after curative resection

Clinico-pathological features

Peritumoral proliferative ductular reaction (PI-DR) is present in HCC and correlates with the degree of inflammation and fibrosis

As reported, the ductular reaction (DR) was positive uniformly and strongly for cytokeratin 7 staining [12]. In peritumoral K7 staining sections of the entire tissue microarray group, DR was frequently found in the fibrous septa and the enlarged periportal areas in HCC patients (Figure 1A). Further, In terms of PI-DR, the proliferating cell nuclear antigen (PCNA) was labeled for the proliferation status of DR (Figure 1B). 37 of 120 patients (30.8%) had PI-DR ≥50%.

Previous studies have shown that DRs correlate with the degree of inflammation and fibrosis in the course of many chronic human liver diseases [9, 10]. Thus, we evaluated the liver necroinflammation and fibrosis and their association with PI-DR. Consistent with prior studies, we observed PI-DR correlated with hepatic necroinflammation grade (r = 0.614, P < 0.001, Figure 1C). Correlations were also found between PI-DR and fibrosis stage (r = 0.527, P < 0.001, Figure 1D).

Expression of hepatic progenitor cell markers in peritumoral tissue of HCC patients and clinicopathologic correlations:relationship with peritumoral PI-DR

Several markers, including EpCAM, OV6, CD133 and c-kit, which are expressed as markers for hepatic progenitor cells (HPCs), have been considered to be the origin of HCCs [19]. However, the link between PI-DR and expression of HPC markers has not been fully elucidated. We firstly conducted an immunohistochemical analysis of four HPC markers (EpCAM, OV6, CD133 and c-kit) in the peritumoral tissue sections of the entire group, the high EpCAM, OV6, CD133 and c-kit expression was seen in 40 of 120 (33.3%), 34 of 120 (28.3%), 33 of 120 (27.5%), and 39 of 120 (32.5%) cases, respectively (Table 2). EpCAM expression was seen in the membrane and cytoplasm of cholangiocytes of all branches of the biliary tree, including canals of Hering, ductules, and small and large bile ducts. The ductular OV-6-positive cells with a more transitional morphology, represented small numbers of individual oval-like cells adjacent to ducts in marginal regions. CD133-positive cells were mostly accumulated as patches with granular appearance in the cytoplasm of a few scattered hepatocytes. C-kit positive cells were diffused in the parenchyma in HCCs (Figure 2A).

Clinicopathologic features	Peritumoral EpCAM density		P Value	Peritumral OV6 density		P Value	Peritumral CD133 density		P Value	Peritumral c-kit density		P Value
	High	Low		High	Low		High	Low		High	Low
Frequency (%)	40 (33.3%)	80 (66.7%)		34 (28.3%)	86 (71.7%)		33 (27.5%)	87 (72.5%)		39 (32.5%)	81 (67.5%)
Gender
Male	35	73	0.747	30	78	0.946	30	78	1.000	37	71	0.363
Female	5	7		4	8		3	9		2	10
Age
≤50	21	42	1.000	16	47	0.453	19	44	0.493	20	43	0.853
>50	19	38		18	39		14	43		19	38
AFP
≤400	25	56	0..408	22	59	0.681	23	58	0.752	24	57	0.333
>400	15	24		12	27		10	29		15	24
Cirrhosis
Absence	20	51	0.149	19	52	0.645	17	54	0.294	17	54	0.016*
Presence	20	29		15	34		16	33		22	27
Satellite lesion
Absence	36	69	0.770	28	77	0.284	29	76	1.000	35	70	0.606
Presence	4	11		6	9		4	11		4	11
Tumor size (cm)
≤5cm	15	37	0.362	15	37	0.913	16	36	0.483	18	34	0.665
>5cm	25	43		19	49		17	51		21	47
Tumor encapsulation
Absence	26	61	0.193	26	61	0.540	23	64	0.672	27	60	0.578
Presence	14	19		8	25		10	23		12	21
Microvascular invasion
Absence	17	38	0.604	17	38	0.565	18	37	0.238	22	33	0.107
Presence	23	42		17	48		15	50		17	48
Major vascular invasion
Absence	39	75	0.657	33	81	0.853	29	85	0.083	38	76	0.687
Presence	1	5		1	5		4	2		1	5
Edmondson grade
Low (I/II)	27	63	0.180	21	69	0.035*	23	67	0.409	28	62	0.574
High (III/IV)	13	17		13	17		10	20		11	19
TNM stage
I/II	25	67	0.009*	23	69	0.142	16	76	<0.001*	29	63	0.678
III	15	13		11	17		17	11		10	18

*p <0 .05 was considered statistically significant.

HCCs were grouped according to the expression status of peritumoral HPC markers, and clinicopathological parameters were further analyzed. As shown in Table 2, patients with a high peritumoral EpCAM and CD133 expression level were more prone to have high TNM stage, high Edmondson grade in patients with the high peritumoral OV6 expression, and high peritumoral c-kit expression patients demonstrated more cirrhosis.

Next, we further identify the correlation between PI-DR and the peritumoral HPC markers’ expression (IOD). PI-DR had a positive correlation with EpCAM (r = 0.441, P < 0.001), OV6 (r = 0.342, P < 0.001), CD133 (r = 0.408, P < 0.001) and c-kit expression (r = 0.291, P = 0.001) (Figure 2B) in peritumoral tissue of the entire group. These findings support the concept that proliferative DR in peritumoral environment may contribute to activation of progenitor cells and related carcinogenesis.

Peritumoral PI-DR as a valid surrogate for intratumoral expression of HPC markers

To assess if the intratumoral expression of HPC markers were correlated with PI-DR in our study, we further performed an IHC analysis of four HPC markers (EpCAM, OV6, CD133 and c-kit) in the intratumoral tissue sections of the entire group as shown in Figure 3A, the high intratumoral expression of EpCAM, OV6, CD133 and c-kit was seen in 39 of 120 (32.5%), 26 of 120 (21.7%), 29 of 120 (24.2%), and 34 of 120 (28.3%) cases, respectively (Table 3).

HCCs were grouped according to intratumoral HPC expression, and clinicopathological features were compared as shown in the Table 3. High intratumoral EpCAM was significantly associated with more major vascular invasion (P = 0.023), high AFP (P < 0.001) and high Edmondson grade (P = 0.005), High intratumoral OV6 expression of the entire group demonstrated more cirrhosis (P = 0.015). High intratumoral c-kit expression was significantly more likely to occur in high Edmondson grade (P = 0.010).

We further analyzed the correlation of PI-DR with the intratumoral HPC markers’ expression (IOD). PI-DR had a positive correlation with intratumoral EpCAM (r = 0.328, P < 0.001, OV6 (r = 0.466, P < 0.001), CD133 (r = 0.356, P < 0.001) and c-kit expression (r = 0.398, P < 0.001) (Figure 3B) in the entire group. Thus, together these results suggest that PI-DR could be used as a valid surrogate for intratumoral expression of HPC markers.

The association of peritumoral high PI-DR in HCC with poor disease-free survival

The survival of the selected HCC patients was analyzed with Kaplan-Meier survival analysis. Patients with high PI-DR (≥50%) were likely to be with significantly poor OS and DFS. The median (95% CI) overall survival time was 25.3 (21.7-28.8) and 48.8 (31.3-66.3) months respectively for patients with high PI-DR (≥50%) and low PI-DR (<50%) in HCC (P < 0.001, Figure 4A). The median (95% CI) disease-free survival time was 10.1 (7.0 - 13.2) and 21.2 (18.4 - 24.1) months respectively for patients with high PI-DR (≥50%) and low PI-DR (<50%) in HCC (P < 0.001, Figure 4B).

Univariate and multivariate analyses of prognostic variables in HCC patients

Factors showing significance by univariate analysis were adopted when multivariate Cox proportional hazards analysis was performed (Table 4). PI-DR (HR: 3.316, 95% CI: 1.961 - 5.610, P < 0.001), K7-DR (HR: 1.863, 95% CI: 1.008-3.443, P = 0.047) were independent prognostic factors for overall survival. Moreover, PI-DR (HR: 2.618, 95% CI: 1.701 - 4.029, P < 0.001) and TNM stage (HR: 1.817, 95% CI: 1.135 - 2.909, P = 0.013) were also independent predictors for disease-free survival.

Patients and clinical data

Archived, formalin-fixed and paraffin-embedded liver specimens used in this study were obtained from the Eastern Hepatobiliary Surgery Hospital, Shanghai, China. From January 1997 to December 2007, patients who underwent hepatectomy and who were post-operatively confirmed as hepatocellular carcinoma were recruited for prospective follow-up. Informed consent was obtained from each patient under a protocol approved by the Hospital Research Ethics Committees. The following tissue microarray included 138 pairs of intratumoral and peritumoral liver tissues from the selected HCC patients. Eighteen cases could not be analysed because of the loss of the samples during the preparation of the tissue microarray. The 120 remain study population consisted of 108 men and 12 women with a mean age of 50.1 ± 11.9 years (range 11–89 years). Tumor size was based on the largest dimension of the tumor specimen. Satellite lesions were defined by the presence of 2 or more nodules including intrahepatic metastases. Microvascular invasion was determined by microscopic examination of the resected specimen, whereas macrovascular invasion was identified by macroscopic examination of specimen. Tumor stage was determined according to the 2009 UICC TNM (International Union Against Cancer, Tumor Node Metastasis) classification system [30]. Tumor differentiation was assessed according to Edmonson and Steiner grading system. The severity of the inflammation at the interface between portal tracts and the parenchyma was assessed on the H&E stained section and graded according to Knodell et al. [31]. Fibrosis was assessed separately, as suggested by Desmet et al. [32]. The specimens were divided into four groups according to the HAI score for necroinflammation grade. The first group included biopsies with HAI 0–4 (absent activity), the second group biopsies with HAI 5–8 (mild activity), the third group biopsies with HAI 9–12 (moderate activity) and the last group biopsies with HAI 13–18 (severe activity). The specimens were also divided into three groups according to the degree of fibrosis, which was staged with scores ranging from 1 to 3, corresponding to: absence or mild fibrosis (stage 1), moderate or severe (bridging) fibrosis (stage 2) and cirrhosis (stage 3).

Follow-up and detection of recurrence

All patients were followed regularly every 2–3 months after surgery until study closure in July 2010 with serum AFP/CA19-9 and abdominal ultrasonography. Progressive elevation of serum AFP/CA19-9 levels and/or ultrasonographic detection of a new hepatic lesion prompted hospitalization for confirmation of diagnosis and appropriate management, including repeat resection, radiofrequency ablation (RFA), transcatheter arterial chemoembolization (TACE) or supportive therapy. Recurrence was confirmed by contrast-enhanced imaging studies according to standard guidelines for HCC [33]. Overall survival (OS) was defined as the interval between the dates of surgery and death while disease-free survival (DFS) was defined as the interval between the dates of surgery and recurrence. If recurrence was not diagnosed, patients were censored on the date of death or the last follow-up. Clinical follow-up was not disclosed to laboratory personnel until statistical analysis.

Tissue microarray (TMA) construction

Tissue microarray (TMA) construction was described in previous reports [34]. Briefly, we constructed the tissue microarray, containing of 120 pairs of intratumoral and peritumoral liver specimens from the enrolled HCC patients. TMA slides with 120 pairs of intratumoral and matched peritumoral samples were constructed. In brief, all archival formalin-fixed and paraffin-embedded specimens used in this study were freshly sectioned and stained with haematoxylin and eosin (H & E). The H & E-stained sections were carefully reviewed by two independent pathologists without knowledge of the patients’ clinical characteristics and outcomes, and the representative regions of the intratumoral and peritumoral liver tissues were defined for microarray, and the peritumoral tissues at the distance (15mm) from the tumor margin were collected. Using a tissue arraying instrument (Beecher Instruments, Sliver Spring, MD), each tissue core with a diameter of 1.5 mm was punched from the marked areas of intratumoral and peritumoral tissues and re-embedded. TMAs containing the tissue cores were then cut into five-micrometre sections for IHC staining.

Immunohistochemical staining

Histopathologic and immunohistochemical evaluation

All slides of positive HPC markers (EpCAM, OV6, C-kit and CD133) staining were observed and photographed with an Olympus microscope (IX-70 OLYMPUS, Japan). Three independent investigators examined the TMA slides without related clinical information. Under 200× magnification, images of positive representative fields were captured. The density of immunostaining was measured using Image-Pro Plus Version 6.2 software (Media Cybernetics Inc., Bethesda, MD), described previously as an established method [4]. The integrated optical density (IOD) in each image was measured, and the density of HPC markers was calculated as IOD/total area of each image. For calculating HPC markers (e.g. EpCAM, OV6, CD133 and c-kit) density, the cutoff for the definition of subgroups was the median IOD to separate high HPC markers’ expression (IOD > median) from low HPC markers’ expression (IOD ≤ median) in microarray tissue samples.

Evaluation of K7-DR and PI-DR

For quantification of the total area of ductular reaction (DR), non-overlapping fields of the biopsy specimen were photographed at 100× magnification after immunohistochemical staining for CK7. The extent of CK7-immunoreactive ductular reaction was semiquantitatively assessed and graded with some modification in a manner analogous to the grading of interface hepatitis in the scoring system of Ishak et al. [31]: In addition, proliferation rate in DR was evaluated by calculating the proliferating cell nuclear antigen(PCNA) labeling index as described in our previous reports [21]. Specifically, 3 400× highpowered fields within each section were randomly chosen and captured. The same fields were captured in sequential serial sections stained with K7 for quantification of the number of reactive ductular cells (RDCs) in reactive ductules [35]. The proliferation index of DR (PI-DR) was calculated as ratio between the number of PCNA immunoreactive nuclei and the total number of RDCs.

Statistical analysis

Statistical analysis was performed using with SPSS 20.0 for Windows (SPSS Inc., Chicago, IL); Differences between categorical variables were assessed by the chi-square test or Fisher’s exact test, when necessary. Pearson’s correlation coefficient was used to determine correlations between continuous normally distributed variables. The degree of association between nonparametric or ordinal variables was assessed by using Spearman nonparametric correlation. Kaplan-Meier analysis was used to determine the survival. Log-rank test was used to compare patients’ survival between subgroups; Significant variables from the univariable analysis were entered in the multivariable analysis, which was performed using the Cox-proportional hazards model with forward stepwise selection. P < 0.05 was considered statistically significant.