Preoperative transcatheter arterial chemotherapy may suppress oxidative stress in hepatocellular carcinoma cells and reduce the risk of short-term relapse

In this study, we aim to investigate oxidative stress in hepatocellular carcinoma (HCC) tissues in patients receiving preoperative transcatheter arterial chemotherapy (TAC) and its association with prognosis. A total of 89 HCC patients enrolled in this study, 39 received preoperative TAC 1 week before surgery (pTAC group) and 50 did not (non-pTAC group). All patients underwent hepatectomy and postoperative TAC and were followed up to 400 weeks. Samples of liver tissue without HCC and hepatitis (n = 15) served as normal controls. Cellular levels of 8-hydroxy-2′-deoxyguanosine (8-OHdG), TP53, and p21waf1/cip1 were measured in both cancer and surrounding tissues using an immunohistochemistry assay. Taken together, our data suggested that preoperative TAC might postpone postoperative HCC relapse within 1 year via suppression of tumor cells by induction of high levels of oxidative stress.


INTRODUCTION
Hepatocellular carcinoma (HCC) is one of the most common malignancies in China [1]. Transcatheter arterial chemoembolization (TACE) is a palliative treatment for patients with unresectable intrahepatic tumors to obtain degradation and possible consequential resection [2]. We found that postoperative adjuvant TACE seems to decrease the short-term HCC recurrence rate among patients with risk factors such as multiple nodules > 5 cm or vascular invasion [3]. When applying transcatheter arterial chemotherapy (TAC) without embolization agents rather than TACE for the treatment of patients with HCC and main portal vein thrombosis(MPVT), TAC as an adjunctive therapy significantly improved survival in HCC patients [4]. The main difference between TACE and TAC is that embolization agents are not used in TAC. Embolization agents play an important role in preoperative TACE to reduce tumor size and induce tumor necrosis by blocking tumor feeding artery. But routine preoperative

Research Paper
Oncotarget 54403 www.impactjournals.com/oncotarget TACE for resectable HCC is not recommended because whether it improves surgical outcome remains controversial, furthermore, preoperative TACE may cause a distinct risk of a resectable HCC becoming unresectable, or increases risk of perioperative complications such as liver failure [5,6]. In contrast, although preoperative TAC is safer than the TACE for resectable HCC patients, but whether it could improve surgical outcomes was not fully studied so far.
The anti-tumor effect of TAC is attributed to oxidative stress induced by anti-neoplastic agents [7] via generation of both high and low levels of reactive oxygen species (ROS) [8]. During cancer chemotherapy, lipid peroxidation induced by oxidative stress generates electrophilic aldehydes that can attack many cellular targets [7]. ROS such as hydrogen peroxide, superoxide anion, hydroxyl group and oxygen are byproducts of cell metabolism [9]. ROS damage tissues and cellular components including membranes, DNA, and proteins. Oxidative stress occurs when the ROS concentration exceeds the antioxidant capability of a cell or tissue [10]. In vitro experiments revealed that effects ROS can stimulate cell growth [11]. ROS regulate genes via protein kinase C activation, oxidative damage, and/or direct activation of transcription factors. The effects of ROS on gene transcription may also inhibit normal cell apoptosis via modulation of myc, bcl-2, and TP53 expression, resulting in an increase in cell number [12]. The most damaging species among ROS is the hydroxyl radical, which is responsible for base modifications such as the generation of 8-hydroxy-2ʹ-deoxyguanosine (8-OHdG) [13], involving oxidation of guanine at the C-8 position [14]. 8-OHdG can induce G-C to T-A transversion during DNA replication, and is thus used as a marker of oxidative DNA damage [15].
TP53 is known as a guardian gene and it is thought that TP53 loss is responsible for a lack of apoptotic signals in tumor cells and thus for their uncontrolled proliferation and recurrence [16]. Chemotherapeutic agents induce oxidative stress in vivo [17]. TP53-dependent and -independent pathways modulate the cytotoxic effects of common anti-neoplastic agents [18]. Moreover, TP53dependent cycle arrest is primarily mediated by the cyclindependent kinase inhibitor p21 waf1 /cip1 [19]. Mutant or absent TP53 status has been associated with resistance to radiation therapy and to apoptosis-inducing chemotherapy [20]. Mutations in the TP53 gene have frequently been detected in patients with recurrent HCC, and the interval between resection and HCC recurrence was significantly later in patients with wild-type TP53 than in those with mutations [21].
Chemotherapy could trigger DNA repair in HCC, raising the hypothesis that the overall efficacy of TAC on HCC depends on the cellular response to oxidative stress caused by chemotherapeutic agents in both cancerous and surrounding tissues. Therefore, we investigated oxidative stress in liver tissues in HCC patients receiving preoperative TAC to determine the effect of oxidative stress on changes in HCC tissue and its association with prognosis.

TP53 expression in tumor tissues
TP53 expression levels in cancer tissues in the pTAC (31.9 ± 32.1%) and non-pTAC groups (29.2 ± 30.0%) were not significantly different (t = 0.41, p = 0.683). TP53 was not detected in surrounding or normal tissues. p21 waf1/cip1 expression in tumor tissues p21 waf1/cip1 expression in cancer tissues was significantly higher in the pTAC group (51.0 ± 33.8%) than in the non-pTAC group (30.0 ± 33.7%; F = 13.459, p = 0.003), and in both cases was significantly lower than in the control group (76.5 ± 17.1%; F = 13.459, p < 0.001). No significant difference in p21 waf1/cip1 expression was found between cancer and surrounding tissues (p > 0.05) in the pTAC and non-pTAC groups. However, p21 waf1/cip1 expression in surrounding tissue was significantly higher in the pTAC group (54.1 ± 28.9%) than in the non-pTAC group (29.6 ± 33.5%; p < 0.001), and in both cases was significantly lower than in the control group (F = 16.613, p < 0.001; Figure 1). The association between p21 waf1/cip1 levels in cancer and surrounding tissues was significant in the non-pTAC group (r = 0.872, p < 0.001) but not in the pTAC group (r = 0.04, p = 0.808).

Alanine aminotransferase (ALT) and aspartate aminotransferase (AST)
There was no significant difference in baseline liver function indices between pTAC and non-pTAC patients Oncotarget 54404 www.impactjournals.com/oncotarget on admission ( Figure 2). However, in the pTAC group, ALT and AST levels after TAC 3 days were significantly higher than at baseline (Z = -4.808, p < 0.001). At 1 week after surgery, there was no significant difference in ALT (p = 0.076) and AST(p = 0.427) levels or in the change in ALT (ΔALT; p = 0.676) and AST (ΔAST; p = 0.281) levels between the pTAC and non-pTAC groups. The 8-OHdG level in cancer tissue was associated with ΔAST (r = -0.348, p = 0.030) in the pTAC group. Similarly, 8-OHdG in cancer tissue was associated with ΔALT (r = -0.281, p = 0.048) in the non-pTAC group (Table 1).
Univariate analysis shows that Barcelona Clinic Liver Cancer (BCLC) classification stage (p = 0.001), HBsAg positive (p = 0.003), AFP levels > 400 ng/ml (p = 0.033), tumor size (p = 0.002) and Vascular invasion (p = 0.028) were significantly influence tumor-free survival time (Table 2). We used a multivariate analysis, including the predictors associated with tumor-free survival time mentioned above, with the Cox proportional hazard method to evaluate the factors influencing tumorfree survival time. The analysis revealed that BCLC stage (p = 0.045) and vascular invasion (p = 0.010) were independent risk factors for the post-operative tumor-free survival time of HCC patients (Table 3).

DISCUSSION
Our study showed that regardless of whether HCC patients received preoperative TAC or not, 8-OHdG Oncotarget 54405 www.impactjournals.com/oncotarget levels were significantly higher in tumor tissues than in control tissues. This indicates an increase in oxidative stress in cancer tissues compared to normal liver tissues. Moreover, 8-OHdG levels were higher in cancer tissues than in surrounding tissues in the pTAC and non-pTAC groups, similar to findings in lung cancer [22]. renal cell carcinoma [23], and carcinoma of the large intestine [24], suggesting higher levels of oxidative DNA damage in cancer cells than in surrounding tissues. This supports that increased 8-OHdG levels were associated with a high risk of hepatocarcinogenesis [25]. Some findings not only indicate the active proliferation and high oxygen metabolism in cancer tissues but also probably reflect the inflammation that occurs in tumor tissues [14,26,27]. 8-OHdG levels are also likely to reflect cell differentiation, proliferation, and invasion in HCC [28]. We found that 8-OHdG levels in HCC tissues were significantly lower in the pTAC group than in the non-pTAC group, probably because the active metabolism and chemoresponsiveness of HCC cells were inhibited by the antitumor pTAC drugs.
However, residual tumor cells might escape from pTAC in a fortified oxidative stress repair mechanism. At the same time, liver aid used to protect patients after TAC may need to be taken into consideration. In peritumoral tissues, there was no difference in 8-OHdG between the pTAC and non-pTAC groups, indicating that the effects of TAC differs between tumor and surrounding tissues: TAC can decrease 8-OHdG levels in cancer tissues, but has no effect on 8-OHdG in peritumoral tissue in HCC. After chemotherapy, a significant decrease in urinary 8-OHdG was found in small-cell lung cancer patients with a complete or partial remission response, which reflects a decrease in tumor mass and thus relief of oxidative stress; by contrast, a significant increase in urinary 8-OHdG was observed after chemotherapy in patients with no response or with progressive disease, and patients with no response to radiotherapy had higher 8-OHdG levels than patients with a response [29]. Toyokuni et al. [30] found that high antioxidant levels induced by persistent oxidative stress increases the chemotherapy resistance of cancer cells.  Note: *Association is significant at the 0.05 level (2-tailed). ALT 1 and AST 1 were the preoperative levels of ALT and AST before hepatectomy. ALT 2 and AST 2 were the post-operative levels of ALT and AST after hepatectomy 1 week.   Oncotarget 54408 www.impactjournals.com/oncotarget During cancer chemotherapy, lipid peroxidation induced by oxidative stress generates numerous electrophilic aldehydes that can attack many cellular targets. These oxidative products can slow the cell cycle progression of cancer cells and cause cell cycle checkpoint arrest that may interfere with the ability of anticancer drugs to kill cancer cells. The aldehydes may also inhibit drug-induced apoptosis by activating death receptors and inhibiting caspase activity [7]. These effects would also diminish the efficacy of treatment [7]. Accordingly, the 8-OHdG level in cancer tissues may remain high after TAC, suggesting that HCC cells may be resistant to a pro-chemotherapy program. This could provide a useful hint for clinicians considering a change in chemotherapy program and for detection of early recurrence after tumor resection.
ROS as signaling molecules can mediate apoptosis via a TP53-dependent pathway. On encountering a DNA defect, TP53 protein encoded by wild-type TP53 immediately stops the cell cycle and starts the DNA repair mechanism; if DNA repair fails, TP53 initiates apoptosis instead. The mechanism by which TP53 inhibits cell cycle progression is likely to involve activation of p21 waf1/cip1  Oncotarget 54409 www.impactjournals.com/oncotarget expression via transcription. The half-life of wild-type TP53 is too short to examine, and mutant TP53 can only be detected by immunohistochemistry. We found that TP53 expression in tumor tissues did not differ between the pTAC and non-pTAC groups, perhaps indicating the presence of cancer cells with mutant TP53 resistance to the TAC program used. This supports the view that tumor cells with mutant TP53 have greater potential for malignancy and invasion [31,32]. The decrease in oxidative stress levels in cancer tissues in the pTAC group can probably be attributed to a repair mechanism mediated by a TP53-independent pathway. Ahn et al. [33] found that HCC occurrence and development were correlated with downregulation of p21 waf1/cip1 protein expression. In our study, p21 waf1/cip1 protein expression was highest in the control tissues, was lower in peritumoral than in control tissues, and was lowest in cancer tissues. This is in agreement with observations by Mise et al. [34], but not with results reported for HCC, hepatic cirrhosis, and normal hepatic tissues by Zhang et al [35]. We do not know whether this contradiction can be attributed to incompetence p21 waf1/cip1 protein expression in tumor tissues or not [36]. In the pTAC group, p21 waf1/ cip1 protein expression increased in cancer tissues because of repair of oxidative DNA damage in cells induced by chemotherapy. We found that although TP53 protein levels in cancer tissues were similar in both HCC groups, p21 waf1/ cip1 expression was significantly higher in the pTAC group than in the non-pTAC group, indicating that in spite of defective TP53 protein in tumor cells, oxidative stress was upregulated in response to chemotherapy via an alternative pathway. p21 waf1/cip1 expression was significantly lower in peritumoral tissues in both the pTAC and non-pTAC groups than in control tissues. It is likely that this is related to inadequate wild-type TP53 function in peritumoral tissues in HCC. Peritumoral p21 waf1/cip1 expression levels were higher in the pTAC group than in the non-pTAC group, so it is possible that p21 waf1/cip1 upregulation involved a TP53-independent pathway or that hOGG1 levels increased after TAC.
Anti-neoplastic agents used in TACE not only cause necrosis of tumor cells but also induce liver dysfunction [37]. Our data revealed that in the pTAC and non-pTAC groups, there was no significant difference in ALT and AST levels between preoperative and post surgery time point. At 3 days after TAC administration, ALT and AST levels were significantly higher than on admission, suggesting that after chemotherapy, patients sensitive to TAC suffer from damage not only to the liver but also to Oncotarget 54410 www.impactjournals.com/oncotarget other organs [38]. Moreover, we found that ΔAST was negatively associated with 8-OHdG in pTAC group, whereas ΔALT was negatively associated with 8-OHdG in the non-pTAC group (Table 1), although these correlations were mild, indicating that the levels of oxidative stress in cancer tissues might affect the liver function.
Survival analysis (Figure 3) showed that the tumorfree survival time did not significantly differ between the pTAC and non-pTAC groups, similar to findings by Kaibori et al. [39]. The median tumor-free survival time was longer in the pTAC group (52 weeks) than in the non-pTAC group (26 weeks), similar to results reported by Choi et al. [40]. 8-OHdG levels have been used as an independent prognostic factor for HCC recurrence, and patients with greater intrahepatic oxidative stress had a higher incidence of HCC recurrence [41,42]. It is interesting to find that 8-OHdG levels in cancer tissues were significantly lower in the pTAC group than in the non-pTAC group. If overlook the interval from pTAC to surgical resection, this might suggest that HCC cells which metabolic active and sensitive to chemotherapy were attenuated by the antitumor drugs used, although it is also possible that the number of cases in our study was not much enough to properly reflect differences in overall survival time. We found that the 1-year cumulative tumorfree survival rate was significantly higher in the pTAC group than in the non-pTAC group. However, the 2-year cumulative tumor-free survival rate was not significantly different between the two groups. These results indicate that preoperative TAC can reduce the HCC recurrence risk in the short term after radical hepatectomy (1 year). Thus, a multimodal therapy strategy to prevent HCC recurrence within the first year after surgery seems to be warranted.
In the pTAC group, tumor-free survival time was inversely correlated with 8-OHdG levels in cancer tissues. This suggests that the levels of 8-OHdG generated by residual tumor cells that escape chemotherapy may reflect the prognosis of HCC patients undergoing preoperative TAC. Other researchers found that high oxidative stress in non-cancerous liver tissue from HCC patients was associated with a higher incidence of HCC recurrence [41,42]. However, we did not find any correlation between 8-OHdG levels in surrounding tissues and survival time. This may be because the population in Guangxi province is exposed to high levels of aflatoxin B 1 (AFB 1 ) [43]. Liver damage caused by hepatitis viruses and AFB 1 may generate more 8-OHdG, which might influence the survival time of HCC patients in Guangxi province.
We found that BCLC stage of HCC, vascular invasion influenced HCC recurrence. This is consistent with previous observations [41] and supports the view that the greater the tumor malignancy and invasion (high pathologic staging and grading), the poorer is the survival rate.
One limitation of this study is the small number. It is because we had applied strict criteria of enrollment. We included only HCCs received radical resection and excluded those received variable dosage of TACE to control the variations. A larger sample size in the future might strengthen the findings in this study. Another concern of bias came from the post-resection TAC/TACE that all our HCCs received. Whether HCC patients could be beneficial from post-resection TAC/TACE remains controversial [44,45]. So when retrospectively studying this population 10-yrs ago, we recruited only HCCs received postoperative TAC at 1, 3, and 6 months after hepatectomy to ensures comparable baseline between the pTAC and non-pTAC groups. Finally, in this study we tried to ensure the patients' clinical baselines in two groups were comparable. It is ideal to get liver tissues before and after TAC therapy and compare their oxidative status. But in practice it is hard to get approval from HCC patients to do biopsy (to get liver tissues) before TAC, especially when the tumors were resectable.
In conclusion, this study clearly showed that after preoperative TAC, the status of oxidative stress in cancer tissue changed in HCC. Oxidative stress levels were higher in HCC tissues than in surrounding tissues and normal liver tissue. Hepatic cancer cells probably escape intervention chemotherapy via a fortified oxidative stress repair mechanism. Differences in the expression of DNA damage biomarkers and cell-cycle regulators in HCC and surrounding tissues suggest that a distinct cellular response to oxidative challenge might have clinical implications. Preoperative TAC postponed early HCC relapse after hepatectomy, for which the mechanism might involve lower levels of oxidative stress in the tumor microenvironment.

Subject recruitment and sample collection
Between April 2006 and April 2008, a total of 463 HCC patients were hospitalized in the Hepatobiliary Surgery Department of First Affiliated Hospital of Guangxi Medical University. Of these, 224 (48.4%) were selected for surgery (including liver transplantation) and 239 (51.6%) received either TACE (n = 180) for unresectable tumors or percutaneous ablation therapy (PAT; n = 59) for single HCC tumors < 2 cm in diameter. Of the 224 surgical patients, 89 (39.7%) underwent radical hepatectomy (cutting-edge to tumor boundary > 2 cm and/or no tumor visible at the cutting-edge under a microscope.) and were divided into two categories: 39 cases who received preoperative TAC (pTAC group) and 50 who did not (non-pTAC group). The 39 pTAC patients underwent transcatheter arterial angiography to differentiate tumor masses and cirrhotic nodules (lipiodol was not used) before TAC administration. A fixed regime (pirarubicin 50 mg + cisplatin 50 mg + fluorouracil 1 g) was used for all TAC patients. Lipiodol was not used for embolization in these patients because it may cause Oncotarget 54411 www.impactjournals.com/oncotarget tumor necrosis. None of the 89 enrolled HCCs received previous HCC treatment or antiviral treatment before preoperative TAC, none of them received Sorafenib before or after surgical/TACE intervention. And none of other HCCs received radiotherapy regardless its PVTT status. 101 HCC patients who received preoperative TAC with different drug doses were excluded. Liver function indices were measured before and on the 3rd day after TAC. Hepatectomy for pTAC patients was performed 1 week after the chemotherapy, and both the pTAC and non-pTAC groups received postoperative TAC at 1, 3, and 6 months after hepatectomy. Samples of 15 non-HCC liver tissues were obtained during operations for hemangioma, liver harvesting for transplantation, or accidental hepatorrhexis, and served as the control group. All control subjects were negative for hepatitis B (HBV) and hepatitis C virus (HCV) sero-markers. The type of hepatectomy for HCC depended on the location of tumor(s), the severity of concomitant cirrhosis, and liver function reserve. In all cases a surgical margin > 2 cm was achieved. The duration of Pringle's maneuver was not significantly different among the three groups. Demographic data for the 104 cases are summarized in Table 4. Formalin-fixed, paraffin-embedded specimens were collected and independently diagnosed by two pathologists (Z.-L. Lu  Oncotarget 54413 www.impactjournals.com/oncotarget Immunohistochemical assays for 8-OHdG, TP53, and p21 waf1/cip1 Tissues were fixed in formalin (10%) immediately after resection, dehydrated in absolute ethanol and embedded in paraffin. Serial sections (5 μm) were prepared for immunohis tochemical analysis. The 8-OHdG, TP53 and p21 waf1/cip1 levels were measured by hepatic immunohistochemical staining. Using monoclonal antibodies for 8-OHdG (N45.1; Japanese Aged Control Institute, Japan), TP53 and p21 waf1/cip1 (mouse anti-human; Maixin-Bio Fuzhou, China). For 8-OHdG immunostaining, tissue sections were pretreated with 100 µg/ml ribonuclease A in phosphate buffer saline (PBS) in order to inhibit nonspecific binding to RNA. Immunohistochemistry was performed using a modified streptavidin-biotinylated peroxidase technique using Ultra Sensitive TM S-P or EliVision TM plus Kit from Fuzhou Maixin. Nonspecific protein staining was blocked by rabbit or goat serum for 30 min at room temperature. The sections were incubated with primary antibody at 4°C overnight. The sections were rinsed with PBS and incubated with biotinylated secondary antibodies for 30 min. They were then washed and treated with 0.3% hydrogen peroxide in methanol for 30 min in order to inhibit the activity of any endogenous peroxide. The slides were washed, incubated with streptavidinbiotin-peroxidase complex for 30 min, and developed according to the manufacturer's directions. The sections were subsequently counterstained with DAB. Brownyellow staining in cell nuclei was observed for 8-OHdGand TP53-positive cells ( Figure 6A, 6B) and in cytoplasm for p21 waf1/cip1 -positive cells ( Figure 6C). The diagnosis was made by two pathologists independently, and the number of positive nuclei (p21 waf1/cip1 -was positive cytoplasm) in four to five randomly selected fields (400×) were counted and expressed as the ratio of positive nuclei to total nuclei (p21 waf1/cip1 -was the ratio of positive cytoplasm to total cytoplasm) .

Statistical analysis
Data for continuous variables are expressed as mean ± SD. One-way ANOVA and Fisher's least significant difference (for multiple comparisons) were used to evaluate differences in 8-OHdG and p21 waf1/cip1 levels in liver tissues. A × test was used to examine the ratio of gender, race, smoking, and drinking in different categories. The Wilcoxon rank test was performed to assess changes in liver function. A paired t-test was used to test differences in 8-OHdG, TP53, and p21 waf1/cip1 levels in cancer and surrounding tissues. The Spearman rank correlation coefficient was used to test associations. The Kaplan-Meier method was used to graphically depict the tumor-free survival time for HCC patients, and the logrank test was applied to compare the median tumor-free survival time between the pTAC and non-pTAC groups.
A Cox proportional hazards model was used to evaluate factors influencing tumor-free survival time. A two-tailed p value <0.05 was considered statistically significant.