Prognostic Factors of Disease-Free Survival in Postoperative Patients with Hepatocellular Carcinoma

Panquan Luo The First A liated Hospital of Anhui Medical University Lixiang Zhang The First A liated Hospital of Anhui Medical University Bocheng Ding The First A liated Hospital of Anhui Medical University Lei Chen The First A liated Hospital of Wenzhou Medical University Baichuan Zhou The First A liated Hospital of Anhui Medical University Zhijian Wei The First A liated Hospital of Anhui Medical University Aman Xu (  xamlpqdoctor@163.com ) The First A liated Hospital of Anhui Medical University


Introduction
On the basis of global cancer statistics, primary liver cancer ranks fth among all types of cancer, and ranks fourth in cancer-related death worldwide. According to recent WHO data, as many as 781,631 people die of liver cancer every year in the world 1,2 .
With helical computed tomography (CT), magnetic resonance imaging (MRI) examination and the level of treatment methods have been greatly improved, patients with early liver cancer underwent radical surgical resection, the 5-year survival rate can reach 70% 3 . However, most liver cancer patients who come to the hospital due to dull pain in the upper abdomen or other symptoms are already in the middle and late stages of the disease, the 5-year survival rate is less than 15% 4 . Therefore, it is urgent to study the prognostic factors of HCC.
The factors and pathogenesis of HCC are controversial comprising of multiple genetic, epigenetic alterations, chromosomal aberrations and gene mutations 5 . It has been reported that the occurrence of liver cancer is related to viral hepatitis, especially hepatitis B 6 . At present, radical hepatectomy is the rst choice for the treatment of HCC patients. After the operation, the necessary and appropriate chemotherapy regimen should be given 7 .
Up to date, many clinical studies have been performed to identify the corresponding markers that can accurately predict the prognosis of HCC. Unfortunately, no clear prognostic factors for HCC patients have been identi ed. A great deal of clinical investigations indicated that factors affecting overall survival of HCC patients include age, HBsAg, alcohol consumption history, TNM stage, differentiation degree, tumor number, tumor size, AFP, GGT, NLR and so on 8-11 . Jun Zheng et al 12 suggested NLR and PLR, which were SIR (Systemic in ammatory response) indexes, were new prognostic markers for predicting the prognosis of HCC. Increased NLR or PLR was found to be an independent predictor of higher recurrence and poor survival in HCC patients receiving curative or palliative therapy. Brian K. P. Goh et al. 13 showed number of cancer nodules (more than three nodules), Child-Pugh status, margin positivity and presence of microvascular invasion, but not tumor size, were independent negative predictors of OS (overall survival). In addition, in the retrospective study of 414 patients with liver cancer over a period of 7 years, our team found that serum amma-glutamyltransferase levels and AST/ALT play an important role in the prediction of OS in patients with primary hepatic carcinoma. Low GGT group and low AST/ALT group have a better prognosis 14 . Our group has studied the prognostic factors of OS in patients with liver cancer, but no further study has been conducted on DFS in liver cancer patients, especially hepatocellular carcinoma patients.
The purpose of this study was to explore the clinical factors affecting DFS in patients with HCC after radical resection, so as to take appropriate therapeutic measures for maximizing these patients DFS.

Patients
This study retrospectively recruited 344 cases of hepatocellular carcinoma patients that received curative hepatectomy during the department of gastrointestinal surgery the First A liated Hospital of Anhui Medical College. The inclusion criteria were: 1) All patients with HCC were con rmed histo-pathologically; 2) Resection of liver cancer was necessary; 3) The clinical data of the patients were complete and correct; 4) The survival of patients could be followed up; 5) no other treatment, such as chemoradiotherapy and interventional therapy before surgery. Exclusion of patients criteria were: 1) Postoperative pathology: bile duct cell carcinoma or metastatic liver cancer; 2) Incomplete data; 3) Child-pugh grading of liver function was grade C; 4) Inability to follow up or loss of follow-up. We collected information of patients through the record room of our hospital, including patients gender, age, personal history (smoking history, alcohol use history), BMI, presence of basic diseases (hypertension, diabetes, history of hepatitis B), presence of liver cirrhosis, tumor number, tumor size, differentiation degree, TNM stage, number of cancer nodules, AFP, portal vein thrombosis, ascites, routine peripheral blood examination (neutrophils, lymphocytes and platelets) and anticoagulation test (prothrombin time, brinogen), liver function test (ALB(albuminbilirubin), TBIL(total bilirubin), TC(total cholesterol), ALT(alanine aminotransferase), AST(aspartate aminotransferase), GGT(gamma-gl utamyltransferase)). In our study, the patients were divided into a High group and normal group based on their thresholds levels of serum ALB, TBIL, TC, TG, ALT, AST and GGT. In addition, the median of NLR and PLR were used as cutoff value, respectively. This study was approved by the ethics committee of the 1st A liated Hospital of Anhui Medical College, and all patients signed the preoperative informed consent.

Follow-up and treatment
The follow -up information of the patients was obtained through telephones and outpatient visit. The follow -ups carried out in normal intervals (follow-up began one month after the surgery, each 3 months within two years, and each 6 months from the third years to fth year, and once a year after the fth year). Routine peripheral blood examination, liver function test, liver ultrasound or CT were performed at each follow-up patient after hepatectomy.

Statistical analysis
All patient data were statistically described using spss 19.0 software. Chi-square test or Fisher exact test were used to investigate categorical variables. Univariate analysis was performed for all variables, and then the variables of P<0.05 were selected for multivariate Cox regression analysis. The kaplan-meier method was performed to analyze DFS and survival curves were compared using the log-rank test in each group. Cox proportional hazards regression were utilized to assess the relationship between different groups and DFS in patients. The hazard ratio (HR) and 95% con dence interval (95% CI) were used to show the risks. HR > 1.0 showed an increased risk of death. The R Project for Statistical Computing 3.5.5 software and RMS program package were used to construct the DFS prognostic nomogram model.P value < 0.05 was considered statistically signi cant.

Results
Clinical characteristics of patients A total of 344 HCC patients were included in this study, among whom 279 cases were males and 65 cases were females, accounting for 81.10% and 18.90% respectively. Details of patients' information were shown in table 1. The patients were divided into two groups according to their age, namely the elderly group with age ≥70 years old and the young group with age < 70 years old. 258 (75%) patients had a history of hepatitis B, only 86 (25%) cases did not. Most patients had no history of hypertension or diabetes (76.74% and 88.08%). Nearly half of the patients (45.93%) had a BMI more than 23. Moreover, the patients who had liver cirrhosis were 256 (74.42%), 88 patients without liver cirrhosis, nevertheless, most of them had abnormal liver function, such as hepatitis and fatty liver. 44 (12.79%) cases had ascites, unfortunately, two of them had bloody ascites. The TNM stage of I-II and III-IV were 300 (87.21%) cases and 44 (12.79%) cases. The median of NLR and PLR was 2.19 (inter-quartile range: 1.64-3.31) and 97.67 (72.10-138.17), respectively. The median follow-up time for this study was 52.00 (10.00-78.75) months. During the follow-up period, the 1-year, 3-year and 5-year DFS of the patients were 73.26%, 59.30% and 44.48%, respectively.

Correlation between different age groups and clinicopathological characteristics of patients
The results of the correlation between clinical and pathological indicators in the elderly group and the young group were shown in Table 2. Patients without hypertension and diabetes appeared to be less common in elderly patients than in the young group (45.65% vs. 81.54% and 80.43% vs. 89.26%). As for hematological index, the age was signi cantly connected with HbsAg, AST, ALB, PLR (P<0.05). However, there were no statistical differences between the two groups with regard to smoking, alchohol intake, BMI, tumor number, tumor diameter, differentiation degree, TNM stage, AFP, TBIL, TC, ALT, GGT and NLR.

The prognostic factors for DFS
The results of univariate regression analysis were expressed in Table 3, and Table 4 showed the multivariate analysis of DFS which included the signi cant factors obtained from Table 3. Multivariable Cox regression analysis revealed that age, history of hepatitis B and TNM stage were independent factors for predicting disease-free survival rate of HHCpatients. Besides, the hazard rate (HR) was 0.543 (95%CI:0.328-0.898) for patients older than 70, 0.654 (95%CI:0.472-0.907) for patients without history of hepatitis B, and 0.585 (95%CI :0.423-0.810) for patients with TNM stage of I-II.
Results of DFS in different age groups, AJCC stage and with or without history of hepatitis B As shown in Figure 1, the DFS of the different age groups were compared with Kaplan-Meier plot. The results indicated that patients with age≥70 have a longer DFS (P<0.05). As can be seen from the Kaplan-Meier curves, the survival rate in the young group showed the most signi cant downward trend in the rst 6 months of follow-up, suggested that the patients in the young group had a greater risk of recurrence or even death within 6 months after surgery. Figures 3 and 5 showed DFS at different TNM stage and with or without a history of hepatitis B, respectively. From the gures, we found that the median DFS of the stage I-II group was 68 months, and the median DFS of the stage III-IV group was 12 months. patients with TNM stage of I-II and no history of hepatitis B had longer DFS ( P<0.05).
In order to investigate the association between TNM stage and the different age groups further we performed survival analysis and results were shown in Figure 2A and Figure 2B. Figure 2A showed that in TNM stage of I-II, the elderly group possessed a longer DFS than the young group. However, there was no signi cant difference in TNM stage of III-IV. In addition, we also investigated the association between age and different TNM stage groups through Figure 4A and Figure 4B. The results suggested that the young group also had a better prognosis in TNM stage of I-II. In contrast, the difference was not statistically signi cant in the elderly group. Furthermore, the association between TNM stage and with or without history of hepatitis B was analyzed, however, no statistically signi cance was found.the association between TNM stage and with or without history of hepatitis B was analyzed in Figure 6A and Figure 6B, and no statistically signi cance was found.
The establishment and validation of prediction model Based on the cox regression model, we used independent prognosis factors such as age, TNM stage, history of hepatitis B to establish a nomogram in Figure 7. Each clinicopathological feature of the patient was projected upward to obtain the corresponding score, which was then added to obtain the total score points. The predicted survival rate was correlated with the total points by drawing a vertical line from the Total Points scale to the 3-year or 5-year DFS scale.The C-index of this prediction model was calculated to be 0.604(95%CI:0.561-0.647) through 1000 times of bootstrap resampling method for internal validation, which suggested that this nomogram had good predictive value. The 3-year and 5-year DFS Calibration curve( Figure 8A and Figure 8B) show that the actual survival curve of the nomogram ts the predicted survival curve, indicating the model is reliable.

Discussion
Primary liver cancer is one of the most common gastrointestinal malignancies. Because there are usually no typical symptoms in the early stage, the diagnosis of some patients is delayed 15 . Hepatocellular carcinoma (HCC) has a high incidence rate and a high mortality rate. As a result, more and more attention has been put to the research of HCC prognostic factors 16 .
Our current research retrospectively analyzed the Clinico-pathological information of 344 patients with HCC admitted to the First A liated Hospital of Anhui Medical College from December 2009 to December 2014. The results of univariate and multivariate cox regression analysis indicated that age, history of hepatitis B and TNM stage were independent factors of DFS in patients with HCC(P<0.05). NLR was calculated using the formula absolute neutrophil count (number of neutrophils/ml) divided by absolute lymphocyte count (number of lymphocytes/ml), PLR was calculated as absolute platelet count (number of platelets/ml) divided by absolute lymphocyte count (number of lymphocytes/ml) 17 . Many studies had shown that NLR, AFP and brinogen were related to liver cancer patients OS 18, 19 . In our study, we used the median of NLR and PLR as their cutoff values, the values were similar to that previously reported 20,21 .To our surprise, hematological indicators such as NLR, PLR and AFP were not prognosis factors of patients DFS. It demonstrates that the levels of NLR, PLR and AFP are reliable indicators for predicting the OS of patients with HCC. However, these indicators are probably not play a role in HCC patients DFS.
In our study, the de nition of elderly group in hepatocellular carcinoma patients was age of 70 or over based on previous published studies 22,23 . Age≥70 patients with HCC had longer DFS. In order to further understand the in uence of age on patients DFS under different AJCC stages, we analyzed the DFS of patients in different age groups under stages I-II and III-IV, respectively. Patients aged ≥70 at stage I-II also had a better prognosis compared to the young group. Interestingly, in study of Atsuko Sakakibara et al. 24 , they divided the age groups into the young group and the old group at the age of 40, it was observed that despite the older adult group having a higher proportion of advanced cancer patients, the overall survival rate of stage IIB patients in the young group at the 3-year follow-up was signi cantly lower, This result was similar to what we found in this study. Beside, The similar results were also found in the study of Liying Zhao et al. 25 , this retrospective cohort study included a total of 995 patients with CRC suggested the patients aged 35 years and younger have speci c clinicopathological characteristics that tend to a worse prognosis. On the contrary, W. Faber et al. 26 retrospectively studied 141 patients with liver cirrhosis after curative resection, and found out prognosis of patients less than 70 years old was signi cantly better. The 5-year survival rate reached about 50%, which was signi cantly higher than that of the elderly group. The most likely reason for these two completely opposite conclusions is that different researchers have different research priorities and parameter Settings, resulting in different results. Therefore, it is necessary to further study the mechanisms by which age affects cancer prognosis and the relationship between age of surgery and DFS in HCC patients. It is generally known that the stage of tumor can greatly affect the prognosis of patients, the more advanced the stage, the worse the prognosis. In this study, we found that regardless of the patients age, the prognosis of patients of TNM stage I-II was signi cantly better than that of patients with stage III-IV. The results were the same as that of a decade-long and population-based Analysis by Lu Wu et al. 27 . For patients at the early stage, radical surgery should rstly be considered, this may prolong the DFS and help prevent metastasis.
As is known to own, hepatitis B virus (HBV) is a leading causing factors of HCC. HBV infection is a global health problem and roughly 30% of the world's population shows serological evidence of current or past HBV infection, especially in China. China is a big country with hepatitis B, although the success of neonatal vaccination since 1992, which had achieved a great reduction 28 , there are still great challenges, HBV infection affects at least 2 billion people worldwide, and 97 million people are HBV carriers and at least 20 million of them still suffer from active or chronic HBV infection in China 29 . Lian Li et al. 30 suggested the HBV DNA level was an independent prognostic factor for patients with OS and DFS, in addition, HBV DNA level of 2000 IU/ml or greater before operation indicated a poorer prognosis. As for the pathogenesis of liver cancer caused by HBV, on the one hand, HBV virus gene fragments can be inserted directly into the genomes of normal hepatocytes, causing gene mutations that induce normal cells to transform into tumor cells, especially in patients with high HBV-DNA levels. on the other hand, HBx is a protein encoded by HBV with trans-activation activity which can dysregulate cell transcription and proliferation control and sensitize liver cells to carcinogenic factors 31 . History of hepatitis B was associated with DFS in HCC patients in this study, this may emphasize that we should pay attention to the history of hepatitis B when we ask the history of HCC patients in the course of clinical practice.
As a visual prediction tool, the nomogram has been widely used in the prognostic research of primary liver cancer, breast cancer, cervical cancer and other malignant tumors. It can integrate multiple predictive variables in a weighted manner and intuitively show the in uence of variables on individual predictive values 32 . In this study, we rst attempt to develop a prognostic nomogram which combined serum markers (i.e., age, in ammatory markers and tumor markers) and clinicopathology characteristics for predicting the probability of HCC patients 3-year and 5-year DFS. The calibration curve and Cindex(concordance index) were used to assess the accuracy of the nomogram, the results showed that the nomogram had great predictive ability.
The main advantage of this study was that the analyzed factors are relatively comprehensive and can effectively reduce residual confounding.
There were several limitations of our study such as the number of patients in the elderly group was small and single-center retrospective design. Hence, Increasing the sample size and replacing the single center with a multi-center prospective study could make the results more convincing and representative.

Conclusions
In conclusion, we found that age, TNM stage and history of hepatitis B were independent prognostic factors for patients DFS after radical hepatectomy. Furthermore, the patients older than 70 years old, earlier TNM stage and no history of hepatitis B were associated with a better prognosis. Moreover, we constructed and validated a new nomogram for predicting postoperative DFS in patients with type II and III AEG. The prediction model showed accurate and reliable prediction e ciency, which may help clinical workers to formulate personalized treatment plans.

Declarations
Availability of data and materials The data used and/or analyzed during the current study were obtained from the Department of Gastrointestinal Surgery, the First Hospital of Anhui University. The data are available from the corresponding author on reasonable request.

Ethics declarations
Ethics approval and consent to participate The present study was reviewed and approved by Ethics Committee of The First Hospital of Anhui University (Hefei, China). Informed consent for use of the medical data in this study was obtained from the patients. I con rm that all methods were performed in accordance with the relevant guidelines and regulations in declaration of Helsinki.

Consent for publication
Not applicable.