Predictive Factors for the Post Embolization Fever after TACE for Hepatocellular Carcinoma Patients: A Single-Center Study in China.

Background Fever is one of the main symptoms for post-embolism syndrome (PES). This study aimed to determine and validate a model to predict fever after transcatheter arterial chemoembolization (TACE) in patients receiving platinum as the main regimen. Materials and Methods Clinical data of HCC patients who underwent TACE with platinum was retrospectively collected in the Fudan University Zhongshan Hospital during January 2016 to January 2018. According to post-TACE medical records, patients were divided into fever group and non-fever group. Predictive factors were selected by multivariate logistic regression. The receiver operating characteristic (ROC) curve were then performed to detect accuracy and discriminative ability of these factors using the derivation cohort and an independent validation cohort. Results Fevers were detected in 44 of 252 patients. Demographics, laboratory data were statistically similar within fever group and non-fever group. Strongest predictors identied in multivariate logistic regression included Iopiodol emulsion dose (OR, 1.081; 95%CI, 1.006-1.162), number of hepatoprotectants (OR, 0.619; 95%CI, 0.419-0.914), K + (OR, 2.992; 95%CI, 1.225-7.308), and albumin-bilirubin (ALBI) grade (OR, 2.249; 95%CI, 1.040-4.862). Furthermore, the area under the ROC curve of derivation cohort and validation cohort were 0.798 and 0.874 respectively, which indicated comparative stability and discriminative ability of this model. Conclusions Iopiodol emulsion dose, number of hepatoprotectants, K + , and ALBI grade are strong predictors for PEF. The multivariate logistic model of these factors shows a discriminative ability to predict PEF in the validation cohort.


Introduction
Hepatocellular carcinoma (HCC), the most frequent type of liver cancer, is now the fourth most common malignant tumor and the third most lethal malignant tumor in China (1). Different from western countries, East Asia had especially suffered from a very large burden of HBV infection (2). As a vital identi ed risk factor for HCC, hepatitis B virus (HBV) infection has contributed to more than 60% liver cancer deaths and cases in China (3,4). Besides the rst-line curative treatment-liver resection (LR) for HCC, currently, transarterial chemoembolization (TACE) is the main treatment option for unresectable, large/multifocal HCCs without vascular invasion or extrahepatic spread (5). It has been recommended as the preferred treatment for Barcelona Clinic Liver Cancer (BCLC) patients with stage B HCC according to the current European Association for the Study of Liver guidelines (6). However, TACE is associated with transient post-embolization syndrome (PES) with an incidence range from 30-80% (7,8). It was characterized as fever, unremitting nausea, vomiting, pain in liver region, abdominal distention, and poor appetite (9).
Among them, post-embolization fever (PEF) has been considered to re ect extensive tumor necrosis and represent the e cacy of TACE by physicians. Recent studies have also validated that PEF may be associated with tumor size and the use of embolic agents (10)(11)(12). However, PEF is less predicted by clinical biochemical indicators, and the few existing conclusions are inconsistent (11). The aim of this study was to analyze the predictive factors of fever after TACE in patients with HCC who were treated with platinum as the main regimen, and to help clinicians predict PEF.

Materials And Methods
Inclusion and exclusion criteria Data were retrospectively collected from patients with HCC who underwent TACE in the Zhongshan hospital, Fudan University from January, 2016 to January, 2018. Inclusion criteria included the following: complete medical records and laboratory data for patients before and after TACE, above 14 years old, the survival time of patients above 3 months, no urinary tract, endocardium, pelvic infection in nearly one month, platinum as the main chemotherapeutic regimen in TACE. Excretion criteria included pregnant women or lactating women and had infection and fever before TACE. Besides, another cohort of 115 patients who met the same inclusion and exclusion criteria mentioned above, were separately collected to be the validation cohort.

Data collection
Data were collected and retrieved through hospital information management system. Demographic information and medical records such as name, sex, age, diagnosis, concomitant diseases, operation procedure, combined medication were recorded. All laboratory examinations before and after TACE were recorded, mainly including blood routine, liver and kidney function parameters (such as total bilirubin, hours before operation and sending them for examination in the morning. ALBI grade has been reported to be a new tool for evaluation of hepatic function in HCC patients as compared to the Child-Pugh classi cation (13). ALBI score was used for grading (≤-2.60 = grade 1, greater than − 2.60 to≤-1.39 = grade 2, greater than − 1.39 = grade 3). The aspartate aminotransferase/platelet count ratio index (APRI, APRI=[AST(IU/L)/upper limit normal]/PLT(× 10 9 /L)] × 100), which is thought to be a biomarker of liver brosis and cirrhosis (14), was calculated as well.

Statistical analysis
Data are presented as the mean and standard deviation (SD) or numbers and percentages. For continuous variables, the differences between groups was calculated by independent Student's t-test or Mann-Whitney U test. Chi-squared test or Fisher's exact test were applied for categorical variables.
Multivariate logistic regression model using forward selection procedure was then constructed to identify the independent predict factors for PEF. The receiver operating characteristic (ROC) curve were performed to detect accuracy and discriminative ability of the model using the derivation cohort and a separate validation cohort. All the statistics were bilateral test, P < 0.05 was statistically signi cant. All statistical analyses were performed with SPSS software (IBM SPSS Statistics 22.0).

Results
Comparison of general data of patients between fever group and non-fever group A total of 252 patients were collected, including 209 males and 43 females, aged from 18 to 85 years, with an average age of 57.8 ± 11.3 years. The general data of fever group (n = 44) and non-fever group (n = 208) were shown in Table 1. There was no signi cant difference in demographic characteristics and complications between the two groups (P > 0.05). The number of varieties of hepatoprotectants in fever group was lower than that in non-fever group (1.5 + 0.6 vs 2.4 + 1.3, P = 0.000). In addition, the amount of iodized oil injected in fever group during TACE was higher than that in non-fever group (10.5 ± 5.5 vs 7.1 ± 4.6, P = 0.000).    hepatoprotectants, K + , ALBI grade and predict model ranged from 0.5 to 0.8, and the Cut-off point was 6.5 mL, 2.5, 4.25 mmol/L, 1.5 respectively (Table 3).

Discussion
TACE exploits the preferential hepatic arterial supply of HCC for targeted delivery and embolizes of the feeding artery branches of HCC by lipiodol emulsion, microspheres, polyvinyl alcohol and gelatin sponge with chemotherapeutic drugs. Lipiodol has the unique property of selective uptake and retention in hyperarterialyzed liver tumors (15). Generally, two or three kinds of chemotherapeutic drugs (such as doxorubicin, epirubicin, idarubicin, mitomycin C, or cisplatin), are emulsi ed in the lipiodol, and then followed by particle embolization to improve the overall survival rate of patients with HCC (16). However, TACE inevitably leads to hypoxic damage to hepatoma cells and surrounding liver tissues. PES is thought to be the result of therapeutic cytotoxicity, tumor ischemia, and intrahepatic and extrahepatic in ammation (9). Studies has showed that PES was associated with a worse survival and a two-fold increased risk of death (7). PEF, a common symptom of PES, was de ned as body temperature greater than 38℃ within 3 days after TACE with no evidence of infection (10). Although this fever is self-limiting, which may not be signi cantly related to the long-term survival rate of patients after TACE (10), and symptomatic interventions can be taken if necessary to achieve satisfactory relief (17), PEF often prolongs hospitalization and leads to unnecessary use of antibiotics.
The incidence of PEF reported in the literature ranged from 20-70% (10)(11)(12)18). This variation was likely attributed to measurement bias derived from differences in the de nitions used. Nevertheless, the pathogenesis of PEF is still unclear. Most studies believe that lipiodol-induced embolism may lead to ischemia, hypoxia and necrosis of some normal hepatocytes (10). In addition, TACE itself can lead to in ammatory factors release (19), such stimuli can contribute to stress responses in the human body (10).
Recently, studies have found APRI and ALBI to be predictors of postoperative outcome for patients undergoing liver surgery (20). Hence, the ALBI grade and APRI were introduced in this study to manifest or indicate the hepatic function as well as liver brosis and cirrhosis (14).Analysis of 252 patients in this study showed that the incidence of PEF was 17.5%, which was similar to most of previous studies (10,12). Jun et al. retrospectively analyzed 443 HCC patients who underwent the rst session of TACE and found that PEF developed in 117 patients (26.41%). A multivariate analysis using logistic regression showed that ALT value after TACE and the lipiodol dose ≥ 7 mL were independent predictive factors of PEF (10). Shim et al. found that pre-procedure serum bilirubin, ascites, tumor size and female gender predicted PEF in a cohort without background infective hepatitis patient (12). However, more previous study disclosed that a dosage of doxorubicin plus iodized oil > 23 mL during chemoembolization and tumor size > 3 cm were signi cant predictors associated with the development of PEF (18).
We found the occurrence of PEF was closely related to some clinical and laboratory variables. Among which, Iopiodol emulsion dose, number of hepatoprotectants, K + , and ALBI grade were independent risk factors for PEF. The results of cut-off value indicated that when Iopiodol emulsion dose was greater than 6.5 mL, K + was greater than 4.25 mmol/L, ALBI grade was more than 1.5, special attention should be paid to the occurrence of PEF in these patients, and good monitoring and prevention should be done. Besides that, our limited data also indicated that the number of hepatoprotectants might be a protect factor for occurrence of PEF. In additon, the area under the ROC curve of validation cohort was 0.874, which indicated comparative stability and discriminative ability of this predictive model.
Here, we performed a single center, retrospective study and the race were limited to Asian, while it is necessary to validate a predict model against external centers with different geography and races.
Second, most of our patients were accompanied with infection of HBV and liver cirrhosis which was in accordance with the background of high HBV prevalence rate in China. Detection and controlling for population strati cation in association studies of hepatitis patients are needed in the following researches. Third, we did not consider the tumor size's in uence on the PEF for patients' variant situation for surgical or disease progression. In consideration of situation of hepatoprotectants wide use in China, we added the number of hepatoprotectants in the analysis and found it maybe a potential protect factor for PEF. Further, how the hepatoprotectants actually act in PEF still need further well-designed study.

Conclusion
PEF is a common complication in patients with advanced, unresectable HCC. we found that Iopiodol emulsion dose, number of hepatoprotectants, K + , and ALBI grade are strong predictors for PEF. Moving forward, the multivariate logistic model of these factors shows a discriminative ability to predict PEF in the validation cohort.