Drug-eluting beads versus conventional transarterial chemoembolization for the treatment of unresectable hepatocellular carcinoma: A meta-analysis

Abstract

Background:

Transarterial chemoembolization (TACE) consists of conventional TACE (cTACE) and drug-eluting beads TACE (DEB-TACE). The benefits of the 2 treatments remain controversial. We conduct this meta-analysis to assess the efficacy and safety of the 2 methods for the patients with unresectable hepatocellular carcinoma.

Methods:

In order to get a sound conclusion, we did thorough search all relevant studies with clear and stringent keyword criteria on the main databases. Objective tumor response rate, overall survival (OS) rate and adverse events were calculated and analyzed by RevMan 5.3 software. The random-effects or fixed-effects model was applied to pool the estimates according to Cochran Q test and I² statistics.

Results:

Twenty-four studies involving 2987 patients were eligible. DEB-TACE significantly improved objective tumor response rate (OR) (risk ratio [RR] = 1.27, 95% confidence interval [CI] [1.08, 1.48]; P = .003). While as for 1-year, 2-year, 3-year, 5-year OS rates, there were no evidences to indicate that DEB-TACE was significantly better than cTACE (RR = 1.05, 95% CI [0.99, 1.11]; P = .08), (RR = 1.02, 95% CI [0.93, 1.11]; P = .68), (RR = 0.92, 95% CI [0.77, 1.10]; P = .37), (RR = 0.92, 95% CI [0.47, 1.80]; P = .81), respectively. Adverse events rate (AE) was also similar in both groups (RR = 1.11, 95% CI [0.99,1.26]; P = .08).

Conclusion:

This meta-analysis demonstrates that DEB-TACE is not superior than cTACE regarding to OS and AE. However, DEB-TACE still be considered to provide a better objective tumor response rate for patients with unresectable hepatocellular carcinoma.

1. Introduction

Currently, hepatocellular carcinoma (HCC) has become the sixth most common cancer worldwide with approximately 600,000 new cases increasing per year and the third leading cause of cancer-related death in the world.^[1,2] Taking a review of the treatment of HCC, the curative therapies are mainly surgical resection, radiofrequency ablation and liver transplantation, which are mainly suitable for patients with early stage HCC. However, once people are diagnosed with HCC, about 70% of them have developed late-stage refractory tumor, with a median survival of only a few months. They have less curative treatment options.^[3–5] Transarterial chemoembolization (TACE) as one of the most important therapies for the advanced HCC patients is widely used in clinical practice. Conventional TACE (cTACE) is performed with injection of an emulsion of chemotherapeutic drug with lipiodol into the tumor-feeding artery. It can make a chemotherapy isolation in the tumor site and block tumor’s blood supply. Although many studies have demonstrated that cTACE provided a significant survival benefit for advanced HCC patients, severe post-TACE complications still exist, such as acute liver, renal failure and upper gastrointestinal bleeding.^[6] In order to improve the concentration of chemotherapeutic agents in the tumor lesions and improve the chemotherapeutic effects, dug-eluting beads TACE (DEB-TACE) has been introduced into clinical practice for the treatment of unresectable HCC during these years.^[7] DEB-TACE uses microspheres as embolic agents, which will deliver higher chemotherapeutic agent and prolong the time of contact with target tumor.^[8,9] Some clinical trials have shown that DEB-TACE reduced the rate of systemic doxorubicin concentration and improved the higher intratumor drug concentration.^[10–13] Despite the seemingly promising results of earlier research, whether DEB-TACE is superior to cTACE in efficacy and safety is still controversial for unresectable HCC. Therefore, the aim of this meta-analysis was concluded to investigate and compare the efficacy and safety between DEB-TACE and cTACE in unresectable HCC.

2. Methods

2.1. Studies strategy and including criteria

The methods used in this study included a computerized bibliographic search. We followed the protocol reported previously for reporting this meat-analysis.^[14,15] The search was performed on PubMed, Web of Science, Embase databases by 2 authors independently, using the following key words: (“advanced hepatocellular carcinoma” OR “HCC” OR “liver cell carcinoma” OR “unresectable hepatocellular carcinoma”) OR (“cancer” OR “tumor”); (“DEB-TACE” OR “drug-eluting microsphere”); (“cTACE” OR “traditional transcatheter arterial chemoembolization”) and (“survival rate” OR “OS”). In order to identify possible additional studies, we performed complementary manual search by checking the references of all the main review articles on this topic.

Eligible studies were randomized controlled trials (RCTs), prospective and retrospective cohort studies meeting the inclusion criteria: all the patients were diagnosed with advanced liver cancer by clinical or pathology diagnosis, and they lost the chance of surgery or refused to operation; using the method of random assignment or according to the wishes of patients divided into cTACE treatment group and DEB-TACE group; reported at least one of following data: survival rate, objective tumor response rate and adverse events (AE) rate; and articles written in English. Excluding criteria: non-human or experiment-based studies, or non-research-based articles, or raw data not suitable for computational analysis.

2.2. Data extraction and quality assessment

From each retrieved article, 2 researchers independently extracted patient characteristics, trial characteristics and outcome, using an extraction form, according to including and excluding criteria. Disagreements were solved by consensus, and if disagreements persisted, by a third investigator. We evaluated the quality of RCTs and other observational studies by the JADAD scale^[16] and Newcastle-Ottawa scale,^[17] respectively.

2.3. Statistical analysis

We used the RevMan 5.3 statistical software provided by Cochrane collaboration network to perform the meta-analysis. Data of overall survival (OS), OR, AE, were pooled and analyzed by using risk ratio (RR) and 95% confidence intervals (CI). Statistical heterogeneity was evaluated using Cochran Q test and I² statistics.^[18] The P value of Q test < 0.10 or I² value > 50% indicated substantial heterogeneity among those articles, a random-effects model was applied to pool the estimates, otherwise, a fixed-effects model was used.^[19] Once heterogeneity was noted, we would use subgroup analysis by stratifying original estimates according to study characteristics (RCTs vs non-RCTs [NRCT]) or sensitivity analysis to assess potential sources of heterogeneity. Publication biases was assessed by funnel plots.

3. Results

3.1. Eligible Studies

Six hundred and eighty-eight potential articles from databases and 10 studies through manual searching were included. After reading the articles and titles, 189 studies which were duplicate reports were excluded. Three hundred and forty-three studies were eliminated, because they were animal studies, reviews, expert opinions, comments or conference reports. The remaining 166 studies were subjected to a full-text review. Seventy-two articles absence of available date and 70 studies failed comparing DEB-TACE with cTACE were excluded. Finally, a total of 24 studies with 2987 patients included. Figure 1 showed the detailed process of the studies selection.

Figure 1.

Eligibility of studies for inclusion in this meta-analysis.

3.2. Study characteristics

Main characteristics of the included studies are reported in Table 1, Figures 2 and 3. The parallel group designs comparing DEB-TACE with cTACE for HCC were performed in all studies, which were published from 2000–2001 to 2018–2012. Among them, 4 studies were RCTs,^[20–22,24] 6 were prospective cohorts^{[23,27,33,35,39,42]} and 14 were retrospective cohort reports.^{[25,26,28–32,34,36–38,40,41,43]} The great majority of patients were in Child–Pugh A/B and BCLC A/B stage. None of the studies showed statistically significant differences in baseline demographic, clinical and tumoral parameters between the 2 groups. The objective tumor response rate (OR) was defined as the percentage of patients of complete response plus partial response.

Table 1.

Characteristics and quality of studies comparing DEB-TACE with cTACE in patients with unresectable HCC.

Study (yr)	ARM	Sample size	Age	Child–Pugh (A/B/C)	BCLC (A/B/C)	Etiology (HBV/HCV)	Study type	Quality
Van MH, et al 2011[20]	DEB-TACE CTACE	16 14	67.3 ± 9.8 56.±13.4	14/2/0 14/0/0	41/26/22 41/23/24	4/4 4/0	RCT	M
Sacco R, et al 2011[21]	DEB-TACE CTACE	33 34	71.3 ± 7.2 68.7 ± 8.1	29/4/0 25/9/0	22/11/0 22/12/0	4/22 4/25	RCT	M
Lammer J, et al 2010[22]	DEB-TACE CTACE	102 110	67.0 ± 9.2 67.3 ± 8.8	77/16/0 89/19/0	24/69/0 29/79/0	14/20 13/12	RCT	M
Recchia F, et al 2012[23]	DEB-TACE CTACE	35 70	72 70	NA NA	NA NA	NA NA	P	M
Golfieri R, et al 2014[24]	DEB-TACE CTACE	89 88	68.9 ± 8.0 68.3 ± 8.0	75/14 71/11	41/26/22 41/23/24	NA NA	RCT	H
Dhanasekaran R, et al 2010[25]	DEB-TACE CTACE	45 26	60.±11.5 59.±13.3	22/11/12 11/11/4	NA NA	5/20 3/14	R	H
Wiggermann P, et al 2011[26]	DEB-TACE CTACE	22 22	70.3 ± 7.06 67.7 ± 9.0	22/0/0 22/0/0	1/17/3 4/15/2	NA NA	R	H
Ferrer Puchol MD, et al 2011[27]	DEB-TACE CTACE	47 25	69.3 ± 11.8 68.4 ± 8.5	NA NA	NA NA	NA NA	P	M
Song MJ, et al 2012[28]	DEB-TACE CTACE	60 69	61.7 ± 9.8 59.4 ± 11.2	56/4/0 62/6/0	27/33/0 28/41/0	44/8 46/8	R	H
Kloeckner R, et al 2015[29]	DEB-TACE CTACE	76 174	NA NA	51/22/3 103/64/7	8/34/30 30/59/77	10/18 7/7	R	M
Petruzzi NJ, et al 2013[30]	DEB-TACE CTACE	33 59	61 66	19/10/4 43/16/0	NA NA	6/20 9/40	R	M
Duan F, et al 2016[31]	DEB-TACE CTACE	47 159	64 66	35/10/2 113/41/5	9/35/3 45/101/13	7/34 48/68	R	M
Arabi M, et al 2015[32]	DEB-TACE CTACE	35 19	67.1 ± 9.6 66.7 ± 9.6	24/11/0 17/2/0	24/11/0 17/2/0	10/19 7/8	R	H
Facciorusso A, et al 2016[33]	DEB-TACE CTACE	145 104	67 67	129/16/0 93/11/0	58/81/6 41/63/0	29/86 41/63	P	M
Liu YS, et al 2015[34]	DEB-TACE CTACE	53 64	64.6 ± 11.2 66.2 ± 9.6	0/53/0 6/58/0	53/0/0 64/0/0	26/15 26/28	R	L
Malagari K, et al 2012[35]	DEB-TACE CTACE	41 43	70.4 ± 7.4 70.4 ± 7.4	23/18/0 26/17/0	26/17/0 28/15/0	NA NA	P	M
Massani M, et al 2017[36]	DEB-TACE CTACE	28 54	68.2 ± 11.21 72.24 ± 6.77	24/4/0 45/9/0	3/4/21 10/27/17	5/9 6/16	R	M
Do Minh D, et al 2017[37]	DEB-TACE CTACE	26 122	61.4 57.3	NA NA	NA NA	NA NA	R	M
Wu B, et al 2018[38]	DEB-TACE cTACE	24 30	56.2 ± 7.47 52.83 ± 6.13	10/14/0 14/16/0	13/11/0 17/13/0	15/5 18/4	R	M
Vogel TJ, et al 2011[39]	DEB-TACE cTACE	93 108	67.0 ± 9.2 67.3 ± 8.8	26/76/0 89/19/0	77/16/0 29/81/0	14/20 13/12	P	M
Scartozzi M, et al 2010[40]	DEB-TACE cTACE	32 50	68 (41–79) 74 (42–89)	14/18/0 26/24/0	13/19 19/21	NA NA	R	M
Zhou GH, et al 2018[41]	DEB-TACE cTACE	99 39	57.9 ± 10.3 58.2 ± 10.0	91/8 NA	24/35/40 NA	94/2 NA	R	M
Eldorry AK, et al 2018[42]	DEB-TACE cTACE	25 25	NA NA	19/6 21/4	3/22 2/23	NA NA	P	M
Liu YS, et al 2018[43]	DEB-TACE cTACE	72 201	61.7 ± 10.3 65.3 ± 10.7	NA NA	16/56 73/128	40/26 87/82	R	L

cTACE = conventional transarterial chemoembolization, DEB-TACE = drug-eluting bead transarterial chemoembolization, H = high quality, HCC = hepatocellular carcinoma, L = low quality, M = moderate quality, NA = not available, P = prospective, R = retrospective, RCT = random-controlled trial.

Figure 2.

Risk of bias summary: review of authors, judgments on the risk of bias of each item in each included study.

Figure 3.

Risk of bias graph: review of authors, judgments on the risk of bias in each item presented as percentages in all included studies, as describe in Figure 2.

3.3. OS rates

One-year OS of 1766 patients from 14 studies was assessed by using Kaplan–Meier curves. As described in Figure 4, a fixed-effects model was used, due to a middle statistical heterogeneity observed (χ² = 22.13, df = 13 (P = .05), I² = 41%). The RR gave a non-significant estimate in favor of DEB-TACE (RR = 1.05, 95% CI [0.99,1.11]; P = .08) than cTACE.

Figure 4.

Forest plot comparing 1-year OS rate between DEB-TACE and cTACE. cTACE = conventional transarterial, DEB-TACE = drug-eluting bead transarterial chemoembolization, OS = overall survival.

Two-year OS was reported in 14 studies. Pooled data from these articles showed no significant difference between the 2 treatments (RR = 1.09, 95% CI [0.95,1.25]; P = .22). A random-effects model was used, because of high heterogeneity among the included studies (χ² = 31.78, df = 13 (P = .003), I² = 59%). Sources of heterogeneity were explored by means of subgroup analysis and sensitivity analysis. Subgroup analysis retrieved separately the results of RCTs and NRCTs. RCTs confirmed no difference between the 2 groups with an RR = 1.06 (0.90–1.24, P = .47), and no evidence of heterogeneity was presented (χ² = 0.01, df = 1 (P = .92), I² = 0%). On the other hand, the meta-analysis of NRCTs reported a non-significant trend in favor of DEB-TACE (RR = 1.11, 95% CI [0.94, 1.33]; P = .22), a funding not highly reliable due to the great level of heterogeneity (χ² = 32.04, df = 11 (P = .008), I² = 66%). So, we conducted a sensitivity analysis by excluding articles which leaded a high heterogeneity of estimate (Fig. 5). Two studies were excluded in the group of NRCTs. Sensitivity analysis of NRCTs confirmed a non-significant superiority in DEB-TACE group (RR = 1.01, 95% CI [0.91, 1.11]; P = .86), and a middle level of heterogeneity was detected (χ² = 13.48, df = 9 (P = .14), I² = 33%). The RR of overall estimate from RCTs and NRCTs was no difference in 2 groups (RR = 1.02, 95% CI [0.93, 1.11]; P = .68) with a low heterogeneity (χ² = 14.28, df = 11, P = .22, I² = 23%).

Figure 5.

Forest plot comparing 2-year OS rate between DEB-TACE and cTACE. cTACE = conventional transarterial chemoembolization, DEB-TACE = drug-eluting bead transarterial chemoembolization, NRCT = non-randomized controlled trial, OS = overall survival, RCT = randomized-controlled trial.

Ten papers assessed data on 3-year OS with no statistically significant difference between the 2 groups (RR = 1.08, 95% CI [0.91, 1.28]; P = .38). Accounting for a high level of heterogeneity (χ² = 25.63, df = 9, P = .002, I² = 65%), meta-analysis was performed by using a random-effects model. Then, we conducted stratifying analysis according to the tumor size, pathological type, clinical baseline characteristics and study characteristics, but we did not gain the new results of OS, due to high heterogeneity and missing data (details not shown), and sensitivity analysis performed by means of eliminating 2 papers. The outcome of 3-OS did not differ significantly with a middle level of heterogeneity (χ² = 12.043, df = 7 (P = .10), I² = 42%) (Table 2). Six articles estimated 5-year OS and confirmed similar efficacy between the 2 groups (RR = 0.92, 95% CI [0.47, 1.80]; P = .81). A random-effects model was used, owing to a high level of heterogeneity (χ² = 14.20, df = 5 (P = .01), I² = 65%). A sensitive analysis was performed, but high heterogeneity did not change.

Table 2.

Risk ratios and heterogeneity of 3-year, 5-year OS rates.

Survival estimate	No. of studies	No. of patients	RR (95% CI)	P	Heterogeneity I2
3-year OS	8	1331	0.92 [0.58, 1.11]	.10	42%
5-year OS	6	1223	0.92 [0.47, 1.80]	.01	65%

CI = confidence interval, OS = overall survival rate, RR = risk ratio.

3.4. Objective tumor response

Fifteen studies provided objective tumor response rate (OR) data, Figure 6. It was classified according to the Modified Response Evaluation Criteria in Solid Tumors for HCC (mRECIST) or the European Association for the study of the LIVER (EASL) criteria. The result suggested that advanced HCC patients with the treatment of DEB-TACE had a significant superiority compared with cTACE (RR = 1.21, 95% CI [1.05, 1.38]; P = .007). As a high heterogeneity was detected (χ² = 59.56, df = 14, P < .00001, I² = 76%), a random-effects model was used. Subgroup analysis was conducted in RCTs and NRCTs. NRCTs confirmed a significant RR in favor of DEB-TACE (RR = 1.32, 95% CI [1.07, 1.62]; P = .01), with a high level of heterogeneity (χ² = 44.99, df = 11, P < .0001, I² = 76%). While RCTs showed a non-significant RR in DEB-TACE (RR = 1.02, 95% CI [0.93, 1.11]; P = .66), with a middle level of heterogeneity (Chi ² = 3.14, df = 2, P = .21, I² = 36%). A sensitivity analysis was performed by excluding each article once per time changed the main summary estimate (Fig. 7). NRCTs reduced heterogeneity by means of eliminating 1 study (χ² = 15.96, df = 10 (P = .10); I² = 37%). Meta-analysis of RR in NRCTs resulted a significant estimate in favor of DEB-TACE (RR = 1.37, 95CI [1.17, 1.61]; P < .0001). The pooled data from all articles were (RR = 1.27, 95% CI [1.08, 1.48]; P = .003), a significant superiority of DEB-TACE over CTACE was registered. No evidence of publication bias was observed by funnel plot (Fig. 8).

Figure 6.

Forest plot comparing objective tumor response rate (OR) between DEB-TACE and cTACE. cTACE = conventional transarterial chemoembolization, DEB-TACE = drug-eluting bead transarterial chemoembolization, NRCT = non-randomized controlled trial, RCT = randomized-controlled trial.

Figure 7.

Forest plot comparing AE between DEB-TACE and cTACE. AE = adverse events, cTACE = conventional transarterial chemoembolization, DEB-TACE = drug-eluting bead transarterial chemoembolization, NRCT = non-randomized controlled trial, RCT = randomized-controlled trial.

Figure 8.

Funnal plot of publication bias of OR between DEB-TACE and cTACE. cTACE = conventional transarterial chemoembolization, DEB-TACE = drug-eluting bead transarterial chemoembolization, RR = risk ratio, SE = standard error.

3.5. Adverse events

Seven studies reported data on AE rate that showed a similar efficacy between the 2 treatments with an RR 0.69, 95% CI [0.38, 1.25]; P = .22. As heterogeneity was high (χ² = 125.29, df = 6, P < .00001; I² = 95%), we used a random-effects model to analyze it. Sensitivity analysis confirmed the non-significant RR (1.11, 95% CI [0.99, 1.26]; P = .08) in favor of DEB-TACE with middle heterogeneity (χ² = 5.66, df = 4, P = .23; I² = 29%). The detailed data of several AEs was described in Table 3, and we also found no significant differences in the post-embolization syndrome (abdominal pain, fever, vomit) (RR = 0.92, 95% CI [0.80, 1.06]; P = .25), and liver failure (RR = 0.84, 95% CI [0.44, 1.61]; P = .60) between the 2 groups. Five articles recorded liver absence data, we used a random-effects model (χ² = 8.47, df = 4, P = .08; I² = 53%). The calculated results confirmed no difference between the 2 arms (RR = 0.51, 95% CI [0.10, 2.53]; P = .41). There were 4 studies reported fatigue results showed similar between 2 groups (RR = 0.89, 95% CI [0.33, 2.35]; P = .81).

Table 3.

Summary of the adverse events between the 2 groups.

Outcomes	No. of studies	No. of patients	RR (95% CI)	P	Heterogeneity I2
Liver absence	3	435	1.06 [0.33, 3.42]	.92	0%
Liver failure	4	285	0.84 [044, 1.61]	.60	0%
Post-embolization syndrome	10	1491	0.92 [0.80, 1.06]	.25	18%
Fatigue	4	613	0.89 [0.33, 2.35]	.81	58%

CI = confidence interval, RR = risk ratio.

4. Discussion

HCC accounts for about 55% of the most common manigant tumors in the digestive system globally.^[44,45] Although the best method of treating liver cancer is radical resection, the 5-year survival rate of the radical resection is only 25% to 50%. Only 18% of the patients could benefit from the operation.^[2,45] cTACE is considered as the main treatment for unresectable advanced HCC. However, many studies reported that DEB-TACE, a novel drug delivery system by using microspheres as embolic agents, is better than cTACE. Unlike cTACE, DEBs can sequester doxorubicin hydrochloride from solution and release it substantially. As a consequence, the toxic reaction would be lower and the efficacy better than cTACE. Hong et al^[11] reported that the blood concentration of chemotherapeutics in whole body was lower with DEB-TACE, most importantly, the blood concentration of chemotherapeutics in tumor area increased slowly and everlastingly in an animal research. Verela et al^[12] also suggested that the DEBs can prolonged time of which to contact chemotherapeutics with tumor. Despite the promising results, comparisons between DEB-TACE and cTACE are still controversial.

In this meta-analysis, we assessed the safety and efficacy between DEB-TACE and cTACE in patients with advanced HCC. A total of 24 clinical studies with 2987 patients were included. Our result suggested that 1-year, 2-year, 3-year and 5-year OS rates between 2 treatments were no difference, as similar as AEs rate. However, this meta-analysis of objective response rate showed significantly increased in DEB-TACE cohort compared with cTACE. According to previous studies, Facciorusso et al^[46] and Chen et al^[47] concluded that DEB-TACE can prolong patients with unresectable HCC in 1, 2, 3-year OS rates. It should be noticed that the pooled estimates with high heterogeneity in these published studies. In the other hand, these studies concluded relatively limited number of articles. In our study, 1, 2 and 3-year OS rates revealed non-significant RR in favor of DEB-TACE with heterogeneity (I² = 41%, 59%, 65%, respectively). In order to obviate low quality data, we adopted sensitivity or/and subgroup analysis. The final results showed that DEB-TACE treatment group did not improve the OS rates compared with traditional TACE with low heterogeneity in our study. Huang et al^[48] reported in their report that DEB-TACE can improve the 1, 2-year OS rates of patients with unresectable HCC. And there was no difference in the 3-year OS rate between the cTACE group and the DEB-TACE treatment group. It’s different from our results, which may be related to the different literature included. They included a total of 6 articles, while we included 14 papers in analyzing the OS rate of 1 and 2 years. The heterogeneity of 1, 2-year OS were 0% and 20%, respectively, which were small in his study. The results of 3-year OS rate in Huang’s report are consistent with ours.^[48] However, there were only 4 articles included in his study. We have a total of 9 articles, and the data were relatively more enough. Five-year OS rate was not reported in their articles, we calculated the result of 5-OS that revealed a result of non-significant superiority in DEB-TACE group. The conclusion is in accordance with clinical experience. Liu et al^[43] reported that DEB-TACE had a nearly similar RR of 5-OS (P = .45) over cTACE.

Some published meta-analysis concluded that the treatment of DEB-TACE for advanced HCC could achieve better objective response rate. In this article, the subgroup of NRCTs confirmed significantly better OR for DEB-TACE compared with cTACE. But this finding was not reliable due to the great level of heterogeneity (P < .0001, I² = 76%). After removing 1 article,^[33] the significant RR in favor of DEB-TACE (RR = 1.27, 95% CI [1.08, 1.48]; P = .003). The subgroup of RCTs with middle heterogeneity did not show any significant differences (RR = 1.02, 95% CI [0.93, 1.11]; P = .66). Despite RCTs reported non-significant difference between the 2 groups, there were 2 studies with a small sample size reported significant differences. Therefore, we adopted the NRCTs result that 2-year OS was significantly higher in the group of DEB-TACE. In the studies of Facciorusso^[46] and Chen,^[47] contrast results were found with high heterogeneity (P = .007, I² = 64%; P = .008, I² = 67%), respectively. Facciorusso et al included a small number of studies. Chen et al did not assessed the included studies.

The data of AEs rate between DEB-TACE and cTACE was similar, which was consistent with the results of the meta-analysis carried out by Huang et al.^[48] The study included in our meta-analysis reported no statistical difference between 2 arms, except for PRECISION V presented significant difference. We noted that the other included articles used lower doses of doxorubicin about 50 to 80 mg in arm of CTACE, but PRECISION V study used high doses of doxorubicin with 100 to 150 mg in cTACE group, which contributed to higher AEs rates.

Our study indicated that the treatment of DEB-TACE cannot prolong OS rate, neither long-term nor short-term, and the 2 treatments have similar degree of safety. Nevertheless, our research concluded that DEB-TACE may have more promising result of objective response than cTACE (P = .009). However, our study may have some limitations. First of all, both prospective and retrospective articles were included which may have introduced selection bias and heterogeneity in outcomes reported. In addition, 5 observational studies included HCC patients with poor liver function (Child–Pugh C). The HCC patients with poor liver function should be offered only best supportive care according to current guidelines. In order to search and analyze possible sources of heterogeneity, subgroup analysis performed according to liver function tumoral stage. Unfortunately, due to the missing data, we could only stratify original estimates into RCTs and NRCTs. Secondly, the number of RCTs is small, and other clinical research from NRCT exited bias, which would influence the pooled data. Thirdly, the reports including our study had small sample size. Overestimation of the treatment effect was more likely in smaller trials compared with larger trials.

In conclusion, this meta-analysis showed that the treatment of DEB-TACE did not provide a better OS rate or a sound degree of safety over cTACE. However, DEB-TACE will provide a better OR for unresectable HCC patients in our study. Due to the weaknesses of our research, our results should be explained with cautions. More RCTs needed to be conducted to explore the efficacy and safety between 2 DEB-TACE and cTACE.

Author contributions

Data curation: Chun-Feng Xie, Chong Zhong.

Writing-original draft: Kun-Liang Feng, Cheng-Ming Xiong, Jun-Hai Huang.

Writing-review & editing: Zi-Yu Wang, Qing-Lian Chen, Zhai-Wen Zhoud.