Atezolizumab plus Bevacizumab with Transcatheter Arterial Chemoembolization (Sandwich Strategy) versus Atezolizumab plus Bevacizumab Alone in Hepatocellular Carcinoma: A Multicenter Retrospective Study Authors: Ko Hashimoto, Tomokazu Kawaoka, Tomoaki Emori, Aiko Tanaka, Yuki Shirane, Ryoichi Miura, Yasutoshi Fujii, Hikaru Nakahara, Kenji Yamaoka, Shinsuke Uchikawa, Hatsue Fujino, Atsushi Ono, Eisuke Murakami, Daiki Miki, C. Nelson Hayes, Akira Hiramatsu, Kei Amioka, Michihiro Nonaka, Yasuyuki Aisaka, Kei Morio, Takashi Moriya, Yuji Teraoka, Hirotaka Kono, Yosuke Suehiro, Keiichi Masaki, Kazuki Ohya, Shintaro Takaki, Nami Mori, Keiji Tsuji, Yumi Kosaka, Takashi Nakahara, Hiroshi Aikata, Masataka Tsuge, Shiro Oka Published: 2025 ABSTRACT AbstractIntroductionThis study aimed to evaluate the effect of sandwiching on-demand transcatheter arterial chemoembolization (TACE) during atezolizumab (Atezo) and bevacizumab (Bev) therapy on overall survival (OS) and progression-free survival (PFS) in patients with unresectable hepatocellular carcinoma (HCC).MethodsWe retrospectively analyzed 398 patients who started Atezo/Bev therapy between October 2020 and April 2024. Overall, 245 patients were included: 51 in the ABC-TACE sandwich group and 194 in the Atezo/Bev alone group. Propensity score matching was performed to balance baseline characteristics, resulting in 49 matched patients per group. OS and PFS were analyzed using the Kaplan-Meier methods, with landmark analysis to adjust for immortal time bias. Multivariate analysis identified independent predictors of OS and PFS.ResultsThe ABC-TACE sandwich group had significantly longer OS (median, not reached vs. 21.0 months, p < 0.05) and PFS (18.7 months vs. 11.2 months, p < 0.05) than the Atezo/Bev alone group. Landmark analysis showed prolonged OS at multiple time points in the ABC-TACE sandwich group (p < 0.05, p < 0.05, p = 0.078). TACE independently contributed to both OS and PFS. No significant differences in adverse events were observed between the groups. No deterioration in the Child-Pugh score was observed before and after TACE (p = 0.976).ConclusionSandwiching TACE during Atezo/Bev therapy may improve survival outcomes for unresectable HCC. Introduction Hepatocellular carcinoma (HCC) is the most common primary malignancy of the liver and the leading cause of cancer-related death worldwide [1]. The Barcelona Clinic Liver Cancer (BCLC) staging system is the most widely used treatment algorithm globally. Patients with intermediate-stage (BCLC stage B) HCC are recommended to undergo transcatheter arterial chemoembolization (TACE), while those with advanced-stage (BCLC stage C) are treated with systemic therapy [2]. However, it has been suggested that BCLC stage B patients who are considered unsuitable for TACE, especially beyond up to seven criteria, should initially receive pharmacotherapy, with TACE added in some cases [3, 4]. After atezolizumab plus bevacizumab (Atezo/Bev) was shown to be superior to sorafenib, it became the first-line systemic therapy for unresectable HCC worldwide [5]. Given that Atezo/Bev alone has a response rate of only 30%, with overall survival (OS) and progression-free survival (PFS) of 19.2 and 6.9 months, respectively, improved systemic therapy strategies are warranted. It has been reported that TACE induces tumor hypoxia, which releases vascular endothelial growth factor and promotes tumor angiogenesis [6]. Furthermore, TACE is expected to release tumor antigens, enhancing the efficacy of immune checkpoint inhibitors (ICIs) by renewing the expression of CD8+ T cells and PD-L1 [7]. Studies such as the LAUNCH study have shown that lenvatinib-TACE (LEN-TACE) prolongs OS in patients with BCLC stage B and stage C, compared to LEN alone [8–12]. A limited number of studies have similarly suggested the clinical utility of combining TACE with Atezo/Bev. Moriyama et al. [13] and Cao et al. [14] retrospectively evaluated the combination of Atezo/Bev with TACE and demonstrated improved survival compared to Atezo/Bev alone. More recently, the IMPACT study (jRCTs051230037) has been initiated; this prospective randomized controlled study aims to evaluate the efficacy of incorporating TACE during Atezo/Bev therapy [15]. This reflects the growing clinical interest in this combination strategy. Kudo [16] has described a treatment strategy in which TACE is incorporated when tumor shrinkage cannot be achieved with atezolizumab plus bevacizumab alone, referring to it as ABC-TACE sandwich therapy. In the present study, we conducted an analysis to evaluate the effectiveness of this ABC-TACE sandwich therapy. To reduce selection bias, we performed propensity score matching. Because sandwich TACE was given on demand, immortal time bias could occur, so we used landmark analysis to minimize this bias. Methods Patients and Study Design Retrospective data were collected from multiple hospitals, including Hiroshima University Hospital, Hiroshima Prefectural Hospital, Hiroshima Red Cross Hospital, Hiroshima City North Medical Center Asa Citizens Hospital, NHO Kure Medical Center, Chugoku Rosai Hospital, JA Hiroshima General Hospital, NHO Higashihiroshima Medical Center, and Hiroshima Memorial Hospital. A total of 398 patients with unresectable HCC who started Atezo/Bev therapy between October 2020 and April 2024 were included. We excluded 35 patients without intrahepatic lesions at the start of therapy and 5 patients with an observation period of less than 6 weeks. To focus on patients for whom TACE could be considered as an additional treatment during Atezo/Bev therapy, we excluded 92 patients who showed progressive disease (PD) or were not evaluable based on Response Evaluation Criteria in Solid Tumors (RECIST) version 1.1 at the first response evaluation and 16 patients who achieved complete response (CR) based on modified RECIST (mRECIST) at the second response evaluation [17]. Among them, 57 patients were treated with Atezo/Bev therapy in combination with TACE or transhepatic arterial infusion (TAI), while 194 patients received Atezo/Bev alone. We defined the ABC-TACE sandwich group as patients incorporating on-demand TACE during Atezo/Bev therapy. Based on this definition, we excluded 6 patients who received a combination of Atezo/Bev therapy and TAI, resulting in 51 patients classified into the ABC-TACE sandwich group and 194 patients into the Atezo/Bev alone group (Fig. 1). To balance patient backgrounds, propensity score matching was performed, resulting in 49 patients in each group. The primary endpoints were OS and PFS comparisons between the two groups. To minimize immortal time bias, landmark analysis was conducted for OS. In addition, multivariate analysis was used to identify independent factors associated with OS and PFS. We also compared adverse events (AEs) between the two groups and evaluated changes in Child-Pugh score at and after TACE. Flowchart of eligible patients with HCC. Atezo/Bev, atezolizumab/bevacizumab; RECIST, Response Evaluation Criteria in Solid Tumors; PD, progressive disease; NE, not evaluated; mRECIST, modified RECIST; CR, complete response; TAI, transhepatic arterial infusion; TACE, transcatheter arterial chemoembolization. Although the IMPACT trial prospectively investigates the benefit of adding TACE specifically for patients with primary stable disease (SD) under Atezo/Bev therapy [15], the objective and design of our study differ. Our retrospective study focuses on the clinical significance of performing TACE on an on-demand basis during Atezo/Bev treatment, encompassing a broader range of clinical scenarios, including cases aiming for curative intent and local control for slow PD. Atezolizumab plus Bevacizumab Treatment Patients received 1,200 mg of atezolizumab and 15 mg/kg of bevacizumab every 3 weeks. AEs were assessed using the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0, and therapy was discontinued if an unacceptable AE occurred. Atezo/Bev therapy was paused at least 14 days before TACE and resumed at least 7 days after these procedures. TACE The timing and purpose of additional TACE were determined by the treating physician based on the tumor’s response just before TACE, as assessed by RECIST and tumor marker trends. The purpose of TACE was classified into four categories: (1) radical cure for tumors with PR, (2) intrahepatic control for tumors with SD showing a tendency toward shrinkage or decreasing tumor markers, (3) intrahepatic control for tumors with SD showing a tendency toward progression or rising tumor markers, and (4) local control for tumors with PD. The choice of drugs for TACE was selected by the treating physician and institution. Conventional TACE (cTACE) used cisplatin, epirubicin, or doxorubicin suspended in lipiodol, which was selectively injected into the tumor vessels and embolized with gel particles. Drug-eluting bead TACE (DEB-TACE) was performed either by loading epirubicin onto DC Bead™ or by loading cisplatin onto Hepasphere. Eleven patients were treated with LEN-TACE. Statistical Analysis All statistical analyses were performed using R version 4.3.1. Propensity score matching was used to align background factors and ensure a 1:1 ratio between the ABC-TACE sandwich and Atezo/Bev alone groups. The matched data included age, gender, treatment line, Child-Pugh score, BCLC stage, Eastern Cooperative Oncology Group performance status (ECOG PS), presence of distant metastases, presence of vascular invasion, maximum tumor diameter, etiology, AFP, and des-γ-carboxy prothrombin (DCP). Fisher’s exact test and the Mann-Whitney U test were used to compare patient backgrounds. The Kaplan-Meier method was used to analyze OS and PFS, and the log-rank test was used to compare survival distributions. Cox proportional-hazards regression was used for univariate and multivariate analyses. Fisher’s exact test was used to compare AEs, and repeated measures analysis of variance was used to analyze changes in hepatic reserve. A p value of <0.05 was considered significant. Results Patient Characteristics A retrospective flowchart is presented in Figure 1. After excluding patients who met the exclusion criteria, the remaining 245 patients were classified into the ABC-TACE sandwich group (n = 51, 20.8%) and the Atezo/Bev alone group (n = 194, 79.2%). Table 1 presents the patient characteristics for the two groups, showing a significant difference in Child-Pugh scores and in tumor number between the two groups (p < 0.05, p < 0.05). After propensity score matching, Table 2 shows the patient characteristics of the two groups, which included 13 of 49 patients (26.5%) in the ABC-TACE sandwich group and 15 of 49 patients (30.6%) in the Atezo/Bev alone group with BCLC C. Table 1.Baseline characteristics of the 245 patients before propensity score matchingCharacteristicABC-TACE sandwich (n = 51)Atezo/Bev alone (n = 194) p valueAge, years74 (49–92)75 (50–93)0.510Sex (male/female)44/7155/390.420Treatment line (1st/2nd or later)36/15142/510.724Child-Pugh score (A/B/C)49/2/0149/41/4 <0.05 BCLC stage (A/B/C)7/31/1323/88/830.066Performance status (0/1/2/3)44/6/1/0154/34/5/10.778EHS (without/with)43/8139/550.073MVI (without/with)44/7153/410.321Tumor size, mm40 (10–160)34.2 (5.4–220)0.212Tumor number (mono/multi)2/4936/158 <0.05 Etiologn (NBNC/hepatitis virus)34/17114/800.337AFP, ng/mL12 (1.4–73465)15.5 (1–307,600)0.416DCP, mAU/mL422 (14–119,154)735 (9.5–572,700)0.385Continuous variables are shown as medians (ranges). Atezo/Bev, atezolizumab/bevacizumab; TACE, transcatheter arterial chemoembolization; BCLC, Barcelona Clinic Liver Cancer; performance status, Eastern Cooperative Oncology Group Performance Status; EHS, extrahepatic spread; MVI, macrovascular invasion; NBNC, non-B, non-C hepatocellular carcinoma; AFP, α-fetoprotein; DCP, des-γ-carboxy prothrombin. Baseline characteristics of the 245 patients before propensity score matching Continuous variables are shown as medians (ranges). Atezo/Bev, atezolizumab/bevacizumab; TACE, transcatheter arterial chemoembolization; BCLC, Barcelona Clinic Liver Cancer; performance status, Eastern Cooperative Oncology Group Performance Status; EHS, extrahepatic spread; MVI, macrovascular invasion; NBNC, non-B, non-C hepatocellular carcinoma; AFP, α-fetoprotein; DCP, des-γ-carboxy prothrombin. Table 2.Baseline characteristics of the patients after propensity score matchingCharacteristicABC-TACE sandwich (n = 49)Atezo/Bev alone (n = 49) p valueAge, years74 (55–89)76 (50–90)0.659Sex (male/female)42/742/71Treatment line (1st/2nd or later)35/1433/160.827Child-Pugh score (A/B)47/249/00.495BCLC stage (A/B/C)7/29/136/28/150.919Performance status (0/1/2)42/6/141/6/21EHS (without/with)41/840/91MVI (without/with)42/741/81Tumor size, mm40 (10–160)30 (9–136)0.311Tumor number (mono/multi)2/472/471Etiology (NBNC/hepatitis virus)33/1638/110.366AFP, ng/mL12 (1.7–73,465)18.1 (1–276,110)0.717DCP, mAU/mL706 (14–119,154)735 (13–67,884)0.837Continuous variables are shown as medians (ranges). Atezo/Bev, atezolizumab/bevacizumab; TACE, transcatheter arterial chemoembolization; BCLC, Barcelona Clinic Liver Cancer; performance status, Eastern Cooperative Oncology Group Performance Status; EHS, extrahepatic spread; MVI, macrovascular invasion; NBNC, non-B, non-C hepatocellular carcinoma; AFP, α-fetoprotein; DCP, des-γ-carboxy prothrombin. Baseline characteristics of the patients after propensity score matching Continuous variables are shown as medians (ranges). Atezo/Bev, atezolizumab/bevacizumab; TACE, transcatheter arterial chemoembolization; BCLC, Barcelona Clinic Liver Cancer; performance status, Eastern Cooperative Oncology Group Performance Status; EHS, extrahepatic spread; MVI, macrovascular invasion; NBNC, non-B, non-C hepatocellular carcinoma; AFP, α-fetoprotein; DCP, des-γ-carboxy prothrombin. Follow-Up Period and Best Response The median follow-up period was 17.5 months (range, 4.2–45.1) in the ABC-TACE sandwich group and 16.6 months (range, 1.9–36.6) in the Atezo/Bev alone group. The distribution of best responses (CR/PR/SD/PD) was 0/20/29/0 in the ABC-TACE sandwich group and 0/16/33/0 in the Atezo/Bev alone group (p = 0.53). Notably, the best responses in the ABC-TACE sandwich group were assessed prior to the administration of TACE. TACE Details The details of initial TACE in the ABC-TACE sandwich group are as follows. Of the 49 patients, 45 were treated with cTACE and 4 with DEB-TACE. Lenvatinib was administered prior to TACE in 10 of the 45 patients who received cTACE and in 1 of the 4 patients who received DEB-TACE. Cisplatin was used in 36 patients, epirubicin in 9 patients, and doxorubicin in 4 patients. The median time from the start of Atezo/Bev therapy to the first TACE was 6.8 months (IQR, 4.1–11.6). The number of times TACE was performed was as follows: 32 patients (65.3%) underwent TACE once, 11 patients (22.4%) underwent the procedure twice, and 6 patients (12.2%) underwent TACE three times. Fourteen patients (28.6%) underwent TACE for: (1) radical cure for tumors with PR, 7 patients (14.3%) for (2) intrahepatic control for tumors with SD showing a tendency toward shrinkage or decreasing tumor markers, 22 patients (44.9%) for (3) intrahepatic control for tumors with SD showing a tendency toward progression or rising tumor markers, and 6 patients (12.2%) for (4) local control for tumors with PD after Atezo/Bev therapy. Treatment Responses, OS, and PFS The OS in the ABC-TACE sandwich group (median, not reached; 95% CI: 26.6 months – not reached) was longer than that in the Atezo/Bev alone group (median, 21.0 months; 95% CI: 16.6–28.7 months; p < 0.05) (Fig. 2a). PFS in the ABC-TACE sandwich group (median, 18.7 months; 95% CI: 12.6–20.7 months) was longer than that in the Atezo/Bev alone group (median, 11.2 months; 95% CI: 7.6–16.4 months; p < 0.05) (Fig. 2b). Kaplan-Meier curves for OS (a) and PFS (b) among patients after propensity score matching. The median OS for the ABC-TACE sandwich group was not reached (26.6 – not reached), and that for the Atezo/Bev alone group was 21.0 (16.6–28.7) months, respectively (p < 0.05, a). The median PFS was 18.7 (12.6–20.7) months for the ABC-TACE sandwich group and 11.2 (7.6–16.4) months for the Atezo/Bev alone group (p < 0.05, b). Atezo/Bev, atezolizumab/bevacizumab; TACE, transcatheter arterial chemoembolization; OS, overall survival; PFS, progression-free survival. To account for immortal time bias in the ABC-TACE sandwich group, landmarks were set at 6 months (the median time of TACE), as well as 3 months before and after the median time of TACE. Patients whose OS did not reach each landmark were excluded to compare OS in both groups (Fig. 3a–c). At the 3- and 6-month landmark analyses, OS in the ABC-TACE sandwich group was significantly prolonged compared to OS in the Atezo/Bev alone group (p < 0.05, p < 0.05, respectively). A similar but non-significant trend was observed at the 9-month landmark (p = 0.078). Kaplan-Meier curves for OS in landmark analysis. Landmarks were set at 6 months (the median time of TACE), as well as 3 months before and after the median time of TACE (a–c). a The median OS for the ABC-TACE sandwich group was not reached (26.6 – not reached), while the median OS for the Atezo/Bev alone group was 21.2 (16.9–28.7) months (p < 0.05). b The median OS for the ABC-TACE sandwich group was not reached (29.8 – not reached), while the median OS for the Atezo/Bev alone group was 22.1 (16.9 – not reached) months (p < 0.05). c The median OS for the ABC-TACE sandwich group was not reached (29.8 – not reached), while the median OS for the Atezo/Bev alone group was 23.2 (20.8 – not reached) months (p = 0.078). OS, overall survival; TACE, transcatheter arterial chemoembolization; Atezo/Bev, atezolizumab/bevacizumab. Independent Factors Contributing to OS and PFS Independent factors contributing to OS and PFS were examined. Univariate analysis was performed for the following variables: sex (male/female), age (<75/≥75 years), treatment line (1st/2nd or later), Child-Pugh score (A/B), BCLC stage (AB/C), performance status (0/1–2), extrahepatic spread (without/with), macrovascular invasion (without/with), tumor size (<35/≥35 mm), tumor number (mono/multi), etiology (non-B, non-C hepatocellular carcinoma/hepatitis virus), AFP (<200/≥200 ng/mL), DCP (<1,000/≥1,000 mAU/mL), and TACE (without/with). Multivariate analysis was performed for factors contributing to OS with a p value <0.05 in univariate analysis, including BCLC stage, macrovascular invasion, DCP, and TACE. Multivariate analysis showed that DCP (HR, 0.488; 95% CI: 0.250–0.955; p < 0.05) and TACE (HR, 0.375; 95% CI: 0.193–0.728; p < 0.05) were independent factors contributing to OS (Fig. 4a). Factors contributing to PFS with a p value <0.05 in univariate analysis included treatment line and TACE. Multivariate analysis showed that treatment line (HR, 0.571; 95% CI: 0.337–0.967; p < 0.05) and TACE (HR, 0.413; 95% CI: 0.239–0.712; p < 0.05) were independent factors contributing to PFS (Fig. 4b). Forest plots of multivariate analyses. a Factors associated with OS. HR and 95% CI are shown for BCLC stage, MVI, DCP, and TACE. DCP (HR, 0.488; 95% CI: 0.250–0.955; p < 0.05) and TACE (HR, 0.375; 95% CI: 0.193–0.728; p < 0.05) were identified as independent factors for OS. b Factors associated with PFS. Treatment line and TACE were included in the model. Both were independent factors: treatment line (HR, 0.571; 95% CI: 0.337–0.967; p < 0.05) and TACE (HR, 0.413; 95% CI: 0.239–0.712; p < 0.05). OS, overall survival; HR, hazard ratio; CI, confidence interval; BCLC, Barcelona Clinic Liver Cancer; MVI, macrovascular invasion; DCP, des-γ-carboxy prothrombin; TACE, transcatheter arterial chemoembolization; PFS, progression-free survival. Description of the Next Treatment following Atezo/Bev Therapy Next-line treatment details for Atezo/Bev therapy are provided. Twenty-six patients in the ABC-TACE sandwich group and 29 patients in the Atezo/Bev alone group progressed to the next treatment. Of these, 20 and 20 patients were treated with systemic therapy, respectively; 3 and 7 with TACE (TAI); and 1/1/1 and 0/1/1 with microwave ablation, stereotactic body radiation therapy, or surgery. Adverse Events To compare the safety of the ABC-TACE sandwich group and the Atezo/Bev alone group, the frequency and severity of treatment-related AEs were analyzed based on CTCAE version 5.0 (Table 3). The most common AEs included high blood pressure, fatigue, diarrhea, loss of appetite, itching, liver damage, and proteinuria. However, there were no significant differences between the ABC-TACE sandwich group and the Atezo/Bev alone group. Table 3.Comparison of AEs in ABC-TACE sandwich group and Atezo/Bev alone groupAEsABC-TACE sandwich (n = 49)Atezo/Bev alone (n = 49)any grade, n (%)grade ≥3, n (%)any grade, n (%)grade ≥3, n (%)High blood pressure22 (44.9)5 (10.2)17 (34.7)5 (10.2)Fatigue11 (22.4)2 (4.1)12 (24.5)2 (4.1)Diarrhea9 (18.4)0 (0.0)7 (14.3)2 (4.1)Loss of appetite8 (16.3)1 (2.0)12 (24.5)0 (0)Itching7 (14.3)1 (2.0)15 (30.6)2 (4.1)Liver damage6 (12.2)1 (2.0)9 (18.4)2 (4.1)Proteinuria5 (10.2)3 (6.1)11 (22.4)2 (4.1)Hyperammonemia3 (6.1)1 (2.0)4 (8.2)0 (0.0)Hypothyroidism3 (6.1)0 (0.0)4 (8.2)0 (0.0)Adrenal insufficiency3 (6.1)1 (2.0)0 (0.0)0 (0.0)Gastrointestinal perforation2 (4.1)1 (2.0)0 (0.0)0 (0.0)Edema1 (2.0)0 (0.0)7 (14.3)3 (6.1)Gastrointestinal bleeding1 (2.0)1 (2.0)3 (6.1)1 (2.0)Kidney damage1 (2.0)0 (0.0)3 (6.1)0 (0.0)Hyperthyroidism1 (2.0)0 (0.0)1 (2.0)0 (0.0)Interstitial pneumonia1 (2.0)0 (0.0)1 (2.0)0 (0.0)Rash0 (0.0)0 (0.0)3 (6.1)0 (0.0)AEs, adverse events; Atezo/Bev, atezolizumab/bevacizumab; TACE, transcatheter arterial chemoembolization. Comparison of AEs in ABC-TACE sandwich group and Atezo/Bev alone group AEs, adverse events; Atezo/Bev, atezolizumab/bevacizumab; TACE, transcatheter arterial chemoembolization. Figure 5 shows the changes in Child-Pugh scores before and after TACE: baseline, at TACE, 3 months after TACE, and 6 months after TACE. No deterioration in the Child-Pugh score was observed at and after TACE (p = 0.976). Changes in Child-Pugh score before and after TACE: baseline, at TACE, 3 months after TACE, and 6 months after TACE. No deterioration in the Child-Pugh score was observed at and after TACE (p = 0.976). TACE, transcatheter arterial chemoembolization. Discussion In the present study, we retrospectively examined the OS and PFS of patients with unresectable HCC who were treated with Atezo/Bev, comparing two groups: the ABC-TACE sandwich group and the Atezo/Bev alone group, with selection bias minimized by propensity score matching. The results showed that the ABC-TACE sandwich group was superior in prolonging OS and PFS, with these findings confirmed by landmark analysis. This suggests that the addition of TACE during Atezo/Bev therapy may offer significant survival benefits. Many studies relating to Atezo/Bev have been published. Kudo et al. [18] showed that conversion is more effective in PR. There are also reports on the characteristics of cases achieving CR, as well as conversion treatments after Atezo/Bev therapy aimed at achieving intrahepatic freedom [19, 20]. We have reported that an increase in AFP in patients with an initial response score of SD results in a PD in 44% of these patients as the next response score [21]. Similarly, there are reports that early changes in AFP are useful for predicting the antitumor effects of Atezo/Bev therapy [22]. Although our study was retrospective and did not directly examine the transition of AFP as an indication for TACE, the results suggest that adding TACE may help delay progression and improve survival outcomes. One of the objectives of the ongoing IMPACT study is to determine whether the addition of intrahepatic control TACE in patients who achieve SD after Atezo/Bev therapy can lead to improved prognosis [15]. In our study, although the timing and purpose of TACE varied, a substantial proportion of TACE procedures were performed in cases classified as SD, particularly those showing signs of progression or rising tumor markers. Therefore, our findings may provide preliminary real-world support for the therapeutic concept being tested in the IMPACT study and highlight the potential value of timely TACE intervention in selected patients receiving Atezo/Bev. The timing of TACE administration also warrants discussion. In the IMbrave150 study, the median time to response was reported to be 2.8 months, and patients who achieved a response (C.R. or P.R.) had a better prognosis compared to those with SD [5]. However, even in cases with PR or SD, disease progression often occurs later. Therefore, curative conversion strategies may help achieve CR or more durable PR or SD. In our study, the indications for TACE were not limited to curative conversion but included a wide range from SD to slow PD. As a result, the overall timing of TACE tended to be later. In the IMPACT study, based on findings from IMbrave150, the timing for intrahepatic control TACE was predefined at 3 months, which may serve as a useful reference for determining the optimal timing of TACE in future clinical practice. Several clinical trials are currently underway to evaluate strategies combining ICIs, anti- vascular endothelial growth factor/TKIs, and TACE. In LEAP-012 and EMERALD-1 trials, a change in treatment regimen is mandated upon disease progression [23, 24]. However, based on findings from the TACTICS and TACTICS-L trials, repeated TACE may still be feasible in cases of local progression or the emergence of new intrahepatic lesions after initial TACE [25, 26]. Notably, the TALENTACE trial adopts TACE-specific PFS as its primary endpoint, thereby defining disease progression as the point at which further TACE becomes unfeasible. In this context, the continuation of ABC-TACE sandwich therapy even in patients with SD or slow PD may prolong TACE-specific PFS and potentially translate into OS benefits in real-world practice [27]. This study included cases in which ABC-TACE sandwich therapy was continued even in patients with SD or slow PD, thus providing results that more closely reflect real-world clinical practice. Moreover, it is well documented that TACE can promote the release of tumor antigens, which may enhance the immune response and lead to the abscopal effect, in which local treatment of a tumor also potentiates the immune response against untreated tumors, potentially improving the efficacy of ICIs such as Atezo/Bev for other intrahepatic HCCs [6, 7, 28]. While this study did not specifically investigate the abscopal effect, future studies should assess the impact of TACE in conjunction with Atezo/Bev on other intrahepatic lesions. The abscopal effect has also been reported in radiation therapy [29, 30]. The comparison between TACE and stereotactic body radiation therapy in combination with Atezo/Bev remains a relevant question as current literature is lacking in this regard. However, we acknowledge that this study has several limitations. As a retrospective analysis, we could not define the precise timing of TACE in relation to patient response, nor could we control for all confounding factors. Moreover, the decision to perform TACE was made at the discretion of the treating physician, potentially introducing selection bias. Even in cases where there was an intention to resume Atezo/Bev therapy after TACE, if the therapy could not be resumed, those patients were classified into the Atezo/Bev alone group in this retrospective analysis. This introduces a potential bias in group allocation. In addition, missing data inherent to the retrospective design may have influenced the results. The inclusion of a relatively small sample size may have also impacted the generalizability of our findings. Nonetheless, the findings of this study are promising, and we advocate for larger, multicenter, prospective studies to validate these observations and explore the optimal timing and role of TACE in combination with Atezo/Bev therapy. In conclusion, this study positions the ABC-TACE sandwich therapy as a promising treatment strategy for unresectable HCC, providing real-world support for ongoing clinical trials such as the IMPACT and TALENTACE trials. Acknowledgment The authors thank JSPS Program for Forming Japan’s Peak Research Universities (JSPS J-PEAKS). Statement of Ethics This study was conducted in accordance with the ethical principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Hiroshima University (E2022-0293). Written informed consent was obtained from each patient after a detailed explanation of the study procedures. Conflict of Interest Statement Tomokazu Kawaoka received a lecture fee from Chugai. The remaining authors have no conflicts of interest to declare. Funding Sources This work was not supported. Author Contributions Ko Hashimoto and Tomokazu Kawaoka designed the study, analyzed the data, and wrote the paper; Ko Hashimoto, Tomoaki Emori, Aiko Tanaka, Yuki Shirane, Ryoichi Miura, Yasutoshi Fujii, Kenji Yamaoka, Shinsuke Uchikawa, Hatsue Fujino, Atsushi Ono, Eisuke Murakami, Daiki Miki, Akira Hiramatsu, Kei Amioka, Michihiro Nonaka, Yasuyuki Aisaka, Kei Morio, Takashi Moriya, Yuji Teraoka, Hirotaka Kono,Yosuke Suehiro, Keiichi Masaki, Kazuki Ohya, Shintaro Takaki, Nami Mori, Keiji Tsuji, Yumi Kosaka, Takashi Nakahara, and Hiroshi Aikata acquired the data; Ko Hashimoto and Hikaru Nakahara performed the statistical analysis; Tomokazu Kawaoka reviewed the results; C. Nelson Hayes supervised English literature; and Tomokazu Kawaoka, Masataka Tsuge, and Shiro Oka revised the manuscript for important intellectual content.