Risk of Variceal Bleeding in Patients With Advanced Hepatocellular Carcinoma Receiving Atezolizumab/Bevacizumab

Hepatocellular carcinoma (HCC) is a major contributor to the global health burden, affecting more than 800,000 new patients annually and comprising the 4th leading cause of cancer-related mortality.1EASL Clinical Practice Guidelines.J Hepatol. 2018; 69: 182-236Abstract Full Text Full Text PDF PubMed Scopus (3664) Google Scholar, 2https://gco.iarc.fr/today/online-analysis-map?v=2020&mode=population&mode_population=continents&population=900&populations=900&key=asr&sex=0&cancer=11&type=0&statistic=5&prevalence=0&population_groupearth&color_palette=default&map_scale=quantile&map_nb_colors=5&continent=0&rotate=%255B10%252C0%255D IAfRoCGI. 2020Google Scholar, 32018 Korean Liver Cancer AssociationKorean J Radiol. 2019; 20: 1042-1113Crossref PubMed Scopus (114) Google Scholar The combination of atezolizumab and bevacizumab (atezo/bev) was approved as a first-line systemic therapy for advanced HCC;4Finn R.S. et al.N Engl J Med. 2020; 382: 1894-1905Crossref PubMed Scopus (1809) Google Scholar,5Cheng A.L. et al.J Hepatol. 2022; 76: 862-873Abstract Full Text Full Text PDF PubMed Scopus (55) Google Scholar however, adverse events need to be validated in a real-world setting. The safety profile of atezo/bev in advanced HCC was similar to the established safety profile of each agent,4Finn R.S. et al.N Engl J Med. 2020; 382: 1894-1905Crossref PubMed Scopus (1809) Google Scholar,6Ikeda M. et al.Ann Oncol. 2020; 31: S698-S699Abstract Full Text Full Text PDF Google Scholar and exploratory analysis reported that the incidence of adverse events was comparable between high-risk and non-high-risk subgroups of atezo/bev-treated patients.7Finn R.S. et al.Cancer Res. 2021; 81 (CT009-CT)Google Scholar However, patients with Vp4 portal vein tumor thrombosis (PVTT) or with varices experienced more variceal bleeding when treated with atezo/bev than with sorafenib.4Finn R.S. et al.N Engl J Med. 2020; 382: 1894-1905Crossref PubMed Scopus (1809) Google Scholar,5Cheng A.L. et al.J Hepatol. 2022; 76: 862-873Abstract Full Text Full Text PDF PubMed Scopus (55) Google Scholar,8Breder V. et al.J Clin Oncol. 2021; 39: 4073Crossref PubMed Google Scholar,9Merle P. et al.IMbrave150: Efficacy and Safety of Atezolizumab Plus Bevacizumab vs Sorafenib by Baseline Varices Status in Patients with Hepatocellular Carcinoma (HCC).Presented at the 11th Asia-Pacific Primary Liver Cancer Expert Meeting. August 15, 2021; Google Scholar Therefore, we aimed to elucidate the incidence and predictors of variceal bleeding in patients with advanced HCC treated with atezo/bev. We evaluated all patients who initiated atezo/bev from June 1, 2020 to October 31, 2021 and followed the patients until death or November 30, 2021. Data from esophagogastroduodenoscopy (EGD) performed within 1 year before atezo/bev initiation were collected. High-risk varices were defined as grade 2 (moderate) esophageal varices with red signs, grade 3 (severe) esophageal varices, or grade 2 or 3 gastric varices regardless of red signs. Endoscopic variceal ligation (EVL) performed within 1 year before or during the atezo/bev therapy in the absence of bleeding was regarded as prophylactic treatment. We used logistic regression analysis to examine the associations between baseline variables and the presence of high-risk varices and occurrence of variceal bleeding. We also reported nonvariceal bleeding events in patients treated with atezo/bev. We identified 194 patients (Supplementary Table 1). The mean age was 62.1 years (standard deviation, 11.4) and 85.6% were male. The proportion of patients with liver cirrhosis was 74.7%; 164 (84.9%) and 104 (53.6%) patients were Child-Pugh A and albumin-bilirubin grade 1, respectively. Among 102 (52.6%) patients who had baseline EGD (median duration from EGD to atezo/bev, 21 days [interquartile range, 7–45]), 17 (16.7%) patients had high-risk varices. On logistic regression analysis, age (odds ratio [OR], 0.94; 95% confidence interval [CI], 0.90–0.98), platelet count (OR, 0.99; 95% CI, 0.98–1.00), Vp4 PVTT (OR, 4.34; 95% CI, 1.39–13.56), infiltrative tumor (OR, 3.63; 95% CI, 1.27–10.40), and Child-Pugh score (OR, 2.45; 95% CI, 1.51–3.98) were associated with high-risk varices (all P < .05; Table 1).Table 1Predictors of High-Risk Varices and Variceal Bleeding on Univariate Logistic Regression AnalysisaAll patients with high-risk varices on baseline esophagogastroduodenoscopy (if available) had liver cirrhosis. All patients who experienced variceal bleeding had high-risk varices on baseline esophagogastroduodenoscopy (if available) and had liver cirrhosis.CharacteristicsOdds ratio (95% confidence interval)P valueHigh-risk varices Age, per 1 year0.94 (0.90–0.98).008 Male sex3.26 (0.40–26.53).27 Previous curative or locoregional treatment1.05 (0.36–3.08).93 Hepatitis B virus infection1.75 (0.57–5.35).33 Albumin, per 1 g/dL0.15 (0.05–0.46).001 Bilirubin, per 1 mg/dL1.45 (0.95–2.21).09 AST, per 1 IU/L1.00 (1.00–1.01).15 ALT, per 1 IU/L1.00 (0.98–1.01).68 Platelet, per 1000/mm30.99 (0.98–1.00).018 PT, per 1 international normalized ratio1630.28 (33.93–78328.03)< .001 Fibrinogen, per 1 mg/dL1.00 (0.99–1.00).25 Portal vein tumor thrombosis, Vp44.34 (1.39–13.56).012 Infiltrative tumor3.63 (1.27–10.40).017 Extrahepatic spread0.84 (0.30–2.34).74 Child-Pugh score, per 1 point2.45 (1.51–3.98)< .001Variceal bleeding Age, per 1 year0.97 (0.91–1.03).29 Male sex1.19 (0.14–10.05).87 Previous curative or locoregional treatment3.02 (0.36–25.09).31 Hepatitis B virus infection0.85 (0.20–3.69).83 Albumin, per 1 g/dL0.33 (0.09–1.17).09 Bilirubin, per 1 mg/dL1.29 (0.80–2.09).30 AST, per 1 IU/L1.00 (1.00–1.01).35 ALT, per 1 IU/L1.00 (0.98–1.02)> .99 Platelet, per 1000/mm30.99 (0.98–1.00).10 PT, per 1 international normalized ratio17.00 (0.32–916.46).16 Fibrinogen, per 1 mg/dL1.00 (0.99–1.01).84 Prophylactic EVLbIn patients who had high-risk varices on baseline esophagogastroduodenoscopy (if available).0.21 (0.02–2.48).22 Portal vein tumor thrombosis, Vp46.75 (1.58–28.80).010 Infiltrative tumor2.26 (0.55–9.37).26 Extrahepatic spread1.05 (0.24–4.54).95 Child-Pugh score, per 1 point1.52 (0.91–2.55).11ALT, alanine aminotransferase; AST, aspartate aminotransferase; EVL, endoscopic variceal ligation; PT prothrombin time.a All patients with high-risk varices on baseline esophagogastroduodenoscopy (if available) had liver cirrhosis. All patients who experienced variceal bleeding had high-risk varices on baseline esophagogastroduodenoscopy (if available) and had liver cirrhosis.b In patients who had high-risk varices on baseline esophagogastroduodenoscopy (if available). Open table in a new tab ALT, alanine aminotransferase; AST, aspartate aminotransferase; EVL, endoscopic variceal ligation; PT prothrombin time. During a median follow-up of 7.4 months (95% CI, 6.6–8.2), 8 patients (4.1%) developed variceal bleeding (Supplementary Table 1). The median duration from atezo/bev to variceal bleeding was 1.8 months (95% CI, 0.0−4.3). The distribution of the Child-Pugh score and albumin-bilirubin grade between the 2 groups was significantly different, with a higher proportion of Child-Pugh B or C (37.5% vs 14.0%; P = .043) and albumin-bilirubin grade 2 or 3 (87.5% vs 44.7%; P = .043) in the bleeding group than that in the nonbleeding group. We provide characteristics of patients with variceal bleeding, and those with nonvariceal bleeding, in Supplementary Table 2. We subsequently analyzed the impact of baseline variables on variceal bleeding (Table 1). All patients with variceal bleeding had liver cirrhosis and had high-risk varices on baseline EGD, if applicable. Patients with Vp4 PVTT were associated with a higher risk of variceal bleeding (OR, 6.75; 95% CI, 1.58–28.80; P = .010). A prophylactic EVL procedure that was performed in 8 out of 17 (47.1%) patients with high-risk varices was associated with lower point estimate for variceal bleeding risk, although this did not reach statistical significance (OR, 0.21; 95% CI, 0.02–2.48; P = .22). The results of our study might guide risk stratification of variceal bleeding and screening strategy. In our cohort, variceal bleeding did occur in patients with nodular tumors without PVTT and preserved liver function, suggesting that endoscopic evaluation and appropriate prophylaxis should be mandated for all patients before initiating atezo/bev therapy. In resource-limited setting, patients with extensive PVTT, lower platelet counts, higher Child-Pugh score, and those with infiltrative tumors could be prioritized for EGD screening, considering the association between these factors and high-risk varices. A recent real-world study reported that the incidence of bevacizumab-related bleeding was not significantly different according to the status of baseline varices or the presence of PVTT.10D'Alessio A. et al.Hepatology. 2022. Mar 21; https://doi.org/10.1002/hep.32468Crossref Scopus (12) Google Scholar However, there are several points to be considered: first, we focused on variceal bleeding, a portal hypertension-related complication, rather than bevacizumab-related bleeding; second, we evaluated the impact of prophylactic EVL on subsequent variceal bleeding; and third, we assessed the impact of Vp4 PVTT, an extreme form of portal hypertension and hemodynamic alteration. Therefore, we believe particular caution should be exercised when treating patients with atezo/bev if they have high-risk varices or extensive PVTT. A limitation of our study is that baseline EGD was not performed in 47% of patients. Although the proportion is comparable with the recent real-world data, it is far from optimal.10D'Alessio A. et al.Hepatology. 2022. Mar 21; https://doi.org/10.1002/hep.32468Crossref Scopus (12) Google Scholar Second, the duration between baseline EGD and start of atezo/bev was variable. The timing of baseline EGD and subsequent endoscopic evaluation after EVL should be standardized before and during systemic anticancer therapy. Third, because of the limited number of patients, the impact of β-blockers on the incidence of variceal bleeding could not be analyzed. Of note, 4 (2.1%) patients in our cohort were started with β-blockers during atezo/bev therapy by managing gastroenterologist’s decision, based on laboratory, imaging, and endoscopic findings. All of these patients did not bleed during atezo/bev therapy. In addition to EVL, the need for medical therapy with β-blockers should be assessed before and during atezo/bev therapy regularly, and evaluating the impact of β-blockers on variceal bleeding is required in future studies. Last, the timing of resuming bevacizumab in those who experienced variceal bleeding should be standardized after analyzing a larger number of patients. In conclusion, our results demonstrated that the incidence of variceal bleeding was not particularly higher in patients who received atezo/bev in clinical practice when compared with the IMbrave150. Baseline EGD examinations should be performed before atezo/bev treatment in all patients with HCC. High-risk varices on baseline EGD and Vp4 PVTT were associated with variceal bleeding. The authors thank Drs Jinju Choi, Min Je Sung, Beodeul Kang, In Kyung Yoo, Sanghoon Jung, Young Eun Chon, Mi Na Kim, Joo Ho Lee, and Seong Gyu Hwang for their valuable contributions. Supplementary Table 1Baseline CharacteristicsCharacteristicsAll (n = 194; 100.0%)No variceal bleeding (n = 186; 95.9%)Variceal bleeding (n = 8; 4.1%)P valueAge, y62.1 ± 11.462.1 ± 11.457.8 ± 9.5.29Female sex, n (%)28 (14.4)27 (14.5)1 (12.5).68Previous curative or locoregional treatment, n (%)137 (70.6)130 (69.9)7 (87.5).26HBV infection, n (%)128 (66.0)123 (66.1)5 (62.5).55Cirrhosis, n (%)145 (74.7)137 (73.7)8 (100.0).09Albumin, g/dL3.9 (3.4–4.3)3.9 (3.5–4.3)3.6 (3.1–4.0).08Bilirubin, mg/dL0.7 (0.5–1.1)0.7 (0.5–1.0)1.2 (0.8–1.6).015AST, IU/L44.0 (30.0–75.0)44.0 (29.8–75.0)54.0 (34.5–73.0).43ALT, IU/L25.0 (18.0–46.0)25.0 (17.8–44.5)35.5 (18.8–59.0).49Platelet, ×1000/mm3153.0 (106.0–208.0)153.5 (109.5–208.0)95.5 (70.5–194.0).06PT, INR1.1 (1.1–1.2)1.1 (1.0–1.2)1.2 (1.0–1.3).36Fibrinogen, mg/dL484.0 (404.8–601.8)487.0 (401.5–607.0)489.5 (465.0–NA).94PVTT, n (%).025 No126 (64.9)122 (65.6)4 (50.0) Vp216 (8.2)16 (8.6)0 (0.0) Vp324 (12.4)24 (12.9)0 (0.0) Vp428 (14.4)24 (12.9)4 (50.0)Infiltrative tumor, n (%)61 (31.4)57 (30.6)4 (50.0).22Extrahepatic spread, n (%)119 (61.3)114 (61.3)5 (62.5).63Child-Pugh score, n (%).043 A5112 (57.7)109 (58.6)3 (37.5) A652 (27.3)51 (27.4)2 (25.0) B718 (9.3)17 (9.1)1 (12.5) B87 (3.6)5 (2.7)2 (25.0) B92 (1.0)2 (1.1)0 (0.0) C102 (1.0)2 (1.1)0 (0.0)ALBI grade, n (%).043 Grade 1104 (53.6)103 (55.4)1 (12.5) Grade 286 (44.3)79 (42.5)7 (87.5) Grade 34 (2.1)4 (2.2)0 (0.0)EVB <1 y before Atezo/Bev, n (%)2 (1.0)2 (1.1)0 (0.0).92EGD <1 y before Atezo/Bev, n (%)102 (52.6)97 (52.2)5 (62.5).72 Esophageal varices35 (34.3)30 (30.9)5 (100.0).004 Gastric varices8 (7.8)6 (6.2)2 (40.0).048 High risk varices17 (16.8)12 (12.5)5 (100.0)< .001Prophylactic EVL, n (%)16 (8.2)15 (8.1)1 (12.5).50ALBI grade, albumin-bilirubin grade; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Atezo/Bev, atezolizumab/bevacizumab; EGD, esophagogastroduodenoscopy; EVB, esophageal variceal bleeding; EVL, endoscopic variceal ligation; HBV, hepatitis B virus; INR, international normalized ratio; NA, not available; PT, prothrombin time; PVTT, portal vein tumour thrombosis. Open table in a new tab Supplementary Table 2Variceal and Nonvariceal Bleeding Events During Atezolizumab/Bevacizumab TreatmentPatient No.AgeSexTumorPVTTPlateletINRChild-Pugh scoreBaseline endoscopic findingProphylactic EVLTime to bleed (mo)Disease status at bleedingVariceal bleeding957MNodularNo107,0001.345EV F2 RCSNo5.2SD2938MInfiltrativeVp478,0001.265EV F3 RCSGV Gr 2No7.6SD6457MNodularNo85,0001.085NANo1.8SD6854FNodularNo26,0001.556NANo3.6PR7868MInfiltrativeVp4223,0001.018NANo1.6PD10361MNodularNo106,0001.258EV F2No0.3PD14868MInfiltrativeVp4254,0001.017EV F3 RCSGV Gr 2No1.2PR15660MInfiltrativeVp468,0001.106EV F2 RCSYes3.4PDNonvariceal bleeding3256FNodularNo35,0001.106NAGastric angiodysplasia0.4Not evaluableaBleeding occurred after 11 days of atezolizumab/bevacizumab administration.12770MInfiltrativeVp4104,0001.436NADuodenal ulcer2.7PR12960MInfiltrativeVp4179,0001.135EV F1, GV Gr 1, PHG, duodenitisDuodenal ulcer1.1PD14971MInfiltrativeVp3108,0001.247NAInvasion of HCC1.0SDEV, esophageal varices; EVL, endoscopic variceal ligation; Gr, grade; GV, gastric varices; HCC, hepatocellular carcinoma; INR, international normalized ratio; NA, not available; PD, progressive disease; PHG, portal hypertensive gastropathy; PR, partial response; PVTT, portal vein tumor thrombosis; RCS, red color signs; SD, stable disease.a Bleeding occurred after 11 days of atezolizumab/bevacizumab administration. Open table in a new tab ALBI grade, albumin-bilirubin grade; ALT, alanine aminotransferase; AST, aspartate aminotransferase; Atezo/Bev, atezolizumab/bevacizumab; EGD, esophagogastroduodenoscopy; EVB, esophageal variceal bleeding; EVL, endoscopic variceal ligation; HBV, hepatitis B virus; INR, international normalized ratio; NA, not available; PT, prothrombin time; PVTT, portal vein tumour thrombosis. EV, esophageal varices; EVL, endoscopic variceal ligation; Gr, grade; GV, gastric varices; HCC, hepatocellular carcinoma; INR, international normalized ratio; NA, not available; PD, progressive disease; PHG, portal hypertensive gastropathy; PR, partial response; PVTT, portal vein tumor thrombosis; RCS, red color signs; SD, stable disease.