Hepatocellular carcinoma — red flags and safety concerns

TL;DR — HCC emergencies arise from tumor, cirrhosis, portal hypertension, infection, thrombosis, procedures, and systemic therapy, often with overlapping symptoms. New hematemesis/melena, confusion, fever, rapidly increasing ascites, jaundice, severe abdominal pain, syncope, dyspnea, or abrupt functional decline requires urgent assessment rather than attribution to “the cancer.” Bevacizumab and VEGF-pathway TKIs add bleeding, thrombosis, hypertension, proteinuria, and wound-healing risk; high-risk varices should be evaluated before atezolizumab-bevacizumab (Finn 2020, PMID 32402160; Singal 2023, PMID 37199193). Checkpoint inhibitors can cause multisystem inflammation and can precipitate fatal graft rejection in transplant recipients; liver-test elevation must be distinguished from progression, obstruction, viral activity, ischemia, and decompensation. After locoregional therapy, expected post-treatment symptoms must be separated from liver failure, infection, biliary injury, non-target embolization, and tumor rupture.

This page is research-oriented and does not replace emergency or clinical assessment.

Immediate danger signals

Signal	Major HCC/cirrhosis concern	First diagnostic frame
Hematemesis, melena, hematochezia, syncope	Variceal/ulcer/tumor-related bleeding	Hemodynamic stabilization and urgent bleeding evaluation
New confusion, somnolence, personality change	Hepatic encephalopathy, infection, bleeding, drug effect, brain event	Identify precipitant and protect airway when needed
Fever/rigors	Spontaneous bacterial peritonitis, cholangitis, post-procedure infection	Cultures, ascitic fluid when present, imaging
Rapid ascites/edema	Decompensation, portal-vein thrombosis, renal injury	Liver/renal/vascular assessment
New jaundice	Liver failure, biliary obstruction, immune hepatitis, tumor progression	Pattern liver tests and image biliary/vascular system
Sudden severe abdominal pain/hypotension	Tumor rupture/hemoperitoneum, ischemia, perforation	Emergency imaging and resuscitation
Dyspnea/chest pain	Pulmonary embolism, effusion, anemia, cardiac/immune toxicity	Cardiopulmonary evaluation
Oliguria/rapid creatinine rise	HRS-AKI, dehydration, contrast/drug nephrotoxicity	Volume, infection, medication and renal work-up
New focal neurologic deficit	Stroke, metastasis, bleeding	Emergency neurologic pathway

Decompensation is a competing emergency

HCC treatment cannot be interpreted separately from liver trajectory. Decompensation includes ascites, encephalopathy, variceal bleeding, jaundice, and renal dysfunction. Infection, bleeding, constipation, dehydration, sedatives, and procedures can precipitate rapid decline.

Change	Possible precipitant	Treatment implication
Rising bilirubin	Progression, obstruction, embolization/radiation injury, immune hepatitis	Reassess all anticancer therapy
Falling albumin	Inflammation, nutrition, liver failure, renal loss	Frailty and dose tolerance
New ascites	Portal progression, TACE/TARE injury, thrombosis	Locoregional/systemic safety boundary
Encephalopathy	Infection, bleeding, constipation, medication	Performance status may be liver-driven
Sodium/creatinine deterioration	Circulatory/renal failure	High near-term competing mortality

AASLD guidance conditions treatment on liver reserve and emphasizes multidisciplinary management because aggressive tumor control can worsen the organ needed to survive it (Singal 2023, PMID 37199193).

Bleeding and anti-VEGF therapy

Bevacizumab and VEGF-pathway TKIs impair angiogenic signaling and can increase bleeding and wound-healing complications. Portal hypertension creates a high baseline variceal risk.

Before a bevacizumab-containing regimen:

• review prior variceal or ulcer bleeding;
• perform guideline-consistent endoscopic assessment;
• treat high-risk varices and allow appropriate recovery;
• review anticoagulant/antiplatelet indications;
• control blood pressure;
• check platelets, renal function, and proteinuria;
• coordinate planned surgery or invasive procedures.

IMbrave150 required variceal risk management in trial selection, so its efficacy cannot be divorced from preparation (Finn 2020, PMID 32402160). Real-world study still documents bleeding and thromboembolic events during atezolizumab-bevacizumab (Ben Khaled 2024, PMID 38798717).

Urgent bleeding symptoms include black stool, vomiting blood, dizziness/syncope, rapidly falling hemoglobin, and hemodynamic change.

Hypertension, thrombosis, renal and wound risk

VEGF-pathway toxicity	Monitoring	Escalation signal
Hypertension	Baseline/home/clinic BP	Severe persistent BP, neurologic/cardiac symptoms
Proteinuria	Urinalysis or quantified protein	Nephrotic-range protein, edema, renal decline
Arterial thrombosis	Vascular history and symptoms	Chest pain, focal deficit, limb ischemia
Venous thrombosis	Swelling, dyspnea, catheter context	Suspected PE/DVT
Wound-healing impairment	Procedure timing	Dehiscence, infection, fistula
GI perforation	Abdominal symptoms and risk anatomy	Acute severe pain/peritonism

Holding intervals around procedures depend on agent half-life, procedure risk, and healing; they should be planned rather than improvised.

Immune checkpoint toxicity

Checkpoint inhibitors can inflame almost any organ.

System	Example presentation	Dangerous differential
Liver	AST/ALT/bilirubin rise	Progression, obstruction, viral flare, ischemia, other drug injury
Bowel	Diarrhea, abdominal pain, blood	Infection, ischemia, portal enteropathy
Lung	Cough, hypoxia, dyspnea	Infection, PE, effusion, metastasis
Endocrine	Fatigue, hypotension, headache	Cirrhosis fatigue, sepsis, adrenal crisis
Heart	Chest pain, arrhythmia, troponin rise	Ischemia, PE, myocarditis
Nervous system	Weakness, neuropathy, confusion	Encephalopathy, stroke, metastasis
Kidney	Creatinine rise	HRS, dehydration, contrast, TKI
Skin	Rash, blistering, mucosal disease	Severe cutaneous reaction

Immune toxicity can occur after treatment stops. Management depends on severity and organ and may require holding therapy and immunosuppression; empiric attribution without excluding infection or obstruction is unsafe.

Transplant and checkpoint inhibitors

Checkpoint activation can break graft tolerance and cause rejection. Solid-organ transplant recipients were generally excluded from pivotal HCC trials. The risk is not captured by standard immune-adverse-event rates.

Critical contexts:

• checkpoint therapy after liver transplant;
• checkpoint exposure before planned transplant;
• uncertain washout interval;
• concurrent immunosuppression;
• differentiating rejection from immune hepatitis or recurrent disease.

Any checkpoint decision in a transplant pathway requires explicit transplant-oncology coordination. Evidence is observational and selection-biased.

Autoimmune disease

Active severe autoimmune disease and baseline immunosuppression complicate checkpoint therapy. Risk varies by organ, activity, prior life-threatening manifestations, and alternative treatments.

The decision should document:

• autoimmune diagnosis and activity;
• current immunosuppressive agents;
• organ damage and flare consequence;
• expected anticancer benefit;
• monitoring and rescue plan;
• alternatives without checkpoint blockade.

TKI toxicity

Sorafenib, lenvatinib, regorafenib, and cabozantinib have overlapping but non-identical toxicity.

Toxicity	Early response
Hand-foot skin reaction	Skin protection, topical care, dose assessment
Diarrhea	Hydration, infection assessment, antidiarrheal plan
Anorexia/weight loss	Nutrition and sarcopenia assessment
Fatigue	Separate drug, cancer, liver, endocrine causes
Hypertension/proteinuria	VEGF-pathway monitoring
Hepatic worsening	Reassess dose, progression, decompensation

Regorafenib evidence came from patients who tolerated sorafenib; extrapolating to sorafenib-intolerant patients ignores trial selection (Bruix 2017, PMID 27932229).

After TACE

Expected post-embolization syndrome can include pain, fever, nausea, and transient liver-test elevation. Red flags include:

• persistent high fever or sepsis;
• severe or worsening pain;
• jaundice/ascites/encephalopathy;
• renal decline;
• biliary ischemia or abscess;
• non-target embolization;
• tumor rupture.

Repeated TACE can produce cumulative liver injury; declining reserve is a stopping signal, not simply a side effect to tolerate (Reig 2022, PMID 34801630).

After TARE or external radiation

TARE planning addresses lung shunt and non-target vessels. Delayed hazards include radioembolization-induced liver disease, biliary injury, gastrointestinal ulceration, and radiation pneumonitis. Personalized dosimetry improves the chance that tumor dose is therapeutic while non-tumor dose remains acceptable (Garin 2021, PMID 33166497).

External-beam radiation can cause liver-function deterioration and injury to bowel, stomach, chest wall, or biliary structures depending on target. Child-Pugh/ALBI change should be a safety endpoint.

After ablation or resection

Procedure	Early red flag
Thermal ablation	Bleeding, infection, bile-duct injury, bowel injury, liver failure
Resection	Hemorrhage, bile leak, infection, thrombosis, post-hepatectomy liver failure
Either	Severe pain, hypotension, fever, jaundice, respiratory decline

After ablation, a small risk of tract seeding or incomplete margin exists; after resection, pathology can reveal microvascular invasion or mixed histology that changes recurrence risk.

Tumor rupture

Spontaneous HCC rupture can present with sudden abdominal pain, peritoneal irritation, hypotension, or hemoglobin fall. It requires urgent hemodynamic assessment and imaging; embolization and surgery are considered according to stability, anatomy, and liver reserve.

Portal-vein thrombosis: bland versus tumor

New portal-vein filling defect can represent bland thrombosis, tumor in vein, or both. Enhancement and continuity with tumor support malignant invasion. The distinction changes:

• stage and transplant eligibility;
• anticoagulation decision;
• locoregional feasibility;
• systemic treatment;
• prognosis.

Do not infer mechanism from “thrombosis” alone.

Infection and viral reactivation

Cirrhosis raises infection risk; procedures, admissions, steroids, and immunotherapy work-up add exposure. HBV status should be known before immunosuppressive cancer therapy, and antiviral management coordinated. HCV cure status does not eliminate decompensation or HCC recurrence risk in cirrhosis.

Fever after a procedure may be inflammatory, but infection must not be dismissed when symptoms persist or liver function worsens.

Medication and supplement hazards

• NSAIDs can worsen renal perfusion and bleeding risk in decompensated cirrhosis.
• Sedatives and opioids can precipitate/worsen encephalopathy and falls.
• Anticoagulant and antiplatelet drugs require individualized bleeding/thrombosis balance.
• Herbal and dietary supplements may cause hepatotoxicity or drug interactions.
• CYP/P-glycoprotein interactions can alter TKI exposure.
• Corticosteroids used for immune toxicity raise infection, glucose, muscle, and viral risks.

Medication reconciliation should include non-prescription products.

Surveillance-lapse hazards

A surveillance order is not protection. Harm occurs when:

• cirrhosis is not recognized;
• ultrasound is repeatedly not scheduled;
• visualization is limited but reported as negative;
• an abnormal AFP is not followed;
• a lesion is lost between services;
• post-treatment recurrence imaging lapses.

Closed-loop tracking is a safety intervention, not merely quality administration.

Quantified high-consequence risks

Hazard	Quantified evidence	Safety implication
Ruptured HCC	Nine-study meta-analysis/681 survivors: emergency hepatectomy had in-hospital mortality RR 2.17 (95% CI 1.03–4.57) versus embolization then delayed resection	Stabilize hemorrhage first when feasible; selection and survivor bias remain (Zhang 2022, PMID 36397082)
Variceal bleeding on atezolizumab-bevacizumab	640 endoscoped patients: 7.0% bled; cumulative incidence 6.3% at six and 7.4% at twelve months	Main portal-vein invasion and prior GI bleeding markedly increased risk (Park 2025, PMID 39871662)
Real-world bleeding	112 Canadian patients: any bleeding 15%; GI bleeding 5%; 41% of scoped patients had varices and 19% required intervention	Absence of mandatory EGD in selected patients did not prove universal safety (Lee 2024, PMID 39199649)
Acute renal failure after TACE	12/140 (8.6%); each prior/additional TACE OR 1.65, Child-Pugh B OR 12.82, severe postembolization syndrome OR 6.64	Hydration, renal risk, contrast burden, liver reserve, and repetition matter (Huo 2004, PMID 15189271)
Post-hepatectomy liver failure	39/361 grade B/C; high ALBI OR 8.68 and greater resection percentage OR 1.10 per percentage point	Couple functional reserve to future-liver-remnant volume (Takahashi 2022, PMID 35213436)
ICI before transplant	91-patient IPD meta-analysis: rejection 26.4%, recurrence 9.9%, death 9.9%	Rejection is a foreseeable graft-threatening event, not a theoretical concern (Rezaee-Zavareh 2025, PMID 38996924)

Bleeding triage is mechanism-specific

Acute hemoperitoneum from tumor rupture, portal-hypertensive variceal bleeding, anti-VEGF-associated mucosal/vascular bleeding, and post-procedural arterial hemorrhage require different responses. Tumor rupture typically presents with sudden pain, hypotension, falling hemoglobin, and intraperitoneal blood; embolization can provide hemostasis and permit delayed definitive assessment (Zhang 2022, PMID 36397082). Portal-vein tumor thrombus can worsen portal pressure while also marking advanced tumor biology, demanding coordinated hepatology, oncology, radiology, and endoscopy decisions (Qiu 2021, PMID 34039819).

For bevacizumab, a negative endoscopy is not a permanent guarantee: portal-vein invasion, tumor progression, thrombosis, falling platelets, and interval decompensation change risk. Conversely, a history of treated varices is not automatically a lifelong exclusion. The decision should record endoscopy date, variceal size/risk signs, prophylaxis, bleeding history, portal-vein status, platelets, anticoagulants, and blood pressure (Park 2025, PMID 39871662).

Liver injury: progression, decompensation, or immune toxicity

Rising aminotransferases during checkpoint therapy have competing explanations: immune-related hepatitis, viral flare/reactivation, tumor progression, biliary obstruction, ischemia, infection, alcohol/drug injury, and spontaneous cirrhosis decompensation. In an HBV-endemic 1,283-patient cohort, any-grade hepatitis flare occurred in 45.8% of HCC versus 25.6% of non-HCC patients; HCC and baseline ALT >40 U/L predicted grade ≥3 flare (Hung 2025, PMID 39582238). This supports diagnostic workup before corticosteroids rather than reflexively labeling every abnormality immune hepatitis (De Martin 2025, PMID 39658265).

Dual-checkpoint regimens increase immune-event frequency relative to monotherapy, while TKIs create a different syndrome of diarrhea, anorexia, hand-foot reaction, hypertension, and proteinuria (Yau 2020, PMID 33001135; Yau 2022, PMID 34914889). In cirrhosis, diarrhea-induced dehydration or anorexia-driven sarcopenia can precipitate decompensation even when the toxicity is not intrinsically hepatic.

Procedure-specific failure signals

Procedure	Early signal	Quantified context
Resection	INR rise, bilirubin rise, encephalopathy, ascites, lactate/renal worsening	Day-2 INR ≤1.20 identified a group without PHLF in one 120-patient cohort; INR >1.60 was strongly concerning (Silva 2022, PMID 35144899)
Resection	Low functional uptake in future remnant	Gadoxetate functional-liver imaging score predicted PHLF with AUC 0.752 in 502 patients (Luo 2022, PMID 35294586)
TACE	Persistent fever/pain, renal injury, bilirubin/INR rise, new ascites	Among 65 refractory patients, 27 deteriorated from Child-Pugh A to B/C within one year (Park 2020, PMID 32209803)
TARE	New ascites/jaundice after bilobar treatment without progression	Whole-liver exposure and cirrhosis are recognized REILD risks; retrospective comparisons cannot define a universally safe volume (Jeschke 2023, PMID 37686549)
Ablation	Hemorrhage, abscess, bile-duct injury, bowel injury	Risk depends on location, approach, and protective techniques; radiation may be an alternative near vulnerable structures (Xi 2025, PMID 39693584)

ALBI predicts PHLF across BCLC stages, but no score replaces volumetry, portal-hypertension assessment, operative extent, and clinical decompensation history (Zhang 2018, PMID 30326907). Repeated TACE should stop before liver injury eliminates systemic and transplant options; Japanese refractoriness criteria use at least two consecutive inadequate responses with imaging/feed-artery reassessment (Kudo 2014, PMID 25427730).

Transplant checkpoint exposure

In 119 pre-transplant ICI-exposed recipients, rejection occurred in 20.2%, median nine days after transplant. Compared with washout >50 days, washout <30 days had OR 21.3 (95% CI 5.93–103) and 30–50 days OR 9.48 (2.47–46.8); six of 24 rejection cases lost the graft (Moeckli 2025, PMID 40042053). These data justify a hard multidisciplinary safety checkpoint before listing/transplanting an ICI-exposed patient, while not proving any interval absolutely safe.

Persistent controversies

• Universal pre-bevacizumab endoscopy. Large real-world cohorts identify clear risk factors, but selection bias prevents concluding that endoscopy can be omitted safely in all low-risk patients (Lee 2024, PMID 39199649; Park 2025, PMID 39871662).
• Emergency versus staged surgery for rupture. Staged embolization lowers in-hospital mortality in pooled observational data, but hemodynamics, tumor resectability, and embolization access determine feasibility (Zhang 2022, PMID 36397082).
• When to stop TACE. Two inadequate procedures is a useful guardrail, not an invariant biological threshold; early bilirubin/ALBI deterioration can justify earlier migration (Kudo 2014, PMID 25427730; Park 2020, PMID 32209803).
• Safe ICI washout before transplant. Risk declines with time but has no verified zero-risk cutoff; organ scarcity makes this a recipient-and-graft safety decision (Rezaee-Zavareh 2025, PMID 38996924; Moeckli 2025, PMID 40042053).

Open questions

• What endoscopic/variceal strategy best reduces bevacizumab bleeding without delaying effective therapy?
• Which biomarkers distinguish immune hepatitis from decompensation or progression early?
• What checkpoint washout and immune profile predict safer liver transplantation?
• Which post-TACE liver-function change should mandate permanent migration to systemic therapy?
• Can electronic symptom monitoring reduce emergency admissions and severe toxicity in HCC?

Related pages

• Systemic therapy — details regimen selection and toxicity.
• Locoregional therapy — explains procedural mechanisms.
• Diagnosis and imaging — distinguishes thrombosis and progression.
• Liver transplantation — covers graft-specific risk.
• Patient experience and advocacy — describes symptom and caregiver burden.

References

1. Singal AG, et al. AASLD Practice Guidance on prevention, diagnosis, and treatment of HCC. Hepatology. 2023;78:1922-1965. PMID 37199193
2. Finn RS, et al. Atezolizumab plus bevacizumab in unresectable HCC. N Engl J Med. 2020;382:1894-1905. PMID 32402160
3. Ben Khaled N, et al. Atezolizumab/bevacizumab or lenvatinib in HCC: real-world bleeding and thromboembolic events. JHEP Rep. 2024. PMID 38798717
4. Bruix J, et al. Regorafenib after sorafenib treatment (RESORCE). Lancet. 2017;389:56-66. PMID 27932229
5. Reig M, et al. BCLC strategy for prognosis prediction and treatment recommendation: 2022 update. J Hepatol. 2022;76:681-693. PMID 34801630
6. Garin E, et al. Personalised versus standard dosimetry in DOSISPHERE-01. Lancet Gastroenterol Hepatol. 2021;6:17-29. PMID 33166497
7. Zhang W, et al. Emergency versus delayed hepatectomy after embolization for ruptured HCC: a meta-analysis. World J Surg Oncol. 2022;20:365. PMID 36397082
8. Qiu G, et al. Multidisciplinary management of HCC with portal-vein tumor thrombus. Biosci Trends. 2021;15:148-154. PMID 34039819
9. Lee CL, et al. Real-world atezolizumab-bevacizumab outcomes, endoscopy, and bleeding. Cancers. 2024;16. PMID 39199649
10. Park K, et al. Risk of variceal bleeding during atezolizumab-bevacizumab treatment. Aliment Pharmacol Ther. 2025;61:1310-1317. PMID 39871662
11. De Martin E, et al. Immune checkpoint inhibitors and the liver. Gut. 2025;74:1165-1177. PMID 39658265
12. Yau T, et al. Nivolumab versus sorafenib in CheckMate 459. Lancet Oncol. 2022;23:77-90. PMID 34914889
13. Yau T, et al. Nivolumab plus ipilimumab after sorafenib in CheckMate 040. JAMA Oncol. 2020;6:e204564. PMID 33001135
14. Hung YP, et al. Hepatic events during checkpoint treatment in an HBV-endemic area. Aliment Pharmacol Ther. 2025;61:501-512. PMID 39582238
15. Huo TI, et al. Acute renal failure after TACE: incidence and risk factors. Liver Int. 2004;24:210-215. PMID 15189271
16. Jeschke M, et al. Bilobar radioembolization and risk of radioembolization-induced liver disease. Cancers. 2023;15. PMID 37686549
17. Takahashi K, et al. Predicting post-hepatectomy liver failure using ALBI and resection percentage. J Am Coll Surg. 2022;234:155-165. PMID 35213436
18. Luo N, et al. Functional liver imaging score for preoperative prediction of liver failure. Eur Radiol. 2022;32:5623-5632. PMID 35294586
19. Silva ANS, et al. Early INR derangement predicts liver failure after HCC resection. Surgeon. 2022;20:e288-e295. PMID 35144899
20. Zhang ZQ, et al. ALBI grade predicts post-hepatectomy liver failure across BCLC stages. World J Surg Oncol. 2018;16:208. PMID 30326907
21. Rezaee-Zavareh MS, et al. Pre-transplant ICI use and post-transplant outcomes: IPD meta-analysis. J Hepatol. 2025;82:107-119. PMID 38996924
22. Moeckli B, et al. Safe checkpoint-inhibitor washout before liver transplantation. Hepatology. 2025;82:1122-1137. PMID 40042053
23. Park KH, et al. Liver-function deterioration after TACE refractoriness. Korean J Gastroenterol. 2020;75:147-156. PMID 32209803
24. Kudo M, et al. TACE failure/refractoriness criteria 2014 update. Oncology. 2014;87 Suppl 1:22-31. PMID 25427730
25. Xi M, et al. RFA versus SBRT for recurrent small HCC: randomized trial. J Clin Oncol. 2025;43:1073-1082. PMID 39693584