Advances in hepatocellular carcinoma (HCC) and biliary tract cancers are rapidly reshaping clinical practice. At the 2026 Hawaii Gastrointestinal Cancers Summit, Dr. Gentry King of Fred Hutchinson Cancer Center reviewed the latest data on FGFR2-targeted therapies in cholangiocarcinoma, immune checkpoint inhibitors in early- and advanced-stage HCC, and the evolving role of systemic therapy across the BCLC staging framework. His analysis highlights both the promise and complexity of modern treatment strategies in liver cancers.

Dr. Gentry King is a medical oncologist specializing in the treatment of gastrointestinal cancers at Fred Hutchinson Cancer Center. At the 2026 Hawaii Gastrointestinal Cancers Summit, Dr. King presented an overview of key abstracts from the past year spanning biliary tract cancer and hepatocellular carcinoma (HCC), with a focus on FGFR2-targeted therapies, immune checkpoint inhibitors across disease stages, and evolving treatment strategies in the BCLC staging framework.

The transcript report below has not been reviewed by the speaker and may contain errors.

FGFR2-Targeted Therapy in Biliary Tract Cancer: The Refocus Study and Lirofugratin

Molecular profiling is essential in the management of biliary tract cancers, which harbor a variety of actionable driver alterations. Approximately 5% of biliary tract cancers carry an FGFR2 rearrangement or fusion, a subtype for which several FDA-approved and NCCN-recommended therapies now exist, including pemigatinib, erdafitinib, and futibatinib, the latter being a more potent covalent inhibitor of FGFR2. Infigratinib was another agent in this class but was ultimately withdrawn from the market, as discussed below.

Lirofugratin (also formerly known as RLY-4008) represents a next-generation FGFR2 inhibitor that was specifically engineered for selectivity against FGFR2, in contrast to pan-FGFR inhibitors that target FGFR1–3. Pan-FGFR inhibition is associated with a distinct toxicity profile, including hyperphosphatemia, nail and skin changes, and nail syndrome. By achieving a much lower IC50 for FGFR2 relative to FGFR1 and FGFR3, lirofugratin was designed with the intent to reduce these off-target adverse effects. Notably, the compound also demonstrated activity against gatekeeper mutations, among the more common resistance mechanisms to currently approved FGFR inhibitors.

The Refocus study reported data from a specific cohort of FGFR2-fusion cholangiocarcinoma patients. The primary cohort (Group 2A) consisted of FGFR inhibitor-naive patients with prior chemotherapy experience. Among 114 evaluable patients, the overall response rate was approximately 46%, modestly outperforming the roughly 40% response rates reported with pemigatinib and futibatinib. In the subgroup of patients who were both FGFR inhibitor-naive and chemotherapy-naive, meaning completely untreated, the response rate climbed to 63%, suggesting that earlier use of targeted therapy may yield greater clinical benefit. This observation aligns with data from the infigratinib program, where patients treated earlier demonstrated higher response rates and longer disease-free survival. In the FGFR inhibitor-experienced subgroup, response rates fell to 22%, as expected.

These findings raise the question of whether FGFR2-targeted therapy should be moved into the first-line setting. However, conducting a randomized phase 3 study in this population presents a significant challenge: FGFR2 fusions represent roughly 5% of an already rare cancer. This was precisely the obstacle that led to the withdrawal of infigratinib's FDA application — the sponsor was unable to complete the required phase 3 confirmatory trial. Similarly, phase 3 attempts by the pemigatinib and futibatinib programs in the first-line setting were also terminated. The prospect of generating robust randomized data in this molecular subtype in the near term, or ever, remains unlikely.

Despite this, the available evidence is compelling enough to support molecular-defined treatment selection. The key clinical takeaway is that patients with cholangiocarcinoma or biliary tract cancer should undergo comprehensive molecular profiling, and if an FGFR2 fusion or rearrangement is identified, treatment with an FGFR2 inhibitor should be strongly considered, particularly earlier in the disease course when response rates appear highest.

It is worth noting that these agents carry a meaningful toxicity burden. Hyperphosphatemia is an on-target effect, while nail changes and hair loss are among the off-target complications. Although lirofugratin was developed with the aim of reducing this toxicity profile through FGFR2 selectivity, the Refocus study reported treatment-related adverse event rates that were not only still substantial but in some cases higher than those observed with pemigatinib. This was a disappointing finding and a reminder that drug development in this space remains challenging, even as the therapeutic landscape continues to grow.

Immune Checkpoint Inhibitors in the Adjuvant Setting for HCC: KEYNOTE-937 and IMbrave050

Treatment of hepatocellular carcinoma has been transformed over the past seven years, largely through the introduction of immune checkpoint inhibitors in advanced disease. More recently, attention has shifted toward whether these systemic therapies can be deployed earlier, specifically in the adjuvant setting following surgical resection or local ablation.

KEYNOTE-937 evaluated pembrolizumab versus placebo in patients with HCC who achieved a radiological response after surgical resection or local ablation. Patients were required to have Child-Pugh A liver function, and the primary endpoints were recurrence-free survival and overall survival. The study was conducted predominantly outside the United States, and as a result, the majority of participants had hepatitis B-related disease, were of Asian origin, and had undergone resection. Notably, 83% of participants were classified as BCLC A, very early stage disease, and most did not have cirrhosis.

The study was negative across all endpoints: recurrence-free survival, overall survival, and metastasis-free survival all failed to demonstrate benefit from one year of adjuvant pembrolizumab. This is not a benign intervention; immune-related adverse events from pembrolizumab can be serious and sometimes irreversible. The result raises a fundamental question about the scientific rationale: in patients who are BCLC A and have undergone complete resection or ablation, is there sufficient tumor antigen remaining to generate a meaningful immune response? There is a real possibility that many of these patients were already cured by their local therapy, and the addition of systemic immunotherapy may simply not be applicable in this context.

IMbrave050 addressed a similar question but enriched for patients at higher risk of recurrence, requiring features such as tumor size greater than 5 cm, poorly differentiated histology, more than three lesions, or AFP greater than 1,200. Despite this higher-risk enrichment, the study also failed to demonstrate an overall survival benefit. The recurrence-free survival curves showed early separation that subsequently converged after approximately one year, coinciding with the end of active treatment, further undermining confidence in durable benefit. Again, 85% of enrolled patients were BCLC A. The adjuvant story for BCLC A HCC, based on current data, appears to be largely concluded without a positive result.

Perioperative and Neoadjuvant Strategies: A More Promising Direction

In contrast to the adjuvant approach, perioperative and neoadjuvant strategies — where systemic therapy is administered while the tumor is still intact — offer a more scientifically compelling rationale. When tumor tissue is present, it can generate tumor-associated antigens capable of priming an immune response. Several single-arm, non-randomized studies using immune checkpoint inhibitor combinations (such as nivolumab plus ipilimumab) or TKI plus immune checkpoint inhibitor regimens in this setting have shown meaningful responses.

The KRAS009 study evaluated the combination of camrelizumab (a PDL1 inhibitor) and rivoceranib (a VEGF TKI) in a perioperative design and demonstrated potential benefit. However, the primary endpoint was recurrence-free survival rather than overall survival. When the treatment intent is curative in a localized cancer setting, overall survival remains the appropriate and preferred endpoint. Overall survival data from this study are still awaited.

Atezolizumab-Bevacizumab Versus TACE in Intermediate-Stage HCC (BCLC B)

Intermediate-stage HCC (BCLC B) represents a clinically heterogeneous population, and the question of whether systemic therapy should supplant or complement locoregional therapy such as transarterial chemoembolization (TACE) remains unresolved. A study presented at this meeting examined atezolizumab-bevacizumab (atezobev) versus a TACE strategy in this patient population. Importantly, prior TACE was not an exclusion criterion, which significantly complicates the interpretation of the results.

The study was designed as a treatment strategy comparison rather than a head-to-head drug trial. Patients on the TACE arm continued receiving TACE until the strategy was deemed no longer viable, at which point strategy failure was declared. The endpoints were complex and somewhat vague; one was defined as "no evidence of clinical benefit as assessed by the investigator," and another as "therapy not further applicable for other reasons," introducing substantial subjectivity into the analysis.

Baseline BCLC staging was not perfectly balanced between arms; approximately 60% of the atezobev arm had BCLC B classification, versus a somewhat higher proportion in the TACE arm. The median number of TACE procedures received was two (range 1–8), reflecting considerable variability. Interim results showed that fewer patients in the atezobev arm experienced strategy failure compared to the TACE arm. However, because prior TACE was permitted and the curves began separating around the time of the second TACE procedure, the data are difficult to interpret cleanly.

The broader takeaway is that intermediate-stage HCC remains an area of active investigation with significant complexity, and the optimal sequencing or combination of locoregional and systemic strategies has not been established. Overall survival continues to be the benchmark that should define success in this setting.

Four-Year Follow-Up of CheckMate 9DW: Nivolumab Plus Ipilimumab in Advanced HCC (BCLC C)

CheckMate 9DW reported four-year follow-up data for the combination of nivolumab and ipilimumab (nivo/ipi) in patients with advanced BCLC C HCC. Patients were randomized to nivo/ipi or investigator's choice of TKI therapy, either lenvatinib or sorafenib. The inclusion of lenvatinib as a comparator raises the bar considerably; a prior study (LEAP-003) evaluating pembrolizumab plus lenvatinib versus lenvatinib alone was negative specifically because the lenvatinib control arm performed exceptionally well.

The study was positive. Overall response rates with nivo/ipi reached 36%, the highest reported in any first-line HCC study to date. The median overall survival and hazard ratio (0.78) were comparable to those seen in the Himalaya trial (durvalumab plus tremelimumab versus sorafenib). However, a notable pattern emerged in the survival curves: during the first year of follow-up, more patients died in the nivo/ipi arm than in the TKI arm. The curves then separated in favor of nivo/ipi over time. This early excess mortality reflects the toxicity profile of dual checkpoint blockade — specifically, immune-related adverse events, most commonly immune-mediated hepatitis, though these were generally manageable and were not attributable to COVID-19.

Whether this early risk is acceptable depends on the individual patient and clinical context, and this discussion should be had explicitly with patients when considering this regimen.

Cross-Trial Comparisons in Advanced HCC: Navigating a Complex Landscape

No head-to-head trials comparing the major first-line regimens in advanced HCC exist, and none are expected given the competitive landscape. Cross-trial comparisons must therefore be approached with an understanding of each study's enrollment criteria, geographic distribution of patients, and endpoint definitions.

Atezolizumab demonstrates early curve separation favoring the systemic therapy arm. Nivo/ipi shows the reverse pattern — early excess mortality followed by late benefit. Durvalumab plus tremelimumab (Himalaya) shows overlapping curves for approximately the first nine months before separation. Response rates differ across regimens, yet hazard ratios are strikingly similar. This convergence in hazard ratios despite the mechanistic differences — one regimen pairs an anti-VEGF agent with a checkpoint inhibitor, while the others combine two checkpoint inhibitors — underscores the complexity of the therapeutic landscape.

Ultimately, treatment selection in advanced HCC must be individualized. Understanding the nuances of each regimen's efficacy and toxicity profile, and engaging patients in shared decision-making about what matters most to them, remains the cornerstone of optimal care.