Cholangiocarcinoma, a cancer arising in the bile ducts, has become an unexpected test of whether precision oncology can deliver on its central promise: matching the right treatment to the right molecular abnormality at the right moment. In a new perspective published in Nature Reviews Clinical Oncology, F. Saj and L. Goyal argue that the field’s next breakthrough may depend less on discovering another target than on redesigning the clinical trials used to evaluate targeted medicines. Their focus is FGFR2-rearranged cholangiocarcinoma, a rare but biologically distinct form of intrahepatic bile duct cancer that has exposed the strengths—and serious limitations—of conventional drug-development models.
FGFR2 rearrangements occur when the gene encoding fibroblast growth factor receptor 2 becomes abnormally fused to another gene. The resulting hybrid can force the receptor into persistent activation, sending continuous growth and survival signals through pathways such as MAPK, PI3K–AKT and STAT. In a healthy cell, FGFR2 activity is tightly regulated by extracellular growth factors. In a tumour carrying a rearrangement, that molecular switch may remain stuck in the “on” position. This dependence creates an identifiable vulnerability: blocking FGFR signaling can suppress tumour growth even when standard chemotherapy has stopped working. Yet the same biological precision that makes the target attractive also makes the patient population unusually small and clinically diverse.
The development of FGFR inhibitors demonstrated how genomic testing can rapidly reshape treatment. Drugs such as pemigatinib and futibatinib have shown that patients with FGFR2 fusions or rearrangements can experience meaningful tumour responses after progression on earlier therapies. These medicines work differently from traditional cytotoxic chemotherapy. Rather than broadly damaging rapidly dividing cells, they inhibit abnormal signaling at the molecular level. However, responses are rarely permanent. Tumours can acquire secondary changes in the FGFR2 kinase domain that prevent drug binding, activate alternative signaling routes or evolve through clonal selection. The central challenge is therefore no longer simply identifying an FGFR2 alteration. It is determining which alteration matters, when it matters, and how the cancer changes during treatment.
Saj and Goyal use this disease setting to question whether standard phase 1, 2 and 3 trial structures are sufficiently sensitive for molecularly defined cancers. Conventional oncology trials are generally organized around tumour location and treatment line, enrolling large groups of patients who may share an anatomical diagnosis but have very different biology. Precision oncology reverses that logic. A single molecular alteration can occur across several tumour types, while one cancer such as cholangiocarcinoma can contain multiple molecular subgroups with distinct drug sensitivities. For rare alterations, the resulting populations are too small for traditional statistical designs, yet their biological coherence may be greater than that of much larger, unselected cohorts.
The authors highlight the importance of defining the treatment population with greater accuracy. “FGFR2-positive” is not necessarily a single clinical category. Rearrangements may differ in their fusion partners, genomic architecture, expression level and functional effect. A test performed on tumour tissue may fail to detect an alteration because the sample contains too few cancer cells, because the fusion is difficult to capture with a particular assay or because the tumour has evolved since the original biopsy. DNA-based sequencing, RNA-based fusion detection and, in some cases, circulating tumour DNA analysis can provide complementary information. RNA testing is especially valuable because it can demonstrate that a rearranged gene is actively transcribed into a fusion product rather than merely identifying a structural change whose biological significance is uncertain.
Trial timing is another critical issue. Many targeted therapies are initially tested in patients who have already received multiple lines of treatment, a population with advanced disease, declining organ function and substantial molecular complexity. A drug may appear less effective in this setting not because the target is unimportant, but because resistant subclones have already emerged or because patients cannot receive treatment at adequate intensity. Conversely, testing a targeted agent too early may obscure its benefit if the trial lacks a control group or if the natural history of the molecular subgroup is poorly understood. The perspective calls for designs that account explicitly for treatment sequence, previous exposure to FGFR inhibitors and the molecular state of the tumour at enrolment.
The conventional endpoint of tumour shrinkage also deserves reconsideration. Objective response rate is useful, but it captures only one dimension of benefit. Some FGFR2-driven tumours may remain stable for long periods without meeting the formal criteria for a partial response. Others may shrink initially but progress rapidly because resistant clones are already present. Progression-free survival, duration of response, overall survival, symptom control and preservation of liver function may each provide important information. In cholangiocarcinoma, where obstruction of the bile ducts can cause jaundice, infection and deteriorating performance status, a treatment that maintains functional health may be clinically valuable even if radiological changes are modest.
The researchers also point to the need for adaptive trials capable of responding to biological evidence as it emerges. Basket trials, which enrol patients from different tumour types on the basis of a shared molecular alteration, can accelerate testing when a genomic target is rare. Umbrella trials, which divide patients with the same cancer according to different molecular features, can compare several targeted strategies within a common clinical framework. Master protocols may allow investigators to add new treatment arms, retire ineffective ones and study combinations without repeatedly rebuilding an entire trial infrastructure. These approaches can conserve patients, time and resources, but they require rigorous statistical planning to prevent small, highly selected studies from producing misleading conclusions.
Resistance monitoring could become one of the most important features of future trial design. A tumour biopsy taken before treatment provides only a snapshot of a dynamic ecosystem. Under the selective pressure of an FGFR inhibitor, sensitive cells may disappear while resistant populations expand. Serial liquid biopsies can sometimes detect emerging FGFR2 mutations or changes in circulating tumour DNA before they become visible on scans. That information could help researchers distinguish primary resistance, in which a drug never works, from acquired resistance, in which the cancer initially responds and later escapes. It could also guide sequential treatment, combination therapies or trials of next-generation inhibitors designed to bind altered forms of the receptor.
The lesson from FGFR2-rearranged cholangiocarcinoma extends far beyond one rare cancer subtype. Precision medicine is often described as a laboratory achievement—finding a mutation and developing a drug against it—but the clinical trial is where that promise becomes meaningful for patients. If trials ignore molecular heterogeneity, treatment history, evolving resistance and patient-centred outcomes, even a powerful targeted therapy may appear less valuable than it truly is. Saj and Goyal’s analysis presents FGFR2-rearranged cholangiocarcinoma as a warning and an opportunity: rare molecular populations should not be treated as miniature versions of common cancers. They require trial designs built around their biology, flexible enough to follow the disease as it changes and precise enough to reveal benefit where conventional methods may miss it.
Subject of Research: Precision clinical-trial design for FGFR2-rearranged cholangiocarcinoma
Article Title: Precision oncology needs precision trial design: lessons from FGFR2-rearranged cholangiocarcinoma
Article References: Saj, F., Goyal, L. Precision oncology needs precision trial design: lessons from FGFR2-rearranged cholangiocarcinoma. Nat Rev Clin Oncol (2026). https://doi.org/10.1038/s41571-026-01194-3
Image Credits: AI Generated
DOI: 10.1038/s41571-026-01194-3
Keywords: FGFR2 rearrangement, cholangiocarcinoma, precision oncology, targeted therapy, clinical-trial design, molecular profiling, drug resistance, basket trials, circulating tumour DNA
