First-Line Trastuzumab Add-On Reaches 55% Response Rate in HER2-Positive Biliary Tract Cancer

Why This Study Matters

Biliary tract cancers (BTC) — including intrahepatic and extrahepatic cholangiocarcinoma and gallbladder cancer — carry a poor prognosis, and the current first-line standard of care (an immune checkpoint inhibitor plus gemcitabine-cisplatin) leaves most patients with a median progression-free survival of roughly 7 months and overall survival just over a year. HER2 overexpression or amplification is present in about 15% of BTCs and is a clinically actionable target, but HER2-directed therapy in this disease had previously been studied almost exclusively in patients who had already progressed on earlier treatment lines. HERBOT is the first prospective trial to test whether adding a HER2-targeted antibody works when introduced immediately, alongside first-line chemoimmunotherapy, rather than held in reserve.

Study Design

HERBOT was an open-label, single-arm, phase 1b/2 trial conducted at 12 academic cancer centers in South Korea. Phase 1b used a standard 3+3 dose-escalation design to establish the recommended phase 2 dose (RP2D); no dose-limiting toxicities were observed at the starting dose, which was carried forward as the RP2D. Phase 2 used a Simon two-stage minimax design, powered against a null objective response rate (ORR) of 27% (drawn from the phase 3 TOPAZ-1 trial) and an alternative ORR of 45%.

The regimen tested was trastuzumab 6 mg/kg (after an 8 mg/kg loading dose) plus nivolumab 360 mg on day 1, and gemcitabine 1,000 mg/m² plus cisplatin 25 mg/m² on days 1 and 8, repeated every 3 weeks. Cytotoxic chemotherapy ran for up to nine cycles, with continuation beyond nine permitted at investigator discretion; trastuzumab and nivolumab continued for up to 2 years. There was no comparator arm.

Patient Population

Between May 12, 2023, and December 3, 2024, 49 participants with centrally confirmed HER2-positive BTC were screened and 40 were enrolled (3 in phase 1b, 37 in phase 2). All were treatment-naive for advanced or metastatic disease (up to one prior cycle of gemcitabine-based chemotherapy was permitted). Median age was 66 years (range 50–81); 87.5% had an ECOG performance status of 1. Gallbladder cancer made up 72.5% of the population, with intrahepatic (17.5%) and extrahepatic (10%) cholangiocarcinoma comprising the remainder. HER2 status was IHC 3+ in 62.5% of participants and IHC 2+/ISH+ in the remaining 37.5%.

Primary Endpoint Results

The primary endpoint — ORR among the 40 participants treated at the RP2D — was met, with an ORR of 55% (22/40; 95% CI 38.5–70.7), exceeding the prespecified Simon two-stage criterion of at least 14 responses among 34 evaluable participants. This comprised one complete response (2.5%) and 21 partial responses (52.5%); the disease control rate was 95.0% (95% CI 83.5–99.4). Median duration of response was 12.6 months (95% CI 5.7–not reached), and median time to response was 2.1 months (95% CI 1.93–2.43).

At a median follow-up of 17.0 months (IQR 11.0–19.4) and a data cutoff of December 23, 2025, median PFS was 10.6 months (95% CI 7.8–17.4), with a 6-month PFS rate of 80.0% (95% CI 64.0–89.5). OS data remained immature, with median OS not reached and a 12-month OS rate of 72.7% (95% CI 55.0–84.4). Two participants underwent curative-intent conversion surgery, 4.54 and 6.97 months after treatment initiation.

The observed ORR of 55%, DCR of 95% and median PFS of 10.6 months compare favorably to historical outcomes reported for standard chemoimmunotherapy alone, supporting the feasibility of integrating HER2-targeted therapy into the first-line setting.
— Discussion, Lee et al., Nature Medicine, 2026 (verbatim; "ORR" = objective response rate, "DCR" = disease control rate)

Subgroup Analyses

All subgroup findings below are explicitly described by the authors as post hoc and exploratory, without adjustment for multiple comparisons; they should be read as hypothesis-generating rather than confirmatory.

Two-sided Fisher's exact test comparing ORR across subtypes gave P = 0.1085 (not statistically significant); the comparison was not adjusted for multiple testing.

HER2 IHC status: ORR did not differ significantly by conventional HER2 IHC status (two-sided chi-square P = 0.4119). However, participants with HER2 IHC 3+ tumors (n=25) showed numerically longer median PFS than those with IHC 2+/ISH+ tumors (n=15): 17.4 vs. 9.7 months (HR 0.46, 95% CI 0.20–1.07). Median OS was not reached in the IHC 3+ group versus 14.2 months in the IHC 2+/ISH+ group (HR 0.77, 95% CI 0.29–2.09).

Exploratory AI-based whole-slide HER2 quantification: using an AI pathology model to quantify the whole-slide proportion of HER2 3+ tumor cells (a 10% threshold), participants with ≥10% HER2 3+ cell proportion (n=15) had a best ORR of 80.0%, versus 40.0% in those with <10% (P = 0.0138). Median PFS was 17.4 months versus 10.5 months, respectively (HR 0.57, 95% CI 0.23–1.34). The authors note this AI-based biomarker is exploratory and has not been externally validated.

Safety Profile

Treatment-related adverse events (TRAEs) of any grade occurred in 36 of 40 participants (90.0%). Most grade 3–4 events were hematologic. Two participants discontinued treatment because of adverse events (drug-unrelated pneumonia and nivolumab-related arthralgia). Cardiac toxicity attributable to trastuzumab was infrequent: one participant had a grade 2 decrease in ejection fraction, though treatment was ultimately discontinued for disease progression rather than the cardiac event.

Table abridged from the article's full adverse-event table (all listed events, hematologic and nonhematologic, are reported in the source). *Grade 3 total row as reported in source; individual event rows above do not sum to the total because not all reported events are shown here.

Interpretation and Broader Context

Current first-line standard of care for advanced BTC — an immune checkpoint inhibitor plus gemcitabine-cisplatin — produces response rates of roughly 27–29% and median PFS around 7 months in the phase 3 TOPAZ-1 and KEYNOTE-966 trials, per the article's own framing. Against that backdrop, HERBOT's 55% ORR and 10.6-month median PFS are described by the authors as comparing favorably, though this is a cross-trial, non-randomized comparison and not a head-to-head result. As a single-arm, uncontrolled, phase 1b/2 study of 40 patients, HERBOT cannot establish comparative efficacy against standard chemoimmunotherapy; the authors are explicit that confirmation requires randomized trials, citing two ongoing phase 3 studies — HERIZON-BTC-302 (NCT06282575) and DESTINY-BTC01 (NCT06467357) — that are evaluating HER2-targeted strategies in the first-line BTC setting.

Limitations

HERBOT is a single-arm, non-randomized, open-label phase 1b/2 trial that enrolled only 40 patients across 12 South Korean centers, with no concurrent control arm; its 55% response rate and 10.6-month median progression-free survival can be set only against historical, cross-trial benchmarks from separate studies of chemoimmunotherapy alone, not a randomized comparison within the trial itself. Overall survival data remain immature, with the median not yet reached at 17.0 months of follow-up. The subgroup findings by cancer subtype, HER2 IHC status, and the exploratory AI-based whole-slide HER2 quantification method are explicitly described by the authors as post hoc and hypothesis-generating, unadjusted for multiple comparisons, and not externally validated; the authors state that confirmation of HER2-targeted therapy's first-line benefit requires the ongoing randomized phase 3 trials HERIZON-BTC-302 and DESTINY-BTC01.