Oncology · Medical Literature
Perioperative tislelizumab for early-stage hepatocellular carcinoma: A phase II trial with integrated tumour microenvironment profiling and predictive modeling.
Abstract
It is controversial whether monotherapy with immunotherapeutics is effective in early-stage hepatocellular carcinoma (HCC). We evaluated immunotherapy monotherapy as an initial neoadjuvant strategy. This prospective, single-arm, phase II trial of an adaptive neoadjuvant strategy enrolled patients with resectable, early-stage HCC. All patients received tislelizumab monotherapy; non-responders received subsequent lenvatinib combination therapy preoperatively. For the overall efficacy of all 36 patients, the objective response rates were 13.9% and 30.6% as per the Response Evaluation Criteria in Solid Tumours v1.1 and modified Response Evaluation Criteria in Solid Tumours, respectively. Among the 33 patients who underwent resection, the major pathological response rate was 33.3% and the pathological complete response rate was 15.2%. Overall, grade ≥ 3 treatment-emergent adverse events occurred in 11.1%, with no surgical delays. Monotherapy responses were predicted by pretreatment tumour microenvironment features. Immune-regulatory cancer-associated fibroblasts colocalized with T cells identified responders, whereas angiogenic-metabolic signatures predicted resistance. A multimodal model integrating blood-based and imaging biomarkers accurately identified monotherapy-sensitive patients (average precision, 0.9167). The specificity for identifying responders was 100%. These preliminary findings suggest that neoadjuvant immunotherapy monotherapy exhibits promising antitumour activity and demonstrates a manageable safety profile in early-stage HCC. The tailored stratification approach appears clinically feasible and may help minimize unnecessary exposure to combination therapy-related toxicity while retaining tentative therapeutic efficacy.
Clinical relevance
In patients with resectable, early-stage hepatocellular carcinoma (HCC), does a perioperative adaptive strategy beginning with tislelizumab (anti-PD-1) monotherapy — with rescue lenvatinib combination added for non-responders — produce meaningful pathological responses, and can pretreatment tumour microenvironment (TME) features predict which patients will respond to immunotherapy monotherapy? __SNAPSHOT__ {"studyDesign":"Prospective, single-arm, phase II trial with integrated tumour microenvironment profiling and predictive modeling (adaptive neoadjuvant design)","setting":"Not reported (inferred: China-based academic hepatobiliary center, based on author names and journal context; not explicitly stated in the abstract)","sampleSize":"n=36 enrolled; n=33 underwent resection","population":"Adults with resectable, early-stage hepatocellular carcinoma; specific staging criteria (e.g., BCLC stage, Child-Pugh class, ECOG PS, HBV status, lesion size) not fully detailed in the abstract","intervention":"Tislelizumab monotherapy as initial neoadjuvant strategy; non-responders received subsequent lenvatinib added to tislelizumab preoperatively (adaptive escalation)","comparator":"Not applicable — single-arm study with no concurrent control group","primaryOutcome":"Not explicitly designated in the abstract as a single pre-specified primary endpoint; reported outcomes include objective response rate (ORR) by RECIST v1.1 and mRECIST, major pathological response (MPR) rate, pathological complete response (pCR) rate, grade ≥3 treatment-emergent adverse events (TEAEs), and TME-based predictive biomarkers","followUp":"Not reported","fundingSource":"Not reported"}
Key findings
Among all 36 enrolled patients, the ORR was 13.9% by RECIST v1.1 and 30.6% by mRECIST. Among the 33 patients who proceeded to surgical resection, the major pathological response (MPR, typically defined as ≥90% tumour necrosis) rate was 33.3% (11/33, calculated) and the pathological complete response (pCR) rate was 15.2% (5/33, calculated). These response rates are notably higher than the radiological ORR by RECIST v1.1, consistent with well-established discordance between imaging and pathological response in HCC. Grade ≥3 treatment-emergent adverse events occurred in 11.1% (4/36, calculated) of patients, and no surgical delays were attributed to neoadjuvant therapy. The proportion of patients who did not undergo resection (3/36, or 8.3%, calculated) is not explained in detail in the abstract. TME analysis identified that immune-regulatory cancer-associated fibroblasts (CAFs) co-localised with T cells in the pretreatment tumour predicted monotherapy response, whereas an angiogenic-metabolic gene signature predicted resistance. No confidence intervals or p-values for efficacy endpoints are reported in the available abstract. NNT cannot be calculated in the absence of a comparator arm.
Practice implications
This phase II single-arm study provides preliminary signal that PD-1 monotherapy with tislelizumab can achieve meaningful pathological responses — including pCR in approximately 1 in 7 resected patients — when used as the initial component of an adaptive perioperative strategy in early-stage HCC. The MPR rate of 33.3% in resected patients is clinically noteworthy if it translates to improved recurrence-free or overall survival, though this association was not established in the available abstract. The safety profile (11.1% grade ≥3 TEAEs, no surgical delays) appears acceptable and consistent with prior experience with anti-PD-1 agents in HCC. The TME biomarker findings — specifically immune-regulatory CAF/T-cell co-localization as a predictor of response and angiogenic-metabolic signatures as predictors of resistance — are hypothesis-generating and potentially clinically useful for patient selection, but require independent prospective validation before they can guide practice. The study does not establish which patients will derive a long-term survival benefit, as recurrence-free survival and overall survival data are not reported in the abstract. The population (early-stage, resectable HCC) is underrepresented in most prior immunotherapy trials, which have focused on unresectable/advanced disease; therefore, the findings address a genuine evidence gap. However, the single-arm design precludes causal inference, and the small sample size (n=36) limits statistical power and generalizability. The abstract does not report whether the adaptive lenvatinib escalation in non-responders contributed disproportionately to the pathological response rates, which is a critical unanswered question. Practice should not change based on this trial alone; it provides justification for larger randomized evaluation. NOTE: This summary is based on the provided abstract only; full-text data including survival curves, subgroup analyses, statistical significance thresholds, and TME methodology details were not available for review. __LIMITATIONS__ Single-arm design with no randomised comparator: causal attribution of pathological responses to tislelizumab (vs. natural history or surgical selection) cannot be established,Very small sample size (n=36 total; n=33 resected): substantially underpowered to detect differences in survival endpoints or to validate biomarker models with adequate sensitivity and specificity,Adaptive design confounds efficacy attribution: the separate contributions of monotherapy versus monotherapy-plus-lenvatinib to the aggregate pathological response rate are not disentangled in the abstract,No survival data (RFS, OS) reported: pathological response (MPR/pCR) is a surrogate endpoint whose prognostic value in perioperative HCC immunotherapy has not been formally validated,Staging criteria, baseline clinicopathological details, etiology (HBV vs. HCV vs. alcohol), and liver function status are not fully described, limiting assessment of generalizability,TME biomarkers derived from the same dataset used to generate the predictive model (potential overfitting); external validation is absent,Follow-up duration not reported: durability of response and postoperative outcomes are unknown,Funding source not reported: potential for undisclosed industry support cannot be excluded,Single-arm phase II evidence base is insufficient to determine whether results are statistically significant or represent a practice-informing effect size in the context of the broader HCC perioperative landscape,Absence of blinded independent radiological or pathological review is not confirmed in the abstract __BOTTOM_LINE_CATEGORY__ promising_but_preliminary __BOTTOM_LINE__ This small, single-arm phase II trial demonstrates that perioperative tislelizumab monotherapy — with adaptive escalation to lenvatinib combination for non-responders — is feasible and achieves pCR in ~15% and MPR in ~33% of resected early-stage HCC patients, with an acceptable toxicity profile and no surgical delays. Pretreatment TME features, particularly immune-regulatory CAF/T-cell co-localization, show promise as predictive biomarkers for monotherapy response. However, the single-arm design, small sample, absence of survival data, and unvalidated biomarkers preclude practice change; these findings should be considered hypothesis-generating and warrant confirmation in a randomised controlled trial.