Neoadjuvant immunotherapy in resectable hepatocellular carcinoma: A meta-analysis of the current evidence

Abstract

BACKGROUND

Hepatocellular carcinoma (HCC) remains a leading cause of cancer-related mortality worldwide. Despite improvements in surgical techniques and systemic therapies, long-term outcomes after liver resection are limited by high recurrence rates. While adjuvant strategies have shown limited benefit, the role of neoadjuvant immunotherapy in resectable HCC is still under investigation.

AIM

To assess the efficacy, feasibility, and safety of neoadjuvant immunotherapy in resectable HCC through a meta-analysis of current literature.

METHODS

A systematic search was conducted across PubMed, Web of Science, EMBASE, Cochrane Library, and Scopus for studies published in the past five years evaluating neoadjuvant immunotherapy in resectable HCC. Primary endpoints included major pathological response (MPR), pathological complete response (pCR), overall response rate (ORR), resection rate, and grade 3-4 treatment-related adverse events (TRAEs). A random-effects meta-analysis was conducted using log odds ratios (ORs) and pooled event rates were calculated to provide absolute estimates of clinical endpoints.

RESULTS

Twelve studies were included in the final analysis. The pooled ORs were 0.28 (95%CI: 0.19-0.41) for MPR, 0.54 (95%CI: 0.25-1.14) for ORR, 0.26 (95%CI: 0.11-0.66) for pCR, 5.37 (95%CI: 2.70-10.66) for resection rate, and 0.33 (95%CI: 0.22-0.50) for grade 3-4 TRAEs. Corresponding pooled event rates were 19% for MPR, 35% for ORR, 22% for pCR, 81% for resection feasibility, and 19% for severe TRAEs.

CONCLUSION

Neoadjuvant immunotherapy appears to be a feasible and safe approach in patients with resectable HCC, achieving moderate pathological responses and high resection rates.

Key Words: Hepatocellular carcinoma; Neoadjuvant immunotherapy; Immune checkpoint inhibitors; Liver resection; Resectable hepatocellular carcinoma; Preoperative treatment

Core Tip: This meta-analysis evaluates the efficacy and safety of neoadjuvant immunotherapy in resectable hepatocellular carcinoma (HCC). Pooled data suggest that this approach is feasible, with moderate pathological response rates and high resection feasibility, without a significant increase in severe toxicities. Although evidence is still limited, early findings support further investigation in prospective trials to define optimal treatment strategies and identify biomarkers predictive of response. Neoadjuvant immunotherapy may represent a promising strategy to improve surgical outcomes and reduce early recurrence in resectable HCC.

INTRODUCTION

Hepatocellular carcinoma (HCC) is the fourth leading cause of cancer-related death worldwide[1,2]. Its development is associated with several risk factors, including chronic alcohol abuse, hepatotropic viruses such as hepatitis C virus and hepatitis B virus, metabolic dysfunction-associated steatotic liver disease and metabolic dysfunction-associated steatohepatitis, and metabolic syndrome, as well as rare genetic disorders[1,3].

Over the past two decades, the therapeutic landscape of HCC has evolved significantly across all disease stages[4]. Major advances include the optimization of liver transplantation protocols, the widespread adoption of minimally invasive and robotic surgical techniques, and the development of novel systemic therapies—most notably, immunotherapy[5-7]. In addition, the implementation of liver hypertrophy-inducing strategies, such as portal vein embolization and associating liver partition and portal vein ligation for staged hepatectomy, has expanded the pool of patients eligible for curative liver resection by increasing the future liver remnant volume[8,9].

Surgical resection remains the mainstay curative option for patients with HCC[7]. However, tumor resectability depends on multiple factors, including the tumor’s size and location, the presence of macrovascular invasion, portal hypertension, liver function (e.g., the presence of cirrhosis), and the patient’s overall clinical condition[2,10]. Despite its curative intent, the long-term outcomes remain suboptimal. The five-year overall survival (OS) rates after resection is around 70%. Tumor recurrence remains a major concern, occurring in up to 50%-70% of patients within five years after surgery, with early recurrence (within two years) being particularly associated with poor prognosis[7,11,12].

While adjuvant therapies have been partially investigated, the role of neoadjuvant strategies in the management of HCC —particularly immunotherapy—has remained largely unexplored. Emerging evidence suggests the potential of neoadjuvant immunotherapy to achieve tumor downstaging in initially unresectable cases and to reduce the risk of recurrence in resectable disease[6,13-15].

Despite these promising preliminary findings, the role of neoadjuvant immunotherapy in resectable HCC remains poorly defined, with no standardized protocols or robust consensus on its clinical utility. In this context, a comprehensive meta-analysis of the available evidence is warranted to better understand the efficacy, feasibility, and safety profile of this emerging strategy. This study aims to fill this gap by synthesizing the current data on neoadjuvant immunotherapy in resectable HCC.

MATERIALS AND METHODS

Search strategy and study selection

A systematic literature search was conducted using PubMed, Web of Science, EMBASE, Cochrane Library, and Scopus to identify studies published in the last five years (January 2021 to May 2025) that evaluated the use of neoadjuvant immunotherapy in patients with resectable HCC[16,17]. The following search string was used: “(neoadjuvant therapy) AND (hepatocellular carcinoma) AND (immunotherapy)”.

Study screening and selection were conducted using Rayyan, an online platform for managing systematic reviews. The eligible studies included clinical trials and retrospective analyses investigating neoadjuvant immunotherapy prior to liver resection in resectable HCC patients.

The primary endpoints assessed were the pathological complete response (pCR), major pathological response (MPR), overall response rate (ORR), grade 3-4 treatment-related adverse events (TRAEs) and the rate of successful resection.

The following exclusion criteria were applied: (1) Non-English language publications; (2) Case reports, editorials, and review articles; (3) Studies not involving neoadjuvant immunotherapy; (4) Studies examining unresectable HCC or other treatment settings (e.g., liver transplantation or transarterial chemoembolization); (5) Studies evaluating neo-adjuvant immunotherapy strategies incorporating cancer vaccines; and (6) Studies that did not report at least three of the five primary endpoints of interest (MPR, pCR, ORR, grade 3-4 TRAEs, resection rate).

Statistical analysis

For studies reporting single-arm binary outcomes (e.g., the pCR, MPR, ORR, or TRAEs), the log odds [ln(odds)] and corresponding standard error (SE) were calculated as follows: Ln(odds) = ln[X/(n - X)], SE = Ö1/X + [1/(n - X)]. Here, X is the number of events and n is the total number of patients. Studies with zero or full event counts were adjusted using continuity correction where necessary[18].

Meta-analyses were conducted using Stata BE 19. The pooled effect sizes were estimated using a random-effects model due to the clinical and methodological heterogeneity among the included studies. Between-study heterogeneity was assessed using Cochran’s Q-test, the I² statistic, and τ². Forest plots were generated to visually summarize the individual study estimates and their corresponding 95% confidence intervals (95%CIs). The pooled effects were computed on the log(odds) scale and subsequently back-transformed into odds ratios (ORs) for interpretability[19]. In addition to the log-odds-based analysis, the pooled proportions were calculated for each binary outcome. This method allowed for the direct estimation of the overall event rates (expressed as percentages) using a random-effects model and a direct estimate of the treatment efficacy and safety as percentages, complementing the OR analysis.

In one included study evaluating two immunotherapy regimens in separate arms, each arm was treated as an independent entry and analyzed separately in the meta-analysis.

Subgroup analyses

To further explore the heterogeneity in the treatment response, a subgroup analysis was conducted for the ORR, pCR, and resection rate based on the type of immune checkpoint inhibitor (ICI) regimen—monotherapy vs combination therapy. The studies were grouped accordingly, and the log ORs and their SEs were pooled using a random-effects model. For interpretability, the results were subsequently back-transformed and reported as ORs with 95%CIs. In a second subgroup analysis, the studies were also stratified based on the number of ICI cycles administered before surgery (≤ 2 cycles vs > 2 cycles). This allowed us to explore whether the treatment duration accounted for the heterogeneity in response outcomes. The log ORs and corresponding SEs were pooled within each subgroup using a random-effects model, and the differences between the subgroups were tested for heterogeneity using a Q-test.

Publication bias assessment

To evaluate the potential small-study effects and publication bias, funnel plots were visually inspected for asymmetry. In addition, Egger’s regression test was performed for each primary endpoint by regressing the standardized effect sizes on their precision. A statistically significant non-zero intercept (β₁) was considered indicative of potential bias, with a P value < 0.05 suggesting asymmetry.

RESULTS

After the removal of duplicates, a total of 395 studies were screened. Following the application of inclusion and exclusion criteria, 12 studies were deemed eligible and included in the final meta-analysis[20-31] (Figure 1; Table 1). Notably, the study by Kaseb et al[23] was included as two separate entries, each corresponding to a distinct treatment arm.

Figure 1 PRISMA flow diagram. Flow diagram summarizing the study selection process. A total of 395 studies were initially identified, of which 12 met the inclusion criteria and were included in the final meta-analysis.

Table 1 Overview of studies included in the meta-analysis.

Ref.	Sample size	Study phase	Study type	Article type	Randomization	NI	Cycles of NI	ICI post-surgery
Bai et al[20], 2022	24	2	Clinical trial	Conference abstract	Randomized	Camrelizumab + apatinib (n = 13)	4	Yes
Chen et al[21], 2022	11	2	Clinical trial	Conference abstract	Non-randomized	Tislelizumab	2	Yes
Cheung et al[30], 2024	20	/	Clinical trial	Full text	/	Nivolumab	3	/
D'Alessio et al[22], 2022	17	1b	Clinical trial	Conference abstract	/	Nivolumab + ipilimumab	2	/
Kaseb et al[23], 2022	30	2	Clinical trial	Full text	Randomized	Nivolumab (n = 13) or nivolumab + ipilimumab (n = 14)	3	Yes
Marron et al[27], 2022	21	2	Clinical trial	Full text	/	Cemiplimab	2	Yes
Ming et al[31], 2024	14	2	Clinical trial	Conference abstract	/	Tislelizumab + lenvatinib	2	Yes
Shi et al[24], 2021	18	1b/2	Clinical trial	Conference abstract	Randomized	Toripalib (n = 10) or toripalib + lenvatinib (n = 8)	1	Yes
Song et al[25], 2023	24	2	Clinical trial	Conference abstract	/	Tislelizumab + lenvatinib	4	Yes
Sun et al[26], 2022	30	2	Clinical trial	Conference abstract	/	Sintilimab + bevacizumab	N/A	/
Wang et al[28], 2022	20	/	Retrospective study	Full text	/	Apatinib + camrelizumab	N/A	/
Xia et al[29], 2022	18	2	Clinical trial	Full text	/	Camrelizumab +apatinib	3	Yes

Summary of study characteristics, including sample size, study type, immunotherapy regimen, number of cycles, and postoperative treatment. NI: Neoadjuvant immunotherapy; ICI: Immune checkpoint inhibitor.

Primary endpoints

The pooled OR for achieving MPR was 0.28 (95%CI: 0.19-0.41), indicating a relatively low but consistent response rate across the studies. For the ORR, the pooled OR was 0.54 (95%CI: 0.25-1.14), indicating a modest overall response to neoadjuvant immunotherapy. Substantial heterogeneity was observed (I² = 71.4%, τ² = 0.89, P = 0.0004). The pooled OR for the pCR was 0.26 (95%CI: 0.11-0.66), suggesting limited complete response rates. Notably, the heterogeneity was high (I² = 76.5%, τ² = 1.57, P < 0.0001; Figure 2A-C). In contrast, the resection rate showed a significantly increased pooled OR of 5.37 (95%CI: 2.70-10.66), suggesting feasibility after immunotherapy. Moderate heterogeneity was present (I² = 62.3%, τ² = 0.87, P < 0.0006). Regarding the grade 3-4 TRAEs, the pooled OR was 0.33 (95%CI: 0.22-0.50), suggesting an overall acceptable safety profile in terms of severe toxicities (Figure 2D and E; Table 2).

Figure 2 Forest plot. A: Major pathological response. Log(odds) forest plot for major pathological response following neoadjuvant immunotherapy in resectable hepatocellular carcinoma. Pooled odds ratio (OR) = 0.28 [95% confidence interval (95%CI): 0.19-0.41]; B: Overall response rate (ORR). Log(odds) forest plot for ORR following neoadjuvant immunotherapy. Pooled OR = 0.54 (95%CI: 0.25-1.14); C: Pathological complete response (pCR). Log(odds) forest plot for pCR in patients treated with neoadjuvant immunotherapy. Pooled OR = 0.26 (95%CI: 0.11-0.66); D: Resection rate. Log(odds) forest plot for resection rate after neoadjuvant immunotherapy. Pooled OR: 5.37 (95%CI: 2.70-10.66); E: Treatment-related adverse events (TRAEs). Log(odds) forest plot for grade 3-4 TRAEs. Pooled OR = 0.33 (95%CI: 0.22-0.50). OR: Odds ratio; 95%CI: 95% confidence interval.

Table 2 Meta-analysis summary.

Endpoint	Pooled OR	95%CI lower	95%CI upper	I² (%)	τ²	P value (Q test)
MPR	0.28	0.19	0.41	0.0	0.0	0.8188
ORR	0.54	0.25	1.14	71.4	0.89	0.0004
pCR	0.26	0.11	0.66	76.5	1.57	< 0.0001
Resection rate	5.37	2.70	10.66	62.3	0.87	< 0.0006
Grade 3-4 TRAEs	0.33	0.22	0.50	3.5	0.02	0.4725

Pooled ORs and 95%CIs for each outcome using a random-effects model. Heterogeneity is assessed via I², τ², and Cochran’s Q test. OR: Odds ratio; 95%CI: 95% confidence interval; pCR: Pathological complete response; MPR: Major pathological response; ORR: Overall response rate; TRAE: Treatment-related adverse event.

In parallel with the OR estimates, the pooled event rates were calculated to provide absolute measures of the efficacy and safety across studies, expressed as percentages, yielding estimated rates of 19% for the MPR, 35% for the ORR, 22% for the pCR, 81% for the resection feasibility, and 19% for the grade 3-4 TRAEs (Table 3; Figure 3).

Figure 3 Radar plot of pooled event rates across clinical endpoints. Radar plot displaying pooled event rates for key clinical endpoints: 19% for major pathological response, 35% for overall response rate, 22% for pathological complete response, 81% for resection rate, and 19% for grade 3-4 treatment-related adverse events. pCR: Pathological complete response; MPR: Major pathological response; ORR: Overall response rate; TRAE: Treatment-related adverse event.

Table 3 Pooled event rates for clinical endpoints (%).

Endpoint	Pooled %	95%CI	I²	P value
MPR	19	13-25	0.0	0.77
ORR	35	18-52	84.1	< 0.001
pCR	22	10-34	85.5	< 0.001
Resection rate	81	72-91	75.9	< 0.001
Grade 3-4 TRAEs	19	11-26	34.2	0.16

Absolute pooled percentages with 95% confidence intervals for each endpoint (major pathological response, overall response rate, pathological complete response, resection rate, and grade 3-4 treatment-related adverse events), estimated using a random-effects model. 95%CI: 95% confidence interval; pCR: Pathological complete response; MPR: Major pathological response; ORR: Overall response rate; TRAE: Treatment-related adverse event.

Subgroup analysis by ICI regimen

The subgroup analysis for the ORR, pCR, and resection rate based on the ICI regimen showed comparable pooled estimates for both monotherapy and combination therapy. For the ORR, the pooled OR was 0.27 (95%CI: 0.10-0.71) for monotherapy and 0.65 (95%CI: 0.26-1.63) for combination therapy. For the pCR, the pooled OR was 0.41 (95%CI: 0.06-2.97) for monotherapy and 0.25 (95%CI: 0.08-0.77) for combination therapy. Regarding the resection rate, monotherapy yielded a pooled OR of 8.94 (95%CI: 2.33-34.35), while combination therapy showed a pooled OR of 4.22 (95%CI: 1.74-10.22). Although the point estimates differed numerically between the subgroups, none of the differences reached statistical significance. These findings suggest that the type of ICI regimen alone may not fully explain the heterogeneity observed across the studies in their response and resectability outcomes (Figure 4).

Figure 4 Subgroup analysis by immune checkpoint inhibitor regimen (monotherapy vs combination). A-C: No significant difference was observed between the subgroups for (A) pathological complete response, (B) overall response rate and (C) resection rate. 95%CI: 95% confidence interval.

Subgroup analysis by number of ICI cycles

An additional subgroup analysis stratified the studies based on the number of ICI cycles administered before surgery (≤ 2 vs > 2). For the ORR, the pooled OR was 0.65 (95%CI: 0.16-2.62) for > 2 cycles and 0.53 (95%CI: 0.09-2.99) for ≤ 2 cycles (P = 0.86). For the pCR, the pooled OR was 0.24 (95%CI: 0.06-0.91) for > 2 cycles and 0.39 (95%CI: 0.11-1.45) for ≤ 2 cycles (P = 0.60). Notably, the resection rate analysis showed a statistically significant difference in favor of shorter regimens: The pooled OR for ≤ 2 cycles was 10.61 (95%CI: 4.57-24.68), compared to 3.49 (95%CI: 2.04-5.95) for > 2 cycles (P for the subgroup difference = 0.03), suggesting that a shorter treatment duration may be associated with improved surgical feasibility (Figure 5).

Figure 5 Subgroup analysis by number of immune checkpoint inhibitor cycles (≤ 2 vs > 2). A and B: No significant subgroup differences were observed for (A) pathological complete response and (B) overall response rate; C: Shorter regimens (≤ 2 cycles) were associated with a significantly higher resection rate compared to longer regimens. 95%CI: 95% confidence interval.

Publication bias assessment

Finally, the assessment of small-study effects using Egger’s test and funnel plots showed no significant evidence of a publication bias for the MPR (β₁ = -2.36, P = 0.109), ORR (β₁ = -0.07, P = 0.969), or grade 3-4 TRAEs (β₁ = +0.11, P = 0.887). In contrast, statistically significant asymmetry was detected for the pCR (β₁ = -5.00, P = 0.0016) and resection rate (β₁ = +3.68, P < 0.0001), suggesting the presence of small-study effects or reporting bias in these outcomes (Table 4; Supplementary Figures 1-5).

Table 4 Assessment of publication bias using funnel plot and Egger’s regression test.

Endpoint	No. of studies	Funnel plot	β1 (Egger)	SE (β1)	P value	Interpretation
MPR	10	Symmetrical	-2.36	1.47	0.109	No evidence of small-study effects; the funnel plot does not suggest publication bias
ORR	9	Symmetrical (2 studies lie outside 95%CI limits)	-0.07	1.81	0.969	No indication of small-study effects; the distribution is balanced across effect sizes
pCR	10	Asymmetrical; small studies cluster to the left	-5.00	1.59	0.0016	Significant small-study effects; suggests potential publication bias favoring negative findings
Resection rate	13	Asymmetrical; small studies cluster to the right	+3.68	0.68	< 0.0001	Strong evidence of small-study effects; indicative of possible reporting bias or selective publication of favorable outcomes
Grade 3-4 TRAEs	11	Symmetrical	+0.11	0.79	0.887	No evidence of small-study effects; results appear robust and not influenced by study size

pCR: Pathological complete response; MPR: Major pathological response; ORR: Overall response rate; TRAE: Treatment-related adverse event.

DISCUSSION

This meta-analysis evaluated the efficacy and safety of neoadjuvant immunotherapy in patients with resectable HCC. The pooled results suggest that this approach is associated with a favorable pathological response without a significant increase in high-grade toxicity (Table 5). Specifically, the pooled MPR rate was 19% (OR = 0.28, 95%CI: 0.19-0.41), the ORR was 35% (OR = 0.54, 95%CI: 0.25-1.14), and the pCR rate was 22% (OR = 0.26, 95%CI: 0.11-0.66). The pooled resection rate was 81% (OR = 5.37, 95%CI: 2.70-10.66), and the incidence of grade 3-4 TRAEs was 19% (OR = 0.33, 95%CI: 0.22-0.50).

Table 5 Grade 3-4 treatment-related adverse events reported in included studies.

Ref.	Neoadjuvant immunotherapy	Cycles of NI	Most common grade 3-4 TRAEs reported
Bai et al[20], 2022	Camrelizumab + apatinib (n = 13)	4	Liver disfunction; thrombocytopenia; and hand-foot syndrome
Chen et al[21], 2022	Tislelizumab	2	No severe events
Cheung et al[30], 2024	Nivolumab	3	No severe events
D'Alessio et al[22], 2022	Nivolumab + ipilimumab	2	ALT/AST elevation
Kaseb et al[23], 2022	Nivolumab (n = 13) or nivolumab + ipilimumab (n = 14)	3	ALT/AST elevation
Marron et al[27], 2022	Cemiplimab	2	ALT/AST elevation; increased blood creatine phosphokinase; and hypoalbuminaemia
Ming et al[31], 2024	Tislelizumab + lenvatinib	2	No severe events
Shi et al[24], 2021	Toripalib (n = 10) or toripalib + lenvatinib (n = 8)	1	Not specified
Song et al[25], 2023	Tislelizumab + lenvatinib	4	No severe events
Sun et al[26], 2022	Sintilimab + bevacizumab	N/A	Not specified
Wang et al[28], 2022	Apatinib + camrelizumab	N/A	Biliary fistula
Xia et al[29], 2022	Camrelizumab +apatinib	3	Rash; hypertension; liver damage; and neutropenia

This table summarizes the most reported grade 3-4 treatment-related adverse events across studies evaluating neoadjuvant immune checkpoint inhibitor regimens in patients with resectable hepatocellular carcinoma. TRAE: Treatment-related adverse event; NI: Neoadjuvant immunotherapy; ICI: Immune checkpoint inhibitor; ALT/AST: Alanine/aspartate aminotransferase; N/A: Not available.

Moreover, several studies included in the analysis also reported encouraging survival outcomes: In a phase II study by Xia et al[29], using camrelizumab plus apatinib as neoadjuvant therapy led to a 1-year recurrence-free survival rate of 53.85%. Similarly, Kaseb et al[23] observed no recurrences among patients achieving an MPR after the perioperative administration of nivolumab (with or without ipilimumab) at 24 months. In addition, a retrospective analysis by Wang et al[28] demonstrated that neoadjuvant treatment with apatinib and camrelizumab significantly prolonged both the recurrence time (17.3 months vs 13.2 months, P = 0.023) and OS (22.4 months vs 19.1 months, P = 0.041) compared to a control group, with no significant increase in perioperative complications. Collectively, these results suggest that neoadjuvant immunotherapy has the potential to improve the pathological response and reduce early recurrence in resectable HCC, with a manageable safety profile.

Subgroup analyses based on the treatment regimen (monotherapy vs combination) were conducted to explore potential sources of heterogeneity. While numerical differences in the point estimates were observed across the subgroups for the ORR, pCR, and resection rate, none of these differences reached statistical significance. These findings suggest that the type of ICI regimen alone may not have been the primary driver of outcome variability, and other factors—such as the tumor burden, treatment duration, patient selection criteria, or surgical eligibility thresholds—may have contributed more substantially to the heterogeneity across studies. However, the small number of available studies, each evaluating different agents or combinations, limited the feasibility of performing direct comparisons between specific immunotherapeutic regimens and highlights the need for more consistent and comparable data. A secondary subgroup analysis stratified the studies according to the number of ICI cycles administered before surgery (≤ 2 vs > 2). Interestingly, while no significant differences emerged in the ORR and pCR between the cycle groups, the resection feasibility appeared to have been higher in the ≤ 2-cycle group (OR of 10.61, 95%CI: 4.57-24.68) than the > 2-cycle group (OR = 3.49, 95%CI: 2.04-5.95), with a significant subgroup difference (P = 0.03). These findings suggest that a shorter treatment duration may be sufficient to achieve tumor downsizing or immune priming, facilitating surgery without compromising its efficacy.

Furthermore, correlative immunological analyses by Kaseb et al[23] provided additional insights into the mechanisms underlying treatment responses. Patients who achieved a major pathological response exhibited a markedly higher density of tumor-infiltrating immune cells, particularly CD8⁺ T lymphocytes, along with an increased CD8⁺/Treg ratio within the tumor microenvironment. These findings suggest that enhanced intratumoral immune activation may play a critical role in mediating the response to neoadjuvant immune checkpoint blockade. Notably, none of the patients who achieved an MPR experienced disease recurrence at a median follow-up of approximately 24 months, supporting the prognostic value of a pathological response in this setting. These data collectively highlight the potential of using immunological biomarkers to refine patient selection and optimize the therapeutic outcomes in resectable HCC. These observations are consistent with findings from Marron et al[27], who reported that patients achieving a major pathological response following perioperative immune checkpoint inhibition demonstrated increased immune infiltration, including higher densities of CD8⁺ T cells, and an enhanced expression of immune activation signatures.

Despite promising results, neoadjuvant immunotherapy has not yet been formally incorporated into the treatment guidelines for resectable HCC. The absence of standardized protocols regarding agent selection, the treatment duration, and the timing of surgery hinders the comparability of study outcomes and complicates the assessment of clinical efficacy. Additionally, definitions of MPR vary across studies, with different thresholds and histological criteria adopted, further limiting interpretability.

Moreover, substantial heterogeneity was observed across the studies for key endpoints such as the ORR, pCR, and resection rate, which may reflect differences in the treatment regimens, patient selection, and surgical criteria. Egger’s test indicated small-study effects suggestive of potential bias for the pCR (β₁ = -5.00, P = 0.0016) and resection rate (β₁ = +3.68, P < 0.0001), while no significant asymmetry was detected for the MPR, ORR, or grade 3-4 TRAEs.

A key limitation of this meta-analysis is the small number of available studies, many of which were preliminary and reported only as conference abstracts, lacking a full, peer-reviewed methodology. This underscores the early stage of clinical development in this field and highlights the need for high-quality, prospective trials to better define the role of neoadjuvant immunotherapy in the curative setting of resectable HCC.

CONCLUSION

This meta-analysis supports the feasibility and potential clinical value of neoadjuvant immunotherapy in patients with resectable HCC. While the pCR and MPR remain limited, the treatment demonstrates a favorable safety profile and high resection rates. These findings suggest that neoadjuvant immunotherapy may be a promising strategy to improve surgical outcomes and long-term disease control. However, larger, well-designed prospective trials are essential to validate these preliminary results and define standardized treatment protocols, optimal timing, and predictive biomarkers for response.