Deep Response as a Novel Efficacy Endpoint in Hepatocellular Carcinoma

The prognostic significance of depth of response (DpR) has been demonstrated in several solid tumors, including colorectal cancer [1‒3], melanoma [4], breast cancer [5], pancreatic cancer [5], renal cell cancer [6, 7], head and neck cancer [8], gastric cancer [9], and non-small cell lung cancer [5, 10‒12]. However, in hepatocellular carcinoma (HCC), the concept of DpR has gained attention only recently [13] with the advent of immune checkpoint inhibitor-based therapies, which have enabled profound tumor shrinkage and long-term overall survival (OS) in a subset of patients achieving deep responses. Owing to recent advances in systemic therapies, particularly immunotherapy-based treatments, not only conventional RECIST-defined objective responses but also profound tumor shrinkage exceeding 50% (DpR50), 60% (DpR60), and even 75% (DpR75) can now be achieved in a considerable proportion of patients. Furthermore, accumulating evidence has demonstrated a strong correlation between DpR and OS [13, 14]. These findings suggest that achieving a deep response through systemic therapy may serve as a predictive biomarker for favorable survival outcomes. In particular, patients achieving deep responses such as DpR50, DpR60, or DpR75 consistently exhibit exceptionally favorable survival outcomes, highlighting the clinical importance not only of response itself but also of the proportion of patients achieving such profound tumor shrinkage.Accordingly, achieving a deep response should itself be regarded as one of the therapeutic goals, and the deep response rate (DRR) may emerge as a novel efficacy endpoint for multidisciplinary treatment strategies. Conventional efficacy endpoints have included objective response rate (ORR), progression-free survival (PFS), time to progression (TTP), OS, time to response (TTR), and duration of response (DOR). However, future clinical trials and real-world clinical practice may increasingly incorporate achievement of deep response and high DRR (e.g., DRR50, DRR60, and DRR75) as additional novel efficacy endpoints for systemic therapies and combination approaches integrating systemic and locoregional treatments.Moreover, it has become evident that even among patients categorized as having stable disease (SD), those who achieve any degree of tumor shrinkage within the SD range (“SD with tumor shrinkage”) have significantly better prognoses than those without any tumor shrinkage (“SD with no tumor shrinkage”) [13, 15]. These observations indicate that SD, which has traditionally been regarded as a single response category, actually comprises two biologically and prognostically distinct subgroups. In other words, even within the population historically considered to derive clinical benefit from treatment because of SD status, there exists a subgroup that gains little or no therapeutic benefit. Early identification of such patients is therefore critical in order to facilitate prompt transition to alternative therapeutic strategies, including locoregional interventions or subsequent systemic therapies. Indeed, patients with SD without tumor shrinkage have also been reported to derive no benefit from conversion surgery [16].Taken together, achievement of a deep response represents an important indicator of favorable survival and should be considered a novel therapeutic objective in HCC management, in addition to existing efficacy endpoints. In this editorial, it is discussed that the potential role of deep response and DRR as emerging efficacy endpoints in future clinical trials and real-world clinical practice.An exploratory analysis of the IMbrave150 trial demonstrated a strong correlation between DpR and OS in patients with unresectable HCC treated with atezolizumab plus bevacizumab (Atezo + Bev) [13]. The IMbrave150 trial was a pivotal global phase III study establishing Atezo + Bev as the standard first-line therapy for advanced HCC by demonstrating a significant OS benefit over sorafenib [17] The DpR analysis, conducted as an exploratory analysis, clearly demonstrated that the “depth” of tumor shrinkage is closely associated with long-term survival [13].DpR was defined as the maximum percentage reduction in tumor volume from baseline according to RECIST 1.1 and represents a quantitative measure of tumor shrinkage beyond the conventional RECIST categories of PR, SD, and PD. In this analysis, tumor shrinkage was observed in 74% of patients treated with Atezo + Bev, markedly higher than the 54% observed in the sorafenib group. Moreover, the magnitude of tumor reduction was substantially greater in the Atezo + Bev arm, with a mean DpR of −42.5% compared with −25.0% in the sorafenib arm.Importantly, DpR showed a strong association with OS. In the 6-month landmark analysis, patients who achieved any degree of tumor shrinkage had significantly prolonged OS compared with those without tumor shrinkage, with an approximately 70% reduction in the risk of death (HR 0.29). Median OS was 24.9 months in patients with tumor shrinkage versus only 14.0 months in those without shrinkage.Furthermore, stratification of DpR into five categories clearly demonstrated that deeper tumor shrinkage translated into progressively better survival outcomes. In particular, patients achieving deep responses (DpR −100% to −60%: DpR60) exhibited remarkably favorable survival outcomes, with a 92% reduction in mortality risk compared with patients exhibiting progressive disease (HR 0.08) (Fig. 1). These findings clearly indicate that prognosis is determined not only by whether a RECIST response is achieved, but rather by how deeply the tumor burden is reduced. Under RECIST v1.1 criteria, both 30% and 60% tumor reductions are categorized simply as PR; however, this analysis demonstrated substantial prognostic differences between these levels of response. Likewise, the SD category encompasses biologically heterogeneous populations, including patients with mild tumor shrinkage and those with slight tumor growth, thereby highlighting the limitations of conventional RECIST criteria in HCC [13, 15].The study also evaluated early tumor shrinkage (ETS). Patients achieving ≥30% tumor reduction at the first radiologic assessment demonstrated a 67% reduction in mortality risk (HR 0.33).Thus, both “rapid tumor shrinkage” and “deep tumor shrinkage” were strongly associated with improved long-term outcomes. This represents an important finding indicating that early response to immunotherapy predicts subsequent therapeutic benefit. Furthermore, these data provide prospective clinical evidence supporting the long-held concept that shorter TTR reflects superior therapeutic efficacy.In addition, the study discussed the clinical relevance of DpR in the context of conversion therapy. With the advent of highly active regimens such as Atezo + Bev, an increasing number of patients with initially unresectable HCC can now undergo curative-intent locoregional therapies (i.e., curative conversion), including surgical resection, ablation, and curative transarterial chemoembolization (TACE), following substantial tumor shrinkage [18‒20]. Particularly among patients achieving deep responses, achievement of cancer-free and drug-free status has extensively been reported [18‒20]. Therefore, achievement of a deep response is clinically meaningful not only as an efficacy endpoint but also as a practical indicator of whether patients may be bridged to potentially curative treatment strategies.Taken together, this study established deep response as an important prognostic marker in unresectable HCC. In particular, it clearly demonstrated that achieving deeper responses with Atezo + Bev is associated with prolonged OS [13, 15]. This relevant analysis further suggested that the achieving deep response as well as the high DRR may emerge, alongside ORR, as a novel and clinically meaningful efficacy endpoint for evaluating HCC therapies in the era of immunotherapy.The HIMALAYA trial [21] was a global phase III study evaluating the efficacy and safety of durvalumab plus tremelimumab versus sorafenib in patients with unresectable HCC. The recently reported 5-year follow-up analysis further demonstrated a clear association between DpR and OS [14]. In particular, among patients treated with the STRIDE regimen (single tremelimumab regular interval durvalumab), deeper tumor shrinkage was associated with progressively improved long-term survival outcomes, suggesting that conventional RECIST v1.1 response categorization alone may be insufficient to fully capture the clinical benefit of immunotherapy, especially anti-CTLA-4 antibody containing regimen [14].Analysis according to the degree of tumor shrinkage demonstrated a stepwise improvement in 5-year survival rates with increasing DpR in the STRIDE arm. Patients achieving 0% to <30% tumor shrinkage had a 5-year survival rate of 11.1%, whereas those with 30% to <50% shrinkage had a 5-year survival rate of 30.7%. Survival outcomes were even more remarkable among deep responders: patients achieving ≥50% tumor shrinkage had a 5-year survival rate of 63.4%, while those achieving ≥75% shrinkage reached an extraordinary 5-year survival rate of 83.3% [14]. These findings clearly established DpR as an important predictor of long-term survival.Furthermore, Kaplan-Meyer curve evaluating the association between DpR and OS demonstrated that median OS was not reached in patients treated with STRIDE who achieved ≥50% or ≥75% tumor shrinkage, indicating exceptionally favorable long-term survival outcomes. Based on these observations, the authors argued that the conventional RECIST v1.1 threshold of 30% tumor reduction for defining response may not represent the optimal efficacy assessment in the era of immunotherapy for HCC. In particular, it was emphasized that the importance of evaluating DpR as a continuous variable, especially for immunotherapeutic regimens, which contains CTLA-4 antibody such as STRIDE. Indeed, patients achieving deep responses of ≥50% tumor reduction following CTLA-4 antibody-based therapy demonstrated outstanding long-term survival, which suggests cases suggestive of potential cure are included.Overall, the 5-year follow-up data from the HIMALAYA trial demonstrated a strong association between DpR and OS and further showed that patients achieving deep responses, including ≥50% and ≥75% tumor shrinkage, experienced exceptionally favorable survival outcomes [14]. These findings indicate that treatment goals and efficacy assessments in immunotherapy for HCC should extend beyond conventional RECIST-based response evaluation and incorporate quantitative assessment of tumor shrinkage degree as a critical determinant of long-term survival benefit.The CheckMate 9DW trial [22] was a global phase III study evaluating the efficacy and safety of nivolumab plus ipilimumab as first-line therapy for patients with unresectable HCC, compared with lenvatinib [23] or sorafenib [24]. The study met its primary endpoint, demonstrating a significant improvement in OS with nivolumab plus ipilimumab, leading to global regulatory approval of this regimen. Tumor response outcomes were also highly favorable. According to blinded independent central review, the ORR was 36% in the nivolumab plus ipilimumab arm, substantially exceeding the 13% observed in the lenvatinib or sorafenib arm. Moreover, the DOR was remarkably prolonged, with a median duration of 30.4 months [22].Exploratory analyses from the CheckMate 9DW trial also evaluated the distribution of DpR [25]. Tumor shrinkage was achieved in 74% of patients overall, while profound tumor shrinkage of ≥50% (DpR50) was observed in 37% of patients. Furthermore, very deep responses with ≥75% tumor shrinkage (DpR75) were achieved in 18% of patients.Importantly, analyses correlating DpR with OS demonstrated that patients achieving ≥60% tumor shrinkage (DpR60) had exceptionally favorable outcomes, with median OS not reached. In contrast, patients with moderate tumor shrinkage of 30%–60% had a median OS not reached, those categorized as SD with tumor shrink (0 to 30% reduction) had a median OS of 30.0 months, those categorized as SD with no tumor shrinkage (0 to 20% growth) had a median OS of 22.9 months and patients with tumor growth >20% had a median OS of only 15.5 months [26]. Thus, progressively deeper tumor shrinkage was clearly associated with increasingly prolonged survival outcomes. Notably, among the 290 evaluable patients who underwent response assessment in the nivolumab plus ipilimumab arm, the DRR for ≥60% tumor shrinkage (DRR60) reached 30%. These findings further support the concept that deep response represents not merely a radiologic observation, but a clinically meaningful surrogate of long-term survival benefit in the era of immunotherapy for HCC.The TALENTACE trial [27] was a multinational, multicenter, randomized phase III study conducted in China and Japan comparing on-demand TACE combined with Atezo + Bev versus on-demand TACE alone. The study population primarily consisted of patients with BCLC-B HCC; however, patients with Vp1/Vp2 portal vein invasion, PS1 status, and solitary large HCC were also included, representing a broad population generally considered candidates for TACE. As a result, approximately 20% of patients had BCLC-A disease, 60% had BCLC-B disease, and 20% had BCLC-C disease.Importantly, the primary endpoint of the trial was not conventional RECIST-based PFS, but rather TACE-specific PFS (TACE-PFS), a composite endpoint defined by the occurrence of any of the following events: (1) ≥50% increase in intrahepatic tumor burden relative to baseline; (2) worsening of macrovascular invasion (MVI); (3) development of extrahepatic metastasis; (4) transient deterioration of liver function to Child–Pugh class B8; or (5) fulfillment of the criteria for TACE failure/refractoriness [28]. This endpoint basically followed those previously implemented in the TACTICS [29‒31] and TACTICS-L trials [32]. It was specifically designed to avoid premature discontinuation of protocol treatment based solely on RECIST-defined progression, thereby minimizing dilution of OS benefit caused by crossover effects resulting from subsequent therapies in both treatment arms. Indeed, both the TACTICS and TACTICS-L studies demonstrated that TACE-PFS correlated well with OS, supporting the clinical relevance of this endpoint [31, 33]. In the TALENTACE trial, TACE-PFS in the TACE plus Atezo + Bev arm was significantly prolonged compared with TACE alone, resulting in this trial a positive one.Most notably, DpR achieved with the addition of TACE to Atezo + Bev demonstrated a striking association with OS [34]. Although based on a 6-month landmark analysis, profound tumor shrinkage of 60%–100% (DpR60) was achieved in 13% of patients (DRR60 = 13%), while objective responses corresponding to modest response (DpR 30–60%) were observed in 65% of patients, with no cases of PD identified (Fig. 2). These findings indicate that the previously established association between DpR and OS achieved by ICI-based systemic therapy alone also extends to deep responses achieved through the addition of TACE to immunotherapy. In this respect, the significance of this analysis is highly novel. Although the DRR60 of 13% was relatively modest, this may partly reflect the fact that the analysis was limited to a 6-month landmark assessment.Importantly, this represents the first analysis demonstrating that deep responses achieved not only by immunotherapy alone, but also through the addition of locoregional therapies such as TACE, are associated with prolonged long-term survival. These findings further raise the possibility that similar associations may also apply to other multidisciplinary strategies combining immunotherapy with locoregional therapies, including surgical resection, radiotherapy, and Yttrium-90 selective internal radiation therapy (Y-90 SIRT).Unlike TALENTACE, DpR analyses have not been performed in the EMERALD-1 [35], LEAP-012 [36], or EMERALD-3 trials [37]. Whether the observed association between deep response and improved clinical outcomes can be replicated in these studies therefore remains to be determined.Emerging evidence suggests that deep radiological responses to immune checkpoint inhibitor-based therapy are closely linked to profound pathological tumor regression. Wang et al. [38] reported that deep responders to Atezo + Bev or PD-1 inhibitor-based therapy frequently achieved pathological complete response (pCR) or major pathological response with little or no residual viable tumor. These responses were associated with an immune-active tumor microenvironment characterized by abundant CD8+ T-cell infiltration, elevated PD-L1 expression, reduced Treg-mediated immunosuppression, and enrichment of tertiary lymphoid structures.Consistent with these findings, Sato et al. [39] demonstrated that some patients treated with Atezo + Bev achieved pCR despite the persistence of residual lesions on imaging. Histologically, these lesions consisted of complete tumor necrosis, hemorrhagic necrosis, lymphocytic infiltration, hemosiderin deposition, and fibrosis, indicating that residual radiographic abnormalities may represent treatment-induced tissue remodeling rather than viable tumor.Similarly, Cheung et al. [40] reported that deep responders to nivolumab frequently exhibited major pathological response (≥60% tumor necrosis), including pCR and near-pCR. These pathological responses were characterized by extensive hemorrhagic necrosis, dense immune-cell infiltration, increased CD8+ cytotoxic T-cell infiltration, reduced CD4 memory resting T cells, and durable recurrence-free survival.Taken together, these observations suggest that deep radiological response may represent the imaging correlate of a robust immune-mediated pathological response. The close association between profound tumor shrinkage (deep response), favorable immune remodeling of the tumor microenvironment, and pathological tumor necrosis provides a biological rationale for the strong link between DpR and long-term survival in patients with HCC receiving immunotherapy.Multiple prospective phase III trials have now demonstrated that the DpR achieved with ICI-based systemic therapy for HCC strongly correlates with OS both in ICI plus Anti-VEGF and ICI plus Anti-CTLA-4 regimens. CTLA-4-containing regimens appear to induce deep responses through enhanced T-cell priming and intratumoral regulatory T-cell depletion. In addition, CTLA-4 blockade promotes the generation of central memory T cells, which are thought to contribute to durable long-term tumor control even in patients who do not achieve a complete response. In contrast, PD-1/PD-L1 blockade predominantly reinvigorates pre-existing exhausted T cells, whereas VEGF blockade promotes vascular normalization, enhances T-cell infiltration, and alleviates immunosuppression within the tumor microenvironment. Together, these complementary mechanisms may account for the deep and durable clinical benefit achieved with combined PD-1/PD-L1 and CTLA-4/VEGF blockade.Patients achieving extremely deep responses, such as ≥75% tumor shrinkage, exhibit more favorable prognoses than those achieving ≥50% or ≥60% tumor shrinkage, while patients with ≥50% or ≥60% tumor shrinkage consistently demonstrate superior OS compared with those achieving only moderate responses of 30%–50% (or 30%–60%). In addition, further deepening of tumor response through the addition of locoregional therapy to systemic treatment can further improve OS (Fig. 3), as demonstrated in the TALENTACE trial (Fig. 2).In summary, achievement of deep response itself should be regarded as an important determinant of favorable long-term survival in both future clinical trials and real-world clinical practice. At the same time, deep response is likely to emerge as one of the key therapeutic goals in the management of HCC in the era of current immunotherapy-based multidisciplinary treatment strategies. In conclusion, deep response as well as high DRR may emerge as a novel efficacy endpoint, complementing conventional efficacy endpoints such as ORR, PFS, TTP, TTR, DOR, and OS.No statement is needed because this study was based exclusively on published data.Lectures: Chugai, Eisai, AstraZeneca. Grants: Otsuka, Chugai, Advisory Consulting: Chugai, Roche, Eisai, AstraZeneca. Masatoshi Kudo is the Editor-in-Chief of Liver Cancer.There was no funding for this editorial.Masatoshi Kudo conceived, wrote, and approved the final manuscript.Data availability is not applicable because this is not a research article.