The “shell game” after objective response in patients with advanced HCC treated with immunotherapy

The advent of immune checkpoint inhibitors (ICIs) has revolutionized the landscape of HCC treatment with longer survivals and higher response rates in comparison to tyrosine kinase inhibitors. Although the detection of complete response (CR) is associated with an obvious positive outcome, its occurrence is still uncommon in patients with HCC. Nevertheless, if its achievement is a clear favorable event, what is the optimal management after the objective response (partial response and CR) remains to be defined in a case-by-case and multidisciplinary setting: (1) conversion to more radical treatments (locoregional treatment, surgery, liver transplantation), (2) ICIs continuation, or (3) discontinuation (Figure 1). Evidence in HCC is scanty, but a recent post hoc analysis of 2 prospective Chinese studies combining 2 locoregional treatments (transarterial chemoembolization and stereotactic body radiotherapy) and ICIs in patients with unresectable HCC, reported the long-term evolution of CR. The study showed that 29 (46%) out of 63 patients achieved CR, and ICIs were discontinued in all of them. The 3 years survival rate was 75.5%, and 10 patients developed tumor recurrence, mostly (60%) in the intrahepatic form.1 However, in patients with HCC and treated with systemic therapy alone, this information was limited.FIGURE 1: Proposed managing algorithm for intermediate/advanced HCC treated with ICI-based therapies according to the objective radiological response. Abbreviations: AFP, alpha-fetoprotein; ICI, immune checkpoint inhibitor; mRECIST, modified Response Evaluation Criteria in Solid Tumors; RECIST, Response Evaluation Criteria in Solid Tumors.In the multicenter, international, and retrospective study published in Hepatology, Scheiner et al2 reported the evolution of patients with advanced HCC that achieved CR after ICIs-based regimens. Of 3933 patients treated with immunotherapy in 28 centers across the world, 174 (4.4%) achieved CR according to modified Response Evaluation Criteria in Solid Tumors (mRECIST), and 97 (2.5%) had CR according to RECISTv1.1 (Response Evaluation Criteria in Solid Tumors). Most of the patients had an advanced HCC (70%), received immunotherapy in the first line (76%), and the most common treatment regimen was atezolizumab and bevacizumab (60%). Of note, the rate of CR is lower than in clinical trials (3%–7% according to RECIST v1.1) and this could be related to the wide heterogeneity of type and line of treatment. However, clinical outcomes were excellent, with survival rates of 98% and 86% at 1 and 3 years, respectively, after a median follow-up of 32.2 months. The median duration of response was 36.5 months, and patients who received immunotherapy for ≥6 months after response had a longer recurrence-free survival than those who discontinued immunotherapy earlier. Of the whole cohort, 128 (74%) patients discontinued immunotherapy, and the most common reason was the presence of CR (33%), followed by adverse events (13%) and tumor recurrence (12%). The results of the study are informative for both physicians and patients considering that this is the first and largest cohort of patients with CR under ICIs-based therapies. The study by Scheiner et al2 is also instrumental to enlighten the shadows hovering over the management of objective responses after ICIs. A part of the scientific community tends to consider feasible radical treatments after response, such as locoregional treatments, liver resection, or liver transplantation. Although this approach seems attractive, it is worth stating that the achievement of response does not necessarily translate into suitability for treatments with more radical intent. In this clinical context, it is crucial to acknowledge that evidence regarding sequential therapies is challenging to obtain through specifically designed randomized controlled trials. Such evidence may more feasibly emerge from well-designed prospective real-world studies.3 Furthermore, responses after ICIs are usually long-lasting, and in the study of Scheiner et al,2 duration of response after CR was 36.5 months. Data from melanoma and lung cancer can be useful since they suggest that discontinuing immunotherapy might be an alternative approach after registering CR by imaging in HCC. Indeed, the 5-year follow-up reports of 2 trials, including pembrolizumab in non–small cell lung cancer (Keynote-010 and Keynote-024 trials), revealed that 83% and 82% of patients were still alive 3 years after completing 2 years of pembrolizumab.4 Similarly, long-term data from patients with melanoma, showed that 86%–90% of them were alive after discontinuation of ICIs and remained in CR after a median follow-up duration of 5 years.5 Ideally, this issue should be properly answered by prospective studies with an intention-to-treat approach comparing the different options. Concerning the setting of liver transplantation, some additional comments are needed. Despite many trials are ongoing to assess its safety after ICIs,6 one should consider (1) the individual benefit/futility of liver transplantation in a patient with CR in comparison to other options (ie, continuing or discontinuing ICIs) while it appears promising in down-staged patients, (2) community benefit/futility and (3) the global safety profile (de novo malignancies, metabolic disorders, and nephrological complications) beside the mere risk of liver rejection. Another relevant point to comment on is the definition of CR. In the study of Scheiner et al,2 it was defined according to both mRECIST and RECIST v1.1. In the field of systemic therapy for HCC, the radiologic criteria commonly used are RECIST v1.1 since antiangiogenic drugs may induce vasoconstriction simulating a CR. This applies also to the combination of atezolizumab and bevacizumab since the general correlation between radiologic and pathologic response is low. In this sense, data from the patients treated with atezolizumab and bevacizumab showed that pathologic complete responses were 57.7% in patients with radiologic partial response and 16.7% in those with stable disease lasting for >6 months, both according to RECIST v1.1.7 Thus, a proportion of patients with partial response or stable disease may present “ghost tumors” at pathology. In the study of Scheiner et al,2 13 out of 174 patients underwent “conversion therapies” (ablation, surgery, or liver transplantation) after ICIs, and 9 of them were surgical treatments. Interestingly, 8 of the 9 with pathology examination exhibited complete pathologic response while in one case, the tumor was still viable. In any case, the number of patients with available pathologic evaluation is extremely low to provide any association between radiologic and pathologic response. Hopefully, baseline or evolutionary biomarkers and artificial intelligence may aid to identify patients with pathologic responses in the future. Besides, as exposed by the authors, one of the main limitations resides in the lack of a central review for both radiology and pathology. In this regard, Rimola et al8 evaluated patients with CR under sorafenib and excluded 40% of the cohort after central revision. This suggests that central review is crucial in the identification of an uncommon event such as CR. Furthermore, it is well known that radiologic intermediate outcomes are suboptimal surrogates for overall survival in the field of HCC since their correlation is not linear.9 One of the best examples is progression-free survival, which does not consider either the pattern of progression or the decompensation of liver cirrhosis.10–12 To conclude, the study by Scheiner et al2 shows for the first time the excellent long-term outcomes of patients with HCC achieving CR with ICIs in a real-life multicenter cohort. In parallel, the study underlines the unmet need for open-ended management after objective response achievement (Figure 1). Likely, this will be one of the challenges of the research in HCC in the next years, as the scientific community strives to establish evidence-based sequential treatment strategies in an era of expanding therapeutic options. Finally, prospective studies are still needed to identify treatment response biomarkers that can better help in clinical practice to guide treatment decisions.