Time to evolve trial design after everolimus failure

Abstract

Hepatocellular carcinoma is a difficult-to-treat cancer and, after numerous phase III trials assessing kinase inhibitors have failed to meet their end points, sorafenib is the only accepted treatment for advanced stages of the disease. Now, the trial EVOLVE-1 has shown a lack of benefit for everolimus in the second-line treatment setting.

Hepatocellular carcinoma (HCC) is a highly complex cancer to treat. The incidence of this disease is growing globally, especially in Asia, currently representing the second-cause of cancer-related death.¹ Apart from curative therapies (resection, transplantation and ablation) established for 30% of patients diagnosed at early stages of the disease, management of more-advanced stages has been challenging.² Randomized controlled studies conducted during three decades (1970–2000) have been mostly under-powered, inconclusive or negative, with the exception of studies assessing transarterial chemoembolization (TACE), in which positive data from two randomized trials and one meta-analysis led to clinical practice guidelines to support this intervention in patients with intermediate-stage HCC.³ Therefore, in the past decade, the landscape of evidence-based medicine in HCC was in stark contrast to the standards adopted in other cancers, such lung, breast or colon cancer.

In 2007, sorafenib emerged as the first systemic therapy for HCC, with proven efficacy at the highest level of evidence, expanding survival from 7.9 months to 10.7 months (HR, 0.69) with manageable adverse events.⁴ This breakthrough changed the standard of care for advance-stage HCC globally and paved the way for testing novel molecular therapies with the hope that the complex and heterogeneous natural history of HCC could be finally curbed. However, the results reported since then have proven otherwise. Twelve phase III trials testing drugs in the adjuvant setting (sorafenib, retinoids), early stage HCC (radiofrequency plus chemotherapy), intermediate-stage HCC (brivanib in combination with TACE), and in advanced stages at first line (sunitinib, linifanib, brivanib, and erlotinib) and second line (brivanib, everolimus and ramucirumab) failed to meet their primary end points. There are several possible reasons for these failures. Firstly, there is a suboptimal understanding of oncogenic drivers and molecular subclasses in HCC. Secondly, liver cirrhosis underlies HCC in 80% of cases, and drugs that can be effective in patients without cirrhosis might induce liver toxicity in patients with this condition (such as sunitinib and linifanib). Thirdly, signs of efficacy from phase II studies are mostly generated from single-arm trials with small sample sizes, in which results are difficult to interpret. Finally, all drugs showed suboptimal antitumoural efficacy to demonstrate survival advantages when ‘all comer’ populations were included (that is, all patients, as opposed to subsets of patients selected according to biomarkers).⁵

On the basis of ample evidence supporting the role of mTOR activation in HCC,⁶ the EVOLVE-1 study tested the efficacy of the mTOR inhibitor everolimus in patients with tumour progression after treatment with sorafenib or who were intolerant to this drug.⁷ This properly powered (n = 546 patients), well-designed study expected a 2-month benefit in overall survival in the patients assigned in a 2:1 randomization mode to receive everolimus when compared with placebo (7 months versus 5 months, respectively). However, the study failed to meet its end point as the final results reported a median overall survival of 7.6 months for the patients in the everolimus group and 7.3 months for the patients in the placebo group (HR, 1.05), with no differences in time to progression.⁷ Treatment–related adverse events were manageable, but led to drug discontinuation in 17% of cases. Noticeably, everolimus led to hepatitis B virus (HBV) reactivation in 37% of cases, despite preventive antiviral therapies. In pre-specified subgroup analysis, patients with HBV-related HCC showed better outcome with everolimus, but further adjustments and multivariate analysis are required to confirm that HBV was not a confounding factor. The authors state that biomarkers predicting response are currently being explored, but they recognize that EVOLVE-1 was not designed to identify molecularly or clinically selected populations that might have benefited from everolimus. Thus, no rescue strategy is envisioned.

This study adds weight to previous investigations testing drugs in HCC and poses a red flag for the current study design. Randomized pivotal phase III trials testing drugs in ‘all comers’ will still be conducted. There might be a rationale to do so, for example, if the tested drug is blocking key drivers of tumorigenesis universally activated in HCC or if randomized phase II trials show outstanding signs of efficacy. For most cases, however, a different strategy needs to be sought. In the first place, we should design proof-of-concept trials to test drugs in subsets of patients with tumours that are ‘addicted’ to specific driver mutations (such as mutations in RAS, or amplifications in FGF19 or VEGFA), just as lung cancers harbouring ALK fusion respond to crizotinib.⁸ A second strategy would be to design trials with enriched populations based on pathway activation. Several critical pathways are known to be activated in subsets of patients with HCC, such as Wnt, JAK–STAT, TGF-β, mTOR, MET, IGF, and Notch signalling, among others.^9,10 Enrichment can be based on immunostaining, gene signatures or other tools. We wonder, for instance, if the outcomes of EVOLVE-1 would have been similar had the study been enriched with patients with mTOR-active HCCs. Finally, trials targeting ‘all comers’ can be designed to identify benefits in pre-specified subpopulations. Thus, if an outstanding benefit emerges in a subset of patients it can be captured. Of course, this strategy would need prior agreement with regulatory agencies to grant approval to use the drug in just that subset of patients.

In terms of impact in clinical decision-making in HCC, another consideration needs to be pointed-out. The HCC research community is facing a situation in which sorafenib has been the sole effective systemic therapy for the past 8 years, and all other attempts to bring novel agents into guidelines and clinical practice have been spurious. This scenario is posing a threat to the management of HCC, which is in peril with non-evidence treatments that are coming onto the scene. Physicians are eager to offer promising alternatives for patients with few hopes, and certainly some encouraging local therapies (such as internal radiation with Y-90) have been documented in phase II studies. Although this understandable attitude may drive individualized decisions, standards of care adopted in guidelines should rely on high level evidence-based data.

In summary, everolimus failure in HCC adds further weight to the need to change the current mindset of trial design. The advancements in the understanding of molecular drivers and classes of HCC should be urgently translated into the design of early clinical trials in order to maximize the chances of positive clinical outcome benefits. All partners involved in this endeavour, including physicians, the pharmaceutical industry and regulatory agencies, should converge in having this goal as a research priority in the field.