On the back of success in melanoma with PD-1 and CTLA-4 checkpoint inhibitors, attention has turned to other checkpoint proteins such as indoleamine 2,3-dioxygenase (IDO). In this editorial, Dr Zeynep Eroglu (Moffitt Cancer Center, USA) discusses the advances and setbacks in the development of IDO inhibitors for melanoma.
Over the past few years, immunotherapy has become standard of care for metastatic or unresectable melanoma as both PD-1 inhibitors (such as nivolumab) and CTLA-4 inhibitors (such as ipilumumab) have produced improvements in progression-free survival (PFS) and overall survival (OS). Combining these checkpoint inhibitors has shown to be particularly efficacious in the first-line setting, improving overall response rate (ORR) and PFS when compared to either agent alone. Building on this success, numerous trials are now exploring combinations of various immunotherapies in order to determine whether they might produce superior response and survival to anti-PD-1 monotherapy, yet have less toxicity than the ipilimumab/nivolumab combination.
One such potential target is indoleamine 2,3-dioxygenase (IDO), a checkpoint protein that contributes to an immunosuppressive tumor environment. Upregulation of the IDO pathway can lead to immunosuppressive effects through metabolism of the amino acid tryptophan to kynunerine; increased kynunerine in the tumor microenvironment can decrease CD8+ T cells and natural killer cells, and increase regulatory T cells and myeloid-derived suppressor cells. Essentially, IDO expression in tumors is associated with a decrease in immune cell infiltration and poor prognosis, and its expression is upregulated in many tumors including melanoma.
Early promise
There have been early phase trials with several small molecule IDO inhibitors such as epacadostat, indoximod, and BMS-986205; however, notably, these drugs have not shown efficacy as monotherapies. In a phase I study of epacadostat in 52 patients with various metastatic solid tumors, no objective tumor responses were observed. However, preclinical in vivo data demonstrated significant synergy between IDO inhibitors and PD-1 inhibitors in combination, with increased tumor shrinkage when compared to anti-PD-1 therapy alone. Thus, the main clinical interest of IDO inhibitors has been in combination with anti-PD-1 antibodies.
A subsequent phase I/II trial of epacadostat with pembrolizumab included a cohort of advanced melanoma patients treated with epacadostat (100 mg BID) with pembrolizumab (standard dose: 200 mg Q3W). In 54 treatment-naïve patients in the melanoma cohort who received the combination, ORR was an impressive 60%, with a very high median PFS of 22.8 months. Tumor responses were observed regardless of PD-L1 expression and BRAF mutation status.
Epacadostat at either 100 mg or 300 mg BID was also combined with nivolumab (240 mg Q2W) in a single-arm phase II trial of 50 patients with advanced melanoma. ORR was again notable at 62%, with higher ORR observed in patients whose tumors had positive IDO and PD-L1 expression, although these levels were not specified (see here). Median PFS was not reached; PFS rate at 12 months was 60%. No significant worsening in toxicity was observed with the combination in either of these trials when compared to that observed with anti-PD-1 monotherapy.
Similar to epacadostat, indoximod is another oral IDO pathway inhibitor; in a single-arm phase II trial, indoximod 1200 mg BID was combined with pembrolizumab in the 85 patients with advanced melanoma who had not received prior anti-PD-1 treatment. ORR remained high at 53% and a complete response rate of 18% was observed.
Across all three of these single-arm, phase II trials, ORR in the 55–60% range was uniformly higher than observed with single-agent pembrolizumab or nivolumab, which has been around 35–40%. The median PFS or PFS-rates reported were also quite a bit higher than expected with anti-PD-1 monotherapy. As such, the results of the first randomized phase III trial of an IDO inhibitor combined with anti-PD-1 therapy were eagerly anticipated, with expectation of superior ORR and PFS data.
Falling at the final hurdle
The double-blind, phase III ECHO-301/KEYNOTE-252 trial randomized patients to either epacadostat 100 mg BID with pembrolizumab 200 mg Q3W or pembrolizumab with placebo. A total of 706 patients with untreated and unresectable or metastatic melanoma were randomized to the study arms. With a median follow-up of 12.4 months, there was surprisingly no difference at all in median PFS, with a median of 4.7 months with the combination, and 4.9 months with pembrolizumab alone (HR=1.00; P=0.517). PFS rate at 12 months was 37% in both arms. Furthermore, there was no difference in OS, with a 12-month OS rate of 74% in both arms. ORRs were also very similar: 34.2% with the combination and 31.5% with pembrolizumab alone. Grade 3 or higher treatment-related adverse events occurred in 21.8% of patients with the combination and 17% with pembrolizumab alone. As a result of these findings, the external data monitoring committee recommended stopping the trial as it did not meet the primary endpoint of PFS, and OS was not expected to reach statistical significance.
It is noteworthy that IDO inhibitors have not been shown to have any single-agent activity in melanoma
The failure of this large, randomized, phase III trial to demonstrate any benefit at all for the addition of an IDO inhibitor to anti-PD-1 therapy prompted quite a few questions. Of note, there were some differences in the patient population compared to the single-arm phase II trials, which had fewer patients with stage 4 M1c melanoma; for example, 40% of patients in the epacadostat with nivolumab trial had M1c disease, as compared to approximately 2/3 of patients in the phase III trial. (Stage 4 M1c in AJCC v7 staging criteria indicates patients who have extra-pulmonary metastases or elevated LDH, and these patients have worse prognosis and outcomes with any systemic treatment.) In addition, the phase III trial used the 100 mg BID dosing of epacadostat, while the phase II trial with nivolumab tested 300 mg BID dosing, raising questions about whether the higher dose should have been used.
Given the complete overlap of the survival curves between the combination and pembrolizumab monotherapy, questions were also raised about whether the IDO inhibitor was inhibiting kynunerine in the actual tumor microenvironment. And while ipilimumab with nivolumab showed significantly higher response rates and PFS in a phase III trial of metastatic melanoma and was subsequently FDA-approved, it is important to note that ipilimumab also has single-agent activity in melanoma, and was previously FDA-approved as a monotherapy. It is noteworthy that IDO inhibitors have not been shown to have any single-agent activity in melanoma.
While there has been a trend in oncology drug development to jump directly from early phase I/II single-arm studies directly to large, randomized phase III trials, these results also raise the question of whether smaller randomized, phase II trials should be conducted before proceeding with a phase III study. This may be especially important for combinations involving drugs without single-agent activity of their own, or when there are no clear answers on whether the drug inhibits its intended target in the tumor.
Looking forward
At this time, it is unclear if there is a path forward for IDO inhibitors in melanoma; after the results of the phase III trial with epacadostat, several other trials with indoximod and BMS-986205 were suspended or placed on hold. Of course, there are numerous other immunotherapy combination trials ongoing in advanced melanoma; time will tell whether any of these combinations will be proven superior to single-agent anti-PD-1 therapy in melanoma, and match the success of the ipilimumab/nivolumab combination with less toxicity. Combinations of IDO inhibitors with anti-PD-1 therapy are still moving forward in randomized trials in other diseases such as bladder and non-small-cell lung cancer to determine whether it may yet prove superior to single-agent anti-PD-1 inhibitors, though it remains to be seen whether these efforts will inform our progress in melanoma.