Abstract
Melanoma has historically been considered a refractory disease with few if any options in the advanced/metastatic setting. Advances in both immune and genetically targeted treatment approaches have revolutionized the spectrum of treatment options for melanoma patients over the last several years. Recently, checkpoint inhibition has become a major focus in the immune-based therapy of cancer, especially melanoma. This concept involves inhibition of regulatory cell surface molecules which act normally to dampen or modulate T-cell activation. Cancer, including melanoma, takes advantage of this physiologic mechanism to turn off T-cell activation and prevent effective T-cell antitumor responses. Checkpoint inhibitors such as anti cytotoxic T-lymphocyte antigen 4 (anti-CTLA-4) and anti programmed death-1 (PD-1) can reverse this immune suppression and release T-cell activation. Nivolumab, a monoclonal antibody to the PD-1 receptor, promotes antitumor immunity by removing this key negative regulator of T-cell activation. In phase I/II studies, promising activity and safety have been observed and ongoing phase III trials are comparing nivolumab with other standard of care therapies (chemotherapy, ipilimumab). Efficacy may be even further increased when used in combination with ipilimumab (albeit with increased toxicity). In contrast to typical short-lived responses with cancer therapy in metastatic solid tumors, many responses induced by nivolumab appear durable. In this review, we discuss the evolution of immune therapy in melanoma leading to the development of nivolumab, the clinical experience with this agent, and its future development and clinical potential.
Background
Advanced melanoma has been resistant to conventional cytotoxic chemotherapy and has historically lacked effective systemic treatment options [Middleton et al. 2000; Korn et al. 2008]. These poor results spurred the intensive pursuit of alternative treatment strategies for the last several decades, which in part led to the understanding that melanoma is a particularly immunogenic tumor. Agents targeting the programmed death-1 (PD-1) receptor and its ligand (PD-L1) are a new, promising class of therapeutics which inhibit a critical negative regulator of T-cell activation and thereby promote antitumor immunity [Hirano et al. 2005, Ott et al. 2013]. Nivolumab (BMS-936558) is a monoclonal antibody to PD-1 being developed for treatment of advanced melanoma and other cancers. Clinical trials have demonstrated promising activity particularly in advanced melanoma, nonsmall cell lung cancer and renal cell carcinoma, as well as a tolerable toxicity profile [Topalian et al. 2012, 2014; Weber et al. 2013]. In this review, we discuss the evolution of immune therapy in melanoma, the clinical experience with nivolumab, and future directions and potential for this agent in melanoma therapy.
Evolution of immune therapy in melanoma
Numerous strategies to stimulate an antineoplastic immune response have been explored. Historically, the cornerstones of immune therapy were high-dose interleukin-2 (IL-2) for metastatic melanoma and high-dose interferon-α for resected melanoma (stage II and III) at high risk of recurrence. High-dose IL-2 induces objective responses in approximately 15–20% of patients with metastatic melanoma and 6–8% of treated patients experience durable (>3 years) complete remissions [Rosenberg et al. 1994; Atkins et al. 1999]. Severe acute toxicities including multiorgan dysfunction, hemodynamic compromise and confusion preclude therapy in patients with marginal functional status, organ dysfunction or advanced age [Schwartzentruber, 2001]. Furthermore, intensive monitoring in an inpatient setting at an experienced center is a requisite for IL-2 therapy. Interferon-α, used in the adjuvant setting for resected, high-risk melanoma, has demonstrated improved relapse-free survival compared with observation [Kirkwood et al. 1996, 2000, 2001, 2004]. However, the effects on overall survival (OS) remain controversial and are modest at best; meta-analyses have demonstrated a relative improvement in OS of approximately 10% [hazard ratio (HR) = 0.89]. Chronic, dose-limiting toxicities are bothersome to nearly all patients and prevent completion of therapy in some. Despite the activity of these therapies, no consistent survival improvement for any agents had been demonstrated in metastatic melanoma prior to 2010 and the need for more effective immune-based therapies remained a clear priority.
Ipilimumab is a fully humanized monoclonal antibody that inhibits cytotoxic T-lymphocyte antigen 4 (CTLA4). CTLA4 engages the antigen presenting cell (APC) receptor B-7.1 and B-7.2 and prevents T cell costimulation, thereby playing a critical modulatory role of immune activation [Leach et al. 1996]. Ipilimumab inhibits this interaction and functions to ‘remove the brakes’ on cellular immune activation, resulting in an antitumor response in some patients. Aberrant T-cell activation against self-antigens may complicate therapy. This was the first agent to demonstrate an improvement in OS in advanced melanoma. In a study of patients progressing on prior therapies, ipilimumab 3mg/kg for 4 doses was compared with a gp100 vaccine [Hodi et al. 2010]. A median OS of 10.1 months was identified with ipilimumab compared with 6.4 months with the vaccine (HR for death, 0.68;
Ipilimumab, while representing a substantial advance over previously available therapeutics, has some clear drawbacks. First, classic Response Evaluation Criteria In Solid Tumors (RECIST) responses are rare and often occur after a prolonged duration of therapy, sometimes after initial evidence of progression of disease with new or existing lesions [Ribas et al. 2009; Wolchok et al. 2009; Hodi et al. 2010]. Second, toxicities may occasionally cause substantial morbidity and even mortality, with severe immune-related adverse events occurring in 10–20% of patients [Weber et al. 2009; Hodi et al. 2010; O’Day et al. 2010]. Colitis, endocrinopathies, dermatitis, hepatitis and neurotoxicity are the most common immune-related manifestations of therapy. These adverse events are quite distinct from the toxicity profile of cytotoxic chemotherapy and are generally reversible with corticosteroids.
Anti-PD-1/PD-L1 therapy
Novel therapeutic targets for immune checkpoint blockade were identified in PD-1 (B7-H1) and its ligand, PD-L1 (B7-DC) [Dong et al. 1999; Tseng et al. 2001]. PD-1, a receptor expressed on CD4+ and CD8+ T cells (as well as B cells and natural killer cells), binds to PD-L1 and induces functional exhaustion of a cytotoxic immune response. While this interaction does have a physiologic role to promote immune tolerance and suppress autoimmunity, many cancers and chronic viral infections exploit this pathway to achieve immune evasion. The interaction between PD-1 and PD-L1 in cancer causes T-cell apoptosis, limits T-cell expansion and inhibits production of IL-2 and interferon-γ [Dong et al. 2002; Pedoeem et al. 2014]. PD-L1 is expressed in a variety of tumor types as opposed to B7.1 or B7.2, but also can be found on inflammatory cells such as T lymphocytes and infiltrating mononuclear cells. Inhibiting this immune modulatory axis, therefore, was hypothesized to induce more specific antitumor responses and mitigate autoimmune toxicity. Several antibodies targeting PD-1 [nivolumab, pembrolizumab (MK-3475)] and PD-L1 (MPDL3280A, MEDI4736, others) are in various stages of clinical development for a variety of cancers and all show promising clinical responses.
Nivolumab
Nivolumab is a fully humanized, monoclonal, immunoglobulin G4 (IgG4) antibody to PD-1. Engaging its target, PD-1, prevents interaction with both PD-L1 and PD-L2. PD-L1 is expressed on approximately 40–50% of melanomas and has limited expression otherwise in most visceral organs with the exception of respiratory epithelium and placental tissue [Dong et al. 1999; Petroff et al. 2003; Kim et al. 2005]. More recently studies have suggested its presence and expression on immune cells infiltrating tumor cells (TIL) [Powderly et al. 2013]. Less is known about the expression and role of PD-L2 in immune tolerance and antitumor immunity. In early nivolumab clinical trials, pharmacodynamic studies revealed that receptor binding was largely dose independent over a 30-fold dose increase [Brahmer et al. 2010]. PD-1 receptor saturation was maintained for several months even in the absence of continued therapy.
Phase I clinical trial
The initial phase I clinical trial of nivolumab was conducted in patients with melanoma, lung cancer, renal cell carcinoma and a limited number of other malignancies [Topalian et al. 2012]. In this report, 94 patients with melanoma were treated in total with nivolumab, with doses ranging from 0.1 to 10 mg/kg every 2 weeks [cohorts of 0.1, 0.3, 1.0, 3.0 or 10 mg/kg]. Among these patients, the objective response rate (ORR) by RECIST criteria across all dose levels was 28%, with the highest response rate observed in 3 mg/kg (41%). Responses were durable in most patients; in the 21 patients that achieved an objective response, 13 continued to respond for more than 1 year. Of note, 8 responding patients experienced transient tumor regression ranging from 1.9 to 5.6 months. Progression-free survival (PFS) at 24 weeks was 41%.
Overall, nivolumab was well tolerated with low grade fatigue, diarrhea, pruritus, nausea and decreased appetite occurring as the most common adverse events. Grade 3/4 treatment related events occurred in 14% of patients and immune-related adverse events were observed in 6%. Pneumonitis, which has emerged as the most serious toxicity, occurred in nine patients and caused three deaths (none in patients with melanoma).
Long-term follow up from this trial in melanoma patients was recently published including 107 melanoma patients [Topalian et al. 2014]; 1 and 2 year survival rates were 62% and 43%, respectively, with a median OS of 16.8 months (Table 1). The updated median PFS was 3.7 months and 27% of patients remained free of progression at 2 years. The updated response rate was similar to the original publication at 31% (33/107) with a Kaplan–Meier estimated median response duration of 2 years (104 weeks). Responding patients also continued to benefit after drug discontinuation; of 17 patients discontinuing therapy for reasons other than disease progression, 12 (71%) had persistent responses lasting ⩾16 weeks. Long-term evaluation of safety was comparable to the original analysis, with 22% of patients experiencing grade 3/4 treatment related events and 5% experiencing grade 3/4 immune-related adverse events. Toxicities were not cumulative and almost exclusively occurred in the first 6 months of therapy. These data were recently updated at the American Society of Clinical Oncology (ASCO) annual meeting in 2014 and demonstrated a 48% survival rate at 2 years and 41% at 3 years [Hodi et al. 2014].
Response rate, survival and adverse events in nivolumab trials.
Cutoff of 5% staining by IHC.
UHC, immunohistochemistry; PD-L1, programmed death ligand-1.
Nivolumab in ipilimumab refractory and naïve patients
Nivolumab was further tested in a second phase I trial to evaluate the impact of prior ipilimumab [Weber et al. 2013]. Nivolumab was administered to patients who were ipilimumab refractory (3 cohorts all with 3 mg/kg dose) and naïve (1 mg/kg, 3 mg/kg or 10 mg/kg cohorts) in combination with a peptide vaccine to melanoma antigens gp100, NY-ESO-1 and MART-1. Across all groups, the ORR response was 25% and an additional 21% had stable disease (SD) as their best response (Table 1). In ipilimumab- naïve patients, the ORR was 24%
The combination of nivolumab and peptide vaccine was generally well tolerated. The most common side effects were reaction at vaccine injection site and fatigue. In ipilimumab naïve patients, grade 3/4 toxicities included one case of bilateral optic neuritis, one case of fever and one case of pneumonitis, all which resolved with steroid treatment. In ipilimumab refractory patients, grade 3/4 toxicities included one case of rash and one case of pneumonitis, both of which were also responsive to corticosteroids.
PD-L1 expression was investigated as a biomarker for response to therapy (see PD-L1 expression below). Other immune biomarkers were also assessed. Although all patients had at least 10% tumor staining for gp100, NY-ESO-1 or MART-1 for trial inclusion, pretreatment antigen-specific CD8+ T cells for NY-ESO-1 and MART-1 were significantly lower in responders compared with nonresponders. With treatment, CD8+ T cells that recognized MART-1 increased in responders but decreased in nonresponders. Regulatory T cells also decreased in responders and increased in nonresponders at the 12 week assessment.
Early results from a larger phase III clinical trial were also presented at the European Society of Medical Oncology (ESMO) meeting. This study compared nivolumab with cytotoxic chemotherapy (dacarbazine or carboplatin/paclitaxel) in patients who had previously received ipilimumab and BRAF inhibitors if a
Results from a randomized phase III trial also became available recently. In this study, previously untreated patients without BRAF mutations were randomized to nivolumab or cytotoxic chemotherapy (dacarbazine). Nivolumab was markedly superior in terms of response rate (40% vs. 13.9%), median PFS (5.1 months vs. 2.2 months), and 1-year overall survival (72.9% vs. 42.1%; all comparisons p<0.001). [Robert et al. 2014]
PD-L1 expression
Since nivolumab inhibits the interaction between PD-1 and its ligand, PD-L1, it was hypothesized that PD-L1 expression by the tumor would be required for response to therapy. To investigate this, 42 pretreatment tumor samples from the phase I trial were assessed by immunohistochemistry (IHC) for expression of PD-L1 (including 18 patients with melanoma). Using a cutoff of 5% IHC as positive expression, 0/17 patients with negative PD-L1 expression achieved an objective response
Other issues also complicate the routine use of PD-L1 as a biomarker for response. The PD-L1 assay developed alongside nivolumab is an IgG1 monoclonal antibody, clone 5-H1, and is now being developed by Dako® [Taube et al. 2012]. Each pharmaceutical company with anti-PD-1/PD-L1 antibodies late in development has distinct companion assays to measure PD-L1 expression. Moreover, PD-L1 expression appears to be dynamic and may be modulated by a number of factors [Akbay et al. 2013; Frederick et al. 2013]. At this time, therefore, the role of PD-L1 expression in treatment decision making for melanoma patients remains yet to be defined.
Other anti-PD-1/PD-L1 antibodies
Pembrolizumab (MK-3475) is another antibody to PD-1 which appears to have comparable efficacy and tolerability. This agent was evaluated in a phase I study of 411 patients, of which 190 patients were naïve to ipilimumab [Ribas et al. 2014]. In this cohort, a response rate of 40% was observed across dose levels. In addition, in 173 patients previously treated with ipilimumab, 26% experienced objective responses [Robert et al. 2014]. Pembrolizumab has recently received approval from the US Food and Drug Administration (FDA) for patients previously treated with ipilimumab and, if harboring a BRAF mutation, BRAF inhibitors.
MPDL3280A is a monoclonal antibody to PD-L1. This agent has also been tested in numerous malignancies and is clinically active in melanoma. Among 35 patients with evaluable responses, 9 (26%) experienced an objective response [Hamid et al. 2013b]. Other agents are previously or currently in development (BMS-936559, MEDI4736, CT-011) but are beyond the scope of this review.
Nivolumab plus ipilimumab
Despite the promising activity of nivolumab as a single agent, less than half of patients achieved a durable response. Nivolumab and ipilimumab remove key negative T-cell regulators at distinct phases of the T-cell activation process, putatively at the level of antigen-presenting cell–T cell interaction, and at the level of effector T cell–tumor interface, respectively. Therefore, it was hoped that combined immune checkpoint blockade would offer superior efficacy compared with either agent alone (albeit tempered by concerns of increased toxicity). Preclinical models demonstrate that these agents in combination decrease regulatory T cells, increase tumor infiltrating effector T cells, and extend survival more effectively than either alone [Curran et al. 2010]. The combination of nivolumab and ipilimumab was evaluated in a phase I clinical trial [Wolchok et al. 2013]. Patients were enrolled in cohorts that either received both drugs concurrently or in sequence. In the concurrent group, regimen dosing of nivolumab ranged from 0.3 mg/kg to 3 mg/kg and ipilimumab dosing ranged from 1 mg/kg to 3 mg/kg.
The maximum dosing level associated with an acceptable level of adverse events was 1 mg/kg of nivolumab and 3 mg/kg of ipilimumab. With this regimen, the ORR was 53% with tumor reduction of ⩾80% in all 9 responding patients. Moreover, this degree of tumor regression occurred rapidly, at the time of first computerized tomography (CT) scan, in all patients. Across all doses, the ORR was 40% and aggregate clinical activity rate (defined as conventional, unconfirmed, or immune-related responses or SD for ⩾24 weeks) was observed in 65% of patients (Table 1). These results indicate that combination therapy may prove to be more effective than monotherapy with either drug alone.
Concurrent therapy had manageable but significant toxicities, with grade 3/4 treatment related adverse events occurring in 53%, higher than in monotherapy trials. The most common toxicities were asymptomatic elevations in lipase (13%), aspartate transaminase (AST) (13%), and alanine aminotransferase (ALT) (11%). The rate of grade 3/4 gastrointestinal toxicity was 9% and serious pneumonitis occurred in 1 patient. Side effects at the maximum acceptable dosing (nivolumab at 1 mg/kg and ipilimumab at 3 mg/kg) included grade 3 uveitis in one patient and grade 3 elevations in AST and ALT levels in another patient. In the majority of cases, toxicities were manageable with corticosteroids (38%), infliximab (2 patients), or mycophenolate mofetil (1 patient).
PD-L1 expression (threshold of 5% staining by IHC) and absolute lymphocyte count (ALC) during treatment (measured between 5 and 7 weeks after therapy initiation) were evaluated as potential biomarkers. Objective responses were observed in patients with both PD-L1 positive tumor samples (6 of 13 patients) and PD-L1 negative tumor samples (9 of 22 patients), indicating that those without PD-L1 expression on tumors may still derive benefits from treatment. This biomarker, therefore, likely has limited value for combination therapy. Increase in ALC, a significant predictor of ipilimumab monotherapy benefit, did not appear predictive in this trial. The ORR was comparable in patients with a low ALC < 1000 cells/cm3 (43%) and normal or high ALC (40%).
Data from this clinical study were recently updated and included an additional 41 patients treated at the nivolumab 1mg/kg plus ipilimumab 3mg/kg dose level [Sznol et al. 2014]. The OS at 1 and 2 years was 85% and 79%, respectively, essentially unprecedented results in melanoma trials (albeit in a small sample size). ORR (42% with 17% CR rate) and rate of grade 3/4 adverse events (61%) both slightly increased from the previous analysis (with 1/94 treated patients dying from a treatment-related adverse event). Median duration of response, OS or PFS had not been reached at the time of analysis. Also of note, aggregate clinical activity rate appeared similar between BRAF mutant (60%,
Future directions and therapeutic potential
Nivolumab and other anti-PD-1/PD-L1 directed therapies represent a major step forward in melanoma therapeutics. While randomized phase III trials have not yet been completed with nivolumab, response rates and 2 year OS observed in early trials have been superior to previously developed immune therapies and molecularly targeted agents. In addition, the incidence of immune-related adverse events appears significantly lower than those observed with anti-CTLA4 directed therapies. Given the favorable efficacy and toxicity profile, nivolumab (along with pembrolizumab) will almost certainly become a standard first-line option for patients with advanced melanoma. The initial regulatory approval of nivolumab, however, may be for previously treated patients (as with pembrolizumab).
One unresolved question currently is whether nivolumab, pembrolizumab, MPDL3280A, or another PD-1/PD-L1-directed therapy earlier in development is the most effective in advanced melanoma. In the initial phase I studies, the ORR of pembrolizumab appears to be somewhat higher than nivolumab (40% and 29%, respectively) [Topalian et al. 2012, Hamid et al. 2013a]. On the contrary, the response rate in patients treated following ipilimumab appeared possibly higher in nivolumab (32% and 26%) [Robert et al. 2014; Weber, 2014]. These agents have not been compared directly and were evaluated in different populations. Moreover, long-term survival has not been evaluated and, therefore, additional studies will likely clarify the relative efficacy of these agents.
Nivolumab in combination with ipilimumab may have still more activity than single-agent nivolumab, although at the cost of increased adverse events [Wolchok et al. 2013]. This combination produced rapid and dramatic responses in a phase I trial assessing this combination and OS was 79% at 2 years. Currently, a three-arm phase III trial is comparing nivolumab, ipilimumab and the combination of nivolumab and ipilimumab [ClinicalTrials.gov identifier: NCT01844505]. This study will give insight into whether combination therapy or either agent alone is the more appropriate upfront treatment option.
With the early success of nivolumab plus ipilimumab, additional immune modulators are also being evaluated in combination with nivolumab. Currently, ongoing early-phase clinical trials (see Table 2) include nivolumab plus either lirilumab (anti-KIR) [ClinicalTrials.gov identifier: NCT01714739], BMS-986016 (anti-LAG-3) [ClinicalTrials.gov identifier: NCT01968109] and BMS-982470 (IL-21) [ClinicalTrials.gov identifier: NCT01629758]. These agents aim to synergize with nivolumab to reverse neoplastic immune evasion through natural killer (NK) cell activation (anti-KIR), dual T-cell checkpoint blockade (anti-LAG-3) or a broad range of immune functions including CD8+ T cell and NK cell activation (IL-21) [Woo et al. 2012; Kohrt et al. 2014; Spolski and Leonard, 2014]. These combinatorial approaches may play a major role, particularly in tumors in which the PD-1/PD-L1 axis is only a partial contributor to neoplastic immune evasion.
Ongoing combination studies involving nivolumab.
KIR, killer cell Ig-like receptors; LAG-3, lymphocyte-activation gene 3; IL, interleukin; N, nivolumab.
Combining nivolumab with targeted therapies (BRAF and MEK-directed therapies) has also generated significant interest [Hu-Lieskovan et al. 2014]. In preclinical models, these agents (particularly BRAF inhibitors) influence the tumor microenvironment, tumor antigen expression and T-cell function. Extinction of BRAF by RNA interference or MEK inhibitors modulated cytokine expression in the tumor microenvironment, specifically decreasing production of immune-suppressive cytokines (IL-6, IL-10, vascular endothelial growth factor) [Sumimoto et al. 2006]. BRAF inhibitors increase melanoma-lineage antigen expression (e.g. MART1, gp100, tyrosinase) which may lead to enhanced immunogenicity and T-cell recognition [Boni et al. 2010; Frederick et al. 2013]. PD-L1 expression may also be influenced by BRAF inhibition although this relationship appears complex [Jiang et al. 2012; Frederick et al. 2013]. In addition, tumor infiltrating lymphocytes (CD8+ T cells) appear to increase with selective BRAF inhibitor therapy [Wilmott et al. 2012; Liu et al. 2013; Frederick et al. 2013].
This type of combination strategy is attractive clinically and suggests the possibility of combining the frequent and rapid responses of targeted therapy with the durability of an immune response. In early clinical trials, experience with immune/targeted regimens gives reason for cautious optimism for nivolumab/BRAF inhibitor combinations. Vemurafenib and ipilimumab induced intolerable transaminitis in the few patients that were treated [Ribas et al. 2013]. A phase I trial of ipilimumab plus dabrafenib ± trametinib, however, appears to have a reasonable side effect profile in recently presented data for the ipilumumab/dabrafenib arm [Puzanov et al. 2014]. The triplet arm, though, led to unacceptable gastrointestinal toxicity and has been discontinued. At this time, no trials combining nivolumab with BRAF or MEK inhibitors are ongoing, although early phase studies of pembrolizumab with dabrafenib/trametinib [ClinicalTrials.gov identifier: (NCT02130466] and MPDL3280A with cobimetinib [ClinicalTrials.gov identifier: NCT01988896] and vemurafenib [ClinicalTrials.gov identifier: NCT01656642[ respectively are enrolling.
Conclusion
Nivolumab and other antibodies blocking PD-1 will almost certainly play a central role in melanoma therapeutics in the future. The potential for durable antitumor immune responses coupled with a favorable toxicity profile makes nivolumab an attractive therapeutic option as a single agent. Furthermore, the comparative lack of toxicities likely will lead to combination strategies with other immune-stimulatory and genetically targeted agents. Identifying predictive biomarkers to assist in therapeutic decision making and determining the most appropriate partners for melanoma therapy are critical next steps in harnessing the potential of nivolumab.
Footnotes
Funding
D.B.J. was supported by NIH grant K12 CA 0906525.
Conflict of interest statement
J.A.S. is on the advisory board for Bristol Myers Squibb and Roche.