Ipilimumab in a real‐world population: A prospective Phase IV trial with long‐term follow‐up

Ipilimumab was the first treatment that improved survival in advanced melanoma. Efficacy and toxicity in a real‐world setting may differ from clinical trials, due to more liberal eligibility criteria and less intensive monitoring. Moreover, high costs and lack of biomarkers have raised cost‐benefit concerns about ipilimumab in national healthcare systems and limited its use. Here, we report the prospective, interventional study, Ipi4 (NCT02068196), which aimed to investigate the toxicity and efficacy of ipilimumab in a real‐world population with advanced melanoma. This national, multicentre, phase IV trial included 151 patients. Patients received ipilimumab 3 mg/kg intravenously and were followed for at least 5 years or until death. Treatment interruption or cessation occurred in 38%, most frequently due to disease progression (19%). Treatment‐associated grade 3 to 4 toxicity was observed in 28% of patients, and immune‐related toxicity in 56%. The overall response rate was 9%. Median overall survival was 12.1 months (95% CI: 8.3‐15.9); and progression‐free survival 2.7 months (95% CI: 2.6‐2.8). After 5 years, 20% of patients were alive. In a landmark analysis from 6 months, improved survival was associated with objective response (HR 0.16, P = .001) and stable disease (HR 0.49, P = .005) compared to progressive disease. Poor performance status, elevated lactate dehydrogenase and C‐reactive protein were identified as biomarkers. This prospective trial represents the longest reported follow‐up of a real‐world melanoma population treated with ipilimumab. Results indicate safety and efficacy comparable to phase III trials and suggest that the use of ipilimumab can be based on current cost‐benefit estimates.


| INTRODUCTION
Ipilimumab is a monoclonal antibody blocking cytotoxic T-lymphocyte antigen-4 (CTLA-4) from binding to its ligands on antigen-presenting cells 1 and was the first immune checkpoint inhibitor (ICI) to improve survival in cancer patients in a randomised trial. 2 Median overall survival (OS) in clinical trials ranges 10.1 to 19.9 months, 2-5 with 20% to 30% of patients obtaining long-term survival, 3,6,7 a rare encounter before the introduction of ICIs. Currently, first line treatment with ICIs or combined BRAF and MEK inhibitors are the standard of care in metastatic melanoma. 8 With the introduction of ICIs, new clinical challenges emerged.
First, a new spectrum of toxicity, known as immune-related adverse events (irAEs), was observed, requiring careful monitoring and prompt intervention such as immunosuppression. 9 Second, clinical and radiological responses could be slow-onset and even mimic progression, followed by a decrease in tumour size, a much debated phenomenon known as pseudoprogression. 10 Third, as illustrated by a median progression-free survival (PFS) of approximately 3 months, 2,11,12 most patients do not respond to treatment, but are still at risk of potentially harmful side effects. No biomarker has yet been discovered to reliably predict treatment benefit, 13 but baseline clinical features such as poor performance status and extensive organ involvement as well as elevated lactate dehydrogenase (LDH) are associated with poor survival in patients receiving ipilimumab. [14][15][16][17][18][19][20] Moreover, a prognostic index involving these characteristics was found to significantly predict OS. 21 Elevated C-reactive protein (CRP) has previously been related to worse clinical outcome in patients treated with ipilimumab. 22,23 Other biomarkers available from routine blood analyses suggested to be associated with an improved prognosis in patients receiving ICIs include a low total white blood cell count (WBC), low absolute neutrophil count (ANC), high absolute lymphocyte count (ALC) and low neutrophil lymphocyte ratio (NLR). 19,24 Additionally, the cost of ipilimumab, when opening the markets postapproval, was perceived a substantial economic burden to the health care system, with an estimated cost-effectiveness of €100 112 per life-year gained. 25 Using drugs in daily clinical practice, there is a concern that efficacy may be decreased and toxicity increased compared to clinical trials due to less stringent eligibility criteria and less intensive monitoring. A number of reports on real-world data from expanded access, named patient and compassionate-use programmes exist, [15][16][17][18]22,[26][27][28][29][30][31] but are restrained by factors such as retrospective study design, patient selection, single centre experiences and/or limited follow-up.
Thus, this national prospective phase IV trial addresses the use of ipilimumab in a real-world population with metastatic melanoma, investigating toxicity and long-term efficacy assessed by clinical oncologists in out-patient departments throughout Norway.

| Study design
This is a prospective, national, multicentre, open label, single-armed, phase IV interventional clinical trial (NCT02068196). The primary objective was to estimate the incidence and severity of adverse events (AEs) in patients with metastatic melanoma treated with ipilimumab in a real-world setting, and to describe the management and outcome of AEs. Secondary objectives were to assess OS, PFS, overall response rate and duration of response.

| Patients
Patients ≥18 years of age with a histologically confirmed diagnosis of unresectable Stage III/IV metastatic melanoma, Eastern Cooperative Oncology Group performance status (ECOG PS) 0-1 and an adequate renal, hepatic and haematological function were recruited from eight sites throughout Norway. Patients were classified according to the American Joint Committee on Cancer version 8. Any previous treatment was allowed. Patients with active brain metastases that required other treatment were not permitted, but patients with known brain metastases that were previously treated, or considered not in need of radiotherapy or surgery, were allowed. No screening for brain metastases was conducted to identify the presence of asymptomatic brain involvement. Patients with a history of autoimmune disease, immunodeficiency, splenic surgery or irradiation, allogeneic stem cell transplantation, known hypersensitivity to recombinant protein products, uncontrolled infectious disease, pregnant or breastfeeding were excluded. All patients provided written informed consent. The intention-to-treat population equalled the safety population and was defined as all patients that received a dose of ipilimumab.

| Treatment and toxicity
The safety population included all patients that had received at least one dose of ipilimumab (N = 151). Treatment is summarised in Figure 2A. Treatment cessation was observed in 32% of patients and was caused by disease progression in 19% of patients and drug-related toxicity in 13%. In 3% of patients, treatment was interrupted due to treatment-related toxicity and later resumed. All but one patient developed immune-related toxicity after resuming ipilimumab.
Treatment-associated AEs were reported in 73% of patients, and are summarised in Table 2 were diarrhoea and colitis, followed by hypophysitis and skin-related toxicity. Additionally, corticosteroids were administered systemically to 13% of patients. Four patients were prescribed corticosteroids due to symptomatic worsening of known brain metastases, and seven patients with previously undetected brain involvement developed symptomatic brain metastases requiring corticosteroids. Other indications for corticosteroids were liver metastases, dyspnoea, nausea and anorexia. Three patients received other immunosuppressive drugs due to insufficient effects of corticosteroids; two patients with colitis received infliximab, and one patient with hepatitis received sirolimus and mycophenolate mofetil.
Brain involvement was associated with a lower incidence of irAEs within the first 3 months (M1d 15% vs M1a 67%, M1b 54% and M1c 53%, P = .041). We did not identify baseline characteristics that significantly predisposed to treatment toxicity.

| Clinical response
In the intention-to-treat population, three patients (2%) achieved a complete response (CR), 11 (7%) partial response (PR), 39 (26%) stable disease (SD) and 86 (57%) progressive disease (PD) as best overall response. All, but one response were confirmed by at least one additional CT scan at a later point in time. This patient had a PR at week 12, but due to contrast allergy later CTs were conducted without IV contrast. Twelve patients (8%) who were considered nonevaluable according to RECIST 1.1 either progressed clinically or died before evaluation and must therefore be considered early progressors. Thus, the overall response rate was 9% and the disease control rate 35%.
No cases of pseudoprogression were identified. Median time to response was 3.2 months (range, 2.5-14.3), and median duration of response 20.8 months (range, 0.9-not reached).
Three patients received retreatment with ipilimumab per protocol. Following retreatment, one patient who had previously achieved PR regained PR, one patient who initially had obtained SD regained SD, and the third patient who had experienced SD progressed.

| Survival
The median follow-up was 68.1 months in patients alive at censoring   ). An irAE was defined as an adverse event that was associated with exposure to the study drug, and that was consistent with an immune phenomenon. Some patients reported multiple irAEs 95% CI: 1.23-2.52, P = .002) (  (Table 3).
Clinical efficacy is outlined in Figure 4A.  An immune-related adverse event was defined as an adverse event that was associated with exposure to the study drug, and that was consistent with an immune phenomenon. c One patient died due to perforated colitis.

| DISCUSSION
The prospective data reported herein represents the longest publi- Hence, there is a concern that patients in daily clinical practice may develop more severe toxicity. This would also increase the costs. A study on the real-world use of ipilimumab in second line reported increased rates of hospitalisation compared to historical controls receiving other second line treatments. 33 In this trial, however, the overall toxicity is in line with phase III trials reporting on ipilimumab in metastatic melanoma, with diarrhoea, rash, pruritus, fatigue and nausea being the most commonly reported AEs. 2,11,12 Safety reports from published real-world data do not contradict our findings, 15,16,18,[28][29][30] but the majority of these reports are challenged by retrospective data collection, and hence, associated with a reporting bias.
Colitis is one of the most serious side effects of ipilimumab, and was observed at rate comparable to phase III trials. 2 Notably, a retrospective report on 740 melanoma patients receiving ICIs, of which the majority received ipilimumab, found that 7.3% of patients developed serious infections. 34 Immunosuppression with corticosteroids or infliximab to treat ICI-induced irAEs was identified as the main risk factor for developing serious infection, 34 supported by a study on melanoma patients developing diarrhoea and colitis from ICIs. 35 In the Ipi4 trial, 10 out of 17 patients with reported associated infections received corticosteroids on study, and were, thus, more susceptible to opportunistic infections. Hence, infections observed in this trial may at least partly be a consequence of immunosuppression of irAEs rather than directly linked to ipilimumab.
Additionally, other comorbidity predisposing to infections, such as diabetes mellitus, has been identified as increasing the risk of developing infections secondary to ICIs. 36 It has been hypothesised that irAEs may be associated with the efficacy of ipilimumab, based on the assumption that increased immune activity may affect both tumour and normal tissues, 37 but it has not been clarified whether such an association exists. In this population, no significant association was observed between an irAE of any grade and OS. High-grade irAEs were, however, significantly associated with a shorter OS despite only one death being attributed to immune-related toxicity. To our knowledge, high-grade irAEs have not previously been related to poor OS. Some previous reports have suggested that irAEs predict a favourable clinical outcome, 15,28,37,38 but the literature is diverging as an analysis of the pivotal phase III trial found no significant association between irAEs and treatment benefit. 39 A meta-analysis studying the association between irAEs and outcomes of ICIs highlighted the use of inappropriate methodology, as the majority of studies failed to recognise irAEs as a time-dependent variable. 40 Thus, it has not been taken into account that patients that are followed for a longer time have an increased risk of experiencing irAEs. Further, these patients may have had greater treatment exposure and are therefore both more likely to obtain a clinical benefit and develop irAEs. Patients that die or exit the study due to progressive disease, on the other hand, may not have had time to develop irAEs. Note: Patients with missing data were excluded from analysis. Abbreviations: HR, hazard ratio; CI, confidence interval; ECOG PS, Eastern Cooperation Oncology Group performance status; NOI, number of organs involved; WT, wild type; ULN, upper limit normal; LDH, lactate dehydrogenase cut-off 205 U/L; CRP, C-reactive protein cut-off 10 mg/L; WBC, white blood cells cut-off 10 Â 10 9 /L; ANC, absolute neutrophil count cut-off 7.3 Â 10 9 /L; ALC, absolute lymphocyte count cut-off 1.1 Â 10 9 /L; NLR, neutrophil lymphocyte ratio. This is known as immortal time bias, and may cause overestimation of the association between irAE and outcome. 40 Landmark analyses have been pointed to as a method of avoiding this bias. 40 A recently published pooled analysis of patients with metastatic melanoma receiving pembrolizumab in three clinical trials concluded with no association between irAEs and OS when applying the landmark approach. 41 We conducted a landmark analysis with a cut-off at 3 months by which the majority of irAEs were encountered. However, this approach  14,16,17,20,22,30 In our data, baseline CRP ≥ 10 mg/L is associated with an increased risk of death. A recent report elucidates the immunosuppressive mechanisms of CRP, suggesting that CRP restricts activated CD4+ and CD8+ T cells by inhibiting proliferation, activation-associated phenotypes and effector functions. 48 Moreover, CRP-treated T cells expressed high levels of interleukin-1β, known to enhance CRP production from the liver, inhibited early events in T cell receptor engagement and downregulated the expression of costimulatory molecules on mature dendritic cells. This included CD80 and CD86, the costimulatory ligands enabled to engage CD28 as ipilimumab blocks CTLA-4. A correlation between improved OS and a normal baseline CRP or decrease in CRP during ipilimumab treatment has previously been identified. 17,22,23,49 No patients received prior PD-1 inhibitors before inclusion in this trial, and our report does therefore not address the use of ipilimumab in this setting. Importantly, less than a year after trial enrolment completed, PD-1 inhibitors were granted reimbursement by national health authorities and replaced ipilimumab as the standard of care in first line treatment of metastatic melanoma. Thus, the current role of ipilimumab is in combination with nivolumab in first line treatment or following progression on PD-1 inhibitors, and BRAF/MEK blockade in BRAF-mutated patients. 8 In conclusion, this prospective phase IV trial indicates that the efficacy and toxicity of ipilimumab are in line with the registrational study, despite concerns that outcomes in a real-world population would be inferior to clinical trials. Long-term survival was similar to a pooled analysis of patients receiving ipilimumab in phase II and III trials. 3 We find that ECOG PS, LDH and CRP are independent predictors of OS in patients with metastatic melanoma, and may be useful in identifying patients benefitting and not benefitting from treatment in everyday clinical practice. Although, ipilimumab monotherapy is no longer a preferred first line treatment in metastatic melanoma, our data supports the use of ipilimumab in a real-world setting. As the current role of ipilimumab monotherapy is second line following progression on PD-1 inhibitors, there is a need for prospective studies investigating the real-world use of ipilimumab in this setting.

ACKNOWLEDGMENTS
We wish to thank patients and their families and the study team for their contribution to this trial.

DATA AVAILABILITY STATEMENT
The datasets generated during and analyzed during the current trial are available from the corresponding author on reasonable request.

ETHICS STATEMENT
The Ipi4 trial (https://clinicaltrials.gov/ct2/show/NCT02068196) was approved by the Regional Committee for Medical and Health