The purpose of this retrospective study is to evaluate tumor response rates and toxicity profiles of IL-2 therapy in a more contemporary cohort of patients with metastatic melanoma and a history of brain metastases.
Purpose: High dose interleukin-2 (IL-2) therapy results in infrequent but durable responses in metastatic melanoma (MM); however, its use in patients with brain metastases (BM) has been controversial due to safety concerns and limited data on efficacy in this population. The purpose of this retrospective study is to evaluate tumor response rates and toxicity profiles of IL-2 therapy in a more contemporary cohort of patients with MM and a history of BM.
Patients and Methods: A total of 320 patients who received IL-2 therapy with stage III or IV MM were included in a prospectively collected database. At the time of study entry, 38 patients had a history of BM, of whom 35 had received previous treatment for BM, which was mostly radiation therapy. A post-hoc analysis of overall survival using Kaplan-Meier (KM) curves, as well as tumor response rates, number of cycles of therapy completed, and toxicity profile by cycle and organ system was conducted. The 282 patients in the database without BM served as a control population.
Results: The average interval from diagnosis of BM to start of HD-IL2 therapy was 3.1 months (range: 0.39 to 13.5). Median follow up (mFU) was 22.0 months. The median overall survival (mOS) in MM patients with BM (n=38) was 12.2 months, versus 19.4 months in those patients without BM (n=282) (p = 0.23). Of the 38 patients with BM, 34 (89.5%) were evaluable for response to HD-IL2. The response rate was 11.8%, as compared to a response rate of 14.0% in subjects without BM. One patient (2.9%) achieved complete response (CR), 3 patients (8.8%) achieved partial response (PR), and 8 patients (23.5%) had stable disease (SD). IL-2 treatment cycle was concluded due to toxicity in the majority of patients in both cohorts and the most commonly affected systems were cardiac, renal, neurological, and gastrointestinal. Treatment-limiting neurologic toxicity was observed in 12.1% of patients with a history of BM and 13.2 % of patients without BM during Cycle 1.
Conclusion: This is the largest analysis reported to date of IL-2 treatment in patients with a history of BM. IL-2 was well tolerated in this cohort and the rate of neurologic complications was similar to the control group. The response rate and median survival are comparable to those observed for patients without a history of BM. IL-2 is safe and has therapeutic activity in melanoma patients with a history of treated BM.
Melanoma is increasing in incidence with approximately 10,000 deaths annually in the United States. Considerable morbidity and mortality from melanoma is related to the potential for disease to metastasize to all locations in the body. Metastasis to the central nervous system (CNS) is especially problematic as it may be associated with significant clinical symptoms and early disease-related mortality. Nearly 40% of patients with stage IV melanoma develop brain metastases (BM), but this may underestimate the prevalence, as 55% to 75% of patients who died from melanoma were shown to have BM on autopsy.1-6Despite modern treatments, such as stereotactic radiosurgery, median overall survival (OS) in MM patients with cerebral metastases is approximately 6 to 9 months.1,7Karnofsky Performance Score as well as the number of BM are key prognostic factors in MM, and can be used to stratify patients into groups with a mOS range of 3.4 months to 13.2 months with the Graded Prognostic Assessment (GPA) scoring system.8Abnormal LDH has also been shown to be a poor prognostic marker in MM patients with BM.9
The mainstay of therapy for melanoma BM has been a localized approach consisting largely of stereotactic radiation and craniotomy with tumor extirpation, when possible. Most systemic chemotherapy, such as dacarbazine, does not cross the blood-brain barrier. Some patients were treated with temozolomide, an alkylating agent that enters the CNS but was not associated with improved survival in patients with MM.10Newer therapies, such as ipilimumab, have shown activity in MM patients with BM, and this is thought to be due to, at least in part, peripheral activation of T cells, which are able to enter and exert an anti-tumor effect in the CNS.11,12Additionally, BRAF inhibitors such as vemurafenib and dabrafenib have shown activity against BM, with dabrafenib demonstrating an intracranial response rate of 39%.13These new agents have brought forth a multi-disciplinary approach to the treatment of melanoma BM involving medical oncology, radiation oncology, and neurosurgery.
High dose interleukin-2 (IL-2) is consistently associated with a response rate of 17% to 20% and a median progression-free survival (PFS) of 13 months in those who responded to therapy.8,14-17The use of IL-2 has often been restricted to patients without brain metastases due to concerns that it would lead to greater toxicity due to cerebral edema from capillary leak syndrome and reduced efficacy due to immunologic privilege of the CNS.18
In contrast to these concerns, several small studies have shown favorable outcomes with IL-2 treatment of patients with BM from metastatic melanoma.1,9,15,19These studies suggested a similar toxicity profile in those with and without BM. In one study, 7 patients with stage IV disease were shown to have BM, and 1 showed a partial response after surgical resection of a solitary brain lesion. Out of a total of 8 patients, 1 was noted to have neurological toxicity, which was thought to be secondary to alcohol withdrawal rather than IL-2 treatment.15Another study demonstrated that 2 patients with BM had some intracranial tumor response; 1 showed a mixed response and 1 showed a complete response; this patient had whole brain radiation and polyvalent vaccination prior to IL-2. There was 1 reported neurological adverse event out of 15 patients, which was characterized as confusion.19A phase III trial of high-dose IL-2 treatment showed an increase in toxicity and response rate compared to low-dose IL-2 when patients were stratified for bone and liver metastases, but not BM.20
This review was undertaken to better evaluate the potential efficacy and toxicity associated with high-dose IL-2 therapy in the contemporary era. A retrospectively and prospectively collected database,Proleukin®Observational Registry to Evaluate the Treatment Patterns andClinical Response inMalignancy (PROCLAIMSM), was established in 2011.21This is the largest IL-2 database and contains demographic, clinical and outcome data on patients with MM receiving IL-2 at 35 different academic and community medical centers. The database was accessed to determine treatment outcomes and adverse event profiles in MM patients with BM treated with high-dose IL-2.
Data were collected from the observational database PROCLAIM. To be included in the prospective cohort of the registry, patients must sign informed consent forms, have been at least 18 years of age, and have received an initial course (2 cycles) of IL-2. Patients who received more than 1 course of IL-2 and/or already undergone a post-treatment scan were not eligible for enrollment, in order to control for bias arising from results of the scan. For patients reporting BM, the sites were queried for the date of BM diagnosis and all treatments for BM. Baseline characteristics on gender, age, tumor staging, ECOG performance status, melanoma subtype, mutation testing, as well as pre-treatment LDH were gathered. The number of BM per patient and the use of steroids were not captured in the database. All sites were required to have approval by their respective institutional review boards (IRBs) prior to subject enrollment.
Treatment and Assessments
All patients were treated with IL-2 as clinically indicated; patients were administered a dose of 600,000 to 720,000 IU/kg IV bolus every 8 hours on days 1 to 5 (Cycle 1) and days 15 to 19 (Cycle 2) with a maximum of 28 doses per two-cycle course.22The total number of cycles and IL-2 doses received by each patient was recorded. Tumor response to IL-2 was captured by imaging of the extracranial and intracranial compartments 4 to 6 weeks after completing 1 course of IL-2. Assessment of tumor response was determined by the treating investigator using either RECIST or WHO criteria per the site’s standard procedure.
The reason for cycle conclusion was recorded; potential reasons included completion of planned doses, toxicity, patient preference, death, etc. If toxicity was given as the reason for cycle conclusion, the site could list up to 3 different systems (ie, cardiac, neurologic, renal, etc.) and the specific toxicity within each system was noted. For example, a neurologic toxicity could be further categorized as agitation, confusion, psychosis, and not otherwise specified. Overall survival data were measured from the start of IL-2 treatment to either the patient’s date of death or date of follow up. Cause of death was recorded.
Analysis was performed from data that were locked on September 24, 2015. Descriptive statistics were provided for baseline characteristics. Follow up time was measured from the IL-2 start date to either the date of death or date of last follow-up. The method of Kaplan and Meier was used to obtain survival curves, mOS, and survival rates with corresponding 95% confidence intervals and was compared using the log-rank test. Tests for association for categorical variables used the Chi square test and given the small sample size the Fisher exact test was utilized when necessary. The two-sample t-test was utilized when testing for an association between the group’s means for continuous variables. Throughout the analysis p-values of <0.05 were considered statistically significant. All statistical computations were performed using SAS 9.4.
Of the patients from the prospective cohort diagnosed with metastatic melanoma (N=320), the patients with BM at baseline (n=38) were identified versus patients without BM at baseline (n=282).TABLE 1provides descriptive statistics for patient demographics. Of the 320 patients identified, 123 (38.4%) were female and the rest were male. The median age was 53 (range, 21-84) and 271 (84.7%) were less than 65 years of age with 49 (15.3%) being over the age of 65 years. The ECOG performance status was 0 in 221 (69.1%), 1 in 90 (28.1%), and 2 in 3 (1%) of the patients. There were no differences in the age, sex or ECOG performance status of patients in the BM cohort compared to the non-BM cohort. While definition of BM places all patients as Stage IV M1c, the non-BM group also had a large number of stage IV M1c patients (120; 37.5%). The majority of patients in both cohorts were reported to have elevated LDH levels prior to starting IL-2 (82.8% for BM group and 67.8% for non-BM group; P = .39).
The average interval from initial diagnosis to diagnosis of BM was 34.9 months (min: 0.03, max: 176). Of the 37 patients with complete data available, there were 7 (18.9%) that were diagnosed with BM during the first year of initial cancer diagnosis. Two of those 7 patients were diagnosed with metastatic disease at the time of their initial presentation. The average interval from diagnosis of BM to start of HD-IL2 therapy was 3.1 months (min: 0.39, max: 13.5). Nine (24.3%) of these patients started HD-IL2 within a month or less from being diagnosed with BM.
Of the 38 BM patients all but 3 patients (7.89%) received local treatment for BM prior to starting HD-IL2. There was 1 patient who had not received any prior oncological treatments prior to starting HD-IL2. Of the 35 BM patients who had received prior treatment for brain metastases, 32 received some form of radiation therapy27 had stereotactic radiosurgery and 6 patients received whole brain radiotherapy (WBRT). Five patients had undergone surgical resection for BM.
Tumor response to IL-2 was available for 34 of the 38 patients (89.5%) in the BM group and 272 of the 282 patients (96.5%) in the control group. In the control population, the objective response rate (ORR) was 14% (7 [2.6%] complete response and 31 [11.4%] partial response,TABLE 2A). An additional 78 patients had SD (28.7%) for a disease control rate (DCR) of 42.7% (TABLE 2B). In contrast, the BM patients had an ORR of 11.8% (1 CR [2.9%] and 3 PR [8.8%],TABLE 2A) An additional 8 (23.5%) patients were stable for a DCR of 35.2% (TABLE 2B). Using Fisher’s exact test, there was no statistically significant relationship between the 2 groups with respect to ORR (P = 1) or DCR (P = .4645,TABLE 2). The patient who achieved a complete response was diagnosed with melanoma metastatic to the brain and the bowel in June 2011. Her BM was treated with stereotactic radiosurgery, followed immediately by brief treatment on a protocol with MAGE-3 vaccine. She then received 2 cycles of IL-2 therapy in September and October 2011 and achieved a CR on her first restaging CT and MRI scans. She remains alive more than 4 years later and has not received any subsequent therapy.
Number of Doses of IL-2 Administered
The total number of doses of IL-2 received by patients for the entire duration IL-2 treatment was calculated for all patients (TABLE 3). In the BM group (n=38) the average number of total doses received was 20.6 (95%CI: 16.6, 24.5), the control patients (n=282) had an average of 24.0 total doses (95%CI: 22.4, 25.5), there was no statistically significant difference between the two groups.
The adverse events observed are shown by organ system for Cycle 1 of IL-2 treatment inFIGURE 1A. In general, there were no significant differences in the BM and control groups with regard to proportion of toxicity by organ system. During Cycle 1, 21.1% of BM patients discontinued IL-2 therapy wholly or in part due to neurotoxicity, including 1 patient who died due to cerebral edema and brain herniation. No patients died during therapy in the control group. In the control group, 16.7% of patients discontinued IL-2 therapy wholly or in part due to neurotoxicity. During Cycle 2, 25.8% of BM patients and 22.1% of control patients discontinued the cycle wholly or in part due to neurotoxicity. Neurotoxicity (FIGURE 1B) was broken down into agitation, confusion, psychosis, and other. In the BM group, the neurotoxicities in cycle 1 were reported as 1 case of vivid dreams, 1 case of agitation, 2 cases of confusion, and 3 cases of fatigue. The neurotoxicities in the BM group in cycle 2 were reported as 2 cases of confusion, 2 cases of psychosis, 1 case of acute encephalopathy, 2 of mental fatigue, and 1 of weakness. No altered sensorium, seizure, or coma was reported in the BM group.
In both groups, slightly over half of the patients were reported deceased at time of last follow up, 60.5% for BM patients and 51.1% for patients without BM. The mOS for all patients was 19.3 months (95%CI: 15.9, 21.5) (TABLE 4A) and median follow-up time was 22 months (95%CI: 18.8, 23.7) (TABLE 4B). The mOS for patients with BM was 12.2 months (95%CI: 6.5, 23.5) and 19.4 months (95%CI: 17.1, 23.2) for patients without BM, which was not significantly different (P = .23). At 1 year, the survival rate for patients with BM was 51.8% (95%CI: 34.2, 66.7) and for patients without BM the 1-year survival rate was 67.6% (95%CI: 61.4, 72.9). The hazard of death was initially lower in the control group, but after 20 months of follow-up there was a crossover of the Kaplan-Meier survival curves (FIGURE 2); past that point, patients with BM had a lower hazard of death than those without BM.
For the BM group there was a statistically significant relationship between tumor response and mOS (P = .0065) (TABLE 2A). The mOS was not reached for patients with complete response (n=1) or stable disease (n=8). Patients with partial response had a mOS of 19.5 months (95%CI: 2.99, not evaluable [NE]) and patients reporting progressive disease had a mOS of 9.2 months (95%CI: 3.88, 12.92). For the control group (TABLE 4B) there was a statistically significant relationship between tumor response and mOS (P <.0001). The mOS for patients with complete response (n=7) was not reached. Patients with PR (n=31) had a mOS of 23.3 (95%CI: 17.1, NE), and patients with SD (n=78) had a mOS 26.8 months (95%CI: 21.3, NE) and patients with progressive disease (n=156) had a mOS of 13.7 (95%CI: 10.29, 17.59).
The development of BM has historically been associated with dismal outcomes for patients with melanoma. Treatment with high-dose IL-2 was considered contraindicated in such patients due to concerns over intracerebral edema secondary to IL-2-induced third space fluid sequestration. Given the overall poor prognosis for such patients and the general exclusion of patients with CNS metastases from clinical trials, the real clinical benefit of IL-2 in patients who are diagnosed with BM remains largely unknown. In an earlier report a 5.6% ORR with high-dose IL-2 was reported in patients with previously untreated BM.9The authors concluded that IL-2 might be safe and appropriate in highly selected melanoma patients. In the current era many things have changed, including improved sensitivity of imaging techniques, increasing use of stereotactic radiosurgery for oligometastases, progress in targeted therapy and immunotherapy in melanoma, and better management of patients on high-dose IL-2. Thus, we sought to determine both the safety profile and clinical benefit of IL-2 in a more contemporary era.
Using a modern, prospectively collected IL-2 data- base, we identified 38 patients with BM at initial diagnosis out of 320 sequentially enrolled melanoma patients. The mean number of IL-2 doses was similar between the two cohorts (20.6 vs. 24; seeTABLE 3). In this series, among the 38 patients who were given high-dose IL-2, an ORR of 11.8% was seen, and this included a complete response rate of 2.9% of the population. These rates are closer to the reported objective response rates of 16% to 17% for metastatic melanoma patients receiving IL-2 without BM.17,23Thus, at least in selected patients, high-dose IL-2 appears to have activity in patients with melanoma BM.
Another argument against IL-2 is that a higher degree of neurotoxicity may be associated with treatment. Although there was 1 probable treatment-related neurologic death in the BM group, there were no other notable differences in the type, frequency, or intensity of adverse events between melanoma patients with and without BM regardless of cycle number or organ system. The most frequently reported toxicity leading to discontinuation of IL-2 was due to cardiac or renal toxicity with comparable rates in patients with and without BM (FIGURE 1). Neurotoxicity was the third most common reason for stopping IL-2 treatment but the rates were roughly similar in both groups. There was a higher rate of gastrointestinal adverse events in patients without BM across all for cycles that were included in the analysis. We attempted to distinguish if the presence of BM might be associated with a different presentation of neurotoxicity. While we saw a slightly higher incidence of agitation in those with BM, the incidence of confusion was actually lower in patients with BM. These differences were minimal and likely suggest that neurologic adverse events, while not infrequent in patients receiving IL-2, are not significantly different in patients harboring BM. Based on these results, no specific recommendations can be made to preferentially monitor patients with BM being treated with high-dose IL-2.
In this study, the median overall survival was 12.2 months for patients with BM compared to 19.4 months for those without BM (P = .237). Although the 1 year survival rate was numerically lower in the BM group than in the control group (51.8% versus 67.6%, respectively), the 3-year survival rate in the BM group was actually higher than the control group (30.2% versus 24.2%, respectively). These data suggest that long term survival is possible regardless of the presence of CNS disease. Further, the median overall survival of 12.2 months is higher than median overall survival reported in other published series of melanoma patients with brain metastases.24,25While it is interesting to speculate that control of CNS disease with IL-2 may prevent subsequent CNS and hematogenous metastasis of melanoma, this hypothesis requires further confirmation.
This study has several important limitations, including the potential for selection bias since the population was derived from expert IL-2 centers, the retrospective nature of the review and the lack of a randomized cohort. Further, this study included an abundance of younger aged and male patients, especially in the brain metastasis group, which may also limit the ability to generalize the findings. We did try to evaluate several important risk stratification factors, such as performance status and serum LDH levels to better define the two cohorts. In the total population, serum LDH was elevated in 159 (69.7%) of the patients with slightly higher levels in the group with BM (82.8%) compared to patients without BM (67.84%). The ECOG performance status was 1 to 2 in 31.6% of the patients with BM compared to 28.72% of those patients without BM. There were slightly younger patients in the BM group (92.1% < 65 years of age) compared to the control group (83.7% < 65 years of age). Thus, the groups appear to be relatively well matched with respect to performance status and LDH, but there was a slightly higher number of younger patients in the BM group. A recent report has demonstrated comparable clinical responses with lower adverse events in older patients so the impact of age is less certain in this population (26). Finally, while the database represents a useful resource, there are certain data that was not collected that may be important, such as the number of BM in each patient and the type of brain-directed treatment.
The impact of immunotherapy on the treatment of melanoma has been significant over the last 5 years with an increasing number of effective agents available. The progress with T cell checkpoint inhibitors targeting CTLA-4 and PD-1 are especially interesting and early studies have suggested activity in the CNS with these agents.11,27This report adds to others suggesting activity with high-dose IL-2 and thus, immunotherapy in general may be an appropriate alternative for patients with CNS disease. The availability of new, effective agents for patients with melanoma is providing new clinical challenges in designing treatment plans. Despite the progress, many patients do not respond to individual drugs, resistance can emerge and an understanding of the potential impact of each approach is needed. This is the largest analysis reported to date of high-dose IL-2 treatment in patients with a history of BM. Although the sample size is small and there are several limitations of the study, these data demonstrate that high-dose IL-2 has therapeutic efficacy in melanoma patients with BM and treatment is associated with an acceptable safety profile. High-dose IL-2 should continue to be carefully considered in appropriate melanoma patients even in the presence of BM. The results support further clinical investigation of high-dose IL-2 alone and in combination for this high-risk cohort of melanoma patients. More broadly, these data prompt reconsideration of the blanket exclusion of patients with BM from treatment with commercial IL-2 or on IL- 2-based clinical trials.