Prognostic significance of autoimmunity during treatment of melanoma with interferon

Abstract

Since the pivotal cooperative group trials in the 1980’s-90’s, high-dose interferon (HDI) has been the standard of adjuvant therapy. Despite multiple other trials evaluating potential new therapies in melanoma, HDI remains the only FDA-approved therapy for stage IIB and III melanoma. Initial reports from the more recent phase III international trials of modifications of the original HDI regimen linked the appearance of autoimmunity with improved outcomes of disease. Trials of high-dose interleukin-2, many years earlier, reported anecdotal observations that were consistent with the hypothesis that autoimmunity and clinical benefit of immunotherapies of melanoma are linked with one another. The only prospectively conducted study examining the appearance of clinical and laboratory evidence of autoimmunity during HDI therapy was published by Gogas and colleagues, demonstrating statistically significant impact on relapse-free survival and overall survival. Retrospectively conducted studies of different intermediate dosage regimens of interferon (IFN) have not fully confirmed the linkage of serological evidence of autoimmunity and improved survival outcomes. With the emergence of new immunotherapies in treatment of melanoma, this review highlights the importance of autoimmunity for future applications in melanoma and reviews significant differences of past studies evaluating the appearance of autoimmunity during IFN therapy in high-risk melanoma.

Introduction

Despite increasing efforts in prevention and therapy, metastatic melanoma at the time of diagnosis remains a very aggressive and deadly disease. Soaring 50% in young women since 1980, the incidence among people under age 30 is increasing faster than any other demographic group. Melanoma is the most common form of cancer for young adults between the ages of 25 and 29 and the second most common cancer in adolescents and young adults between ages 15 and 29 [1].

Melanoma is most common in Caucasians, striking men and women of all ages, races, and skin types. The biology of melanoma differs for intermittent, sun-damaged skin and for acral/mucosal sites where solar radiation appears to be a less important etiologic factor, and the mutation of genes such as c-KIT and BRAF distinguish these forms of melanoma [2, 3]. Malignant transformation of melanocytes in the basal layer of the epidermis gives rise to cutaneous melanoma. Although clearly linked to ultraviolet radiation, other risk factors such as family history and the presence of clinically atypical and histopathologically dysplastic nevi, skin complexion also plays an important role.

Additional risk for development of melanoma is associated with the melanocortin-1 receptor gene variant (MC1R) that results in red hair, fair skin and freckles. In affected individuals, [lighter] pheomelanin pigment is produced instead of the [darker] eumelanin pigment [4]. While eumelanin has the ability to absorb UV radiation, pheomelanin gives rise to free radicals upon contact with UV radiation contributing to further skin damage. The development of more potent genetic analysis tools allows new insights into gene mutations that occur in melanoma. This recently discovered genetic information has given rise to many new drugs that have already made substantial progress in clinical trials [5].

The use of high-dose interferon (HDI) in high-risk melanoma has been controversial as adjuvant treatment for patients with high-risk melanoma. High-risk disease is defined as all melanomas with thick primary lesions (T4 N0 M0, greater than 4 mm depth) or any invasion depth with positive lymph nodes (T1–4 N1 M0) with 5-year recurrence risks of 60% and 75%, respectively (see Table 1). Adjuvant HDI therapy is associated with significant toxicity in relation to the constitutional, hematologic, hepatic, neurologic, cutaneous, musculoskeletal, gastrointestinal, cardiovascular, and renal systems, potentially affecting the ability to deliver the full course of therapy [6]. Despite these toxicities of the HDI regimen, several studies from 1999 to 2001 have documented the feasibility of this treatment for patients treated in cooperative group settings across the US, where 90% of patients were able to complete therapy in the largest and most recently completed E1694 HDI trial [7]. HDI toxicity is now largely manageable, using the published guidelines that have emerged from the USA, Canada, and Europe [6, 8, 9].

Table 1.

Melanoma-risk groups based on 5-year relapse risk

Melanoma risk group	Stage
Low	IA–B
Intermediate	IIA–IIIA
High	IIIB–IV

The immune system plays a crucial role in the progression and pathogenesis of melanoma, but the exact mechanisms necessary to determine the outcome of the early disease are not fully understood. Rare cases of spontaneous regression of metastatic melanoma have been reported. Recent literature reviews in 2009 analyzed 76 published cases but did not find a particular mechanism for this phenomenon [10]. Between 10% and 35% of all melanomas show regression upon pathology review; however, such regression is generally considered to be associated with adverse outcomes [11]. Although other published studies do not uniformly support this belief [12, 13], the most recent 2009 American Joint Committee on Cancer compilation reported that no adverse impact of regression is apparent [14]. This suggests the direction of the tumor evasion from appropriate host immune response.

During the first trials with high-dose interleukin-2 (IL-2) and lymphokine-activated killer cells in patients with metastatic melanoma and renal cell cancer, a percentage of patients developed autoimmune thyroiditis as a result of the treatment [15]. This prompted a clearer focus on the phenomenon of autoimmunity than previous findings associated with favorable outcomes in the context of IL-2 therapy [16]. Autoimmunity can be seen as an example of overreaction of the immune system. A fine balance between regulatory (CD 25⁺, CD4⁺) and self-reactive/effector T cells prevents immune response to the normal host tissue in the human body [17].

Evidence implicates that regulatory T cells fulfill a unique role, controlling immune responses that are, in general, either physiologic or pathologic [18]. Depletion of regulatory T cells in otherwise normal mice causes autoimmune diseases [19, 20]. Regulatory mechanisms involving T regulatory cells that are marked by FOXP3—a transcription factor that is essential for development and function of regulatory T cells—are one component that has only recently been evaluated [21, 22]. Mutation of the FOXP3 gene causes a clinical syndrome called IPEX (immune dysregulation, polyendocrinopathy, enteropathy, and X-linked), that provides further evidence of its importance in the regulation of the immune system [23].

Another important player in immune regulation is the regulatory T cell receptor CTLA-4, already prevalent as a target in ongoing malignant melanoma trials [5]. FOXP3 and CTLA-4 expression on T cells appear to be corollaries. For example, terminally differentiated CD4⁺ T cells express FOXP3 highly and CD25⁺ constitutively expresses CTLA-4. Blockage of CTLA-4 with monoclonal antibodies causes organ-specific autoimmune disease and inflammatory bowel disease in otherwise healthy mice. Furthermore, mice lacking CTLA-4 on their regulatory T cells develop similar autoimmune disease as their FOXP3-deficient counterparts. T cell inhibitory receptor programmed death-1 (PD-1) is a more recent target for antibody blockage (MDX-1106, Bristol–Myers Squibb) currently undergoing a phase-I trial in solid tumors [24]. Recent laboratory evidence suggests that blocking the T cell immunoglobulin mucin (TIM) 3 receptor together with PD-1 receptor is capable of reversing T cell dysfunction in melanoma patients [25]. The importance of immunoregulation and the emergence of pharmacological means to modulate these, in context with current melanoma therapies, make it critical to evaluate the process of autoimmunity.

Interferon

Interferon (IFN) was discovered more than 50 years ago as a substance that showed inhibition of viral replication [26]. Two subtypes have been developed for therapeutic purposes so far, interferon-α−2-a (Roferon-A^®, Roche Pharmaceuticals, Basel, Switzerland) and interferon-α−2-b (Intron-A) Schering Plough, Kenilworth, NJ, USA). These two IFNs differ from each other in only 2 of the 166 amino acids, and are nearly identical in therapeutic effects and toxicity profiles. All type I IFNs bind to a specific cell surface receptor complex known as the IFN-α receptor (IFNAR) [27]. The IFN-α receptor is widely expressed on both tumor cells and immune effector cells. It mediates most of its effects via activation of the Janus kinase (JAK) that further recruits transcription of the STAT pathway [28–30]. The STAT factors form dimers, which are transported to the nucleus where they modulate the interferon response elements (ISREs) in combination with stabilizing molecules. These IFN-stimulated response elements act as transcription factors to activate a number of genes that ultimately lead to interferon-α production. Effects include the inhibition of viral growth, growth inhibition of tumor cells, and immunostimulatory effects by activation of different cytokines [28].

In the early 1980s [31], Bart et al. demonstrated promising effects of IFN-α upon melanoma cells in vitro in the murine B16 melanoma cell line. The observation of clinical anti-tumor effects of IFN in the first phase I and phase II trials ultimately led to the pursuit of this agent in earlier stages of melanoma in the cooperative groups [32, 33]. Three US cooperative group studies have evaluated high-dose IFN-α−2b (HDI) for the treatment of high-risk melanoma in the past [7, 34, 35]. The pivotal ECOG trial E1684 reported in 1996 was the first prospective trial (n=280) that investigated the HDI regimen currently approved in the USA this regimen [of IFN-α−2b at 20 million units (MU)/m²/d intravenously (IV) 5 days/week for 4 weeks]) followed by 10 MU/m²/d subcutaneously (SC) three times weekly for 48 weeks was compared versus observation [34].

The median relapse-free survival (RFS) time in patients receiving IFN was 1.72 years, while median for those in the observation arm was 0.98 years, respectively. The overall median survival time was 3.82 years in the IFN group as opposed to 2.78 years in the observation group. The following ECOG trial E1690 was a prospective, randomized and three-arm study that evaluated HDI, low-dose IFN, and observation in high-risk melanoma (clinical stage IIIB/IV) [35] (see Table 2). Adjuvant IFN-α therapy has been hypothesized to operate through multiple mechanisms including antiviral, antivascular, pro-apoptotic, and immunomodulatory effects. To date, immunomodulatory effects appear to be most closely linked to its clinical anti-tumor effects in node-metastatic melanoma [36, 37].

Table 2.

Comparison of study characteristics analyzing statistical significance of autoimmunity in melanoma in interferon therapy

Study characteristics	Gogas et al. 2006	Bouwhuis et al. 2009	Bouwhuis et al. 2010
Study size/patient samples analyzed	364/200a	1388 +855/187 +356b	1256/296
Pathology review	Centralized	Multiple centers	Multiple centers
Autoantibody testing	Accredited clinical laboratory	Research laboratory	Research laboratory
Interferon dosing	Modified high-dose regimen (IV, SC)	Intermediate-dose (SC only)	Intermediate-dose (SC only)

^aSubstudy of a larger clinical trial that was conducted prospectively at a single institution

^bIn this study, patient samples were obtained from two separate clinical trials (EORTC 18952 and Nordic IFN trial)

Therapy in the adjuvant setting of potentially curative operable disease aims at inducing host response that may eradicate micrometastatic disease, i.e., the presumed source of future disease recurrence [7]. Results showed a dose-dependent and significant benefit on RFS in the HDI group. The ECOG trial E1694 high-dose IFN showed significant benefit in terms of RFS and overall survival when compared to GMK ganglioside vaccine [7]. Less-intensive regimens tested in the melanoma adjuvant setting have not demonstrated durable effects upon relapse or death as shown with HDI.

These include intermediate-dose IFN (given subcutaneously) regimens tested in the European Organization for Research and Treatment of Cancer (EORTC) 18952 (T4, N1–2) [38]. The EORTC 18952 trial randomized patients to intermediate-dose IFN (10 MU/d for 5 days per week for total of 4 weeks followed by 10 MU three times a week for 1 year or 5 MU for three times a week for 2 years) versus observation only. Results of this trial, using lower IFN dosing when compared to the previously mentioned ECOG studies, did not show statistically significant distant metastasis-free interval and overall survival. Despite positive data from recent clinical trials of other immunotherapies in advanced inoperable melanoma, Interferon-α−2b (IFN-α−2b) remains the only agent approved by regulatory authorities worldwide (United States Food and Drug Administration [FDA]) for adjuvant therapy of high-risk melanoma to date.

The first large study that looked at autoimmunity associated with HDI therapy in patients with high-risk melanoma was published by Gogas et al. in 2006 [39]. The parent trial enrolled 364 patients in 13 institutions, and the autoimmunity sub-study analyzed 200 patients enrolled at 1 institution. Results showed significant survival benefit in patients that developed serum auto-antibodies on HDI therapy and will be discussed in more details in a later section.

Pegylated interferon

Pegylated interferon (PEG-IFN), widely used for the treatment of hepatitis, has not been studied as extensively as recombinant HDI in the treatment of high-risk melanoma [40, 41]. Available evidence suggests that the schedule of administration for this agent is more convenient than HDI, but the acute and chronic toxicity do not appear to differ qualitatively. The favorable pharmacokinetic properties of PEG-IFN allow its administration on a weekly basis with sustained exposure to IFN during that entire period at modest ambient levels [42].

The large size of the PEG-IFN molecule results in a 100-fold reduction in renal clearance when compared with conventional IFN [43]. Peak plasma levels of PEG-IFN are achieved approximately 80 h post-administration, whereas conventional intravenous IFN reaches its peak serum levels within minutes; subcutaneous administration does so within 7–12 h after administration, due to absorption and elimination kinetics given by the IV and SC routes. The role of the high concentrations of IFN achieved by the IV route of administration in the induction phase of the HDI therapy has no counterpart in therapy given with lower doses administered by the SC route alone or the kinetics of PEG-IFN, and the induction phase has been a component of both trials that have shown independent positive results in terms of overall survival [34, 35].

The largest clinical trial to date using PEG-IFN reported by Eggermont and colleagues showed a significant improvement in recurrence-free survival, but did not show a significant impact on overall survival. Subset analyses suggest that the impact of this regimen is observed almost entirely among patients with microscopic lymph node disease [40].

Autoimmunity induced during interferon-α therapy

The evasion of host immunity appears to be crucial in the development and progression of melanoma. As reported nearly three decades ago [44–46], clinical observations have suggested that the appearance of autoimmune vitiligo is a favorable prognostic factor in patients with melanoma. Results of later trials using high-dose IL-2 and HDI have further linked the appearance of autoimmunity with improved outcome—although none of these trials were specifically designed to analyze the appearance of autoimmunity [15, 16, 47–51].

Until this time, the only large trial that prospectively investigated the appearance of clinical and laboratory features of autoimmunity during HDI treatment in patients with high-risk melanoma was published by Gogas and colleagues in 2006 [39]. This substudy was partially conducted by the Hellenic Cooperative Group (HECOG) trial at 13 institutions that enrolled 364 patients with high-risk melanoma. Patients received either a month-long or year-long course of a modified HDI regimen. Blood samples from the substudy population of 200 patients enrolled at the First University of Athens, were collected for autoantibody testing against six autoantigens prior to, during, and up to 12 months after initiation of HDI therapy. Patients with preexisting autoantibodies were excluded. Regardless of the IFN arm, autoantibodies (or clinical manifestations of autoimmunity) were detected in 52 patients. At 45.6 months of median follow-up, only 7 of 52 patients in the autoimmunity group developed relapse, whereas 108 of 148 patients without autoimmunity relapsed.

Similarly, outcomes in overall survival were 2 of 52 patients dying in the autoimmunity group, whereas 80 of 148 patients without autoimmunity died. In multivariate analyses, autoimmunity was found to be an independent prognostic marker for improved RFS and overall survival (P<0.001). A retrospective analysis exploring evidence autoimmunity (testing for three autoantibodies) in patients from the EORTC 18952 and the Nordic IFN trial did not demonstrate a statistically significant correlation of autoimmunity and RFS [52].

A more detailed evaluation of the study characteristics reveals some important differences from the Gogas study. Although enrollment criteria for high-risk melanoma patients was very similar, the EORTC study tested only a small subset of patients enrolled in the two clinical trials (see Table 2). Patient samples were obtained from 27 different collaborating centers for central testing of autoantibodies, but each enrolled patient had local pathology assessment of their melanoma at the collaborating center. Patient sera were analyzed by the same enzyme-linked immunoabsorbent assay (Quanta Lite; Inova Diagnostics, San Diego, CA). Tests by the Gogas study were performed in an accredited hospital laboratory, whereas those in the Bouwhuis study were completed in a non-accredited/non-standardized research laboratory.

Unlike the Gogas et al. study, Bouwhuis and colleagues did not report data on clinical manifestations of autoimmunity, only three autoantibodies were screened (see Table 3) and seroconversion occurred in 34% and 32% of the EORTC 18952 and Nordic IFN trials, respectively. This is a significantly higher number than the 26% rate of overall seroconversion and clinical autoimmune disorders reported by Gogas et al. Bouwhuis’ study further analyzed results using three different statistical models. Model 1, analyzed for “any positive autoantibody test” (time-independent Cox model), showed statistical significance for recurrence-free survival. To address guarantee-time bias phenomenon, they used time-dependent Cox models that tested in model 2 for “latest positive autoantibody status” and in model 3 for “latest autoantibody status.” Analysis using models 2 and 3 did not enable this group to confirm the previously reported statistical significance for correlations between autoimmunity and improved recurrence-free survival.

Table 3.

Laboratory and clinical parameters assessed in each study to detect autoimmunity

Gogas et al. 2006	Bouwhuis et al. 2009
Antithyroglobulin antibodies	Antityroglobulin antibodies
Antinuclear antibodies	Antinuclear antibodies
Anticardiolipin antibodies (IgM and IgG)	Anticardiolipin (IgA, IgM, and IgG)
Anti-DNA antibodies
TSH
Thyroxine
Triiodothyronine
Clinical evaluation for manifestations of autoimmunity done prospectively	No prospective clinical evaluation

Further differences between these studies include the doses of IFN tested: Gogas used high dosages of interferon in the induction phase (IV) and maintenance dosages (SC), whereas the Bouwhuis group studied therapy employing intermediate doses of interferon for induction and maintenance phases. A more recent study by Bouwhuis examined the prognostic value of autoantibody responses in patients with high-risk melanoma receiving PEG-IFN, which also did not detect improved outcomes associated with autoimmunity [53]. As with the prior study of this group, only a small subset of patients from the original study cohort was analyzed using research laboratory tests for autoantibodies, rather than GLP clinical laboratory assays. Direct comparison between the two studies is difficult, as PEG-IFN has different pharmacokinetics compared to non-pegylated IFN.

Discussion

Despite ongoing clinical trials and reports of new promising therapeutics, HDI remains the only regimen approved for adjuvant treatment of high-risk melanoma (Table 4). Based on the results of three ECOG and intergroup studies with HDI, the therapeutic efficacy of interferon appears to exceed other regimens, and is associated with treatment of 1 year’s duration. Rate-limiting features of this regimen include the side effects for patients, although more than 90% of patients now complete the planned regimen.

Table 4.

Interferon-α-2b-dosing regimens used in treatment of high-risk melanoma

Interferon regimen	Duration/dosing schedule
High-dose IFN
Induction	IFN, 20 MU/m2 iv, 5 times per week for 4 weeks
Maintenance	IFN, 10 MU/m2 sc, 3 times per week for 48 weeks
Intermediate-dose IFN
Induction	IFN, 10 MU/m2 iv, 5 times per week for 4 weeks
Maintenance	IFN, 5–10 MU/m2 sc, 3 times per week for 48 weeks
Low-dose IFN	IFN 3 MU/m2 sc, 3 times per week (varying duration)
PEG-IFN
Induction	PEG-IFN 6 μgram/kg sc, weekly for 8 weeks
Maintenance	PEG-IFN 3 μgram/kg sc, weekly for up to 5 years

Overall, clinical benefits from HDI therapy amount to a reduction in the vulnerability for relapse by one-third (E1684, E1690, E1694) and reduction of mortality by one- quarter to one-third (E1684, E1694). Attempts to select a population of melanoma patients that might respond more favorably to HDI treatment has led to the exploration of autoimmunity, which has anecdotally been associated with improved outcome of autoimmunity as reported by multiple investigators [15]. Gogas and colleagues conducted a well-designed prospective study that analyzed 200 patients and showed a significant benefit of HDI therapy in patients who developed autoimmunity during therapy.

Subsequently, Bouwhuis and colleagues reported that a subset of patients from larger EORTC studies did not show the same statistical correlation of favorable outcomes associated with autoimmunity. Stuckert and colleagues presented further evidence during the 2007 ASCO annual meeting. Their findings showed autoimmunity as a predictive biomarker in recurrent-free survival in HDI of the E2696 trial [54]. Final analyses of this trial are pending. Staging of malignant melanoma specimens in the trials of HECOG, ECOG, and EORTC include centralized staging, whereas Bouwhuis et al. conducted staging in various institutions.

Differences are also seen in laboratory tests of specimens from the previously discussed clinical studies. While Gogas and colleagues were using an accredited clinical laboratory for all their specimen testing, Bouwhuis and colleagues used non-accredited experimental laboratories. Thus, it is difficult to ascertain if the analyses limited to only three autoantibodies were able to generate enough statistical data for comparison with more rigorous tests from the Gogas experiment.

Interestingly, the rates of seroconversion are much higher in the Bouwhuis study. Bouwhuis acknowledges these differences between the two studies, and while the same assays were reportedly used, there is no clear explanation for this discrepancy. Finally, the trials have employed different dosage regimens. The Gogas study used a higher-dose IFN regimen, while E18952 used intermediate-IFN dosages and E18991 used PEG-IFN at intermediate dosages. Several points require exploration in future studies. Larger trials using prospective enrollment with centralized pathology review and homogenous evaluation of specimens from all patients in clinically approved laboratories would avoid several pitfalls of the study design. In the absence of an overall survival impact of PEG-IFN, EORTC 18991 may not be the most promising regimen in which to validate a role of predictive biomarkers of therapeutic benefit.