Maintenance treatment with the immunomodulator MGN1703, a Toll-like receptor 9 (TLR9) agonist, in patients with metastatic colorectal carcinoma and disease control after chemotherapy: a randomised, double-blind, placebo-controlled trial

In recent years, a number of agents have been developed with the aim of harnessing the inherent anti-cancer capabilities of the immune system. Most efforts have focused on the activation of adaptive immunity, as seen with ipilimumab—a cytotoxic T lymphocyte-associated antigen 4 inhibitor active in melanoma—and with other checkpoint inhibitors that target the programmed cell death 1 pathway (Hodi et al. 2010; Robert et al. 2011; Topalian et al. 2012; Tournigand et al. 2012; Brahmer et al. 2012). An additional immunomodulatory approach involves stimulation of innate immunity. MGN1703 is a novel, synthetic, dumb-bell-shaped, covalently closed DNA molecule (Weihrauch et al., submitted), which activates the innate immune system via Toll-like receptor 9 (TLR9) (Fig. 1). The TLR9 receptor is expressed on plasmacytoid dendritic cells (pDC) and B cells and recognises non-methylated cytosine–guanine dinucleotide (CG) motifs in bacterial, viral and mitochondrial DNA (Schmidt et al. 2006a; Krieg 2002; Kanzler et al. 2007; Ahmad-Nejad et al. 2002). TLR agonists can stimulate innate antitumour mechanisms, including activation of natural killer T (NKT) cells, monocytes and macrophages, and induction of cytokines (Kanzler et al. 2007; Carpentier et al. 2006; Pashenkov et al. 2006; Friedberg et al. 2005). TLR agonists containing non-methylated CG motifs have exhibited limited, transient adverse effects and some efficacy in clinical trials in various tumour types (Kanzler et al. 2007; Carpentier et al. 2006; Pashenkov et al. 2006; Friedberg et al. 2005).

MGN1703 has been investigated as adjuvant therapy in two trials (Weihrauch et al. 2005; Wittig et al. 2001). In the first, MGN1703 was used in combination with chemoimmunotherapy in patients with metastatic colorectal carcinoma (CRC) (Weihrauch et al. 2005); in the second, patients with metastatic solid tumours of colon, renal cell, or melanoma origin received autologous gene-modified cells with adjuvant MGN1703 (Wittig et al. 2001). In both cases, regimens were well tolerated with mild, transient side effects, although MGN1703 doses were low (Weihrauch et al. 2005; Wittig et al. 2001). More recently, a phase I trial showed that twice-weekly subcutaneous applications of single-agent MGN1703 (up to 60 mg) were well tolerated in 28 patients with metastatic solid tumours (Weihrauch et al., submitted). Six patients had stable disease (prolonged in three cases), and one with refractory CRC had a partial response.

The current study evaluated MGN1703 as maintenance treatment in patients with metastatic CRC who had disease control after induction chemotherapy. Failure of a different TLR9 agonist to improve outcomes in advanced non-small cell lung cancer when added to chemotherapy suggests that the supportive immune response initiated by TLR9 agonists may firstly require release of tumour-associated antigens (Hirsh et al. 2011). A decreased tumour burden together with a chemotherapy-free interval allowing recovery of immune cells may also facilitate a more effective immune response (Hirsh et al. 2011). This type of immunotherapy may therefore prove successful when given sequentially after chemotherapy. Treatment for metastatic CRC has improved significantly over the last two decades (Chu 2012), and the potential value of maintenance therapy is being extensively investigated (Strickler and Hurwitz 2012; Díaz-Rubio et al. 2012; Tournigand et al. 2012; Koeberle et al. 2013; Koopman et al. 2014).

IMPACT, a randomised, placebo-controlled, double-blind, phase II study, evaluated the efficacy and safety of subcutaneous MGN1703 (60 mg twice-weekly) as maintenance therapy following successful first-line induction therapy in metastatic CRC. This was the first placebo-controlled trial to prospectively investigate the impact of an immunomodulator as maintenance therapy in metastatic CRC and was based on the hypothesis that patients with disease control could benefit from immunotherapy.

The IMPACT trial showed that MGN1703 maintenance treatment was well tolerated and appears to induce durable PFS and disease control in selected patients with advanced CRC following induction therapy.

Findings from the study reflect the complications and challenges faced in assessing clinical response to immunotherapies. In the primary endpoint analysis, Kaplan–Meier curves separated after the median point; similar findings have been reported in other recent studies in metastatic CRC, for example, in populations with mixed KRAS mutation status treated with epidermal growth factor receptor inhibitors (Saif and Shah 2009; Ramos et al. 2008). Therefore, the activity of MGN1703 may be limited to a specific subpopulation of patients who achieve a prolonged response to treatment. Indeed, results from the secondary analyses indicate that three possible subgroups existed. Firstly, the greatest benefit with MGN1703 was observed in patients who achieved shrinkage of their tumour after induction chemotherapy. This is consistent with the theory that immunomodulatory treatment is more likely to be successful in patients with relatively low tumour burden. Similarly, normal CEA values after induction therapy also had predictive value. Finally, results from the immunological biomarker analysis reported a significant effect of MGN1703 on PFS relative to placebo in patients with levels of activated NKT cells ≥3.08 % but not in those with levels below this cut-off. Activated NKT cells may be a potential biomarker for selecting patients more likely to benefit from MGN1703 treatment.

Two assessments of response were used in the trial (investigator and independent review). Primary endpoint results with the two assessment methods were similar, although P-values decreased with the investigator assessment relative to independent review (0.04 versus 0.07), reflecting a decrease in statistical uncertainty caused by a reduction in the number of censored patients. In the independent review, 13 patients were censored (eight local relapses as assessed by investigators could not be confirmed on scans by the independent review). The investigator assessment included these eight patients so that only five patients were censored. Such inaccurate identification of relapse is indicative of the challenges involved in measuring responses to immunotherapy. Immunomodulators are associated with unique patterns of response versus conventional therapies. In studies of ipilimumab, for example, disease regression was sometimes not seen for several months and, in other cases, lesion growth was observed due to immune-related inflammation not disease progression (Weber et al. 2012).

Immune-related response criteria have now been defined (Wolchok et al. 2009), although traditional RECIST criteria, which may not detect all responses to immunotherapy (Hales et al. 2010), were used in the current trial. Therefore, in subsequent studies on MGN1703, immune-related response criteria may be more appropriate.

Data from several randomised trials investigating maintenance treatment for metastatic CRC with agents such as bevacizumab have recently been reported (Díaz-Rubio et al. 2012; Tournigand et al. 2012; Koeberle et al. 2013; Koopman et al. 2014). Interpretation of these data is complex, as all the trials differ in design and in how the maintenance or de-escalation phase is managed. Therefore, many questions remain regarding whether, and in what form, maintenance treatment should be used in this setting. Comparison of PFS data from the current trial and other trials of maintenance therapy is made difficult by the different assessment schedules employed. In the study reported here, the first assessment was performed at month 3 to allow patients to receive adequate immunomodulatory therapy; this is reflected in the steep decay of the upper portions of the Kaplan–Meier curves. With regard to other immunomodulators, the current data appear to be consistent with observations from some trials showing no effect in terms of median PFS but lasting benefit in some patients (Chung et al. 2010; Kirkwood et al. 2010). It may be of interest to observe, as with other immunomodulators, whether mature OS results for MGN1703 point to a broader activity than seen on the shorter-term PFS endpoint.

Structural and preclinical studies have shown that MGN1703 is a potent TLR9 agonist with limited capacity for interactions with molecules outside its target structure (Schmidt et al. 2006b, 2008; Kapp et al., submitted). In addition to this specificity, the relevant cellular activation profile of MGN1703 is distinct to that of other TLR9 agonists, such as PF-3512676, which failed to show benefit compared with chemotherapy alone in two phase III trials in NSCLC (Hirsh et al. 2011; Manegold et al. 2012). Additional immunotherapy approaches have been investigated in patients with tumour types other than CRC. These include agents that target specific immune regulatory checkpoints and increase the endogenous antitumour immune response. For example, ipilimumab which binds to CTLA-4, has been shown to result in 2-year survival rates of 23.5 % in patients with metastatic melanoma (Hodi et al. 2010). The majority of adverse events with ipilimumab are consistent with the proposed mechanism of action and can be severe, and long-lasting (Hodi et al. 2010). Other immune checkpoint inhibitors have been developed to target the programmed cell death 1 receptor and prevent T-cell inactivation, with studies investigating their potential in melanoma and NSCLC (Pardoll and Drake 2012). The combination of MGN1703 with a checkpoint inhibitor could offer a potentially synergistic immunotherapeutic approach in this setting.

Safety data for MGN1703 were in accordance with those from the phase I trial (Weihrauch et al., submitted). MGN1703 was generally well tolerated, with most drug-related adverse events being mild or moderate. Common adverse events included injection-site reactions, fever and fatigue. Good tolerance of MGN1703 was confirmed by the lack of treatment interruption despite therapy for up to 30 months in patients experiencing prolonged response. In general, the tolerability profile of MGN1703 appears similar to that of therapeutic vaccines such as sipuleucel-T, rather than the checkpoint inhibitors (Frohlich 2012).

Overall, the results from this trial are encouraging and suggest a potential role for TLR9-targeted immunoactivation with MGN1703 in maintenance therapy of metastatic CRC after an effective first-line regimen inducing a tumour response. It is important, however, that the current results are not over-interpreted because of the very limited sample size due to the early study termination. A randomised, phase III trial of MGN1703 in the first-line/maintenance setting in patients with metastatic CRC is planned.