Phase 1b/2a study of galunisertib, a small molecule inhibitor of transforming growth factor-beta receptor I, in combination with standard temozolomide-based radiochemotherapy in patients with newly diagnosed malignant glioma

Glioblastoma (GB) is the most common and aggressive brain cancer representing approximately 15% of all primary brain tumors, and about 55% of all gliomas [1]. The standard of care for patients with GB consists of maximal surgical resection followed by radiotherapy (RTX) with concomitant and maintenance temozolomide (TMZ). This therapy results in progression-free survival (PFS) at 6 months of 53.9% and median overall survival of 14.6 months [2]. Most of the therapy-responsive patients will die within a period of 2 years and the 2- and 5-year overall survival rates are 27% and 9.8%, respectively [3].

GB is characterized by persistent angiogenesis at the tumor site, and decreased peripheral immune responsiveness in patients [4]. GB microenvironment is enriched in immunosuppressive molecules such as transforming growth factor (TGF)-β that plays a specific role in cancer cell growth [5], in addition to affecting immune cell response, and endothelial cell and fibroblast differentiation [6, 7].

TGF-β is a multifunctional cytokine that is involved in a variety of cell functions including cell proliferation, migration, survival, and death that influence tumor growth in advanced forms of cancer [6, 7]. Upon binding to their ligands (TGF-β1, 2, and 3), the TGF-β kinase receptors are phosphorylated triggering phosphorylation of SMAD2 and SMAD3, and formation of SMAD complexes [8, 9].

Galunisertib is an oral small molecule inhibitor of TGF-β kinase receptor type I (TGF-β RI/ALK5) [10] and selectively inhibits the serine/threonine activity of the receptor, thereby preventing the phosphorylation of downstream proteins, SMAD2 and SMAD3 [10]. The antitumor activity of galunisertib has been demonstrated in three different in vivo tumor models; two breast cancer models, MX1 and 4 T1; and a non-small cell lung cancer model, Calu6 and fibrosis [11‐13].

Based on the role of TGF-β in patients with malignant GB, evidence of antitumor effects of TGF-β inhibitors such as galunisertib (including in a monotherapy study in glioblastoma), and a favorable short- and long-term toxicity profile [14, 15], a multicenter Phase 1b/2a clinical trial was initiated to investigate the clinical benefit of combining galunisertib with standard TMZ-based radiochemotherapy (TMZ/RTX) in patients with newly diagnosed malignant glioma.

Patient disposition, demographic, safety, drug-related treatment-emergent adverse events (TEAEs), and response data were summarized using patient number, frequency counts, or percentages as appropriate. The safety analysis was based on summaries of AEs reported in Common Terminology Criteria for Adverse Events (CTCAE) version 4.0, and possibly drug related.

All time-to-event variables were analyzed using the Kaplan-Meier method with 90% confidence interval (CI). Univariate Cox models were used to evaluate results for potential prognostic markers by considering their impact on OS and PFS. Continuous markers were first converted to 2-level categorical variables by dichotomizing at the median and hazard ratios between treatment arms estimated for each level.

The absolute cell number of Treg cells, CD4+, and CD8+ T cells from each patient were presented in profile plots together with the geometric mean and 90% CI at baseline, Day 42 and Day 182 for each cell type. Pair-wise t-Tests were used to estimate the change from baseline to Day 42 for T cell subsets in each arm.

We here report the efficacy, safety, and PD of galunisertib combined with radiochemotherapy in newly diagnosed malignant glioma. The overall toxicity and PK results from the Phase 1b study (Online Resource: Supplemental Results) were used to determine the recommended Phase 2a dose. Additionally, PK studies done during Phase 2a showed that the plasma levels of galunisertib were not altered when combined with TMZ and radiation (Online Resource: Supplemental Fig. 4) and achieved the targeted biologically effective dose level.

While both treatments showed comparable results for median OS (18.2 vs 17.9 months), the galunisertib plus radiochemotherapy group had a shorter estimated PFS than the radiochemotherapy group (7.6 vs 11.5 months). This difference might be explained by the small number of patients in both arms or the earlier withdrawal of patients from galunisertib plus radiochemotherapy arm. For example, 55% of patients from the experimental arm were moved to subsequent therapies, versus 37.5% of patients from control received other therapies (Fig. 1).

The overall safety data across treatment arms was similar; however, the frequency of grade 3–4 toxicities was higher in the galunisertib plus radiochemotherapy arm. There was a severe case of myeloablative marrow aplasia during the first cycle of treatment in a Phase 1a patient; this finding is more likely related to the known side effect of TMZ and radiation than to galunisertib treatment [20]. Galunisertib was not associated with bone marrow side effects in preclinical toxicology studies evaluating galunisertib in human bone marrow assays or in other combination studies with chemotherapeutic agents [14].

In addition, because cardiovascular toxicities are associated with small molecule inhibitors of TGF-β signaling in preclinical toxicology studies [21], cardiac toxicity was monitored in all patients. Galunisertib treatment did not show any clinically significant cardiac safety concerns, which are consistent with previous reports for a TGF-β small molecule inhibitor [22].

Biomarker studies did not find any correlation between baseline T cell subsets (including Tregs) and OS or PFS (Online Resource: Supplemental Fig. 3A and 3B). As reported by others, the CD4+, CD8+, and Treg cells count at 10 weeks post radiochemotherapy treatment were numerically decreased [23, 24]. The longitudinal analysis shows early decrease of CD4+ and CD8+ T cell counts during radiation, followed by a steady phase or a slight recovery in these cells over time within the galunisertib plus radiochemotherapy arm, while the pattern of CD4+, CD8+, and Treg cell counts were steady over time in the radiochemotherapy arm during both radiation and post-radiation phases. In other diseases, such as lung cancer, transient decreases in CD8+ T cells followed by an increase is associated with better OS [25]. Hence, the relevance of our observation needs further examination in order to decide whether such a response would be also expected in GB patients.

An exploratory analysis was performed to examine whether any two clusters of patients with respect to OS and PFS emerged after 200 days of galunisertib plus radiochemotherapy treatment (Fig. 3a). This limited analysis set showed that the 5 patients with a low number of CD8+ T cells had a mean OS of 25.5 months and mean PFS of 14.4 months, and the 11 patients with a higher number of CD8+ cells had a mean OS of 19.4 and mean PFS of 13.1 months. However, these observations need further confirmation in larger cohorts of patients.

Furthermore, we found no association between OS and MDC/CCL22 contrary to reports for second line treatment of GB patients [26]. It is possible that baseline levels of MDC/CCL22 are different between first and second line patients. Additionally, we found no association between pSMAD2 levels in tumor tissue and OS. The presence of CD3+ T cells in tumor tissue was not associated with OS changes. These findings are different from those reported for the second line patients treated with galunisertib [17].

TGF-β is a major driver of glioma progression, via its role in tumor cell proliferation and invasion, angiogenesis, and immune suppression within the tumor microenvironment [27]. Blocking TGF-β signaling by inhibition of its receptor TGF-β RI is one strategy for abrogating its pro-tumorigenic effects. Galunisertib is one of the only drugs in development designed to specifically target TGF-β RI; it is furthest along the clinical trial pipeline [28], not only in the setting of recurrent glioma/GB (in combination with TMZ/RTX in the current study and in combination with lomustine in [15, 26]), but also for other solid tumors. Galunisertib is currently being evaluated in Phase 1/2 or Phase 2 studies in combination with immune checkpoint inhibitors [29, 30], sorafenib [31], and gemcitabine [32]. In a Phase 2 study in patients with pancreatic cancer, the first line treatment of galunisertib in combination with gemcitabine resulted in an OS benefit of 8.9 months compared to 7.1 months for patients receiving gemcitabine alone (hazard ratio [HR], 0.79; 95% CI, 0.59–1.09) [32]. In HCC patients with elevated alpha-fetoprotein prior to treatment and who had previously progressed on sorafenib or were considered not eligible to receive sorafenib, galunisertib monotherapy achieved a median OS of 7.3 months (95% CI, 4.9–10.5). OS was longer for those patients who showed reduced alpha-fetoprotein (>20% from baseline) compared to non-responders (21.5 months vs 6.8 months) [33]. While these signals were encouraging, the sponsor discontinued future clinical development for galunisertib in mid-2017 [34]. Other TGF-β RI inhibitory drugs in early clinical development (Phase 1) include LY3200882 [35] and vactosertib [36], however there are no published reports of efficacy of these compounds as of yet. In contrast to these small molecule approaches, large molecule development has had advances, including M7824 (bintrafusp alfa), which is currently being evaluated in registration studies, including for NCSLC [37].

In conclusion, the combination of galunisertib with standard radiochemotherapy did not accentuate the toxicity profile of the radiochemotherapy. Even though survival probability was unchanged between the two treatments and PFS was reduced in the galunisertib plus radiochemotherapy arm, the disease control rate was higher in the galunisertib plus radiochemotherapy treatment when compared to radiochemotherapy treatment alone. Due to new R&D priorities, Eli Lilly discontinued the development of galunisertib in 2017 [34].