Dijana Djureinovic, PhD, MSc

Anti-VEGF therapy enhances PD-1 inhibitor activity in pre-clinical models and has been used to treat peri-lesional cerebral edema and radiation necrosis. We conducted a two-institution phase 2 trial of bevacizumab and pembrolizumab in untreated melanoma brain metastases (MBM) patients (NCT02681549). Patients were anti-PD-(L)-1-naïve, had ≥1 asymptomatic, non-hemorrhagic 5-20mm MBM, not requiring immediate local therapy or steroids. Thirty-seven patients received four doses of bevacizumab and pembrolizumab every 3 weeks. The brain metastasis response rate (primary endpoint) was 54.1%. The extracranial response rate was 56.3%. Median intracranial progression-free survival was 2.2 years (95% CI 0.41-not reached). Median overall survival was 4.3 years (95% CI 1.6-not reached). Grade 3 treatment-related adverse event rates from bevacizumab and pembrolizumab were 10.8% and 18.9%, respectively. Higher pre-treatment vessel density in metastatic tumors and lower on-therapy increases in circulating angiopoietin-2 were associated with response to therapy. Pembrolizumab with bevacizumab was well-tolerated and demonstrated substantial activity in patients with untreated MBM with promising overall survival, justifying further evaluation of this regimen.

Stimulating innate immunity can potentially enable us to overcome resistance to PD-(L)1 blockade. We previously conducted a phase 1 trial of cabiralizumab (anti-CSF1R) with sotigalimab (CD40 agonistic antibody) and nivolumab. Our purpose was to determine safety and the effects of this regimen on circulating and tumor-infiltrating immune cells and to determine the activity of this regimen in a phase 1b trial for melanoma patients whose disease had progressed on anti-PD-(L)1. CyTOF analysis on circulating immune cells taken before and during treatment revealed a reduction in non-classical monocytes and an increase in dendritic cells. Patients with prolonged stable disease had less T-regulatory cells and more circulating antigen presenting cells after treatment compared to patients that were treated for a shorter time. In the phase 1b component of the trial in 13 melanoma patients, objective response rates were: 1 confirmed partial response (7.7%), 1 unconfirmed partial response (7.7%), 5 stable disease (38.5%) and 6 disease progression (42.6%). Despite therapy-induced changes in circulating immune cells and previous preclinical studies supporting rationale for this combination, responses in humans were insufficient to proceed to the second stage of the phase 1b trial. Given the challenges with translating doses from mice to humans, we proceeded to study various doses of anti-CSF1R in combination with CD40 agonist and anti-PD-1 in a murine model. Higher dose anti-CSF1R in mice was associated with increased tumor growth, worse survival and by single-cell RNA-sequencing analyses, we identified a more suppressive monocyte/macrophage profile in murine tumors. Our study suggests that more anti-CSF1R might not be better. Further optimization of cabiralizumab dosing is necessary to evaluate the clinical potential in combination with anti-PD-1 and anti-CD40 in a difficult-to treat patient population whose therapeutic options are limited.

Background: Cancer testis (CT) genes are expressed in various types of cancer but otherwise restricted to normal tissues of testis and placenta. Several CT genes have shown to encode immunogenic proteins that are able to induce an anti-tumor response in cancer patients. The presence of autoantibodies towards expressed CT proteins could indicate which CT proteins are more suitable for immunotherapeutic interventions, as these are recognized by the patient’s immune system. Methods: Suspension bead arrays (Luminex) were used to analyze the presence of autoantibodies towards expressed CT proteins in plasma samples from patients with non-small cell lung cancer (NSCLC). The technology enables to screen for autoantibodies in minute amount of patient plasma. Protein fragments with an average length of 80 amino acids, produced within the Human Protein Atlas, were coupled to unique beads, allowing multiplex analysis of 244 different autoantibodies towards antigens representing 198 unique genes in each sample. The primary sample set included 51 samples from 34 individuals taken before radiation therapy and 17 samples taken after radiation therapy. Longitudinal plasma samples taken during radiation therapy were available for most individuals resulting in a total of 89 samples. Results: Of 198 analyzed CT genes, autoantibodies against antigens representing 25 genes were detected in at least one of the 51 samples from the primary study set. The autoantibody detection ranged from five different autoantibodies in two individuals to no detected autoantibodies in seven individuals. Among those individuals with samples available both before and after radiation therapy (n¼13), the autoantibody profiles were not altered by the treatment. Three individuals however showed autoantibodies towards one additional protein in the sample taken after radiation therapy compared to the sample before radiation. In two individuals, autoantibodies detected towards one protein in the sample taken before radiation were not detected in the sample taken after radiation. Unsupervised hierarchical clustering with 25 detected autoantibodies and all 89 samples showed that samples from the same individual cluster based on the autoantibodies’ profile. There was no apparent association of autoantibody profiles with clinical parameters (histology, gender, age, stage). However, patients with detected autoantibodies showed a longer overall survival than patients without autoantibodies. Conclusions: This study provides a first comprehensive analysis of autoantibody detection against antigens representing 198 CT genes. Among the identified autoantibodies only AKAP4 has been reported previously in NSCLC. The individual autoantibody profiles showed only minor differences between samples taken before, during and after radiation therapy.