PHILADELPHIA—An antibody against PD-L1 on tumor cells has shown good activity against malignant melanoma and was well tolerated in a Phase I trial, achieving an objective response rate of about 30 percent, according to data reported here at the 10th International Meeting of the Society for Melanoma Research. Patients whose tumors expressed more PD-L1 had a higher rate of disease control than did patients with less tumor PD-L1.
In his presentation, which was an update of findings reported at the American Society of Clinical Oncology Annual Meeting (OT 6/25/13 issue), Jeffrey Sosman, MD, Professor of Medicine and Director of the Melanoma and Tumor Immunotherapy Program at Vanderbilt-Ingram Cancer Center, explained that PD-L1 on tumor cells and in the tumor microenvironment interacts with PD-1 on T cells and allows cancer cells to evade T cell-mediated killing. MPDL3280A (made by Roche) is an antibody engineered to block the PD-L1/PD-1 interaction but to avoid killing activated T cells.
There is theoretical reason for an advantage of blocking PD-L1 as opposed to blocking PD-1, he noted: “In that setting PD-L2 is not blocked, and there's some evidence that PD-L2 expression on lung dendritic cells and macrophages may be important in immune homeostasis.”
For this dose-escalation trial, patients had incurable or metastatic disease measurable by RECIST 1.1 criteria, ECOG performance status of 0 or 1, and no prior treatment with an immune checkpoint inhibitor, including ipilimumab. MPDL3280A was given intravenously every three weeks for 16 cycles with periodic computed tomography scanning to assess response by RECIST 1.1 and immune-related response criteria. Thereafter, patients with a complete or partial response or stable disease were followed every 12 weeks until disease progression.
MPDL3280A doses were escalated from 0.01 to 20 mg/kg, and the expansion-phase cohort included 43 patients on doses of 1 to 20 mg/kg. Patients had a median age of 63, two-thirds were male; 78 percent had cutaneous melanoma; 11 percent had mucosal lesions; nine percent had ocular melanoma; and two percent had an unknown type or the location was unavailable.
Of the 35 with cutaneous melanoma, 25 presented with stage M1c, and five each with M1a or M1b. A BRAF mutation was detected in 11.
Good Efficacy, Excellent Safety
Among the 43 patients assessed, 12 (28%) had a confirmed RECIST 1.1 objective response in this single-agent trial, Sosman reported. These results are roughly consistent with the estimated 40 percent prevalence of PD-L1 in melanoma samples (not within the clinical trial). Twenty-four (56%) of the patients had a complete or partial response or stable disease.
For the 39 patients with non-ocular melanoma, the objective response rate was 31 percent, but none of the four ocular melanoma patients' tumors responded. Response rates were independent of BRAF status, but patients who had had prior immunotherapy had about double the response rate of patients without prior immunotherapy (44% vs. 19%).
“For actual RECIST response, whether you had a low expression or a higher expression of PD-L1 did not affect your response rate,” Sosman said. Combined complete and partial responses occurred in six of 20 patients (30%) with PD-L1 immunohistochemistry (IHC) 2-3 and in five of 19 (26%) patients with IHC 0-1.
“On the other hand, there was a very significant difference in terms of stable disease, and the overall number, because of the very big difference in stable disease, favored those with higher PD-L1 IHC scores,” he said. Of the patients with IHC 2-3 tumors, 10 of 20 had stable disease, but there were no patients with stable disease in the IHC 0-1 group.
Tumors shrank by about 35 to 100 percent—i.e., the sum of the lesion diameters—where there were complete or partial responses, Sosman reported. For progressive disease, lesions increased by up to 60 percent, but a few shrank by as much as 25 percent.
One partial response occurred at six weeks of treatment, but most occurred between 12 and 24 weeks. The first complete response was at 36 weeks. Only one patient of the 12 had progressive disease, and that was between 24 and 27 weeks of treatment.
Increased T Cell Activation
Sosman showed data indicating that PD-L1-positive tumors that respond to treatment have increased T cell activation based on IHC and on gene expression of T cell markers. Possible mechanisms of action of response to MPDL3280A are pre-existing intratumoral CD8+ T cells, increased trafficking or proliferation of such cells, and increased T cell activation and cytotoxicity, as indicated by upregulation of gene expression of cytotoxic compounds—for example, granzymes and perforin.
Patients tolerated treatment with MPDL3280A well, he said. “The toxicity was really quite mild, the majority being grades 1 and 2.”
No maximum tolerated dose or dose-limiting toxicities were identified, and most adverse events were grade 1 or 2 and did not require intervention, he said. The most common adverse event was fatigue, occurring in 40 percent of patients. No pneumonitis, colitis, or treatment-related deaths were seen. Most of the grade 3-4 toxicities were laboratory abnormalities.
Summing up, Sosman said that MPDL3280A treatment in this trial was associated with a 31 percent confirmed objective response rate in non-ocular melanoma patients and that responses are continuing in all but one of the 12 responders. The responses may be rapid or delayed, a higher objective response occurred among patients who received prior immunotherapy, and the drug is well tolerated.