Encouraging Activity Seen With Glembatumumab Vedotin in Advanced Melanoma

Patrick Ott, MD, PhD

The antibody-drug conjugate (ADC) glembatumumab vedotin (GV) demonstrated an 11% overall response rate (ORR) in patients with refractory advanced melanoma who had received prior treatment with a checkpoint inhibitor and duration of response (DOR) in a phase II study presented at the 2017 ASCO Annual Meeting.

In addition to the 11% ORR (95% CI, 4.7-21.9), there was 1 complete response. Stable disease (SD) at 6 weeks or more was observed in 53% of patients; the disease control rate (SD or better for at least 3 months) was 52%. The median DOR was 6.0 months. Median PFS was 4.4 months, and median OS was 9.0 months, reported Patrick Ott, MD, PhD, clinical director, Melanoma Center, Dana-Farber Cancer Institute during the meeting.

GV contains a fully human IgG2 antibody to glycoprotein NMB (gpNMB), an internalizable, transmembrane glycoprotein overexpressed in multiple tumor types, including melanoma. High tumor gpNMB expression is associated with worse prognosis. The ADC delivers monomethyl auristatin E (MMAE), a potential antimitotic agent to gpNMB-expressing tumor cells. After internalization of the conjugate, the linker is cleaved, releasing free MMAE within those tumor cells, causing tubulin inhibition and cell death.

Prior clinical experience with glembatumumab in a phase I/II study in patients with unresectable stage III or IV melanoma and at most 1 prior therapy that was conducted prior to the availability of checkpoint inhibitors, showed a 13% ORR. In those earlier studies, the most significant treatment-related toxicities included rash, neutropenia, and neuropathy. Development of rash, particularly in the first cycle, was associated with greater progression-free survival (PFS) and ORR.

This single-arm, open-label, phase II study (CDX011-05; NCT02302339) assessed the efficacy and safety of GV monotherapy in patients with measurable, unresectable stage III or IV melanoma who had received ≤1 cytotoxic chemotherapy, ≥1 checkpoint inhibitor (CPI), and if the tumor had a BRAF V600 mutation, ≥1 BRAF/MEK inhibitor.

Patients received 1.9 mg/kg GV every 3 weeks via a 90-minute intravenous infusion until disease progression or unacceptable toxicity. Tumor assessments were conducted every 6 weeks for 6 months, then every 9 months. The primary endpoint was ORR assessed by RECIST 1.1. More than 6 responses in 52 patients defined the threshold for a positive study. Secondary endpoints included PFS, overall survival (OS), DOR, safety, and correlation of tumor gpNMB expression with outcome. A centralized laboratory determined gpNMB expression using immunohistochemistry (IHC) in archival and/or pre-treatment tumor samples.

Of the 62 patients enrolled, 55% were male, with a median age of 67 years. All patients had stage IV melanoma, and 89% had M1c disease, which is associated with a worse prognosis. The duration of advanced and/or metastatic disease was 25.3 months (range 3.1 to 318.8 months). Patients had a median of 3 prior regimens, including CPI in 100%, anti—CTLA-4 and anti–PD-1/PD-L1 each in 94%, and BRAF or BRAF/MEK inhibitors in 24% (all patients with BRAF mutations).

Most patients experienced treatment-related adverse events (AEs), including neuropathy, alopecia, rash, fatigue, neutropenia, nausea, decreased appetite, pruritus, and diarrhea. The most frequent grade 3/4 toxicity was neutropenia (19%). Other ≥3 treatment-related AEs included rash (8%), neuropathy (6%), fatigue (3%), pruritus (2%), and diarrhea (2%). The only treatment-related discontinuations occurred in 2 patients due to grade 2 sensory neuropathy. One patient died of pneumonia that was considered possibly treatment-related.

Treatment-related rash in cycle 1 was associated with improved ORR (rash 21%; no rash 7%), PFS (median 5.5 months if rash in cycle 1, vs 4.0 months if no rash in cycle 1; P = .006), and OS (median 15.8 months vs 6.6 months; P = .026).

All patients with available tissue (n = 59) had gpNMB-positive tumors; 78% had 100% gpNMB-positive epithelial cells. There was no clear correlation of gpNMB expression with outcome in this population due to the high levels of expression. For 10 patients with available matched primary and metastatic tumors, gpNMB expression was consistent between samples.

Ott concluded that GV showed encouraging antitumor activity in patients with poor-prognosis, CPI-refractory, advanced melanoma. The primary study endpoint of ORR was met, with a promising ORR and DOR. The majority of patients experienced disease control, defined as SD for at least 3 months, with a tolerable safety profile and no new safety signals compared with prior studies. High treatment-related rash in cycle 1 may be a predictive marker of response as seen in previous studies. The high gpNMB expression in this population does not allow correlation between expression and outcome.

Evidence that GV has immunomodulatory properties as well as cytotoxic activity forms the rationale for ongoing studies evaluating GV in additional cohorts in combination with the costimulatory CD27 agonist antibody varlilumab, or in combination with PD-1 inhibitors.

Anthony W. Tolcher, MD, South Texas Accelerated Research Therapeutics, San Antonio, Texas, commented on the study, saying that the response rate is borderline for any new therapy given only 1 complete response. He said one has to assume there is intrinsic resistance of melanoma to anti-tubulin agents based on old studies as well as the current one. The lack of correlation with gpNMB expression makes it impossible to select patients who might best respond to treatment. The side effect profile raises the question of linker stability. If the rash is an on-target effect, it may be possible to use masking technology. However, if the rash is off-target, it presents a serious issue, and may prevent dose escalation. Tolcher believes that although the study was well-executed, the payload or toxin linked to the antibody is not appropriate for the indication.

Ott responded, “Melanoma is a different disease. It is known to be chemoresistant, so matching the payload to the target disease is a little bit harder if the payload is a cytotoxic agent. It is an intriguing thought to have different payloads that are not cytotoxic. I am looking at this from the angle of an immunotherapist,” he continued, pointing out that there is good preclinical evidence for the combination of GV with immunotherapies, providing the rationale for ongoing trials. Acknowledging the low response rate in the current study, Ott said, “We were excited about the activity, which is better captured by the clinical benefit.” He said patients experienced quality of life benefits not reflected in the response rate.

Ott PA, Pavlick AC, Johnson DB, et al. A phase II study of glembatumumab vedotin (GV), an antibody-drug conjugate (ADC) targeting gpNMB, in advanced melanoma. [ASCO abstract 109]. J Clin Oncol. 2017;35(suppl; abstr 109).