Atezolizumab Plus Bevacizumab and Chemotherapy Misses PFS End Point in Ovarian Cancer Trial


In patients with newly diagnosed stage III/IV ovarian cancer, adding atezolizumab to a backbone combination of bevacizumab and chemotherapy did not improve progression-free survival, missing the primary end point of the phase 3 IMagyn050/GOG 3015/ENGOT-OV39 trial presented during the virtual 2020 European Society of Medical Oncology Congress.

Kathleen N. Moore, MD

In patients with newly diagnosed stage III/IV ovarian cancer, adding atezolizumab (Tecentriq) to a backbone combination of bevacizumab (Avastin) and chemotherapy did not improve progression-free survival, missing the primary end point of the phase 3 IMagyn050/GOG 3015/ENGOT-OV39 trial presented during the virtual 2020 European Society of Medical Oncology (ESMO) Congress.

Results showed that the PFS in the intent-to-treat population (ITT; n = 1301) was 19.5 months with the atezolizumab combination versus 18.4 months with bevacizumab/chemotherapy (HR, 0.92; 95% CI, 0.79-1.07; P = .2785). In the PD-L1–positive population (n = 784), the median PFS with the atezolizumab regimen was 20.8 months versus 18.5 months with bevacizumab plus chemotherapy (HR, 0.80; 95% CI, 0.65-0.99; P = .0376).

However, results from exploratory PFS analyses demonstrated a trend favoring atezolizumab in a subgroup of patients with PD-L1 immune cells of 5% or greater, which could warrant further evaluation. Here, the median PFS with the atezolizumab regimen was not been reached versus 20.2 months in the bevacizumab/chemotherapy arm (HR, 0.64; 95% CI, 0.43-0.96).

“Despite notable successes with the incorporation of atezolizumab and bevacizumab into the treatment of other solid tumors, IMagyn050, the first such study done in ovarian cancer, did not meet its first primary end point for extending PFS in either the ITT or the fully powered PD-L1–positive population,” said Kathleen Moore, MD, lead author of the study, as well as the director of the Oklahoma TSET Phase I Program and associate professor in the Section of Gynecologic Oncology at Stephenson Cancer Center, said in a presentation during the meeting. “The signal of clinical benefit for atezolizumab in the immune cell PD-L1–high subgroup may warrant further evaluation.”

The PD-L1 inhibitor atezolizumab has proven efficacy in several cancers, and immune cell PD-L1 expression has been linked with even stronger activity in select tumors, noted Moore. To date, the standard of care for the frontline treatment of patients with ovarian cancer is platinum/taxane chemotherapy in combination with bevacizumab. Blocking VEGF has been shown to increase cytotoxic T-cell trafficking into several solid tumor types, thus justifying the combination of the antiangiogenic agent bevacizumab with atezolizumab, added Moore.

“The pronounced antitumor effects of combining antiangiogenic and anti–PD-1 or anti­–PD-L1 therapies have been demonstrated in metastatic non–small cell lung cancer, unresectable hepatocellular carcinoma, and advanced endometrial cancer,” said Moore.

Investigators launched IMagyn050 to determine whether that benefit would extend to patients with ovarian cancer. Specifically, the objective of the trial was to examine whether the incorporation of the PD-L1 inhibitor atezolizumab into a frontline bevacizumab-containing chemotherapy regimen would lead to significant clinical efficacy with tolerable safety.

The trial included patients with previously untreated epithelial ovarian, primary peritoneal, or fallopian tube cancer, either post primary cytoreduction or who were scheduled for neoadjuvant chemotherapy followed by interval cytoreductive surgery. Patients had to have an ECOG performance status from 0 to 2.

In the trial, participants were randomized 1:1 to receive paclitaxel, carboplatin area under the curve (AUC)6, bevacizumab, and atezolizumab for 6 cycles followed by bevacizumab and atezolizumab for an additional 16 cycles, or paclitaxel, carboplatin, bevacizumab, and placebo for 6 cycles.

Randomization was stratified by disease stage (III vs IV), performance status (0 vs 1/2), treatment approach (adjuvant vs neoadjuvant), and PD-L1 status (immune cells of less than 1% vs 1% or higher via the VENTANA SP142 assay).

The co-primary end points were PFS tested simultaneously in the PD-L1–positive and ITT populations, and OS, which would use hierarchical testing, examining the PD-L1 population first followed by the ITT population. The proportion of patients scheduled for neoadjuvant chemotherapy was capped at 20% of the enrolled population, added Moore.

At the time of the data cutoff, which was March 30, 2020, a total of 1301 patients had undergone randomization; 650 were enrolled to the control arm and 651 were enrolled to the experimental arm. A total of 645 patients in the control arm received treatment, and 641 patients on the experimental arm received treatment. Forty patients in the control arm remained on active treatment versus 44 patients in the experimental arm. The median duration of follow-up was 19.9 months.

The neoadjuvant population of the trial represented 25% of the participants on the trial. Moreover, 60% of patients’ tumors were classified as PD-L1–positive, with immune cells of 1% or higher. Thirty-one percent of participants had stage IV disease. When examining the distribution of histologies, approximately three-quarters of patients had high-grade serous histology, noted Moore.

Additional results from the trial showed that the Kaplan–Meier curves for the interim analysis of the median overall survival (OS) in the ITT and PD-L1 populations, reported at 41% and 37% maturity, respectively, did not appear to be significantly different. In the ITT population, the median OS had not been reached in either of the treatment arms (HR, 0.96; 95% CI, 0.74-1.26; P = .7887). The 2-year event-free survival (EFS) rate was 80.6% in the investigational arm and 79.4% in the control arm.

In the PD-L1–positive population, the median OS was not reached in the investigational arm versus 31.2 months in the control arm (HR, 0.98; 95% CI, 0.68-1.41; P = .9083). The 2-year EFS rates were 82.1% versus 82.5%, respectively.

“When looking at subgroup analyses for PFS by baseline characteristics, there is a trend for improvement in PFS with atezolizumab in those with stage III disease and the opposite effect in those with stage IV disease,” said Moore. “These and other analyses are unadjusted for other important co-variates.”

When examining subgroup analyses of PFS by histology, results revealed a trend for improvement in PFS with atezolizumab among tumors classified as clear cell, noted Moore. However, she warned that this analysis was limited in that a very small sample size was used and there were wide confidence limits.

Investigators utilized the VENTANA SP142 assay to determine PD-L1 status. PD-L1 positivity in the primary analysis was defined as at least 1% staining of immune cells per immunohistochemistry as a function of the tumor area, according to Moore. Predefined exploratory analyses for efficacy were conducted on subgroups that were defined as the proportion of positive immune cells in the presence or absence of tumor staining.

Subgroup analyses of PFS by PD-L1 status showed an improvement in PFS that favored the atezolizumab arm for those with tumors that had at least 5% of PD-L1 positive immune cells (HR, 0.64; 95% CI, 0.43-0.96). The median PFS for this subgroup was not estimable in the atezolizumab arm versus 20.2 months in the control arm. This subset represents 20% of the overall study population.

An improvement in PFS favoring the atezolizumab arm was also observed for patients whose tumors had at least 1% of PD-L1–positive tumor cells (HR, 0.41; 95% CI, 0.19-0.90). The median PFS was not estimable in the atezolizumab arm versus 15.0 months in the control arm. However, this subset represents only 6% of the total study population, noted Moore.

With regard to safety, the 2 arms were found to be fairly well balanced, according to Moore. However, more serious adverse effects (AEs) were reported in the atezolizumab arm versus the control arm, at 47% versus 33%, respectively. Grade 3/4 AEs were reported in 79% versus 73% of patients, respectively.

Slightly more patients in the atezolizumab arm experienced AEs that resulted in treatment discontinuation compared with the control arm, at 26% versus 22%, respectively. Additionally, more AEs of special interest were observed in the atezolizumab arm versus the control arm, at 73% versus 52%, respectively.

“AEs of fatal outcomes and grade 3/4 AEs were comparable between the arms,” said Moore. “The incorporation of atezolizumab did not appear to compromise the delivery of the backbone chemotherapy and the frequencies of AEs for the atezolizumab arm are consistent with those previously reported.”

The most frequently reported AEs in the investigational and control arms included alopecia (60.0% vs 63.7%), nausea (50.5% vs 52.5%), anemia (44.4% vs 41.8%), arthralgia (41.4% vs 41.5%), fatigue (37.9% vs 39.0%), and hypertension (35.0% vs 41.0%), among others. Severe AEs that had an incidence of 2% or higher in either arm included febrile neutropenia (8.4% with atezolizumab vs 3.7% with the control) and pyrexia (4.0% vs 1.2%, respectively).

Additionally, AEs of special interest in the atezolizumab and the control arms included rash (41.3% vs 25.6%, respectively), hypothyroidism (25.9% vs 12.9%), hepatitis (2.6% vs 2.2%), pneumonitis (1.9% vs 0.6%), hyperthyroidism (7.9% vs 3.6%), adrenal insufficiency (0.8% vs 0.3%), infusion-related reactions (12.1% vs 7.6%), colitis (3.3% vs 1.7%), severe cutaneous reactions (2.3% vs 0.5%), myositis (0.6% vs 0.8%), meningoencephalitis (0.5% vs 0.5%), and pancreatitis (0.8% vs 0%).

“Looking closer at AEs of special interest for atezolizumab, we see a greater incidence of thyroid abnormalities in those women randomized to the PD-L1 inhibitor and more rash; this does include more serious or severe cutaneous reactions,” said Moore. “You also would see colitis, pancreatitis, and infusion-related reactions that occurred relatively infrequently but were higher in the atezolizumab group. As such, while imbalances were observed across several AEs of special interest, this does appear to be consistent with what we have seen for this combination in other solid tumors.”


Moore K, Bookman M, Sehouli J, et al. Primary results from IMagynp050/GOG 3015/ENGOT-OV39, a double-blind, placebo-controlled randomized phase 3 trial of bevacizumab-containing therapy +/- atezolizumab for newly diagnosed stage III/IV ovarian cancer. Presented at: 2020 ESMO Virtual Congress; September 19-21, 2020; Virtual. Abstract LBA31.

Related Videos
Related Content