Angiogenesis inhibition

In the ECOG 4599, a total of 878 patients with recurrent or advanced NSCLC were randomized to receive first-line treatment with carboplatin and paclitaxel, or carboplatin and paclitaxel plus bevacizumab. Patients with squamous histology were excluded based on previous phase 2 data where squamous patients suffered higher rates of serious hemorrhagic events, including fatal pulmonary hemorrhage. Platinum chemotherapy was given for six cycles, and patients in the bevacizumab arm continued to receive bevacizumab thereafter until disease progression or toxicity. Response rate, progression-free survival (PFS), and OS were all significantly improved with the addition of bevacizumab to platinum chemotherapy (relative risk, 35% vs 15%, P <.001; PFS, 6.2 months vs 4.5 months, P <.001; OS, 12.3 months vs 10.3 months, P = .003). Class effects of angiogenesis inhibition, such as hypertension, proteinuria, and bleeding, were all significantly higher in the bevacizumab-containing arm. In addition, rates of neutropenia, febrile neutropenia, and thrombocytopenia were also higher in the bevacizumab-containing arm.

Key exclusions in the ECOG 4599 study included squamous histology, history of hemoptysis, and the presence of brain metastases. Subsequent to the positive results from ECOG 4599, multiple studies were undertaken to better define a “bevacizumab-eligible” population. BRIDGE was an open-label phase 2 study for patients with squamous lung cancer. Baseline risk factors for pulmonary hemorrhage, such as tumor cavitation or tumor involvement of major blood vessels, and history of hemoptysis or a bleeding diasthesis, were exclusionary. Patients with squamous histology were treated initially with carboplatin/paclitaxel for two cycles, with bevacizumab added with the third cycle. The primary end point of the trial was the incidence of grade 3 or higher pulmonary hemorrhage. One of 31 patients (3.2%) who received bevacizumab had a grade 3 to 4 pulmonary hemorrhage; there was one additional grade 1 pulmonary hemorrhage that occurred on study. A European study in which patients with squamous NSCLC were treated with bevacizumab from cycle 2, after a 3-week course of radiation treatment before start of chemotherapy, was terminated early after 2 of the first 20 patients had grade 3 to 4 pulmonary hemorrhage. Overall, given the potentially catastrophic consequences of pulmonary hemorrhage, treatment of squamous patients with bevacizumab is considered contradindicated and has not been adopted by the oncology community.

Other exclusions to “bevacizumab-eligibility” have been loosened since the original ECOG 4599 study as more data support safety in these populations. PASSPORT was an open-label phase 2 study of bevacizumab in combination with chemotherapy among patients with nonsquamous NSCLC with treated brain metastases. The primary end point was the incidence of symptomatic grade 2 or higher central nervous system hemorrhage. Patients received bevacizumab in conjunction with first- or second-line therapy, and multiple different chemotherapy regimens including platinum combinations or single-agent therapies were allowed. A total of 115 patients were enrolled in this study, among whom 80% had received whole-brain radiotherapy, 19.1% had received radiosurgery alone, and 0.9% had neurosurgery alone. There were no reports of grade 1 to 5 central nervous system hemorrhage, supporting the safety of bevacizumab in treated brain metastases. Indeed, it may be that bevacizumab could be considered even in untreated small asymptomatic brain metastases. BRAIN was a phase 2 study that enrolled patients with nonsquamous NSCLC with untreated asymptomatic brain metastases. Patients were treated with carboplatin and paclitaxel with bevacizumab in the first-line setting or erlotinib with bevacizumab in the second-line setting. Among 91 patients enrolled, there was one grade 1 intracranial hemorrhage observed, suggesting that even in small asymptomatic untreated brain mets, treatment with bevacizumab could potentially be considered.

Specific subsets of patients may accrue more benefit from the addition of bevacizumab to their treatment regimen. In a retrospective analysis of elderly patients in ECOG 4599, with elderly defined as age greater than 70, the addition of bevacizumab to chemotherapy was associated with a higher response rate and PFS, but OS between the two arms was not different (median OS, 11.3 months for the bevacizumab-containing arm, 12.1 months for chemotherapy alone; P = .4). Moreover, in the elderly population, the addition of bevacizumab was associated with a significantly higher toxicity rate, with grade 3 to 5 toxicities occurring in 87% of those receiving bevacizumab as compared with 61% of those receiving chemotherapy alone. These results suggest that caution should be applied when considering bevacizumab in elderly patients, because side effects and toxicities may outweigh potential benefits. Finally, an analysis by sex showed that although the improvements in relative risk and PFS were seen in men and women in ECOG 4599, the improvement in OS was seen in men only. Median OS in men was significantly better with bevacizumab compared with chemotherapy alone (11.7 months vs 8.7 months), whereas women had comparable OS regardless of treatment arm (13.1 months with chemotherapy alone, 13.3 months with chemotherapy plus bevacizumab; P = .87).

Other platinum doublets besides carboplatin and paclitaxel have been combined with bevacizumab, with varying results. AVAIL was a randomized phase 3 trial investigating cisplatin and gemcitabine as first-line chemotherapy with bevacizumab at 7.5 mg/m ² , bevacizumab at 15 mg/m ² , or placebo. The study was originally designed with a primary end point of OS, but after ECOG 4599 was reported, the protocol was amended to a primary end point of PFS. The study was designed to allow comparison of the bevacizumab-containing arms with placebo, but was not powered to compare the two doses of bevacizumab. A total of 1043 patients were randomized to the three arms. Response rates were higher for the bevacizumab-containing arms as compared with placebo, and PFS was significantly prolonged with bevacizumab. Median survival, which was greater than 13 months in all groups, was not significantly different with the addition of bevacizumab. The authors hypothesized that the high degree of subsequent therapies and favorable prognostic factors (younger age, higher proportion of dry stage IIIB, high proportions of adenocarcinoma and never-smokers) may have influenced the OS results.

POINTBREAK was a phase 3 study in first-line stage IIIB or IV nonsquamous NSCLC where patients were randomized to treatment with carboplatin, paclitaxel, and bevacizumab followed by bevacizumab, versus carboplatin, pemetrexed, and bevacizumab followed by pemetrexed and bevacizumab. The study was designed with a primary end point of OS to demonstrate superiority of the pemetrexed-containing arm, bearing in mind that pemetrexed had shown particular activity in the nonsquamous NSCLC population. However, OS was not improved with pemetrexed/carboplatin/bevacizumab as compared with paclitaxel/carboplatin/bevacizumab. Response rates were similar (34.1% vs 33.0%) between the two arms. PFS was improved in the pemetrexed-containing arm (HR, 0.83; median PFS, 6.0 months vs 5.6 months; P = .012), but OS, the primary end point, was not different. The toxicity profile differed between the two regimens, with more anemia, thrombocytopenia, and fatigue seen in the pemetrexed-containing arm and more neutropenia, febrile neutropenia, sensory neuropathy, and alopecia in the paclitaxel-containing arm. In a prespecified exploratory analysis of patients who reached the maintenance portion of the trial, median PFS and OS favored the patients who were in the pemetrexed-containing arm rather than paclitaxel (ie, those patients receiving pemetrexed/bevacizumab during maintenance rather than bevacizumab alone): median OS was 17.7 months versus 15.7 months, and median PFS 8.6 months versus 6.9 months. However, because of the design of this study, POINTBREAK cannot directly answer the question of which may be the better maintenance regimen. Given the lack of an OS difference in the study as a whole, it also does not establish carboplatin/pemetrexed as a superior regimen to combine with bevacizumab as compared with carboplatin/paclitaxel.

Overall, these studies show that in a selected group of patients, the addition of bevacizumab to platinum-based first-line chemotherapy has benefit. The trials suggest carboplatin and paclitaxel and carboplatin and pemetrexed are both reasonable regimens to combine with bevacizumab. Although the lack of an OS benefit in AVAIL may be caused by prognostic factors related to the population rather than chemotherapy regimen itself, the net effect has been that platinum/gemcitabine is a potential, but not necessarily favored, choice in combination with bevacizumab. The studies leave open the question of optimal maintenance strategy because this question was not directly tested in any of these studies.

Epidermal growth factor receptor inhibition

A series of studies have evaluated the addition of EGFR monoclonal antibodies to chemotherapy in the first-line setting. Two large studies, FLEX and BMS099, investigated the addition of the EGFR antibody cetuxiumab to chemotherapy. BMS099 was a phase 3 trial that enrolled 676 patients for treatment with carboplatin and taxane versus carboplatin, taxane, and cetuximab. Patients were enrolled regardless of histologic subtype and EGFR expression. The primary end point was PFS. Although response rate was higher with the addition of cetuximab (25.7% vs 17.2%; P = .007), there was no significant difference in PFS (HR, 0.902; 95% CI, 0.761–1.069; P = .236). Similarly, OS was not significantly different between the two arms (median OS, 9.69 months with cetuximab vs 8.38 months without; HR, 0.89; 95% CI, 0.754–1.051; P = .169). Biomarker analysis did not show any specific biomarker (KRAS or EGFR mutation, EGFR immunohistochemistry or fluorescence in situ hybridization) that seemed to predict for efficacy. FLEX was a randomized phase 3 trial where patients with advanced NSCLC whose tumors expressed EGFR were treated with chemotherapy plus cetuximab or chemotherapy alone. The trial included all histologies and PS 0 to 2. A total of 1125 patients were treated with cisplatin and vinorelbine versus cisplatin and vinorelbine with cetuximab, which was continued after chemotherapy until disease progression or unacceptable toxicity. The main cetuximab-related side effects were rash, diarrhea, and infusion reactions, and grade 3 to 4 febrile neutropenia was higher with cetuximab. Response rate was higher in the cetuximab-containing arm (36% vs 29%; P = .01). PFS was not different between the two groups (median PFS, 4.8 months in both groups; HR, 0.943; 95% CI, 0.825–1.077). The study did meet its primary end point of improvement in OS (median, 11.3 months vs 10.1 months; P = .044). However, despite meeting the primary end point of an OS benefit, FLEX did not lead to the widespread adoption of cetuximab in addition to chemotherapy, perhaps because of the relatively small increment in benefit, and the incongruence between the PFS and OS results.

More recently, the EGFR antibody necitumumab has been investigated in combination with chemotherapy. In the nonsquamous setting, necitumumab does not have any benefit. INSPIRE, a phase 3 trial of cisplatin and pemetrexed with or without necitumumab in the first-line treatment of nonsquamous NSCLC, was stopped early because of toxicity. An imbalance was observed in nonfatal and fatal thromboembolic events and overall number of deaths from all causes, with more adverse events noted in the necitumumab arm.

However, necitumumab does show a significant benefit when used in squamous NSCLC. Indeed, this benefit may have been presaged in prior studies, because analysis of FLEX showed that the OS benefit was most pronounced among patients with squamous histology. SQUIRE was a randomized phase 3 study of necitumumab in the first-line treatment of stage IV squamous NSCLC. Patients with squamous histology with PS 0 to 2 were randomized to receive cisplatin and gemcitabine with or without necitumumab. A total of 1092 patients were enrolled. The primary end point was OS. Although objective responses were similar between the two groups, the disease control rate was higher for the necitumumab group. Both PFS and OS were significantly improved with the addition of necitumumab (median OS, 11.5 months vs 9.9 months; HR, 0.84; 95% CI, 0.74–0.96; P = .01). Grade 3 or higher adverse events that were more common with necitumumab included hypomagnesemia and rash. There was a higher rate of venous thromboembolic events in the necitumumab arm, both of any grade (9% vs 5%) and grade 3 or higher (5% vs 3%), but unlike in the nonsquamous setting, the rate of fatal thromboembolic events was not different between the two groups. Evaluation by immunohistochemistry for EGFR protein expression (high expression defined as H-score ≥200, low as H-score <200) was performed as a prespecified analysis and was not predictive of differential effect. In November 2015, necitumumab was approved by the Food and Drug Administration for use in combination with cisplatin/gemcitabine chemotherapy for the first-line treatment of metastatic squamous NSCLC.