This article is mentioned in:

Introduction

In several clinical trials, first-line combination chemotherapies containing bevacizumab (Bev) were revealed to improve clinical outcomes in patients with advanced non-squamous non-small cell lung cancer (non-sq NSCLC) (1–3). Sandler et al (3) reported significant survival benefits, such as overall survival (OS) of >1 year (12.3 months), with addition of Bev to paclitaxel plus carboplatin in the treatment of non-sq NSCLC. Although Bev was approved for NSCLC in 2009 in Japan, there are insufficient data regarding the efficacy, toxicity and predictive markers for Bev treatment. The present study evaluated the efficacy and safety of Bev-containing combination chemotherapy in patients with non-sq NSCLC in clinical settings in Japan. Landmark survival analysis, or disease control at 8 weeks, was reported to be a more powerful predictor of subsequent survival compared with the traditional tumor response rate in advanced NSCLC (4). This may provide an early assessment of subsequent outcome. Since treatment with bevacizumab occasionally results in cavitary lesion without tumor shrinkage (5), stable disease may also be important for understanding drug efficacy. Therefore, landmark analysis was utilized in the present study. In addition, the identification of predictive markers for Bev-containing chemotherapy efficacy was attempted. The mutation of epidermal growth factor receptor (EGFR) is a key factor in predicting the response and survival rate following EGFR-tyrosine kinase inhibitor (EGFR-TKI) treatment (6,7), but no data have demonstrated the importance of EGFR mutations in predicting the effect of Bev treatment. In the present study, multivariate analysis using Cox's regression model revealed that specific EGFR mutations are predictive markers for Bev treatment efficacy.

Patients and methods

Patient eligibility

Patients scheduled for Bev treatment between August 2010 to July 2012 were prospectively enrolled in the present study. Eligible patients had histologically or cytologically confirmed inoperable advanced, stage IIIB-IV, or recurrent non-sq NSCLC, an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0–2, and adequate organ function for cytotoxic chemotherapy. The main exclusion criteria were diagnoses of squamous cell carcinoma or symptomatic brain metastasis, although patients with controlled brain metastases were eligible, surgery or surgical biopsy within 4 weeks, a history of significant hemoptysis, of >2.5 ml per episode, coagulation treatment, bleeding tendency, active concomitant malignancy, presence of significant comorbidities such as uncontrolled hypertension, interstitial pneumonia, active gastrointestinal ulcer, angina pectoris, pregnancy or lactation or other factors as judged by a medical oncologist. Protocol-specified demographics, disease characteristics including EGFR mutation status and patient medical history were collected at baseline. Patients were treated at four major hospitals participating in the Shinjuku Thoracic Oncology Group (STOG), Keio University School of Medicine (Tokyo, Japan), National Center for Global Health and Medicine (Tokyo, Japan), Tokyo Medical University (Tokyo, Japan), Tokyo Women's Medical University (Tokyo, Japan). The study protocol was approved by the institutional review board at each institution. All patients provided written informed consent prior to inclusion in the present study.

Study design and treatment

The primary endpoints were progression-free survival (PFS) and safety. The secondary endpoints were the response rate (RR), time to response and landmark survival (4). Patients received 15 mg/kg Bev every 3 weeks in conjunction with the chemotherapy prescribed by attending physicians, followed by 15 mg/kg Bev every 3 weeks with or without chemotherapy as maintenance. Any line of chemotherapy was permitted. Bev-containing regimens were as follows: Carboplatin (CBDCA) + pemetrexed (PEM); cisplatin (CDDP) + PEM; CBDCA + paclitaxel (PTX); PEM; docetaxel (DTX); CBDCA + DTX; and CBDCA + PEM + erlotinib (n=41, n=21, n=18, n=9, n=9, n=3 and n=1, respectively). All chemotherapy regimens contained either PEM (n=72) or a taxane (DTX or PTX; n=30). Treatment was continued until tumor progression or the patient experienced unacceptable toxicities, such as grade 2 or severe hemoptysis, grade 3 or severe bleeding, or by the decision of attending physician.

Evaluation

Tumor response was assessed according to the Response Evaluation Criteria in Solid Tumors version 1.1 (8). Tumors were assessed at baseline by computed tomography, magnetic resonance imaging, bone scintigraphy and/or fluorodeoxyglucose positron emission tomography. During treatment, a radiographic evaluation was performed subsequent to at least every two courses of treatment and/or at the time of suspected disease progression. To confirm response, a partial response (PR) or complete response (CR), radiographic evaluations were recommended 4 weeks subsequent to the original evaluations. The disease control rate (DCR) was defined as [CR + PR + stable disease (SD)]/total patients. Toxicities were graded according to the National Cancer Institute Common Toxicity Criteria version 4.0 (9).

Statistical analysis

The primary endpoints were PFS and safety. Using the 13-week PFS rates of 80.1% in the chemotherapy plus Bev arm and 67.7% in the chemotherapy-alone arm from the ECOG 4599 study (3), the sample size was calculated to achieve a power of 80% with a two-sided α of 0.05, and the expected and threshold values for PFS were 6.2 and 4.5 months, respectively. The estimated minimum sample size was 77, which was calculated by SWOG statistical tools using a one-arm binomial setup. Allowing for a maximum dropout rate of 30% and considering that the present study was an observational cohort study with a minimum of 12 months of follow-up, 100 patients were sought for enrollment. PFS and OS subsequent to the initiation of the Bev-containing chemotherapy were estimated by the Kaplan-Meier method. PFS was defined as the time from the initiation of Bev therapy to investigator-assessed disease progression or mortality from any cause. OS was defined as the time from the initiation of the Bev therapy to mortality from any cause. OS, PFS and responses were assessed in all eligible patients on an intent-to-treat basis. Patients without an event (progression or mortality) were censored at the last follow-up or data cutoff date, whichever occurred first. Univariate and multivariate analyses were performed using Cox's proportional hazards model to assess the independent effects of patient and disease characteristics on PFS and OS. To avoid possible confounding effects of treatment with EGFR-TKIs prior to accrual of the present study (‘pre-treatment with EGFR-TKIs’) multivariate analyses of PFS and OS were adjusted for ‘pre-treatment with EGFR-TKIs’. Landmark survival analyses (4) of PFS and OS were performed by comparing the patients who achieved PR or SD 8 weeks subsequent to Bev administration. All P-values in the present study are two-sided. P<0.05 was considered to indicate a statistically significant difference. Statistical analysis was conducted using IBM SPSS Statistics version 19.0 software (IBM SPSS, Armonk, NY, USA). The present study is registered with the University Hospital Medical Information Network (UMIN) Clinical Trials Registry (www.umin.ac.jp/ctr) as trial number UMIN000004609.

Results

Patients characteristics and treatment

Between August 2010 and July 2012, a total of 102 patients were enrolled in the present study, with a data cutoff date of September 30, 2014. The duration of follow-up subsequent to final registration was 14 months. The median follow-up time was 1,021 days (33.6 months). Patient characteristics are summarized in Table I. Histological analysis revealed that 98 patients (96.1%) exhibited adenocarcinoma, three patients exhibited NSCLC that was not otherwise specified, and one patient exhibited pleomorphic carcinoma. The majority of the patients were receiving first-line treatment (n=57, 56%), but patients receiving second-line (n=22, 22%), third-line (n=12, 12%) and later lines of treatment, fourth, fifth, sixth and seventh line, n=5, n=4, n=1, and n=1, respectively, were also included. At the data cutoff date, the median numbers of cycles of Bev administration were 9, 6, and 7 for the first-, second-, and third- and later lines of treatment, respectively. EGFR mutation status was available for the majority of the patients (98%), with mutations identified in 44.1% of the patients. Amongst the patients with EGFR mutations, mutation types were characterized for 39 patients. The most common type of mutation was an exon 19 deletion (n=28, including one case with T790M), followed by the exon 21 mutations L858R, L858R and T790M, and L861Q (n=11 total; n=8; n=2; n=1, respectively). Amongst the patients with major EGFR mutations, 14 of 28 patients (50%) with the exon 19 deletion and 8 of 11 patients (73%) with the exon 21 mutations received ‘pre-treatment with EGFR-TKIs’. In total, 12 of 28 patients (43%) with the exon 19 deletion and 8 of 11 patients (73%) with the exon 21 mutations were treated with EGFR-TKIs subsequent to completion of the present study, that is, ‘post-treatment with EGFR-TKIs’. Only five patients with the exon 19 deletion lacked information regarding EGFR-TKI treatment, but all patients with the exon 21 mutations received EGFR-TKIs prior and/or subsequent to the present study.

Table I. Patient characteristics (n=102).

Table I.

Patient characteristics (n=102).

Characteristic	Classification	n (%)
Age, years	(median, range)	64	(36–85)
Sex	Male	60	(58.8)
	Female	42	(41.2)
Smoking history	+	61	(59.8)
	−	41	(40.2)
Performance status	0	66	(64.7)
	1	34	(33.3)
	2	2	(2.0)
Histology	Adenocarcinoma	98	(96.1)
	Others	4	(3.9)
Stage	IIIA	2	(2.0)
	IIIB	8	(7.8)
	IV	77	(75.5)
	Recurrent	15	(14.7)
Brain metastasis	+	11	(10.8)
	−	91	(89.2)
Platinum combination	+	84	(82.4)
	−	18	(17.6)
Treatment line	First	57	(55.9)
	Second	22	(21.6)
	Third and later	23	(22.5)
EGFR mutation	+	45	(44.1)a
	−	55	(53.9)
	Unknown	2	(2.0)

a Type of EGFR mutation, Exon 19 del n=28, Exon 21 n=11, no information n=6.

Toxicity

The toxicity profile is summarized in Table II. The most frequent grade 3 or 4 adverse events were neutropenia and hypertension. Severe grade 3 or 4 hematological toxicities included leukopenia (25.5%), neutropenia (42.2%), anemia (6.9%), thrombocytopenia (4.9%) and febrile neutropenia (3.9%). Severe grade 3 or 4 Bev-associated adverse events included hypertension (30.4%), proteinuria (5.9%), thromboembolism (4.9%) and epistaxis (1.0%). Overall, the majority of adverse events were manageable. No treatment-associated mortality was observed.

Table II. Toxicities observed.

Table II.

Toxicities observed.

	All n (%)	G3 or more n (%)
Anemia	58(56.9)	7(6.9)
Thrombocytopenia	47(46.1)	5(4.9)
Leukopenia	69(67.6)	26(25.5)
Neutropenia	65(63.7)	43(42.2)
Febrile neutropenia	4(3.9)	4(3.9)
Hypoalbuminemia	42(41.2)	1(1.0)
Increased AST	45(44.1)	3(2.9)
Increased ALT	40(39.2)	2(2.0)
Increased ALP	19(18.6)
Increased creatinine	15(14.7)
Proteinuria	49(48.0)	6(5.9)
Declining PS	56(54.9)	10(9.8)
Nausea	42(41.2)	2(2.0)
Vomiting	16(15.7)	1(1.0)
Appetite loss	52(51.0)	2(2.0)
Diarrhea	10(9.8)
Constipation	40(39.2)
Mucositis oral	11(10.8)
Fatigue	27(26.5)	1(1.0)
Malaise	39(38.2)
Macular rash	14(13.7)
Fever up	18(17.6)
Dyspnea	10(9.8)
Peripheral motor neuropathy	3(2.9)	1(1.0)
Peripheral sensory neuropathy	27(26.5)	4(3.9)
Dysgeusia	15(14.7)
Alopecia	13(12.7)
Pain	11(10.8)
Headache	5(4.9)	1(1.0)
Hypertensiona	79(77.5)	31(30.4)
Epistaxis	16(15.7)	1(1.0)
Thromboembolic event	7(6.9)	5(4.9)
Bleeding	5(4.9)	1(1.0)
Infection	7(6.9)	2(2.0)
Duodenal ulcer	1(1.0)	1(1.0)
Gait disturbance	1(1.0)	1(1.0)
Meningitis	1(1.0)	1(1.0)

a A total of 47 cases had high blood pressure before bevacizumab treatment. AST, aspartate aminotransferase; ALT, alanine aminotransferase; ALP, alkaline phosphatase.

Efficacy

The objective response to Bev-containing chemotherapies is summarized in Table III. With 102 evaluable patients, the RR was 44.1% and the DCR was 92.2%. Only 1 patient achieved a CR. The median PFS was 8.3 months (95% CI=6.4–10.2 months), which is longer compared with the predefined expected and threshold values for PFS (6.2 and 4.5 months, respectively), as demonstrated in Fig. 1A. The primary PFS endpoint was met. The median OS was 26.3 months (95% CI=22.2–30.4 months), as illustrated in Fig. 1B. There was no significant difference in the time to response to BEV between chemotherapy-naïve: Median, 73 days; range, 25–519 days, and previously treated patients: Median, 40 days; range, 23–535 days (P=0.265). No significant difference was observed in PFS between patients with SD, 268 days (95% CI=196–416 days) and those who achieved PR, 266 days (95% CI=125–431 days) by landmark analysis (P=0.97). There was also no significant difference observed in OS between patients with SD, 860 days (95% CI=640–1079 days) and those who achieved PR, 664 days (95% CI=329–999 days) by landmark analysis (P=0.35). Univariate and multivariate analyses were performed to identify the variables significantly associated with PFS and OS, as summarized in Tables IV and V. Univariate analysis revealed that several clinical factors were associated with PFS (P<0.15), as illustrated in Table IV. In the crude model, multivariate analysis indicated that a higher T factor, EGFR exon 21 mutations, and poor PS were linked to significantly shorter PFS. Subsequent to adjustment for pretreatment with EGFR-TKIs, multivariate analysis revealed that a higher T factor [adjusted hazard ratio (HR)=1.33] and poor PS (adjusted HR=1.63) were associated with significantly shorter PFS, whilst EGFR exon 19 deletion (adjusted HR=0.47) was associated with prolonged PFS (P<0.05), as demonstrated in Table V. In the crude model for OS, multivariate analysis indicated that a lower T factor and EGFR exon 19 deletion were significant favorable prognostic factors. Subsequent to adjustment for pretreatment with EGFR-TKIs, multivariate analysis revealed that EGFR mutations (adjusted HR=0.20), better PS (adjusted HR=0.45), and the absence of primary site lesions (adjusted HR=0.29) were significant favorable prognostic factors (P<0.05). The Kaplan-Meier estimation of OS was significantly higher for patients with the EGFR exon 19 mutations compared with the estimation for patients with the EGFR exon 21 mutations or wild-type EGFR (P=0.037).

Figure 1. Survival analyses. (A) PFS. (B) OS. PFS, progression-free survival rate; OS, overall survival rate.

Table III. Response to intervention.

Table III.

Response to intervention.

Response	n (%)
Complete response	1(1.0)
Partial response	44(43.1)
Response rate	(44.1)
Stable disease	48(47.1)
Non-CR/non-PD	1(1.0)
Disease control	(92.2)
Progressive disease	5(4.9)
Not evaluable	3(2.9)

Table IV. Progression-free survival by univariate Cox's regression analyses using an adjusted model for pretreatment of EGFR-TKI (P<0.15).

Table IV.

Progression-free survival by univariate Cox's regression analyses using an adjusted model for pretreatment of EGFR-TKI (P<0.15).

Variable	HR	95% CI	P-value
Performance status (0, 1, 2)	1.94	1.26–2.26	0.003
T factor (0, 1, 2, 3, 4)	1.30	0.08–1.08	0.005
N factor (0, 1, 2, 3)	1.25	1.05–1.05	0.01
Brain metastasis (no/yes)	2.22	1.16–4.16	0.02
Target lesion (no/yes)	2.62	1.19–5.19	0.02
Recurrence subsequent to surgery vs. IIIA-IV	2.17	1.12–4.12	0.02
History of surgery (no/yes)	0.58	0.35–0.35	0.03
Primary site (no/yes)	2.01	1.03–3.03	0.04
EGFR exon 19 deletion (no/yes)	0.57	0.33–0.33	0.04
Combined treatment (taxane/PEM)	0.61	0.38–1.38	0.05
Chemotherapy regimen (PEM/docetaxel/Pt + PEM/Pt + taxane)	1.28	0.96–1.96	0.09

[i] TKI, EGFR-tyrosine kinase inhibitor; Taxane, paclitaxel or docetaxel; PEM, pemetrexed; Pt, platinum agent; CI, confidence interval.

Table V. Progression-free survival by multivariate Cox's regression analyses using an adjusted model for pretreatment of EGFR-TKI.

Table V.

Progression-free survival by multivariate Cox's regression analyses using an adjusted model for pretreatment of EGFR-TKI.

Variable	HR	95% CI	P-value
T factor (0, 1, 2, 3, 4)	1.33	1.10–1.10	0.003
EGFR exon 19 deletion (no/yes)	0.47	0.25–0.25	0.02
Performance status (0, 1, 2, 3, 4)	1.63	1.02–2.02	0.04
N factor (0, 1, 2, 3)	1.17	0.97–1.97	0.11
Combined treatment (taxane/PEM)	0.73	0.41–1.41	0.28
Brain metastasis (no/yes)	0.87	0.39–1.39	0.74

[i] Taxane, paclitaxel or docetaxel; PEM, pemetrexed; CI, confidence interval.

Discussion

The present study evaluated the efficacy and safety of Bev-containing chemotherapy regimens in clinical practice in Japan, as first and later lines of chemotherapy. The RR of the present cohort study was 44.1%, and the median PFS was 8.3 months, data that demonstrates better efficacy for patients compared with those of previous reports (E4599, RR 35%, PFS 6.2 months; ARIES, RR 49%, PFS 6.6 months) (2,3). The primary endpoint of the present study, PFS, was met, suggesting that Bev-containing chemotherapy is effective in clinical settings in Japan. The median OS, 26.3 months, was also improved compared with those of previous trials (E4599, 12.3 months; ARIES, 13.0 months; SAiL, 14.6 months) (1–3). The effect of Bev was similar to that observed in a Japanese phase II trial that used Bev as first-line chemotherapy (JO19907) (10), which reported an RR of 60.7%, a median PFS of 6.9 months and a median OS of 22.8 months, although the present study included patients receiving first-line and subsequent Bev treatment.

With respect to the severe adverse events of special interest due to Bev treatment, grade III or higher hypertension, proteinuria, thromboembolism and epistaxis occurred in 30.4, 5.9, 4.9 and 1.0% of patients, respectively, in the present study compared with rates of 7, 3.1, 0.2 and 0.7%, respectively, in the E4599 trial (3) and 6, 3, 8 and 1%, respectively, in the SAiL study (1). A total of 11% of patients exhibited severe hypertension in the JO19907 study conducted in Japan (10). The higher incidence of hypertension in the current study may reflect the baseline characteristics of the patients, such as the higher incidence of comorbid hypertension at the time of Bev-containing chemotherapy initiation: 46.1% of patients exhibited hypertension at baseline. Although the present study revealed a higher incidence of hypertension, all toxicity was manageable.

The EGFR mutation type was a predictive marker for Bev-containing chemotherapy efficacy in the present study, with the EGFR exon 19 mutation being a favorable predictor of PFS. Previously, several studies suggested that the effects of EGFR-TKI treatment differ according to the type of EGFR mutation (11–13). Regarding treatment with afatinib, a second-generation irreversible EGFR-TKI, the EGFR mutation type was associated with OS in patients with EGFR mutation-positive lung adenocarcinoma (11). Patients with the EGFR exon 19 deletion experienced a significant survival benefit with afatinib treatment compared to chemotherapy, although no such benefit was observed in patients with the EGFR exon 21 mutations (11). In two phase II studies of EGFR-TKIs, erlotinib or gefitinib, plus Bev, improved PFS was seen upon addition of Bev to the EGFR-TKI regimens in the patients with the EGFR exon 19 deletion, but not for the patients with the exon 21 mutations (12,13). The reason for the Bev-mediated survival benefit in patients with the EGFR exon 19 deletions remains unclear. Several studies have reported distinct biochemical properties of different EGFR mutations that may explain the different responses to EGFR-TKIs (14,15). An association between EGFR and vascular endothelial growth factor (VEGF) has also been reported. EGFR-mutated tumors display higher VEGF expression levels than wild-type EGFR tumors (16). EGFR-TKI-resistant tumors also produce greater levels of VEGF, and the amount of VEGF production varies according to the mutation type (17,18). These data may explain the favorable efficacy of Bev-containing chemotherapy in patients with EGFR exon 19 deletions.

There were several limitations to the present study. Firstly, the sample size was small compared with those in previous studies (SAiL, n=2212; ARIES, n=1967) (1–2). Secondly, the treatment was not restricted to first-line chemotherapy, although Bev-containing chemotherapy was used for first-line treatment. Concerning the OS, the higher frequency of EGFR mutations in the Japanese population may favorably affect patient outcomes compared with previous reports (SAiL and ARIES). In a Japanese phase II study (10), OS was 22.8 months and PFS was 6.9 months for first-line treatment with CBDCA + PTX + Bev. In that trial, the EGFR mutation data were not available, but 41% of patients received EGFR-TKIs as post-protocol therapy. In the present study, EGFR mutations were exhibited in 43.1% of patients. A considerable percentage of patients may have benefited from post-progression treatment with EGFR-TKIs. A major strength of the present study was that EGFR mutation status data were available for the majority of patients. This allowed the importance of EGFR exon 19 mutation as a possible predictive marker for Bev treatment efficacy to be elucidated.

In conclusion, Bev-containing combination chemotherapy was effective in treating patients with non-sq NSCLC in clinical settings in Japan. Adverse events were well-tolerated and acceptable. Even though these are the results of ad hoc analyses, multivariate analysis revealed that a lower T factor, better PS, and the EGFR exon 19 mutations were associated with prolonged PFS.

References