Abstract
Aim: The German ML21165 study evaluated bevacizumab-containing therapy for metastatic breast cancer (mBC) in routine oncology practice. Patients and Methods: Patients received bevacizumab with chemotherapy until disease progression, unacceptable toxicity or consent withdrawal. Pre-specified end-points were safety and efficacy [response rate, progression-free survival (PFS) and overall survival (OS)]. Results: Between May 2007 and September 2009, 865 patients received first-line bevacizumab plus paclitaxel for mBC, of whom 16% were aged ≥70 years and 9% had ECOG performance status of 2 or more. At data cut-off (median of 15.9 months' follow-up), the median PFS was 9.6 months [95% confidence interval (CI)=9.0-10.4 months] and the median OS was 21.6 months (95% CI=19.4-23.5 months). The most common non-haematological adverse drug reactions of grade 3 or more were pain (9%), hypertension (5%), sensory neuropathy (3%) and proteinuria (3%). Prolonged bevacizumab was well-tolerated. Conclusion: The efficacy and safety of first-line bevacizumab–paclitaxel in routine oncology practice is consistent with results from randomized trials.
The humanized monoclonal antibody bevacizumab, that targets vascular endothelial growth factor-A, was first introduced into clinical practice for the treatment of colorectal cancer in 2004 (1). Subsequent randomized phase III trials in advanced renal, non-small-cell lung, breast, ovarian and cervical cancer and glioblastoma demonstrated significant improvements in efficacy with the addition of bevacizumab to standard therapy (reviewed in 1). In human epidermal growth factor receptor (HER)2-negative metastatic breast cancer (mBC), three randomized phase III trials demonstrated significantly improved progression-free survival (PFS; primary end-point) and response rate when bevacizumab was combined with first-line chemotherapy (2-4). The most substantial benefit was seen with the combination of bevacizumab and paclitaxel in the E2100 trial (2).
It is commonplace that controlled clinical trials provide the highest level of evidence. However, it is also undisputed that clinical trials are prone to selection bias and do not always reflect clinical reality. The use of bevacizumab in combination with paclitaxel is recommended in German treatment guidelines [Arbeitsgemeinschaft Gynäkologische Onkologie (AGO) recommendations (5)] as first-line treatment for HER2-negative mBC. Thus in many institutions, bevacizumab in combination with chemotherapy has been integrated into the routine first-line treatment of mBC, particularly for patients with triple-negative breast cancer (TNBC) or hormone receptor-positive disease requiring more intensive therapy.
To determine how treatment results compare with data generated in controlled clinical trials, we conducted the present non-interventional study in routine oncology practice.
Patients and Methods
ML21165 was a prospective multicentre single-arm non-interventional study evaluating bevacizumab in combination with first-line chemotherapy given in routine clinical practice in Germany. Eligible patients had mBC, were aged ≥18 years and had received no prior chemotherapy for their advanced disease. Patients with contraindications for bevacizumab were ineligible. There were no specific requirements with respect to baseline blood pressure (BP) levels.
Bevacizumab was given in combination with chemotherapy according to the treating physician's standard practice. A minority of patients received bevacizumab with chemotherapy agents other than paclitaxel. For the purposes of this report, we focus on the large and more homogeneous subset of patients treated with bevacizumab in combination with paclitaxel. The paclitaxel treatment schedule was at the discretion of the treating clinician. Bevacizumab was given according to the licensed label in Germany at the time the study was designed (10 mg/kg every 2 weeks or 15 mg/kg every 3 weeks). Both agents were continued until disease progression, unacceptable toxicity or withdrawal of consent. If paclitaxel was stopped due to toxicity, bevacizumab could be continued as a single agent (or with the addition of endocrine therapy at the investigator's discretion) until disease progression.
Pre-specified end-points were safety [adverse drug reactions, adverse events (AEs) of special interest with bevacizumab based on experience from previous clinical trials] and efficacy [overall response rate, PFS and overall survival (OS)]. Toxicity was graded according to National Cancer Institute Common Terminology Criteria for Adverse Events (version 3.0) (6). The case report form (CRF) included specific questions relating to the occurrence and grade of haematological abnormalities (haemoglobin, leucocytes, granulocytes and thrombocytes) and selected AEs (bleeding, nausea, vomiting, diarrhoea, sensory neuropathy, hand–foot syndrome, fever, pain, hypertension and proteinuria). Other AEs were to be recorded as free text on the CRF at each cycle. The CRF also included tick boxes to record any suspected cases of gastrointestinal perforation, arterial thromboembolic events and reversible posterior leucoencephalopathy syndrome (RPLS), although grading of these three AEs was not specifically requested.
Response was assessed according to local practice and did not require confirmation 4 weeks after the first recorded response. However, at several participating centres, it was standard practice to confirm responses before recording them as such. Therefore reported response rates in this non-interventional study represent a mixture of unconfirmed and confirmed responses. Disease progression was determined according to investigator assessment. The study was monitored centrally and any queries were sent to the relevant investigator for clarification.
As this was a non-interventional study governed by strict rules for such studies in Germany, diagnostic approaches and the frequency of follow-up visits varied according to the routine practice in each participating centre. In general, diagnosis and follow-up were performed according to the German guidelines available at the time of the study (6). Of note, neutrophil counts were usually recorded before the start of a cycle and therefore do not reflect nadir measurements. Data were collected for 1 year after the start of bevacizumab therapy. Three additional requests for follow-up information were made in January 2010, September 2010 and November 2011 for patients who had been off study for at least 1 year and for whom neither death nor loss to follow-up had been reported. At these follow-up time-points, data were collected on PFS, OS, continuation of bevacizumab after the end of the study and further lines of treatment, but not safety.
The planned sample size of 1,000 patients provided 99% probability of recording at least four occurrences of an adverse drug reaction with a true incidence of 1%. The probability of capturing at least one occurrence of an adverse drug reaction with a true incidence of 0.5% exceeded 99%. PFS and OS were estimated using the Kaplan–Meier method with corresponding 95% confidence intervals (CIs) for the medians calculated according to the Brookmeyer and Crowley method (7). The duration of follow-up was reported descriptively as the median of reported values.
The study was approved by the relevant Ethical Committee at each participating centre and all patients provided written informed consent.
This non-interventional study was listed in the local register created in collaboration with the German health authorities – see http://www.vfa.de/de/arzneimittel-forschung/datenbanken-zu-arzneimitteln/nisdb/nis-details/_625
Results
Patient population. A total of 1,049 patients starting bevacizumab treatment between May 2007 and September 2009 were eligible for the study. Of these, 865 patients from 495 centres were treated with bevacizumab in combination with paclitaxel in the first-line chemotherapy setting. The cohort comprising the analysis population for the present report were generally representative of patients presenting in routine clinical practice (Table I). Baseline BP was reported for 661 (76%) of the 865 patients. Systolic BP was ≥140 mmHg in 195 (29%) of the 661 patients (24% of those aged <65 years vs. 41% aged ≥65 years) and diastolic BP was ≥90 mmHg in 53 patients (8%).
Treatment exposure. The data cut-off date was 31 July 2012. At this time, the median duration of bevacizumab exposure was 7 months (range=<1-46 months). Bevacizumab therapy was continued for ≥6 months in 564 patients (65%) and for ≥12 months in 167 patients (19%). In 360 patients (42%) in whom chemotherapy was discontinued before disease progression, bevacizumab was continued as a single agent.
Reasons for discontinuing bevacizumab treatment/ observation (in descending order of frequency) were: disease progression or death from mBC (55% of patients); end of observation period (17%); unacceptable toxicity (11%); refusal/non-compliance (6%); death from other cause (2%); or loss to follow-up (2%). The reason given for bevacizumab discontinuation was ‘other’ in the remaining patients, most commonly because of maximal response (4%), patient's wish (3%), deterioration of health (3%) or end of planned therapy (2%).
Efficacy. The median duration of follow-up was 15.9 months. At this time, 83% of patients had experienced a PFS event and 61% had died.
The best overall response rate (confirmed or unconfirmed, as mentioned above) was 62% (95% CI=59–65%), including complete responses in 11% of patients. Only 8% of patients had progressive disease as their best response (indicating primary resistance to bevacizumab–paclitaxel therapy). The median PFS was 9.6 months (95% CI=9.0-10.4 months) in the overall population (Figure 1). The median OS was 21.6 months (95% CI=19.4-23.5 months) (Figure 2). The 1-year OS rate was 73% (95% CI=70-76%).
Exploratory sub-group analyses of response rate, PFS and OS in populations of particular clinical importance are summarized in Table II. There was little variation in response rate according to clinical characteristics. Similarly, there was no relevant difference in PFS in older versus younger patients. However, PFS and OS were shorter in patients with TNBC or three or more metastatic organ sites. The median PFS was 7.9 months in patients with TNBC and 8.3 months in patients with three or more metastatic sites; the median OS was 15.8 months and 19.2 months, respectively.
Safety. The most common all-grade AE was pain (51%), that could have been related to underlying disease or treatment. Hand–foot syndrome (any grade) was reported in 31% of patients, sensory neuropathy in 30% and hypertension in 27%. The most common non-haematological adverse drug reactions of grade 3 or more were pain (9%), hypertension (5%), sensory neuropathy (3%) and proteinuria (3%). Grade 3/4 neutropenia was reported in 6% of patients and anaemia in 4%, although it should be noted that nadir measurements were not usually recorded. Focusing on clinical AEs of grade 3 or more considered by the investigator to be at least possibly related to bevacizumab, the most common were hypertension and thrombosis/ embolism (Table III).
Arterial thromboembolic events occurred in eight patients (1%), of which three cases were considered by the investigator to be related to bevacizumab. Six patients (1%) experienced gastrointestinal perforation; all cases were considered by the investigator to be related to bevacizumab. One suspected case of RPLS was observed, but this did not meet all criteria for RPLS.
Twelve patients (1.4%) died during the study from AEs considered at least possibly related to bevacizumab. The causes were reported as intestinal perforation (n=4), gastrointestinal bleeding and suspected gastric perforation with haemorrhagic shock (n=1), hepatic cirrhosis with gastrointestinal bleeding due to gastritis (n=1), cerebral haemorrhage (n=1), bleeding (n=2), cerebral infarction/ confusional state/hemiparesis accompanied by oesophageal ulcer (n=1), left ventricular dysfunction (n=1) and circulatory collapse/sudden cardiovascular failure (n=1).
Figure 3 shows AEs in patients aged <65 versus ≥65 years. Patients aged ≥65 years were at slightly increased risk of grade 3/4 anaemia, hypertension, sensory neuropathy and proteinuria compared to younger patients, but grade 3/4 pain was less frequent in the older sub-group.
Safety of prolonged bevacizumab therapy was assessed in an exploratory sub-group analysis of the 167 patients treated for 1 year or more. This patient population was characterized by a marginally younger age (median 57 years vs. 58 years in the overall population) and fewer patients with a disease-free interval <12 months (14% vs. 24%, respectively), although similar proportions of patients had three or more metastatic organ sites, prior chemotherapy exposure and Eastern Cooperative Oncology Group (ECOG) performance status 2 or 3. The majority (78%) of patients treated for 1 year or more received bevacizumab as a single agent (or with endocrine therapy) after discontinuation of chemotherapy. The incidence of grade 3 or more hypertension was higher in patients treated for at least 1 year (11.4%; 10.8% grade 3, 0.6% grade 4) than in the overall population (5.1%). However, there was no difference in the incidence of other grade 3 or more AEs of special interest, such as proteinuria, bleeding, arterial thromboembolism, cardiac function abnormalities and gastrointestinal perforation.
Additional follow-up questionnaires were available from 785 patients (91%) and revealed that 234 patients (30% of those for whom information was available) continued bevacizumab after completing the observation period of the study. Among the 212 patients for whom information on therapy duration was available, the median bevacizumab exposure after completing the study was 11.1 months, with 42% continuing for more than 1 year after completing the study observation period. At least one further line of chemotherapy was administered to 169 (40%) of the 424 patients with information available (and up to five further lines in a small proportion of patients).
Discussion
In this analysis of a large observational study, which to the best of our knowledge represents the largest reported population of patients treated with bevacizumab–paclitaxel combination therapy, first-line bevacizumab-containing treatment was active and well tolerated in a broad population of patients with mBC representative of those treated in routine oncology practice. Data collection in non-interventional studies typically differs from that in prospective clinical trials as it is based on routine clinical practice, which varies according to local practice. Nevertheless, the efficacy results from this study are generally consistent with those from randomized phase III trials and the ATHENA safety study in routine practice (2-4, 8-10). The median PFS of 8 months in patients with TNBC is noteworthy and compares favourably with other treatment options reported in the literature (11). Likewise, efficacy is particularly encouraging in the sub-group of patients with three or more metastatic sites, representing patients with a high tumour burden.
The median OS was slightly lower than that reported in the literature, potentially attributable to the broader patient population and, in particular, the older age of these patients compared with previously reported studies (median 58 years in the present study compared with 56 years in E2100 and 53 years in ATHENA) and the typically poorer performance status (40% ECOG performance status 0 in the present study vs. 58% in the ATHENA study). The 73% 1-year OS rate in the present analysis is within the range reported in the E2100 trial and in sub-groups of patients treated with bevacizumab–paclitaxel combination therapy in the ATHENA study (10, 12), although lower than the 81% 1-year OS rate in the more recently reported TURANDOT trial (9).
The safety profile of bevacizumab–paclitaxel was consistent with the established safety profile of bevacizumab-containing therapy in the context of randomized prospective clinical trials, despite inclusion of a broader patient population. Of note, thromboembolic events and cardiac toxicity were infrequent. The exceptionally low rate of neutropenia in a population treated with taxane-containing therapy may be explained by differences in data collection and reporting in this study compared with, for example, the E2100 and ATHENA studies. However, it is also important to note that only selected AEs were included in the CRF, and information on other AEs was derived from free text entered by the treating physician on the CRF. Furthermore, those events reported as AEs specifically covered with individual questions in the CRF were not duplicated in the overall AE reporting. This limits the ability to directly compare safety data from this non-interventional study with findings from conventional prospective clinical trials.
The efficacy of and tolerance to bevacizumab-containing therapy do not appear to diminish with increasing patient age, although possible physician bias when selecting older patients for inclusion in this study must be considered. Grade 3/4 hypertension and proteinuria were more common in older than younger patients, consistent with observations among patients aged 70 years or more in the ATHENA study (13). Slightly more sensory neuropathy may be expected in older patients regardless of bevacizumab administration. Overall, bevacizumab-containing regimens appear to provide an effective and well-tolerated alternative to chemotherapy doublets, which are often not feasible in elderly patients (14, 15).
Sub-group analyses suggest that bevacizumab-containing therapy for at least 1 year is well tolerated. There was an increase in grade 3 or more hypertension in patients with prolonged bevacizumab exposure but this side-effect was manageable. Furthermore, this observation is consistent with reports in the literature in both breast cancer and ovarian cancer (10, 16, 17). Hypertension can occur in later as well as earlier cycles and therefore an increased incidence with longer exposure is not unexpected. The results suggest that patients with mBC can receive first-line bevacizumab-containing therapy for prolonged periods, benefiting from encouraging long-term disease control without major toxicity, although the lack of systematic safety data collection beyond 1 year should be considered when interpreting this exploratory analysis.
A limitation of this study is the less comprehensive efficacy follow-up and less strict definition of response rate (representing a combination of confirmed and unconfirmed responses) compared to prospective trials. However, the trial was designed pragmatically to reflect routine oncology practice and as such, rigorous tumour assessment was not realistic. The sample sizes of populations treated with other bevacizumab-containing regimens and/or in the second-line setting are too small for meaningful analyses.
In summary, safety and efficacy results from the largest reported population of patients treated with first-line bevacizumab–paclitaxel combination therapy for mBC provides reassuring evidence of the tolerability of this regimen. A second non-interventional study in Germany (AVANTI), that includes more intensive data collection (treatment decision-making, management of selected AEs, quality of life), is assessing the safety and efficacy of bevacizumab with either of its licensed chemotherapy partners – paclitaxel or capecitabine. Interim results were recently reported (18).
Acknowledgements
The Authors would like to thank all patients, their families, the investigators and the nurses who participated in this trial. The Authors thank Jennifer Kelly (Medi-Kelsey Ltd, Ashbourne, UK) for medical writing support in drafting the manuscript with funding from Roche Pharma AG, Germany, before review, revision and the decision to publish by the Authors.
This non-interventional research was sponsored by Roche Pharma AG, Grenzach-Wyhlen, Germany.
Footnotes
- Received December 31, 2015.
- Revision received February 15, 2016.
- Accepted February 16, 2016.
- Copyright© 2016 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved