First-line chemotherapy for metastatic colorectal cancer (mCRC) is frequently combined with antiangiogenic agents, namely, bevacizumab, a monoclonal antibody targeting vascular endothelial growth factor (VEGF) A. This is based on the results of several studies demonstrating a benefit in progression-free survival [
Prolonged duration of bevacizumab treatment
These observations led to the setup of clinical trials, which investigated the use of bevacizumab beyond progression in patients who had undergone a bevacizumab-based first-line treatment. Two trials (TML [
More recently, randomized trials investigated other drugs with antiangiogenic properties in second and further lines of treatment in patients with metastatic colorectal cancer after pretreatment with bevacizumab-based regimens. For instance, the phase-3 VELOUR trial investigated the addition of aflibercept (a fusion protein trapping VEGF-A, VEGF-B, and placental growth factor [PlGF]) in combination with 5-fluorouracil and irinotecan (FOLFIRI) treatment in patients who had been pretreated with oxaliplatin-based regimens [
In the present meta-analysis we sought to investigate the concept of treatment with antiangiogenic drugs in multiple lines beyond progression by analyzing aggregate data of randomized trials. Special emphasis was given on describing potential improvements of progression-free and overall survival related to specific subgroups including KRAS wildtype patients.
Primary objective of the present analysis was to investigate progression-free survival (PFS) and overall survival (OS) in patients with metastatic colorectal cancer who had been pretreated with an antiangiogenic treatment and underwent antiangiogenic treatment beyond progression.
Secondary objectives were to assess the effects of the continued or repeated antiangiogenic treatment in subgroups (stratified by age, sex, ECOG status, and tumor KRAS mutational status) and in studies using anti-VEGF treatment (i.e., bevacizumab and aflibercept) versus tyrosine kinase inhibitors (TKI). Moreover, we investigated the response rates (i.e., the rate of evaluable patients achieving complete or partial remissions) and the rate of “tumor stabilization,” that is, the rate of evaluable patients without primary progression while receiving treatment.
Only randomized phase II and III trials were included in the current meta-analysis. The inclusion of subgroups of randomized trials was allowed provided sufficient information was given in the trial reports. Only studies performed with the approval of an appropriate ethics committee and conducted in compliance with the declaration of Helsinki were included in this meta-analysis.
Antiangiogenic treatment was defined as the use of drugs targeting at least one important angiogenic pathway, for instance, monoclonal antibodies targeting VEGF or VEGF-receptors, or (multi)TKI targeting angiogenic pathways.
Searches in PubMed and proceedings of international meetings were conducted to identify studies with information relevant for the current analysis.
Eligible studies were phase II or III, randomized, controlled trials comparing (i) antiangiogenic drugs in combination with either active treatment (i.e., chemotherapy) or placebo with (ii) active treatment or placebo alone in patients who had previously been treated with antiangiogenic drugs for metastatic colorectal cancer. We used MeSH and full-text search terms for metastatic colorectal cancer and molecular targeted therapies, limiting our results to English language articles published in PubMed between January 1, 2007, and October 11, 2015. For PubMed, the search was ((((“molecular targeted therapy” [All Fields] OR (“molecular” [All Fields] AND “targeted” [All Fields]) AND (“therapy” [All Fields] OR “therapies” [All Fields]) AND (“colorectal neoplasms” [All Fields] OR “colorectal cancer” [All Fields]) OR (“colorectal” [All Fields] AND “cancer” [All Fields]) AND (“randomized” [All Fields] OR “randomized study” [All Fields]) AND English [lang])))).
In addition to computerized search, references of retrieved papers were also screened for missing trials.
To minimize publication bias we conducted a manual search of conference abstracts. For conferences, the search was “colorectal cancer” or “advanced colorectal cancer”, manually limited to abstracts on targeted therapies. The proceedings of the following meetings were examined for presented abstracts limiting the search to the years 2007–2016: (i) American Society of Clinical Oncology (ASCO) annual meetings; (ii) ASCO Gastrointestinal Cancer Symposium; (iii) European Society for Medical Oncology (EMSO) and European multidisciplinary cancer congress (ECCO) meetings; (iv) World Congress on Gastrointestinal Cancer. Two independent reviewers (RDH, UR, or UH) assessed title, keywords, and abstracts of retrieved studies. If studies met the inclusion criteria, they assessed full texts and mutually decided on inclusion.
Data extraction was conducted independently by three investigators (RDH, UR, and UH) in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance [
The impact of antiangiogenic treatment on overall survival (OS) and progression-free survival (PFS) was measured in terms of the hazard ratio (HR). In all studies included in this meta-analysis, the HR was calculated as ratio of the active therapy divided by the hazard of the control group. Therefore, a HR lower than 1 indicates a benefit of the active treatment, whereas a HR higher than 1 indicates a higher risk of death or progression of the active treatment, respectively. The estimates of the HR were extracted directly from the publications and variance of the estimates was calculated from published confidence intervals of the HR. For the TML study, unstratified as well as stratified HRs were reported. As the unstratified analysis was the primary analysis of this study, the unstratified estimates were included in the meta-analysis.
Heterogeneity of the individual HRs was tested using Cochran’s
Meta-analysis of the remission, progression, and rates of “tumor stabilization” was performed using the same statistical methods described above. In the remission analysis and in the progression analysis patients with unknown status were counted as progressive. Finally, in the “prevention of progression” (POP) analysis patients with unknown status were excluded.
ML-estimates of the odds ratio (OR) were determined from reported contingency tables and the within-trial variance was computed from the inverse of the matrix of second derivatives of the log-likelihood (Woolf’s formula).
The Medline search was done on June 15, 2016. It resulted in 3,252 articles. Hand-searches of conference proceedings were conducted including all conferences until June 2016.
Based on the criteria defined above the following eight studies were selected for inclusion into the current meta-analysis (PRISMA diagram; cf. Figure
PRISMA diagram.
Trial characteristics are depicted in Table
Overview of studies included in meta-analysis.
Study | First author/year | Investigational treatment | Control treatment | Primary endpoint | Number of patients (ITT) |
---|---|---|---|---|---|
TML | Bennouna 2012 [ |
Chemotherapy (oxaliplatin or irinotecan based) |
Chemotherapy (oxaliplatin or irinotecan based) alone | OS | 819 (409/410) |
CORRECT | Grothey 2013 [ |
Regorafenib | Placebo | OS | 760 (505/255) |
BEBYP | Masi 2013 [ |
Chemotherapy (FOLFIRI, mFOLFOX) |
Chemotherapy (FOLFIRI, mFOLFOX) alone | PFS | 184 (92/92) |
VELOUR subgroup | Allegra 2012 [ |
FOLFIRI plus aflibercept | FOLFIRI | OS | 373 (186/187) |
AGITG subgroup | Siu 2013 [ |
Cetuximab plus brivanib | Cetuximab | OS | 310 (152/158) |
FOSCO | Hoehler 2013 [ |
FOLFOX/FOLFIRI plus sorafenib | FOLFOX/FOLFIRI plus placebo | PFS | 69 (32/37) |
RAISE | Tabernero 2015 [ |
FOLFIRI plus ramucirumab | FOLFIRI plus placebo | PFS | 1072 (536/536) |
CONCUR | Li 2015 [ |
Regorafenib | Placebo | OS | 81 |
Table
Overview of number of studies and patients available for primary and secondary analyses.
Progression-free survival | Overall survival | |
---|---|---|
|
|
|
|
||
|
||
Gender (male/female) |
|
|
Age (cut-off 65 years) |
|
|
ECOG status (0 versus ≥1) |
|
|
KRAS status (WT/MUT) |
|
|
|
||
|
||
Remission rate |
|
|
Progression rate |
|
|
Prevention of progression rate |
|
The use of antiangiogenic drugs beyond progression improved progression-free survival and overall survival over control (i.e., best supportive care or active treatment alone). The hazard ratio for overall survival (
Meta-analysis for overall survival. CI: confidence interval; HR: hazard ratio.
Meta-analysis for progression-free survival. CI: Confidence interval; HR: hazard ratio.
The results of the subgroup analyses for PFS and OS according to sex, age, ECOG status, and KRAS status (WT versus MUT) are given in Table
Results of subgroup analyses for progression-free survival.
Subgroup |
|
HR | 95% CI | Test of heterogeneity |
Test of heterogeneity |
|
---|---|---|---|---|---|---|
Age | ||||||
<65 years | 1670 | 0.62 | 0.47–0.82 |
|
|
89.21% |
≥65 years | 1165 | 0.74 | 0.66–0.84 |
|
|
0% |
|
||||||
Sex | ||||||
Male | 1907 | 0.68 | 0.54–0.84 |
|
|
77.40% |
Female | 1110 | 0.66 | 0.49–0.89 |
|
|
80.51% |
|
||||||
ECOG status | ||||||
0 | 1440 | 0.61 | 0.48–0.80 |
|
|
81.64% |
≥1 | 1385 | 0.73 | 0.60–0.88 |
|
|
54.09% |
|
||||||
KRAS status | ||||||
Wildtype | 1219 | 0.60 | 0.47–0.76 |
|
|
76.62% |
Mutation | 1326 | 0.66 | 0.53–0.83 |
|
|
80.67% |
Results of subgroup analyses for overall survival.
Subgroup |
|
HR | 95% CI | Test of heterogeneity |
Test of heterogeneity |
|
---|---|---|---|---|---|---|
Age | ||||||
<65 years | 1578 | 0.80 | 0.71–0.90 |
|
|
0% |
≥65 years | 1073 | 0.84 | 0.73–0.97 |
|
|
0% |
|
||||||
Sex | ||||||
Male | 1604 | 0.82 | 0.73–0.92 |
|
|
48.59% |
Female | 1047 | 0.81 | 0.70–0.94 |
|
|
41.30% |
|
||||||
ECOG status | ||||||
0 | 1290 | 0.76 | 0.67–0.88 |
|
|
0% |
≥1 | 1351 | 0.86 | 0.76–0.97 |
|
|
0% |
|
||||||
KRAS status | ||||||
Wildtype | 1157 | 0.74 | 0.64–0.85 |
|
|
0% |
Mutation | 1260 | 0.89 | 0.78–1.02 |
|
|
0% |
PFS was improved both in studies using mAB targeting the VEGF-axis (
The use of antiangiogenic drugs beyond progression did not increase response rates (
Analysis of response rate. Number of patients achieving tumor remissions according to treatment.
Study name | Active treatment | Control | OR | 95%-CI | ||
---|---|---|---|---|---|---|
Remission | No remission | Remission | No remission | |||
TML | 22 | 382 | 16 | 390 | 1.40 |
|
CORRECT | 5 | 500 | 1 | 254 | 2.54 |
|
BEBYP | 19 | 73 | 16 | 76 | 1.24 |
|
VELOUR | 22 | 164 | 16 | 171 | 1.43 |
|
RAISE | 72 | 464 | 67 | 469 | 1.09 |
|
|
||||||
Overall | Test of heterogeneity: |
|
|
Analysis of tumor progression. Number of patients experiencing tumor progression according to treatment.
Study name | Active treatment | Control | OR | 95%-CI | ||
---|---|---|---|---|---|---|
Progression | No progression | Progression | No progression | |||
TML | 129 | 275 | 186 | 220 | 0.55 |
|
CORRECT | 298 | 207 | 217 | 38 | 0.25 |
|
BEBYP | 28 | 64 | 39 | 53 | 0.59 |
|
RAISE | 139 | 397 | 167 | 369 | 0.77 |
|
|
||||||
Overall | Test of heterogeneity: |
|
|
Analysis of tumor stabilization rate. Number of patients experiencing at least stable disease according to treatment.
Study name | Active treatment | Control | OR | 95%-CI | ||
---|---|---|---|---|---|---|
No progression | Progression | No progression | Progression | |||
TML | 275 | 87 | 220 | 142 | 2.04 |
|
CORRECT | 221 | 284 | 38 | 217 | 4.44 |
|
BEBYP | 64 | 27 | 53 | 37 | 1.65 |
|
RAISE | 397 | 87 | 369 | 134 | 1.66 |
|
|
||||||
Overall | Test of heterogeneity: |
2.25 |
|
Meta-analysis of remission rate. CI: confidence interval; OR: odds ratio.
Meta-analysis of progression rate. CI: confidence interval; OR: odds ratio.
Meta-analysis of prevention of progression rate. CI: confidence interval; OR: odds ratio.
Using aggregate data from eight randomized trials we found a clinically relevant and significant improvement of PFS and OS for the use of antiangiogenic drugs beyond progression with a cumulative hazard ratio of 0.64 for PFS and 0.83 for OS. The test for heterogeneity of study results was significant for both OS and PFS. The latter was mainly influenced by the results of the FOSCO trial for PFS and OS and additionally of the CORRECT trial for PFS. While the relative risk reduction for PFS was 35%, the survival benefit was 17% (HR 0.83). The main outlier in the OS analysis was the small FOSCO trial, which, however, had a relatively strong impact on the hazard ratio. Excluding FOSCO from the meta-analysis would result in an even lower hazard ratio underlining a significant benefit of continuing antiangiogenic drugs after first progression. FOSCO is one of a couple of studies investigating the addition of multityrosine kinase inhibitors to a chemotherapy doublet. All of these trials—regardless of the treatment setting, that is, 1st or 2nd line—have resulted in negative study results mainly caused by increased toxicity, decreased dose intensity, and compromised quality of life (e.g., CONFIRM or HORIZON studies). Thus, with the exception of FOSCO we found relatively homogeneous results regarding the patient-relevant endpoint of overall survival with a relative risk reduction for death of approximately 20% for continuation of antiangiogenic treatment beyond progression. Interestingly, the effects of different concepts (i.e., continuing bevacizumab beyond progression or switching to other antiangiogenic agents, for instance, aflibercept or ramucirumab) resulted in comparable hazard ratios for OS ranging between 0.77 and 0.86. Therefore, either approach might be considered for patients progressing after bevacizumab-based first-line therapy and the decision should be made in light of toxicity, patient preference, and drug approval status.
We also investigated the effects of continued antiangiogenic treatment in subgroups. The decision to carry out analyses stratified by age, ECOG status, and gender as well as KRAS mutational status was mainly taken due to the fact that other subgroups of interest (for instance, liver only metastases and time interval between last bevacizumab treatment) have not been reported by a sufficiently high number of trials to enable a meta-analysis. In all subgroup analyses an improved PFS was found. The HR of the respective dichotomized groups (age <65 versus ≥65 years; women versus men, ECOG 0 versus ≥1, and KRAS wildtype versus KRAS mutation) were comparable. Similarly, overall survival results are comparable for dichotomized subgroups regarding age and ECOG performance status. However, the effect of using antiangiogenic drugs beyond progression regarding OS was weaker in the subgroup of women (
No improvement in response rates was seen; however, the rate of progression was decreased. Additionally, we analyzed the potential to prevent tumor progression (defined as the number of patients with evaluable remission status achieving at least stable disease). In this analysis a clinically relevant benefit for the use of antiangiogenic drugs beyond progression was found. This is in line with earlier findings from the first-line setting, indicating that the activity of bevacizumab in combination with chemotherapy with respect to a prolongation of PFS is predominantly driven by disease stabilization [
In all, the current meta-analysis demonstrates the usefulness of continued antiangiogenic drugs beyond progression regarding their potential to improve PFS and OS in a clinically meaningful manner. The benefit was seen in the subgroups stratified by age, gender, and ECOG performance status.
Limitations of the current analysis are that no individual patient data were used and toxicity could not be assessed. Furthermore, although the antiangiogenic activity of the drugs included in the analysis may differ, we decided to include studies with multityrosine kinase inhibitors such as regorafenib as well, because these drugs are believed to exert their main activity via antiangiogenic mechanisms. Moreover, the size of some subgroups assessed in our analyses was small. Therefore, statistical power of single analyses might have been too low to show a significant difference in these subgroups, namely, women and KRAS-mutant tumors. Our meta-analysis, synthesizing data from several trials, indicates that using antiangiogenic drugs beyond and after progression can meanwhile be regarded as an established strategy in the treatment of patients with metastatic colorectal cancer. Future research should especially focus on the optimal sequence of using these antiangiogenic drugs, for instance, the timing of the switch from bevacizumab to broader active drugs such as aflibercept or regorafenib. Clearly, in this aspect biomarkers will be needed to elucidate if inhibition of alternative angiogenic pathways or additional tyrosine kinases would be required for continued antiangiogenic activity or if the patient could remain on bevacizumab treatment.
The authors have declared no competing interests.
All authors made contributions to the acquisition and interpretation of data and critical revision of the manuscript for intellectual content and approved the final version for publication. I. Burkholder performed the statistical analysis. R.-D. Hofheinz and U. Ronellenfitsch contributed equally.
The authors are indebted to Professor Dr. Thomas Hoehler, Recklinghausen, for proving data from the clinical trial report of the FOSCO trial.