Benefit of taxanes as adjuvant chemotherapy for early breast cancer

Pooled analysis of 15,500 patients

Authors


  • Presented as a Poster Discussion at the European Cancer Conference (ECCO), Paris, France, October 30th to November 3rd, 2005.

Abstract

BACKGROUND

The magnitude of the survival benefit of taxanes as adjuvant chemotherapy for early breast cancer is still unclear. A pooled analysis of Phase III trials was performed to assess the advantages that adjuvant taxane chemotherapy has over standard chemotherapy.

METHODS

All Phase III trials were considered eligible. A pooled analysis was accomplished and event-based relative risk ratios (RR) with 95% confidence intervals (95% CI) were derived. The significant differences in disease-free survival (DFS) and overall survival (OS) were explored. Magnitude outcome measures were absolute benefits and the number of patients needed to treat. A heterogeneity test was applied as well. A sensitivity analysis in 6 subpopulations was also performed.

RESULTS

Nine trials designed to assess if paclitaxel or docetaxel improve survival (15,598 patients) were gathered. One of the 9 trials did not report OS results. Significant differences in favor of taxanes were seen in DFS in the overall (RR: 0.86; 95% CI, 0.81–0.90 [P<.00001]) and lymph node-positive population (RR: 0.84; 95% CI, 0.79–0.89 [P<.0001]), and in OS in the overall (RR: 0.87; 95% CI, 0.81–0.83 [P<.0001]) and lymph node-positive population (RR: 0.84; 95% CI, 0.77–0.92 [P<.0001]). The absolute benefits in DFS and OS in favor of taxanes ranged from 3.3% to 4.6% and from 2.0% to 2.8%, respectively.

CONCLUSIONS

Considering all the available Phase III trials, taxane-based adjuvant chemotherapy for early breast cancer seems to add a significant benefit in both DFS and OS over standard chemotherapy. The lack of significant heterogeneity in the sensitivity analysis underscores the homogeneous effect across all trials. Cancer 2006. © 2006 American Cancer Society.

Breast carcinoma is the second most frequent cause of cancer-related death for women in the U.S.1 In the firstline treatment of patients affected by metastatic breast carcinoma (MBC), polychemotherapy produces advantages in terms of response rate and median time to disease progression (TTP) over single-agent chemotherapy.2 Anthracyclines (doxorubicin and epirubicin) are an important treatment, regardless of other drug(s) added to a combination regimen. Furthermore, anthracycline combination regimens improve the relative risk and TTP compared with nonanthracycline combination regimens.2

In firstline chemotherapy, paclitaxel and docetaxel in combination with anthracyclines improve the overall response rate and time to progression when compared with anthracycline-based chemotherapy.3, 4 The main limitation of the anthracycline/taxane combination is hematologic toxicity.3

The magnitude of the benefit of taxanes as part of adjuvant chemotherapy for early breast cancer after surgery is still unclear.

Nine recent clinical trials were conducted to compare standard adjuvant chemotherapy with anthracyclines versus a concomitant or sequential regimen with taxanes in patients with nodal-involved or high-risk node-free breast cancer treated surgically.5–13 Although the benefit in disease-free survival (DFS) was present in nearly all trials, the eventual benefit in overall survival (OS) was unclear because of small trial cohorts and, consequently, inadequate statistical power. Nevertheless, 3 of these trials showed a significant benefit in both DFS and OS.6, 11, 12

In today's evidence-based medicine, the strongest contribution to care changes is provided by recommendations from large randomized clinical trials (RCTs) or metaanalyses.14

Although conflicting results are not provided regarding the improvement in DFS produced by the adjunct use of taxanes, a metaanalysis would better help to explore the eventual benefit in OS and to explore all survival outcomes in different subpopulations or strategies (i.e., sequential or concomitant).

In addition to some degree of uncertainty concerning benefit in DFS in some subgroups and, above all, in OS, is the question of the magnitude of such benefit, especially if balanced with complexity of administration, toxicity of the regimen, and economic cost.

To clarify how much taxanes improve outcome over standard treatment, we performed a pooled analysis of all published or presented Phase III trials.

MATERIALS AND METHODS

Our comprehensive analysis was conducted after 4 steps: definition of the outcomes (definition of the question the analysis was designed to answer), definition of the criteria applied for the selection of the eligible trials, definition of the search strategy, and a detailed description of the statistical method used.3, 15

Outcome Definition

We considered taxane-based adjuvant chemotherapy arms as experimental and control chemotherapy as standard comparator arms. Analysis was conducted in order to assess eventual significant differences in survival outcomes.

The primary outcome for the magnitude of eventual benefit analysis was the DFS defined as the time expressed in months between randomization and appearance of recurrence (local or distant or both) or death for any cause; the secondary endpoint was OS defined as the time expressed in months between randomization and death for any cause.

For both outcomes 6 subgroups of interest were identified for examination, helping to decrease biases and differences between trials (sensitivity analysis): 1) overall population; 2) node-positive population (considering only trials enrolling patients with nodal disease involvement); 3) sequential population (considering only trials in which taxanes were given sequentially after standard comparator chemotherapy); 4) concomitant population (considering only trials where taxanes were concurrently administered); 5) paclitaxel population (considering only trials with paclitaxel); and 6) docetaxel population (considering only trials with docetaxel).

Search Strategy

The deadline for a trial publication to be eligible for this analysis was October 2005. Eventual update of Randomized Clinical Trials (RCTs) was gathered through Medline (PubMed, available at URL: www.ncbi.nlm.nih.gov/PubMed), American Society of Clinical Oncology (ASCO, available at URL: www.asco.org), European Society for Medical Oncology (ESMO, available at URL: www.esmo.org), Federation of European Cancer Societies (FECS, available at URL: www.fecs.be), or San Antonio Breast Cancer Symposium (SABCS, available at URL: www.sabcs.org) website searches were performed until October 31, 2005. Key words used were: adjuvant chemotherapy, paclitaxel, docetaxel, taxanes, early, breast cancer, review, metanalysis, meta-analysis, pooled analysis, randomized, Phase III, comprehensive review, systematic review. In addition to computer browsing, review and original articles were also scanned in the references chapter to look for missing trials. Furthermore, lectures of major meetings (ASCO, ESMO, ECCO, and SABCS) having “adjuvant chemotherapy for breast cancer” or “early breast cancer” as topics were checked. No language restrictions were applied.

Trial Identification Criteria

We gathered all Phase III prospective and randomized trials, published as formal articles in peer-reviewed journals or presented at the ASCO, ECCO, ESMO, or SABCS meetings until October 2005, in which previously untreated patients had undergone curative surgical resection for early breast cancer and were subsequently randomized to receive chemotherapy either with or without sequential or concomitant taxanes (paclitaxel or docetaxel).

All trials enrolling node-positive or node-negative and positive early breast cancer patients were considered eligible for this metaanalysis (i.e., trials enrolling only node-free patients were not included).

Data Extraction

From each arm within each trial the number of death and recurrence (local of distant) events were obtained. All data were reviewed and separately computed by 2 different independent investigators.

Data Synthesis

The log of relative risk ratio (RR) was estimated for each considered endpoint and 95% confidence intervals (95% CI) were derived.16 By this method, it was possible to apply a fixed-effect model as well as a random-effect model according to the inverse variance and the Mantel–Haenszel method. To test for heterogeneity between trials the Q statistic was used. Patients were accounted and filled in the 2 × 2 tables using the intention-to-treat (ITT) assignment where applicable. The significance of heterogeneity test suggests preferring the random-effect estimation for a more appropriate evaluation of the results. Absolute benefits for each outcome were calculated (i.e., absolute benefit = exp {HR × log[control survival]} − control survival).17 The number of patients needed to treat for a single beneficial patient was determined (NNT: 1/[(Absolute Benefit)/100]).

The results are depicted in all figures as conventional-metaanalysis forest plots. The pooled analysis calculations were accomplished using the Comprehensive Meta-Analysis Software, v. 1.0.23 (CMA, Biostat, Englewood, NJ).3

RESULTS

Selected Trials

Nine trials (15,598 patients) were gathered.5–13 All were specifically designed to assess if chemotherapy with taxanes (paclitaxel or docetaxel) improved survival. One of the 9 trials did not report OS results. Selected trial features, single populations, and regimens are listed in Table 1.

Table 1. Selected Randomized Clinical Trials
DrugRCTs (Reference)PatientsMedian Follow-Up, MonthsArmsDFS (%)OS (%)
  • RCTs: randomized controlled trials; DFS: disease-free survival; OS: overall survival; MDACC: M. D. Anderson Cancer Center: F: 5-fluorouracil; A: doxorubicin; C: cyclophosphamide; P: paclitaxel; NR: not reported; CALGB: Cancer and Leukemia Group B; NSABP: National Surgical Adjuvant Breast Project; HeCOG: Hellenic Cooperative Oncology Group; E: epirubicin; P: paclitaxel; D: docetaxel; ECTO: European Cooperative Trials in Operable Breast Cancer; PACS: Programme Adjuvant Cancer du sein; BCIRG: Breast Cancer Interantinal Research Group; ECOG: Eastern Cooperative Oncology Group.

  • *

    Significant difference.

PaclitaxelMDACC5259608FAC83NR
2654P+4FAC86
CALGB1551694AC6568
9344615704AC+4P70*74*
NSABP B-2871528654AC7285
15314AC+4P76*85
HeCOG8297624E+4CMF6881
2984E+4P+4CMF7084
ECTO9453604A+4CMF7387
4514AP+4CMF80*91
DocetaxelUS oncology10510434AC8791
5064DC9194
PACS 01111003606FEC73.286.7
9963FEC+3D78.3*90.7*
 BCIRG 00112745556FAC6881
  7466DAC75*87*
 ECOG 2197131441594AC8793
  14444AD8794

Combined Analysis

All obtained results are displayed in Tables 2 and 3 and Figures 1–4. RRs are to be interpreted as follows: RRs <1.0 favor the taxane-based chemotherapy group, whereas RRs >1.0 favor the standard arm. The sensitivity analyses performed in all 6 subpopulations accrued different patient sample sizes as shown in Tables 2 and 3.

Table 2. Disease-Free Survival-Sensitivity Analyses*
SubpopulationPatients (No. Randomized to Clinical Trials)RR (95% CI)PHeterogeneity test (P)Absolute Benefit (%)NNT
  • RR: relative risk; 95% CI: 95% confidence intervals; NNT: number needed to treat for 1 patient to benefit.

  • *

    Fixed-effect model.

Overall15,598 (9)0.86 (0.81–0.90)<.00001.283.330
Lymph node- positive9670 (4)0.84 (0.79–0.91)<.00001.574.323
Sequential9298 (5)0.87 (0.82–0.92).00001.433.628
Concomitant6300 (4)0.80 (0.72–0.89)<.0001.253.429
Paclitaxel8203 (5)0.87 (0.81–0.93)<.0001.213.628
Docetaxel7395 (4)0.83 (0.75–0.91)<.0001.393.231
Table 3. Overall Survival - Sensitivity Analyses*
SubpopulationPatients (No. Randomized to Clinical Trials)RR (95% CI)PHeterogeneity test (P)Absolute Benefit (%)NNT
  • RR: relative risk; 95% CI: 95% confidence intervals; NNT: number needed to treat for one patient to benefit.

  • *

    Fixed-effect model.

Overall15,074 (8)0.87 (0.81–0.93)<.0001.292.049
Lymph node-positive9670 (4)0.84 (0.77–0.92)<.0001.152.836
Sequential8774 (4)0.89 (0.82–0.96).003.212.147
Concomitant6300 (4)0.80 (0.68–0.93).004.472.050
Paclitaxel7679 (4)0.90 (0.83–0.98).013.382.050
Docetaxel7395 (4)0.78 (0.68–0.90).001.452.442
Figure 1.

Disease-free survival in the overall population (relative risk plot). Citation: trial [reference number]; Year: publication/presentation year; NTotal: number of patients; fixed combined: fixed effect model; random combined: random effect model.

When the data were pooled and plotted, significant differences in favor of taxanes were seen in DFS in the overall population (RR: 0.86; 95% CI, 0.81–0.90 [P<.00001]) and the lymph node-positive group (RR: 0.84; 95% CI, 0.79–0.89 [P<.0001]) (Table 2) (Figs. 1 and 2). Heterogeneity was never significant, suggesting a homogeneous trend in favor of taxanes across all trials and subgroups (Table 2). Even including the Greek trial8 in the lymph node-positive population (which enrolled 98% of lymph node-positive patients), DFS was improved in the taxane arm (10,265 patients in 5 trials; RR: 0.86; 95% CI, 0.82–0.91 [P<.0001]; heterogeneity, P = .24).

Figure 2.

Disease-free survival in the lymph node-positive population (relative risk plot). Citation trial; [reference number]; Year: publication/presentation year; NTotal: number of patients; fixed combined: fixed effect model; random combined: random effect model.

Moreover, significant differences in favor of taxanes were seen in OS in the overall population (15,074 patients; RR: 0.87; 95% CI, 0.81–0.93 [P<.0001]) and in the lymph node-positive group (RR: 0.84; 95% CI, 0.77–0.91 [P<.0001]) as well (Table 3) (Figs. 3 and 4). Again, heterogeneity was never significant.

Figure 3.

Overall survival in the overall population (relative risk plot). Citation: trial [reference number]; Year: publication/presentation year; NTotal: number of patients; fixed combined: fixed effect model; random combined: random effect model.

Figure 4.

Overall survival in the lymph node-positive population (relative risk plot). Citation: trial [reference number]; Year: publication/presentation year; NTotal: number of patients; fixed combined: fixed effect model; random combined: random effect model.

Including the Greek trial8 in the lymph node-positive population as well, OS was improved in the taxane arm (10,265 patients in 5 trials; RR: 0.87; 95% CI, 0.81–0.93 [P = .0003]; heterogeneity, P = .10).

DFS and OS were always improved whatever the strategy, either sequential or concomitant, or drug, either paclitaxel or docetaxel.

With regard to our major objectives (i.e., the magnitude of benefit), the absolute benefit in favor of chemotherapy ranged from 3.2% to 4.6% for DFS, and from 2.0% to 2.8% for OS (Tables 2 and 3). The number of patients needed to treat for a single beneficial patient ranged from 23 to 31 for DFS and from 36 to 50 for OS (Tables 2 and 3).

All the results obtained with the Mantel–Haenszel estimation were comparable with regard to effect, intervals, significance, and heterogeneity.

DISCUSSION

The introduction of taxanes (paclitaxel and docetaxel) in breast cancer treatment has produced significant improvement in the management of such disease in the advanced/metastatic setting. The adjunction of these drugs as part of combination chemotherapy for both first-line and second-line chemotherapy for metastatic disease has become a new standard of care, thanks to the increase in response rate (both partial and complete) and time to progression,3, 4 which has to be considered the most realistic endpoint in the firstline setting,18 and in OS in the secondline setting.19, 20

Moreover, the significant increase in complete response has suggested the introduction of these agents into an earlier setting of treatment such as neoadjuvant chemotherapy; indeed, the link between pathologic complete response and OS improvement is the emerging issue.21, 22 Therefore, the more attractive counterpart of breast cancer treatment, which is the adjuvant approach, has become a challenging field for paclitaxel and docetaxel. The recent update of the Early Breast Cancer Trialists' Collaborative Group (EBCTCG) individual patient data (IPD) metaanalysis confirmed the milestone role of anthracyclines as adjuvant chemotherapy for early breast cancer.23, 24

For all these perspectives, several RCTs have been conducted to explore the role of taxanes in adjuvant chemotherapy for early breast cancer. Most of those trials have demonstrated the benefit in terms of DFS (and in some cases in OS) in favor of taxane adjunction. However, concerns exist regarding the suboptimal control arm in some trials.6, 7, 10, 13

To summarize and quantify the magnitude of the benefit of taxanes as adjuvant chemotherapy for breast cancer, we performed a pooled analysis based on published of presented data. Both formal articles and abstracts were included, based on literature searches, allowing results to be gathered from all studies meeting inclusion criteria. This may also help in avoiding publication biases related to the arbitrary exclusion of not-published trials producing negative results. This low-cost and short-duration approach allows for the rapid inclusion of results, especially important when the value of treatment in an important clinical setting is uncertain.

Our results show that, whatever population is explored out of the 6 explored specific sample sizes in the sensitivity analysis, the adjunction of taxanes significantly improves both DFS and OS. In particular, the absolute gain in DFS and OS ranged from 3.3% to 4.6% and from 2.0% to 2.8%, respectively (Tables 2 and 3) (Figs. 1–4). A statistically homogeneous trend across all populations was found; this phenomenon stresses the always positive behavior of taxanes in these trials. Even considering the most “clean” population (i.e., the lymph node-positive sample size), the absolute gains in DFS and OS were 4.3% and 2.8%, respectively.

The benefit is present even comparing whatever the sequential or concomitant approach, which strongly opens the issue regarding the toxicity analysis. The differential subanalysis exploring the 2 different taxanes demonstrated a benefit for both chemotherapeutics, even if a strong benefit for docetaxel in OS is suggested, although obtained in a smaller and perhaps not highly powered population.

These absolute improvements are relatively similar to those produced by the anthracyclines adjunction in the EBCTCG metaanalysis, which allowed their introduction as a gold standard in the adjuvant treatment of breast cancer.23, 24

Conversely, the obtained absolute benefits determine the number of patients needed to treat for a single beneficial patient ranging from 23 to 31 for DFS and from 36 to 50 for OS. This is obviously a fundamental issue, which needs to considered for all the pros and drawbacks of such experimental treatment.

The gain in survival is crucial for any adjuvant treatment goal, which translates, when significant, into saved lives. Unfortunately, most of the included randomized trials did not report detailed results regarding toxicities and costs; the same metaanalysis is needed for at least febrile neutropenia and toxic deaths to better balance the obtained results with the survival improvement.

Although metaanalyses based on individual patient data are considered to be more rigorous, controversy continues as to whether large inclusive pooled analyses should be considered definitive or to be better for hypothesis generation.25

The sensitivity analysis performed to decrease biases, which actually produces positive results regardless of the subpopulation, and the lack of significant heterogeneity do strengthen the benefit provided by adjuvant taxanes. Furthermore, one of the most controversial variables in pooling results, which is the difference in the single trial median follow-up data, did not significantly differ between all gathered studies (about 50–60 months) (Table 1).

Moreover, our pooled analysis actually has enough statistical power, because if we designed a randomized clinical trial in order to obtain an absolute benefit in favor of taxanes of 3.3% (starting from the obtained control rate of 72.8%), we would need approximately 3000 patients. We actually gathered data on >15,500 patients, so we consider our results reliable. Actually, the same power issue is in favor of the lymph node-positive sample size (theoretical and actual size 2000 and 9670 patients, respectively).

Some issues need to be explored more extensively: as suggested by the EBCTCG metaanalysis, the benefit of chemotherapy is smaller in hormonal receptor-negative patients.24 An imbalance between collected trials on hormonal receptor status is present, and a subgroup analysis adjusted for this finding is mandatory, because the magnitude of the benefit could be lower in a hormonal receptor-negative population. Indeed, when analyzing in detail the results of each single gathered trial, some differences are present: whereas Mamounas et al.7 and Martin et al.12 did not show clear differences between hormonal receptor-positive and -negative patients, Henderson et al.6 reported a multivariate analysis showing less efficacy for paclitaxel adjunction for patients expressing hormonal receptors or receiving tamoxifen. In the same direction, the trial presented by Goldstein et al.13 suggested a favorable trend (although not significant for sample size) for docetaxel in patients with progesterone-negative receptor, whereas a negative trend for hormonal receptor-positive patients was observed. The trial presented by Roche et al.11 did not report this specific outcome, but this subanalysis could be misinterpreted if a small but significant difference between the 2 arms for hormonal receptor-positive and -negative patient characteristics existed.11

Conversely, the overexpression of HER-2 could enhance the magnitude of the benefit provided by adding taxanes, as suggested in metastatic disease.26 Unfortunately, data are lacking across all randomized trials regarding the eventual link between survival outcome and hormonal receptor-negative status and HER-2 overexpression; this last finding opens attractive perspectives regarding the combination with trastuzumab.27–29

With >15,500 patients enrolled in randomized clinical trials showing positive data in favor of taxanes, and considering our combined results consistent with all studies and demonstrating advantages in both DFS and OS in all subpopulations, the combination of standard chemotherapy plus paclitaxel or docetaxel (either sequential or concurrent) is beneficial for patients undergoing surgery for early breast cancer, and should be considered one of the standard options in this setting.

Whereas a consistent benefit was observed, the magnitude of the benefit was not large: overall, an approximately 2% to 4% absolute benefit, and approximately 35 to 45 patients need to be treated for 1 to benefit. Both the fact of survival benefit and the magnitude of that benefit should be balanced with toxicity and costs, and, moreover, discussed in clear terms with patients when considering adding taxanes to postoperative chemotherapy.

Ancillary