Abstract
A recent survey analysis suggests an increased reluctance amongst oncologists to employ adjuvant therapy for breast cancer with advancing age [1]. Although there are numerous potential explanations for this observation, one might surmize that the poor representation of older adults in relevant data sets contributes to this trend [2]. In fact, to date, only two prospective, Phase III trials evaluating the role of adjuvant therapy in older adults have been reported. The first, French Adjuvant Study Group (FASG) 08, randomized patients over the age of 65 years with node-positive, operable breast cancer to either tamoxifen alone for 3 years or tamoxifen for the same duration with six cycles of epirubicin-based chemotherapy [3]. The trial identified a decreased risk of relapse with the addition of epirubicin (hazard ratio [HR]: 1.93; 95% CI: 1.70–2.17) with acceptable toxicity. Supporting this data, several large meta-analyses and Surveillance, Epidemiology and End Results (SEER) registry studies have concluded that age alone should not serve as a contraindication to receipt of adjuvant chemotherapy [4–6].
More recently, Muss et al. have compared standard adjuvant chemotherapy (doxorubicin [trade name: adriamycin]-cyclophosphamide [AC] or cyclophosphamide-methotrexate-fluorouracil [CMF]) to an oral, single-agent regimen (capecitabine) in women 65 years of age or greater with stage I–IIIB breast cancer (Cancer and Leukemia Group B [CALGB] trial 49907) [7]. Supporting this comparison, a randomized, Phase II trial comparing capecitabine and CMF in the metastatic setting yielded similar survival with the two regimens [8]. Furthermore, AC and CMF appear to have similar efficacy in Phase III studies [9,10]. In the current trial, four cycles of AC and six cycles of CMF were administered at standard doses [11,12]. Patients randomized to capecitabine received six cycles of therapy, although dosing was modified through the course of the trial. In an initial cohort of 56 patients, capecitabine was dosed at 2000 mg/m2 in two divided doses for 14 days every 3 weeks. This dose was escalated to 2500 mg/m2 on the same schedule if no adverse events were encountered during the first cycle of therapy. Owing to unacceptable toxicity, this dose escalation was ultimately abandoned. Interestingly, a similar phenomenon was observed in a study conducted by the National Cancer Institute of Milan, Italy, albeit in the setting of metastatic disease [13]. In this Phase II effort, patients over the age of 65 years with stage IV breast cancer received capecitabine dosed at 2500 mg/m2. In light of two treatment-related deaths observed in 30 patients enrolled at this dose, the trial was amended to examine a dose of 2000 mg/m2 – ultimately, capecitabine appeared to be safe and efficacious with this modified regimen.
Employing a Bayesian statistical design, CALGB 49907 was projected to include between 600 and 1800 patients with a primary end point of relapse-free survival. After enrollment of 600 patients, capecitabine reached a prespecified likelihood for inferiority compared with standard chemotherapy. Accrual was suspended subsequent to this analysis, with a total of 633 patients enrolled. Baseline characteristics of patients randomized to standard chemotherapy (n = 326) and capecitabine (n = 307) were well balanced with respect to age, performance status, race and most clinicopathologic characteristics; however, a slight preponderance of patients with tumors of 2 cm or less were randomized to receive standard chemotherapy rather than capecitabine (49 vs 39%, respectively; p = 0.04). Approximately 60% of patients included in the trial were between the age of 70 and 79 years, and the vast majority (>95%) demonstrated an Eastern Cooperative Oncology Group (ECOG) performance status between 0 and 1. With respect to biologic characteristics, HER2-positivity was noted in approximately 10% of the study population, mirroring frequencies observed in previous SEER database analyses [14]. It is noteworthy that the frequency of hormone-receptor positivity (67%) in this older population was lower than anticipated by previous observational studies (in excess of 80%) [14].
With a median follow-up of 2.4 years, both the risk of relapse (HR: 2.09; 95% CI: 1.38–3.17; p < 0.001) and the risk of death (HR: 1.85; 95% CI: 1.11–3.08; p = 0.02) were significantly higher with capecitabine compared with standard chemotherapy. Multivariate analysis further demonstrated a higher risk of relapse in patients with a greater degree of lymph node involvement (four vs one positive lymph nodes), larger tumor size (5 vs 2 cm) and negative hormone-receptor status. A thought-provoking exploratory analysis stratified patients by both treatment arm and hormone-receptor status. Significant interaction between these variables was found for both progression-free survival and overall survival. A pronounced difference in benefit from the two treatments was observed in estrogen receptor-/progesterone receptor-negative patients compared with other subsets, with capecitabine quadrupling the risk of relapse (HR: 4.39; 95% CI: 2.9–6.7; p < 0.001) and tripling the risk of death (HR: 3.76; 95% CI: 2.23–6.64; p < 0.001) relative to standard chemotherapy. These results may be reflective of an earlier relapse pattern with hormone receptor-negative breast cancer, an observation documented in large experiences, such as the Oxford Overview [15]. Nonetheless, caution should be exercised in interpreting this data given the post hoc nature of the analysis and the relatively brief follow-up reported thus far.
It is noteworthy that efficacy results for AC and CMF in CALGB 49907 have not been separately reported. Although the authors allude to a lack of qualitative differences in these data, a more detailed understanding may provide insight into the relative benefit of anthracycline-based chemotherapy in this population. At present, data related to the frequency of anthracycline-induced cardiotoxicity in older adults is derived from retrospective experiences. In a SEER–Medicare database review examining nonmetastatic breast cancer patients between the ages of 66 and 90 years, the incidence of congestive heart failure was higher amongst recipients of anthracycline-based adjuvant therapy compared with recipients of nonanthracycline-based chemotherapy (HR: 1.45; 95% CI: 1.19–1.76) [16]. Considering this potential risk, attention has been focused on exploring nonanthracycline regimens in older adults. The US Oncology trial 9735 demonstrated superior overall survival with four cycles of docetaxel-cyclophosphamide compared with four cycles of AC as adjuvant therapy of stage I–III breast cancer (87 vs 82% at 7 years median follow-up; p = 0.032) [17]. An exploratory analysis suggested comparable benefit from this regimen in patients over the age of 65 years compared with younger patients.
Differences in toxicity were observed in each of the treatment arms in CALGB 49907. Grade 3–4 hematologic toxicity was more frequent with CMF (52%) and AC (54%) compared with capecitabine (2%). Grade 3–4 nonhematologic toxicity events occurred in 41, 25 and 33% of patients receiving CMF, AC and capecitabine, respectively. Importantly, two deaths (due to colitis and infection in separate cases) were attributed to capecitabine monotherapy. Although the increased rates of hematologic toxicity observed with standard chemotherapy are of concern, when taken together this consideration is counter-balanced by the treatment-related mortality associated with capecitabine. Furthermore, it is unclear whether more judicious use of growth factor support could have mitigated the hematologic toxicity (primarily neutropenia and leukopenia) observed with AC and CMF. Emerging evidence suggests that an age of over 65 years represents a significant risk factor for chemotherapy-induced neutropenia, and various guidelines have suggested colony-stimulating factor use as a standard in this demographic [18,19].
Ultimately, 62, 92 and 80% of patients receiving CMF, AC and capecitabine, respectively, completed all the prescribed cycles of therapy. While this is informative for standard intravenous chemotherapy, these data do not account for deviations in compliance with oral capecitabine. This issue is specifically addressed by a companion study, CALGB 60104 [20], which involved a total of 161 patients who were randomized to capecitabine. Each patient was monitored with microelectronic devices adherent to the caps of capecitabine pill bottles, designed to count the number of times the bottle was opened. Using this technique, it was estimated that 24% of study participants took less than 80% of the expected doses – these patients were thus categorized as nonadherent. Several other studies of oral anticancer therapy point to poor compliance amongst older adults, and further suggest that compliance is impaired with longer durations of therapy. For example, an age-stratified assessment of patients allocated to receive 5 years of letrozole in the Breast International Group (BIG) trial 1–98 suggested that 38.4% of patients aged 75 years or older did not complete the planned duration of therapy, in contrast to 22.6% of patients younger than 75 years of age (p < 0.0001) [21].
In summary, Muss et al. are to be applauded for the completion of an ambitious and innovative trial. A widening gap in survival has been observed between middle-aged and older adults with cancer, with evidence suggesting that the phenomenon is due, in part, to undertreatment of the latter group [22,23]. CALGB 49907 provides substantial evidence to support the use of standard chemotherapy (as opposed to less aggressive regimens) in older adults with early-stage breast cancer.
Future perspective
With data from CALGB 49907 in mind, investigators are now charged with several tasks. Deciphering the benefit associated with adjuvant anthracycline-based chemotherapy in older adults remains a relevant clinical question; already, prospective efforts exploring taxane-based therapy in this setting are underway. Developing novel metrics to assess the older adult is also of critical importance – it is now well recognized that chronologic age and physiologic age may have substantial discordance [24]. Finally, it is crucial not to overlook the biology of breast cancer in older adults. Although there are distinctions in the frequency of hormone-receptor and HER2 status in age-based cohorts, molecular assays incorporating a wide panel of genetic markers (i.e., the 21-gene recurrence score) have a predictive capability that is independent of age [14,25]. When considered together, these represent challenging, broad-based goals for the oncology community. Nonetheless, Muss et al. have demonstrated the feasibility of prospective analyses in older adults with breast cancer – this data set can represent both a paradigm and platform for future work.
Executive summary
To date, only two trials examining adjuvant therapy for breast cancer in older adults have been published (the French Adjuvant Study Group [FASG] 08 and the Cancer and Leukemia Group B [CALGB] 49907).
CALGB 49907 randomized patients to standard chemotherapy (doxorubicin [trade name: adriamycin]-cyclophosphamide or cyclophosphamide-methotrexate-fluorouracil) or an oral regimen (capecitabine).
The study employed a novel Bayesian design, and ultimately enrolled 633 patients.
The risk of relapse (hazard ratio: 2.09; 95% CI: 1.38–3.17; p < 0.001) was significantly higher with capecitabine compared with standard chemotherapy.
The risk of death (hazard ratio: 1.85; 95% CI: 1.11–3.08; p = 0.02) was significantly higher with capecitabine compared with standard chemotherapy.
Investigators must now determine the role of anthracycline-based regimens in older adults.
Investigators are also challenged to juxtapose physiologic age against chronologic age in future trials of adjuvant therapy.
Footnotes
Sumanta Pal's efforts are supported by CBCRP 15IB-0140 (California Breast Cancer Research Program Junior IDEA Award). The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.
No writing assistance was utilized in the production of this manuscript.