• © 2005 by American Society of Clinical Oncology

Chemotherapy and Radiation Therapy for Breast Cancer: What Is the Optimal Sequence?

The optimal integration of systemic and local therapy in the treatment of early-stage breast cancer has been controversial. Unlike the treatment for many other cancers, chemotherapy and radiation therapy (RT) are typically not given concurrently in patients with breast cancer because of the widespread use of anthracycline-based chemotherapy regimens and the concern for excessive radiation toxicity with concurrent treatment. Hence it is necessary to decide how best to sequence systemic and radiation therapies. This question arises for both patients treated with lumpectomy and those treated with mastectomy.

In patients treated with lumpectomy, there is a concern that delaying the initiation of RT until after chemotherapy might lead to higher rates of local (breast) recurrence (LR). Theoretically, delays in RT may allow progression of microscopic disease beyond that which can be treated with standard RT doses. This concern is heightened by the use of more prolonged chemotherapy regimens, such as the addition of the taxanes to doxorubicin and cyclophosphamide (AC). Several retrospective studies have found higher rates of LR in women for whom RT is delayed,^1-3 but this finding has not been consistent across studies.^4-6 A recent abstract by Pierce and co-investigators⁷ from Southwest Oncology Group and Eastern Cooperative Oncology Group reported a retrospective analysis of the 10-year results from four randomized trials involving chemotherapy and RT in patients treated with lumpectomy. The sequencing was determined by the treating oncologists. There were no differences seen in the rates of local, regional, or distant recurrence based on sequence. Although such data are somewhat reassuring, biases in treatment selection confound the interpretation and generalizability of these retrospective reviews.

The only prospective randomized trial designed to study the sequencing of chemotherapy (doxorubicin, cyclophosphamide, methotrexate, fluorouracil, prednisone) and RT was performed by our group. It was first reported in 1996⁸ and later updated in 2001.⁹ Because only 244 patients were randomly assigned, the trial was underpowered to detect small but potentially important differences in outcome. At 10 years, the updated report showed no statistically significant differences in overall patterns of first failure or event-free survival. As noted in the 5-year report, there were more LRs in the arm of patients who received chemotherapy first and more distant recurrences in the arm of patients who received RT first as sites of first failure; however, the distribution of sites of first failure between the two arms was no longer significant. In a regression model that examined interactions, there was a statistically significant interaction between treatment sequence and margin status. In this subset analysis by margin status, patients with close margins had a low LR rate in the RT-first arm, compared with a substantially higher rate when RT was delayed. Patients with positive margins had high rates of LR, independent of sequence, and patients with negative margins had low rates of LR with either sequence. This subset analysis requires confirmation by other studies. However, it suggests the hypothesis that close margins indicate the possibility of a greater residual tumor burden,¹⁰ a risk factor for LR that is compounded by any delay in initiating RT.

Concerns about the delay in RT must be balanced against the beneficial effects of chemotherapy on local control when combined with RT either given concurrently or in sequence. In the National Surgical Adjuvant Breast and Bowel Project (NSABP) B-13¹¹ randomized trial, 760 estrogen receptor–negative and node-negative patients were randomly assigned to methotrexate and fluorouracil (MF) or to no adjuvant chemotherapy. For patients treated with lumpectomy, RT was given concurrently with MF. At 8 years, the cumulative LR rate among women treated with lumpectomy and adjuvant MF chemotherapy was 2.6%, compared with 13.4% among women who received RT alone (P = .001). Similarly, NSABP-19¹¹ compared MF with cyclophosphamide plus MF (CMF) in estrogen receptor–negative and node-negative patients. Again, for patients treated with lumpectomy, RT was given concurrently both with MF and CMF. Among patients treated with lumpectomy, the 5-year cumulative LR rate was 5.6% for MF patients, compared with 0.6% for CMF patients. Thus there was a suggestion of an even further decrease in local recurrence with the inclusion of cyclophosphamide. Similarly, we reported the results of a prospective single-arm trial of concurrent CMF and reduced-dose RT in a group of 112 women with early-stage breast cancer.¹² The crude 5-year rate of LR was only 4%, including only one LR in 25 patients with positive margins and one LR in 16 patients with close margins. The positive interaction between chemotherapy and RT resulting in decreased LR has also been seen in retrospective studies. A retrospective study from our group using polychotomous logistic regression¹³ on 533 patients treated with conservative surgery and RT showed that only margin status and the use of systemic therapy were associated with LR. The relative risk of LR for patients also treated with systemic therapy was 0.30 compared with patients treated with RT alone (P = .002). Among a subset of patients who received systemic therapy and had focally positive margins, the 8-year rate of LR was only 7%. Similar findings have been noted in retrospective studies from other institutions.^14-16 There have been no formal trials comparing concurrent versus sequential RT and chemotherapy. Indirect comparisons among different NSABP trials suggest that LR is lower with concurrent treatment (approximately 0.5% per year) compared with sequential treatment (> 1% per year).¹⁷ However, given the potentially powerful interaction between RT and the now common doxorubicin-based regimens, it is not feasible to use concurrent treatment. Hence sequential treatments have become standard. Unresolved to this point has been whether increasing durations of chemotherapy before RT compromise local control.

In this issue of the Journal of Clinical Oncology, Sartor et al¹⁸ from the Cancer and Leukemia Group B (CALGB) provide an important contribution to the literature with their retrospective study of the prospective randomized trial, CALGB 9344.¹⁹ A total of 3,170 patients were randomly assigned to three dose levels of doxorubicin given with cyclophosphamide, and then underwent a second randomization to four cycles of paclitaxel (AC+T) or to no further systemic therapy. All patients treated with lumpectomy were intended by protocol to receive RT; however, in 9% of patients, RT was not given, and in 13% of patients, it was not known whether RT was given. The use of RT after mastectomy was left to the discretion of the treating oncologists. RT was given at the conclusion of chemotherapy in all patients. Among the patients treated with lumpectomy, 125 patients were treated with AC followed directly by RT, and 144 patients were treated with AC+T and then RT. Despite the further delay in initiating RT, the addition of paclitaxel was associated with a decreased rate of isolated locoregional recurrence (LRR) at 5 years (9.7% v 3.7%; P = .04). Moreover, the addition of paclitaxel did not alter the delivery of RT, with no increase in RT interruptions or in the time to complete RT. Although RT toxicity data were not gathered prospectively, substantial differences in acute side effects would presumably have an impact on the ability to delivery RT in a timely fashion. On the basis of the report from Sartor et al, we believe that the best available data indicate that when patients are treated with lumpectomy and AC+T, RT should be used after completion of all chemotherapy. On the basis of our previous experience,^8,9 we believe that one should still be cautious in the subset of patients with close surgical margins. In these patients, a re-excision to obtain adequate margins seems prudent before starting chemotherapy.

The article by Sartor et al¹⁸ also provides new data on the use of RT after mastectomy in the setting of taxane-based therapy. Ninety-eight patients in the AC arm were treated with postmastectomy RT compared with 87 in the AC+T arm. Among the mastectomy patients who also received RT, the 5-year rate of LRR was very low after either AC (4.3%) or AC+T (3.5%). Two hundred fifty-three patients were treated with mastectomy and AC without RT, and 245 patients were treated with mastectomy and AC+T without RT. Overall, the isolated 5-year LRR rate without RT was 10.8% in the AC arm compared with 8.8% in the group receiving AC+T (P = not significant). These rates of LRR are comparable to those seen in other large cooperative group studies.^20,21 Sartor et al correctly suggest that the rules for postmastectomy RT may change in the setting of more effective chemotherapy. Not only can more effective systemic therapy better eliminate systemic micrometastases, but it may also be more successful at decreasing the local tumor burden. The risk of LRR may decrease to a level that might obviate the need for RT. However, the addition of paclitaxel to AC did not achieve this aim in the present report.

The risk of LRR, both with and without RT, will need to be re-examined as systemic therapy becomes more effective. It is also possible that even more effective systemic therapy might increase the utility of postmastectomy RT. A survival benefit for postmastectomy RT is observed only when used in the presence of effective systemic therapy.^22,23 This finding suggests that current systemic therapies may be more successful in treating systemic involvement than in eradicating locoregional disease. The possible explanations for this observation include a greater locoregional microscopic disease burden than elsewhere in the body or perhaps diminished blood flow or relative hypoxia in the surgical bed. Eventually, of course, improved systemic therapy may lead to decreased reliance on local therapies. At the present time, however, optimal locoregional treatment remains an important part of multidisciplinary breast cancer management.