Do Recent Advances in Diagnostic and Therapeutic Procedures Negate the Benefit of Postmastectomy Radiotherapy in N1 Patients With a Low Risk of Locoregional Recurrence?

Recent advances in breast cancer management might make the use of postmastectomy radiotherapy (PMRT) redundant in the treatment of pT1/T2N1 patients. We investigated the impact of PMRT on disease-free survival (DFS) in these patients who have a low risk of locoregional recurrence (LRR) after contemporary multidisciplinary management.Between 1998 and 2011, 1123 patients underwent upfront surgery for pathologically diagnosed pT1/T2N1 breast cancer, at a single institution. A retrospective review was performed on 692 patients who had a mastectomy with axillary lymph node (LN) clearance. Most patients received adjuvant systemic chemotherapy and/or endocrine therapy. PMRT was administered to 17.8% of the patients. The median follow-up time was 98 months.The entire cohort was divided into 2 groups, the early-era (1998-2003) and late-era (2004-2011) cohorts. Grouping was based on the use of modern therapies since 2004 including sentinel LN (SLN) biopsy, anthracycline/taxane-based chemotherapy, and aromatase inhibitors. Late-era patients had a significantly lower 5-year LRR compared with early-era patients (3.2% vs 10.3%, respectively; P < 0.001). In late-era patients, although PMRT did not significantly reduce the 5-year LRR rate (1% vs 3.8%, respectively), it did improve the 5-year DFS rate (96.1% vs 87.5%, respectively). After controlling for all clinicopathological variables, PMRT was independently associated with improved DFS. In subgroup analysis, depending on the presence of micro- or macrometastasis in the axillary nodes, the benefit of PMRT was most apparent in patients with macrometastasis (hazard ratio, 0.19). In the late-era cohort with no PMRT, the 3-year distant metastasis risk increased according to LN tumor burden (0%, 5.2%, and 9.8% in micrometastasis, SLN macrometastasis, and non-SLN macrometastasis, respectively).Advanced surgical and systemic therapies might not negate the benefit of PMRT in recently diagnosed pN1 patients who have a very low risk for LRR. Our data indicate that the overall recurrence risk combined with the LRR should be considered for an indication of PMRT, and raises the question of whether the receipt of PMRT would improve outcome in patients with micrometastasis.

INTRODUCTION

In breast cancer patients, anatomic staging, especially nodal status, is considered a significant factor for the prognosis of locoregional recurrence (LRR) and selection of adjuvant radiation therapy after mastectomy. The survival benefit of postmastectomy radiotherapy (PMRT) in node-positive breast cancer patients has been well established through multiple-randomized trials.[1-3] The results of Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) meta-analyses confirmed that PMRT consistently reduced the risk of LRR by two-thirds and increased disease-free survival (DFS) and cancer-specific survival.[4] Although there is an international consensus that PMRT should be indicated for patients with tumors that measure >5 cm or for those with ≥4 positive lymph nodes (LNs), the role of PMRT in patients who have tumors that measure ≤5 cm and 1 to 3 positive LNs (pT1–2N1) is highly controversial because axillary LN dissection seems likely to outweigh the potential benefit of PMRT.

Recently updated EBCTCG reports have reaffirmed the benefit of PMRT in a subset of N1 patients who had axillary dissection at least level II, irrespective of adjuvant systemic therapy (mostly cyclophosphamide, methotrexate, and fluorouracil).[1] However, resistance to apply the results of older studies to present practice remains, and the routine use of PMRT has not been recommended. This is because the absolute risks of any recurrence or death have decreased during recent decades because of improved screening and treatment protocols. In other words, the characteristics of patients with pT1/2N1 have changed favorably overtime. The widespread use of sentinel LN (SLN) biopsy combined with extensive pathologic analysis has resulted in frequent identifications of nodal micrometastasis, and a higher proportion of patients have now been treated using modern systemic agents. Studies in the 1980s reported the LRR rate of patients who did not undergo PMRT to be 17.7%, whereas recent studies reported rates of 6% to 10%.[1-3] In this respect, the present absolute benefits of PMRT for patients with T1/T2N1 breast cancer are likely to be small.

A subgroup analysis of the Danish Breast Cancer Cooperative Group 82 b and c trials suggested that reduced LRR in response to PMRT translated as a larger reduction in cancer mortality in women with N1 breast cancer compared with those with ≥4 positive LNs.[4] Recent data from the National Cancer Institute of Canada Clinical Trials Group MA.20 and the European Organization for Research and Treatment of Cancer 22922 trials indicated that optimized locoregional control is crucial for long-term survival, especially in patients with a relatively lower competing risk of distant metastasis (DM).[5,6] From their standpoint, PMRT does not only limit itself to locoregional control, but also to the prevention of systemic progression. Here, we verified the hypothesis that modern improvements in diagnostic and therapeutic procedures have resulted in a lower risk of LRR and superior survival in patients with T1/T2N1 breast cancer who were treated with mastectomy and axillary LN dissection. We subsequently evaluated the contribution of PMRT to survival outcomes regarding overall recurrence as well as LRR.

METHODS

Patients

This retrospective observational study was approved by the Institutional Review Board of Severance Hospital in Seoul, Korea. Patient consent was not required, because the collected data were existing information. We identified 1123 consecutive patients who underwent upfront surgery and who were diagnosed with pathological T1N1/T2N1 breast cancer between January 1998 and December 2011. Patients who underwent breast conservation surgery (n = 431) were excluded. The data from the remaining 692 patients were reviewed retrospectively. Preoperative evaluation consisted of a complete history, a physical examination, complete blood counts, mammography, breast ultrasonography, a bone scan, and computed tomography scans or magnetic resonance imaging, if indicated. Systemic staging with [18F]-fluorodeoxyglucose-positron emission tomography was performed in 109 patients (15.8%). Immunohistochemical staining for estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) was performed. Breast cancer molecular subtypes were categorized as luminal A (ER+, moderate to strong PR+, and HER2−); luminal B (ER+, negative or weak PR+, and HER2+/−); HER2 (ER− and PR− and HER2+); and triple negative (ER− and PR−, HER2−).[7,8]

Treatment

All 692 patients underwent modified radical mastectomy. Although SLN sampling was first implemented at our institution in 2000, it was not commonly employed until 2003; thus, SLN sampling was only performed in 357 patients (51.6%). Those with biopsy-confirmed or suspicious axillary nodal metastasis on positron emission tomography imaging were spared. Surgeons did not refer all T1/T2N1 patients to radiation oncologists; PMRT was offered to patients with high-risk features such as multiple positive LNs. Three-dimensional conformal PMRT was applied to the chest and regional nodal areas (supraclavicular, axillary, and internal mammary nodes) using the reverse hockey stick technique with a total dose of 50.4 Gy in 28 fractions.

Before the publication of the Cancer and Leukemia Group B 9344 study (1998–2003), if patients had comorbidities such as cardiac disease or a poor performance status, our institutional policy for adjuvant chemotherapy for node-positive patients was to offer 6 cycles of fluorouracil, doxorubicin, cyclophosphamide or 6 cycles of cyclophosphamide, methotrexate, and fluorouracil.[9] After the Cancer and Leukemia Group B study (2004–2011), most node-positive patients were offered a combination of cyclophosphamide (600 mg/m2) and doxorubicin (60, 75, or 90 mg/m2) for 4 cycles with or without an additional 4 cycles of paclitaxel (175 mg/m2) (cyclophosphamide, doxorubicin, and/or paclitaxel [AC ± T]). Trastuzumab therapy was indicated for patients with HER2-overexpressing tumors. In accordance with these policies, adjuvant systemic therapy was selected after discussion with a medical oncologist.

Statistical Analyses

The cumulative incidences of LRR and DM were estimated using the 1–Kaplan–Meier method, and the survival probabilities of DFS and overall survival (OS) were estimated using the Kaplan–Meier method. The time to recurrence or death was calculated from the date of the mastectomy. Comparisons were performed using a t-test (continuous variables) or the Chi-square test (nominal variables). Univariate analyses for survival outcomes were performed using the log-rank test. Multivariate analyses for survival outcomes were performed using the Cox regression model with stepwise backward elimination (alpha = 0.20). The level of statistical significance was set at P < 0.05. All statistical analyses were performed with SPSS version 20.0 (IBM SPSS Statistics, IBM Corporation, Armonk, NY).

RESULTS

The clinicopathological characteristics of the 692 patients are shown in Supplementary Table 1, http://links.lww.com/MD/A350. Briefly, SLN-positive patients underwent axillary LN dissection (mean number of LN resected, 17.1 ± 6.8). Most patients received adjuvant systemic chemotherapy (632, 91.3%) and/or endocrine therapy (477, 69.4%). Trastuzumab was administered in 26 (13.2%) patients with HER2-overexpressing tumors. PMRT was administered to 123 (17.8%) patients.

The median follow-up time among surviving patients was 98 months (range, 2–197 months). Recurrence was observed in 126 patients (18.2%). A total of 87 (12.6%) patients died. The overall incidence of LRR and DM as the first failure pattern was 7.5% (n = 52) and 12.9% (n = 89), respectively, including 31 patients who suffered from simultaneous LRR and DM. The 5-year cumulative incidences of LRR and DM were 5.9% and 10%, and the 10-year rates were 9.3% and 13%, respectively (Figure 1). The 5-year DFS and OS were 86.5% and 91.9%, respectively, and the 10-year rates were 79.5% and 86%, respectively.

FIGURE 1

Kaplan–Meier (KM) overall survival (OS) and disease-free survival (DFS) rates, and the 1–KM cumulative incidence of locoregional recurrence (LRR) and distant metastasis (DM).

Entire Cohort (1998–2011)

The entire cohort was divided into 2 groups, the early-era (1998–2003) and late-era (2004–2011) cohorts, because modern therapies including SLN biopsy, taxane-based chemotherapy, and aromatase inhibitors have been commonly used since 2004 at our institution (Table 1). Comparisons of patient characteristics for each cohort are listed in Table 1. Patient characteristics have changed favorably overtime (eg, older patients, less T2, smaller tumor size, lower grade tumors, increased detection of micrometastasis, fewer positive LNs, and a higher proportion of hormone-receptor positive tumors). In addition, in late-era group SLN biopsy use was widespread facilitating increased detection of micrometastasis in LNs, and fewer numbers of positive LNs, postoperatively. Furthermore, in the late-era group, patients were more likely to be treated using a modern chemotherapy regimen (AC ± T), anti-HER2 targeted agents, and PMRT (all P < 0.05). The median follow-up of the early-era and late-era cohorts were 155 (range, 124–197) months and 73 (range, 2–122) months, respectively.

TABLE 1

Comparison of Patient, Tumor, and Treatment Characteristics According to the Treatment Era (N = 692)

Compared with the early-era cohort, the late-era cohort had significantly better 5-year OS (88.6% vs 94.2%, P = 0.009) and 5-year DFS (82.1% vs 89.3%, P = 0.007) (Figure 2A, B). Improvement in DFS in the late-era cohort was a consequence of a significant reduction in LRR and DM (P < 0.001, Figure 2C, D). Notably, the 5-year LRRs were 10.3% (95% confidence interval [CI] 6.8%–13.8%) and 3.2% (95% CI 1.4%–5.0%) for the early-era and late-era cohorts, respectively.

FIGURE 2

(A) Overall survival, (B) disease-free survival, (C) locoregional recurrence, and (D) distant metastasis in all patients with T1N1/T2N1 breast cancer (N = 692) according to treatment era (1998–2003 vs 2004–2011).

Multivariate Cox regression analyses were performed to analyze the factors independently associated with LRR, DM, DFS, and OS (Table 2). Older age (≥45 years), smaller LN ratio (<25%), use of SLN biopsy, and administration of AC ± T were significantly and independently associated with improved DFS. T-stage, tumor grade, and use of PMRT were related to DFS, but with borderline significance. Administration of the AC ± T was significantly and independently associated with improved LRR control, and having a nonhigh grade tumor showed borderline significance. In an analysis of DM, old age (≥45 years), T1 tumor stage, nonhigh grade tumor, a smaller LN ratio (<25%), SLN biopsy, and the chemotherapy regimen administered were significantly and independently associated with improved DM control.

TABLE 2

Multivariate Cox Proportional Hazards Survival Analysis in the Entire Cohort According to Each Clinicopathological and Treatment Variable

Late Cohort (2004–2011)

To provide more clinically relevant information for current practice from a relatively homogenous patient group, we assessed the effect of PMRT on clinical outcomes limited to the data from late-era group. Based on our previous experience, LN status was the first consideration when determining the use of PMRT. Understandably, patients treated with PMRT had more LN macrometastasis and positive LNs than those who were not treated with PMRT (all P < 0.05, Table 3). AC ± T chemotherapy was more frequently administered to patients who underwent PMRT (85.6% vs 68.3%, P = 0.003). Otherwise, disease and treatment characteristics were well balanced between those who underwent PMRT and those who did not (all P > 0.05).

TABLE 3

Comparison of Patient, Tumor, and Treatment Characteristics According to Postmastectomy Radiotherapy Use in 412 Patients Who Were Treated in the Late Era (2004–2011)

The 5-year LRR rates were not significantly different between those who underwent PMRT and those who did not at 1% and 3.8%, respectively (Figure 3A). The 5-year DM rates were not significantly different (Figure 3B). The 5-year DFS rates were 96.1% and 87.5%, respectively, and the 10-year DFS rates were 96.1% and 79.5%, respectively (P = 0.015) (Figure 3C). The 5-year OS rates for patients who underwent PMRT and those that did not were 100% and 91.9%, respectively (P = 0.013) (Figure 3D).

FIGURE 3

(A) Locoregional recurrence, (B) distant metastasis, (C) disease-free survival, and (D) overall survival in the late-era group (2004–2011) patients with T1N1/T2N1 breast cancer (n = 412) according to the use of postmastectomy radiotherapy (PMRT).

In the multivariate analysis, PMRT (P = 0.038), nonhigh grade tumor (P = 0.025), and AC ± T chemotherapy (P = 0.007) were significantly and independently associated with improved DFS (Table 4). To better assess the impact of PMRT, we analyzed the data by excluding patients who did not undergo chemotherapy. In these patients, PMRT (hazard ratio [HR], 0.22; 95% CI 0.05–0.91) and macrometastasis (vs micrometastasis; HR, 2.89; 95% CI 1.20–6.94) were significantly and independently associated with DFS. Multivariate analysis could not be applied to assess the effect of PMRT on OS, since there were no deaths in the PMRT group. To gain insight into associations between the PMRT effect and size (macro- vs micrometastasis) or number of positive LNs, we performed subgroup analyses. The benefits of PMRT in DM control (HR, 0.13; 95% CI 0.02–0.99) and DFS (HR, 0.19; 95% CI 0.05–0.78) were most apparent in patients with macrometastasis compared to those with micrometastasis (Table 5).

TABLE 4

Stepwise Univariate and Multivariate Analyses Using Cox Regression Method for Disease-Free Survival in 412 Patients Who Were Treated in the Late Era (2004–2011)

TABLE 5

Univariate Cox Proportional Hazards Survival Analysis Among Patients in the Late-Era Cohort (2004–2011) According to Metastasis Size and Positive Lymph Node Number

In the late-era cohort, in patients with N1 disease who did not undergo PMRT, the effect of LN tumor burden on the risk of the tumor recurrence was assessed and is summarized in Supplementary Table 2, http://links.lww.com/MD/A350. The 3-year LRR rate was not significantly different as a function of LN tumor burden, whereas the DM risk increased according to LN tumor burden. The 3-year DM rate was 0%, 5.2%, and 9.8% for those with micrometastasis, SLN macrometastasis, and non-SLN macrometastasis, respectively.

DISCUSSION

The 2005 EBCTCG pooled analysis showed a survival benefit of PMRT when the 5-year LRR risk exceeded 10%.[10] Their analysis evaluated the benefit of PMRT in patients who underwent a mastectomy according to node positivity. Although PMRT yielded similar proportional reductions in LRR irrespective of node positivity, the absolute reductions in LRR in node-positive women were large and resulted in a clear survival benefit. Therefore, PMRT was indicated in N2 disease; however, the benefit of PMRT in N1 disease, which is assumed to have a relatively low absolute LRR risk, remained debatable. Recently, an updated meta-analysis showed that LRR reduction in response to PMRT mediated a significant survival benefit in patients with N1 disease, regardless of systemic chemotherapy.[1]

Here, patients with T1/T2N1 disease who were treated more recently (2004–2011) had lower risks of DM and LRR, and had better survival compared with those who were treated earlier (1998–2003). In recently treated patients who were at very low risk for LRR, PMRT significantly improved DFS. These findings indicate that recent advances in surgical and systemic therapies might not negate the benefit of PMRT in N1 patients, and that a very low LRR rate should not be a surrogate endpoint to exclude patients from PMRT.

A recent meta-analysis indicated that the 5-year LRR in a series of patients who underwent PMRT between 1964 and 1986 was 2.8%,[1] which was similar to that for the recently treated patients in the present study (1%). In contrast, in the past, the 5-year LRR in patients who did not undergo PMRT was as high as 16.5%, whereas this has continually decreased over time and was considerably low in our late-era cohort (3.8%).[2,3,11] A number of possibilities could underlie improvements in the 5-year LRR, DFS, OS, and the rate of DM. First, the Korean national cancer-screening program for breast cancer became active during 2003–2005, meaning that the patient characteristics of those with N1 disease have changed favorably over time.[12] Second, improvements in pathological diagnostic methods have increased the detection of micrometastasis, which might have been classified previously as N0 disease. Third, the median age of Korean breast cancer patients has increased overtime, which might affect treatment outcomes.[12] Furthermore, increased use of more effective adjuvant systemic therapies likely resulted in superior outcomes in recently treated patients.[13,14]

To assess the impact of PMRT in the late-era cohort rather than entire cohort was more clinically relevant, although the cut-off point of 2003 was somewhat arbitrary. In the late-era cohort, PMRT was not associated with a significant reduction in LRR or DM risk, but was significantly associated with improvements in DFS and OS, which resulted from possible reduction of both LRR and DM by PMRT. In addition, PMRT prevented subsequent progression of subclinical disease within the chest wall and regional nodes (eg, supraclavicular and internal mammary nodes) to systemic disseminations. These findings are congruent with 2 recent clinical trials that addressed the clinical benefit of adjuvant radiotherapy in breast-conserved patients. The National Cancer Institute of Canada Clinical Trials Group MA.20 trial enrolled patients with high-risk N0 and N1 disease and randomly assigned them to treatment with or without regional RT.[5] In the preliminary report, regional radiotherapy reduced the 5-year DM rate by 4.3% and reduced the LRR rate by 2%. The European Organization for Research and Treatment of Cancer 22922/10925 trial showed similar results; regional radiotherapy reduced the 10-year DFS rate by 3%.[6] In our study, both AC ± T and PMRT were independently associated with an improvement in DFS. This supports the hypotheses that radiotherapy and systemic therapy have different time-dependent effects[15] and that the effects of radiotherapy and systemic therapy are synergistic or additive rather than competitive.[16]

The benefit of PMRT on DFS was most apparent in patients with macrometastasis compared with that in those with micrometastasis. In addition, the risk of DM or any recurrence was lower in patients with low LN tumor burden (micrometastasis) compared to that in those with macrometastasis. Mittendorf et al[17] reported similar survival in patients with micrometastasis compared with node-negative patients. Since SLN biopsy was validated as standard operating procedure in 2003, efforts have been made to assess the SLN using serial sectioning of entire SLN rather than using 3 sections per axillary LN.[18,19] Such extensive pathologic analysis results in the frequent identification of micrometastatic foci (≤2 mm, N1mi), and in our opinion, the use of PMRT in patients with small-volume LN disease should be conservative, potentially avoiding local therapy overtreatment.

Risk-adaptive-personalized treatment is the current standard treatment in mastectomy patients with N1 disease. The current international consensus does not recommend routine use of PMRT unless there are additional adverse features.[8] However, the indication and frequency of PMRT use varies significantly between institutions, and many investigators have identified various risk factors of LRR including young age (eg, ≤45 years),[20,21] large tumor size (eg, T2 or >3 cm),[11,21] LN ratio (eg, >15%–25%),[20,22-24] ER-status,[20,25] lymphovascular invasion,[25,26] extranodal extension, a medially located tumor,[20] inadequate axillary surgery,[11] and an earlier treatment era (eg, before 1997).[11] However, our results suggest that a risk-adaptive approach should be considered based on overall recurrence risk rather than LRR risk. In this regard, the results of the United Kingdom Medical Research Council-SUPREMO trial, which is a large randomized trial investigating the benefit of PMRT in modern treatment regimens that has the primary end-point of OS, are eagerly awaited.[27]

Our study has several limitations including those inherent in retrospective analysis. The statistical analyses are incomplete because multivariate analysis for LRR and OS could not be performed because none of the patients undergoing PMRT died. In addition, patients who underwent PMRT had more LNs that were positive and were, therefore, more likely to receive AC ± T chemotherapy, leading to potential selection bias. However, it is noteworthy that PMRT had a protective effect after controlling for these potentially confounding factors. Routine use of internal mammary node irradiation in our study should be taken into consideration to interpret the benefit of PMRT, which may be important because all previous positive trials of PMRT used internal mammary node irradiation.

In summary, patients with T1/T2N1 breast cancer who were treated recently had a lower risk of LRR and DM with excellent survival outcomes compared with those treated in the past. PMRT significantly and independently improved DFS, although reductions in LRR were not large or significant in those treated with contemporary multidisciplinary management. We believe that the present study does not convincingly support the routine use of PMRT in all patients with T1/T2N1 breast cancer. However, it does support the results of the recent EBCTCG meta-analysis that PMRT improves LRR-free survival, DFS, and cancer-specific survival in node-positive patients in contemporary practice.[1] Similar concerns regarding regional radiotherapy in breast-conserved patients might be meaningful. Further studies, ideally, well-designed and controlled, need to be conducted to confirm our findings.