Risk of early recurrence among postmenopausal women with estrogen receptor-positive early breast cancer treated with adjuvant tamoxifen

Authors

  • Hagen Kennecke MD, MHA,

    Corresponding author
    1. Division of Medical Oncology, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    2. Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
    3. Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    • Vancouver Cancer Clinic, 600 West 10th Avenue, Vancouver, BC, V5Z 4E6, Canada
    Search for more papers by this author
    • Fax: (604) 877-0585.

  • Heather McArthur MD,

    1. Division of Medical Oncology, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    2. Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Ivo A. Olivotto MD,

    1. Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
    2. Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    3. Division of Radiation Oncology, British Columbia Cancer Agency, Victoria, British Columbia, Canada
    Search for more papers by this author
  • Caroline Speers BA, CHIM,

    1. Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    2. Population and Preventive Oncology Program, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Chris Bajdik MD,

    1. Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    2. Cancer Control Research, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Stephen K. Chia MD,

    1. Division of Medical Oncology, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    2. Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
    3. Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Susan Ellard MD,

    1. Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
    2. Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    3. Division of Medical Oncology, Cancer Center of Southern Interior, British Columbia Cancer Agency, British Columbia, Canada
    Search for more papers by this author
  • Brian Norris MD,

    1. Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
    2. Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    3. Division of Medical Oncology, British Columbia Cancer Agency, Surrey, British Columbia, Canada
    Search for more papers by this author
  • Malcolm Hayes MD,

    1. Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
    2. Department of Pathology, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    Search for more papers by this author
  • Jeff Barnett BSC (Pharm),

    1. Systemic Therapy Program, British Columbia Cancer Agency, Victoria, British Columbia, Canada
    Search for more papers by this author
  • Karen A. Gelmon MD

    1. Division of Medical Oncology, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    2. Department of Medicine, University of British Columbia, Vancouver, British Columbia, Canada
    3. Breast Cancer Outcomes Unit, British Columbia Cancer Agency, Vancouver, British Columbia, Canada
    Search for more papers by this author

  • Presented in part at the 28th Annual San Antonio Breast Cancer Conference, San Antonio, Texas, December 7–11, 2005.

Abstract

BACKGROUND

Adjuvant aromatase inhibitors (AIs), instead of or after tamoxifen, are effective in decreasing recurrence in postmenopausal women with estrogen receptor (ER)-positive breast cancer. An understanding of which patients are at risk of early recurrence while they are receiving tamoxifen may improve clinical decision making.

METHODS

The patients who were included in this study were women aged ≥50 years with early-stage, ER-positive breast cancer diagnosed between 1986 and 1999 and had been treated with tamoxifen. Characteristics of the patients with early recurrences (within 2.5 years of diagnosis), late recurrences (between 2.5 years and 5 years) and no recurrence within 5 years were compared. Logistic regression analyses were conducted to identify which groups were at risk of early recurrence.

RESULTS

Among 3844 women, 304 women (7.9%) developed disease recurrence within 2.5 years. Higher than average rates of recurrence within 2.5 years were observed in cohorts with lymph node (N)-positive tumors (11.5%), grade 3 histology (14.3%), or low-positive ER levels, ie, 10–49 fmol/mg or 10%–20% staining (14.9%). In multivariate analyses, only pathologically N-positive tumors (1–3 vs 0 positive lymph nodes: odds ratio [OR], 1.6; 4–9 vs 0 positive lymph nodes: OR, 2.23 [P = .03]) and low-positive ER status (OR, 2.04; P = .01) were associated with recurrence within 2.5 years compared with recurrence between 2.5 years and 5 years. Other clinical and pathologic variables were not predictive of early recurrence.

CONCLUSIONS

Subgroups of women with early ER-positive breast cancer may be identified who are at increased risk of recurrence within 2.5 years of diagnosis despite tamoxifen. It remains to be proven whether upfront AI therapy results in an advantage to these women. Cancer 2008. © 2008 American Cancer Society.

Clinicians choose among different adjuvant endocrine strategies for postmenopausal women with estrogen receptor (ER)-positive, early-stage breast cancer. Aromatase inhibitors (AIs) are superior to 5 years of tamoxifen monotherapy when they are given instead of tamoxifen, after 2 to 3 years of tamoxifen, or after 5 years of tamoxifen.1–5 American Society of Clinical Oncology guidelines recommend that all eligible women receive therapy with AIs either instead of or after tamoxifen.6 To our knowledge, the optimal treatment strategy has not been determined to date, and cross-trial comparisons have not demonstrated conclusively whether AI monotherapy or sequential tamoxifen-AI therapy is superior. A potential advantage of sequential therapy is a reduced risk of toxicity because of a shorter exposure to either agent. Trials of sequential tamoxifen-AI therapy have reported an overall survival benefit,7, 8 whereas trials of AI monotherapy versus tamoxifen1, 9 have yet to report a survival advantage. A disadvantage of initiating therapy with tamoxifen may be that women who experience an early recurrence while they are receiving tamoxifen may lose the opportunity to benefit from adjuvant AI therapy.

Understanding the timing of breast cancer recurrence may be helpful for clinicians in choosing among various adjuvant endocrine options. The 5-year and 10-year risks of breast cancer recurrence rates are well defined,10, 11 and it is known that patients with ER-positive breast cancer have a relatively constant risk of recurrence between 5 years and 10 years after diagnosis.12, 13 In this study, we evaluated factors associated with early versus late recurrence among postmenopausal women with ER-positive breast cancer who received tamoxifen to identify those at risk of early recurrence.

MATERIALS AND METHODS

Study participants were women residents of British Columbia aged ≥50 years who had ER-positive breast cancer diagnosed between 1986 and 1999 and who received a prescription for tamoxifen from the British Columbia Cancer Agency (BCCA) within 1 year of diagnosis. The date of diagnosis was defined as the earlier of Class IV/V cytology, positive pathology, or, in the absence of the former, clinical diagnosis. Women were included if they had breast cancer with T1/T2 pathologic tumor classification, N0 to N3a lymph node status, and negative for metastasis (M0).14 Women were excluded if they had a prior or synchronous contralateral breast cancer, noninvasive disease alone, ER-negative or ER-unknown status, advanced stage (including clinical or pathologic T3/T4 and clinical N2, N3, or M1 disease), or if they were referred to the BCCA only after recurrence. Patients who died of a nonbreast cancer death or developed a contralateral, new, primary breast cancer within 5 years of diagnosis and before a recurrence event also were excluded. Patients underwent definitive surgery, including axillary lymph node dissection, and received adjuvant radiation and systemic therapy as described in the BCCA cancer management guidelines, which evolved over time.15–17

Data were abstracted from the prospectively assembled BCCA Breast Cancer Outcomes Unit database, which contains demographic, pathologic, staging, treatment, and outcomes information on all patients who have been diagnosed with breast cancer since January 1, 1989 and referred to a BCCA facility. Pathology characteristics abstracted included lymph node status (positive or negative), the number of axillary lymph nodes removed and involved with cancer, tumor size, histologic grade, ER status, and the presence of lymphatic or vascular invasion in the primary tumor (Table 1). Before April 1991, ER status was determined by dextran-coated charcoal assay, and tumors were considered ER-positive if they scored >10 fmol/mg cytosol protein. After April 1991, ER staining was changed to the Abbott enzyme immunoassay monoclonal biochemical method, and tumors were considered ER-positive if they scored >15 fmol/mg. ER staining using the immunohistochemical (IHC) method replaced the biochemical method after April 1992 and was graded according to the following scale, which was modified from the Allred scoring system: ER 0, <10% of cells stained positive (corresponding to Allred score, 0–2); ER 1+, from 10% to 20% of cells stained positive (Allred score, 3–4); ER 2+, from 20% to 60% of cells stained positive (Allred score, 5–6); and ER 3+, from 60% to 100% of cells stained positive (Allred score, 7–8). The majority of patients had a pathology review on referral to the BCCA and, if it was done, then the review diagnosis was used to describe pathologic characteristics.

Table 1. Selected Clinical, Pathologic, and Treatment Characteristics of 3844 Women With Early-stage Breast Cancer Diagnosed in British Columbia Between 1986 and 1999
CharacteristicNo. of patients (%)Chi-square P
Recurrence within 2.5 years from DX: Group 1Recurrence after 2.5 years and within 5 years from DX: Group 2No recurrence within 5 years of DX: Group 3Group 1 vs Group 2Group 1 vs Group 3
  1. DX indicates diagnosis; ER, estrogen receptor; IHC, immunohistochemistry; fmol/mg, femtomoles per milligram cytosol protein; Chemo, chemotherapy; BCS, breast-conserving surgery; ALND, axillary lymph node dissection; Mx, mastectomy; RT, radiotherapy.

Total304 (100)283 (100)3257 (100)  
Age at DX, y     
Median, y64.964.965.56.007
 50–5995 (31.3)92 (32.5)1073 (32.9)  
 60–6998 (32.2)100 (35.3)1163 (35.7)  
 70–7982 (27)72 (25.4)860 (26.4)  
 ≥8029 (9.5)19 (6.7)161 (4.9)  
Histology   .23<.001
 Ductal286 (94.1)261 (92.2)2894 (88.9)  
 Lobular15 (4.9)21 (7.4)356 (10.9)  
 Other3 (1)1 (0.4)7 (0.2)  
Tumor size, cm   .24<.001
Median, cm2.42.32  
 0.1–119 (6.3)28 (9.9)418 (12.8)  
 1.1–2112 (36.8)95 (33.6)1479 (45.4)  
 2.1–5173 (56.9)160 (56.5)1360 (41.8)  
Grade   .003<.001
 16 (2)16 (5.7)473 (14.5)  
 2112 (36.8)128 (45.2)1729 (53.1)  
 3166 (54.6)122 (43.1)874 (26.8)  
 Unknown20 (6.6)17 (6)181 (5.6)  
Lymphovascular invasion   1.0<.001
 Absent120 (39.5)110 (38.9)1828 (56.1)  
 Present169 (55.6)154 (54.4)1275 (39.1)  
 Not applicable/unknown15 (4.9)19 (6.7)154 (4.7)  
ER level   .008<.001
 Low [10–49 fmol/mg; IHC 1+]54 (17.8)29 (10.2)279 (8.6)  
 Moderate/high [≥50 fmol/mg; IHC 2+/3+]196 (64.5)208 (73.5)2369 (72.7)  
 Positive unquantified54 (17.8)46 (16.3)609 (18.7)  
Surgical margin status   .89.27
 Positive30 (9.9)27 (9.5)273 (8.4)  
 Negative249 (81.9)240 (84.8)2834 (87)  
 Not applicable/unknown25 (8.2)16 (5.7)150 (4.6)  
No. of lymph nodes removed   .53.45
Median no.10910  
  <652 (17.1)42 (14.8)469 (14.4)  
 6–10121 (39.8)129 (45.6)1258 (38.6)  
 11–1583 (27.3)68 (24)975 (29.9)  
 ≥1644 (14.5)41 (14.5)527 (16.2)  
 Not applicable/unknown4 (1.3)3 (1.1)28 (0.9)  
Lymph node status   .05<.001
 Negative71 (23.4)87 (30.7)1661 (51)  
 Positive233 (76.6)196 (69.3)1596 (49)  
No. of positive lymph nodes   0.10<.001
 071 (23.4)87 (30.7)1661 (51)  
 1–3140 (46.1)125 (44.2)1257 (38.6)  
 4–971 (23.4)48 (17)286 (8.8)  
 ≥1021 (6.9)22 (7.8)51 (1.6)  
 Unknown1 (0.3)1 (0.4)2 (0.1)  
Systemic therapy   .51.18
 Tamoxifen alone249 (81.9)238 (84.1)2766 (84.9)  
 Chemo + tamoxifen55 (18.1)45 (15.9)491 (15.1)  
Initial surgery   .05<.001
 BCS + ALND117 (38.5)133 (47)1850 (56.8)  
 Total Mx + ALND187 (61.5)150 (53)1407 (43.2)  
Local therapy   .10<.001
 BCS + RT103 (33.9)122 (43.1)1720 (52.8)  
 BCS alone14 (4.6)11 (3.9)130 (4)  
 Mx + RT82 (27)58 (20.5)365 (11.2)  
 Mx alone105 (34.5)92 (32.5)1042 (32)  

The primary endpoint was recurrence-free survival. A breast cancer recurrence was defined either as a local (breast or chest wall), regional (ipsilateral axillary, infraclavicular, internal mammary, or supraclavicular), or distant recurrence (any other site, including contralateral lymph nodes) or as death from breast cancer in the absence of a recorded recurrence. Subsequent contralateral breast cancer was not included as an endpoint, because the objective of the study was to characterize patterns of recurrence in the primary breast only. Patients were assigned to 1 of 3 groups according to their recurrence status. Group 1 (early recurrence) experienced a breast cancer recurrence within 2.5 years of diagnosis, Group 2 (late recurrence) experienced a recurrence between 2.5 years and 5 years after diagnosis, and Group 3 (no recurrence) included women who had no documented recurrence within the first 5 years. Women who experienced a recurrence >5 years after diagnosis were included in Group 3. Univariate analysis was performed on selected patient, tumor, and treatment characteristics. Chi-square tests and analyses of variance were conducted to compare the distribution of characteristics between the 3 groups.

Because the proportional hazards assumption required for a Cox regression model was not met, binary logistic regression analyses were chosen to estimate the likelihood of an early event by comparing Group 1 (early recurrence) with Group 3 (no recurrence) and by comparing Group 1 (early recurrence) with Group 2 (late recurrence). A variable's effect on early recurrence was estimated by the model's odds ratio (OR). A 95% confidence interval (95%CI) was calculated for each OR. Women were excluded from the multivariate analyses if values for any of the selected variables were unknown.

Annual recurrence ratios (for each year after diagnosis) were defined as the fraction of patients who recurred during each 1-year interval. All statistical analyses were performed with SPSS (version 14.0) software (SPSS Inc., Chicago, Ill). This study was approved by the Research Ethics Board of the BCCA, University of British Columbia.

RESULTS

There were 3844 women who met the study eligibility criteria and were included in the study. The median follow-up of the patients who remained alive was 9.1 years, and all patients had a minimum 5 years of follow-up. The median age was 65 years and was similar between the 3 recurrence-timing cohorts. All patients received tamoxifen, and 15.4% of patients received chemotherapy. In total, 28.6% of patients had stage I breast cancer, 58.3% of patients had stage II breast cancer, and 13% of patients had stage III breast cancer.14 Table 1 shows the distributions of age and pathologic and treatment characteristics for the 3 recurrence-timing cohorts. There were 304 patients (7.9%) who developed disease recurrence within the first 2.5 years of diagnosis (Group 1), 283 patients (7.4%) who developed disease recurrence between 2.5 years and 5 years of diagnosis (Group 2), and 3257 patients (84.7%) who experienced no recurrence within the first 5 years. Compared with patients who developed disease recurrence between 2.5 years and 5 years, patients with early recurrence were more likely to have higher histologic grade, to have low ER positivity, to have higher lymph node burden, and to undergo mastectomy (Table 1). The distribution of all variables differed significantly between patients who recurred early (Group 1) and those who did not develop a recurrence (Group 3), with the exception of surgical margin status, the number of lymph nodes removed, and the receipt of adjuvant chemotherapy.

Table 2 shows the proportions of patients that developed disease recurrence within 2.5 years, developed disease recurrence between 2.5 years and 5 years, and did not recur within 5 years, overall and for selected subgroups defined by lymph node involvement, grade, and ER positivity. Overall, 7.9% of patients developed disease recurrence within 2.5 years of diagnosis. A higher rate of recurrence was observed in patients with lymph node-positive disease, grade 3 histology, or low-positive ER status.

Table 2. Crude Rates of Breast Cancer Recurrence for Select Subgroups
CharacteristicNo. of patients (%)
TotalRecurrence within 2.5 years of DXRecurrence between 2.5 years and ≤5 years of DXNo recurrence within 5 years of DX
  1. DX indicates diagnosis; ER, estrogen receptor; fmol/mg, femtomoles per milligram cytosol protein; IHC, immunohistochemistry.

All3844304 (7.9)283 (7.4)3257 (84.7)
Lymph node negative, all181971 (3.9)87 (4.8)1661 (91.3)
Lymph node positive, all2025233 (11.5)196 (9.7)1596 (78.8)
1–3 Positive lymph nodes1522140 (9.2)125 (8.2)1257 (82.6)
4–9 Positive lymph nodes40571 (17.5)48 (11.9)286 (70.6)
Grade 3, all1162166 (14.3)122 (10.5)874 (75.2)
Grade 3, negative lymph nodes46634 (7.3)27 (5.8)405 (86.9)
Grade 3, 1–3 positive lymph nodes48578 (16.1)57 (11.8)350 (72.2)
Grade 3, 4–9 positive lymph nodes15835 (22.2)27 (17.1)96 (60.8)
ER low [10–49 fmol/mg or IHC 1+]36254 (14.9)29 (8)279 (77.1)
ER moderate/high [≥50 fmol/mg or IHC 2+/3+]2773196 (7.1)208 (7.5)2369 (85.4)

Multivariate comparisons of the characteristics between Groups 1 and 3 and between Groups 1 and 2 are shown in Tables 3 and 4, respectively. Grade 3 tumors (OR, 2.53; 95% CI, 1.88–3.42), low ER level (OR, 2.27; 95% CI, 1.56–3.32), and positive lymph nodes were associated significantly with an increased risk of recurrence within 2.5 years compared with no recurrence within 5 years (Table 3). Women who had low-positive ER tumors and those with involved axillary lymph nodes were significantly more likely to have an early recurrence rather than a later recurrence (Table 4). Compared with the lymph node-negative cohort, patients who had from 4 to 9 positive lymph nodes were twice as likely to have an early recurrence rather than a later recurrence (OR, 2.23; 95%CI, 1.22–4.09). With from 1 to 3 positive lymph nodes, the OR for early recurrence was 1.60 (95% CI, 0.99–2.56). Grade, tumor size, the presence of lymphovascular invasion, and the receipt of adjuvant chemotherapy were not associated with the timing of recurrence among patients who had a recurrence within 5 years.

Table 3. Odds Ratio for Recurrence Within 2.5 Years of Diagnosis Compared With No Recurrence Within 5 Years
CharacteristicNo. of patientsOR (95% CI)P
  1. OR indicates odds ratio; 95% CI, 95% confidence interval; ER, estrogen receptor; fmol/mg, femtomoles per milligram cytosol protein; IHC, immunohistochemistry.

Grade  <.001
 1 or 217771.0 
 37982.53 (1.88–3.42) 
ER level  <.001
 Moderate/high [50 fmol/mg; IHC 2+/3+]23131.0 
 Low [10–49 fmol/mg; IHC 1+]2622.27 (1.56–3.32) 
Lymphovascular invasion  .08
 Absent14961.0 
 Present10791.31 (0.97–1.78) 
Tumor size, cm  .20
 0.1–214431.0 
 2.1–511321.22 (0.90–1.64) 
No. of positive lymph nodes  <.001
 012681.0 
 1–310422.27 (1.59–3.22) 
 4–92654.59 (2.95–7.13) 
Adjuvant chemotherapy  .19
 Yes4201.0 
 No21551.31 (0.88–1.94) 
Table 4. Odds Ratios for Early Breast Cancer Recurrence (Within 2.5 Years) or Later Breast Cancer Recurrence (From 2.5 Years to 5 Years)
CharacteristicNo. of patientsOR (95% CI)P
  1. OR indicates odds ratio; 95% CI, 95% confidence interval; ER, estrogen receptor; fmol/mg, femtomoles per milligram cytosol protein; IHC, immunohistochemistry.

Grade  .11
 1 or 21971.0 
 32091.29 (0.86–1.95) 
ER level  .01
 Moderate/high [≥50 fmol/mg; IHC 2+/3+]3401.0 
 Low [10–49 fmol/mg; IHC 1+]662.04 (1.16–3.60) 
Lymphovascular invasion  .32
 Absent1721.0 
 Present2340.81 (0.53–1.23) 
Tumor size, cm  .47
 0.1–21901.0 
 2.1–52160.86 (0.57–1.29) 
No. of positive lymph nodes  .03
 01281.0 
 1–31941.60 (0.99–2.56) 
 4–9842.23 (4.09) 
Adjuvant chemotherapy  .53
 Yes631.0 
 No3430.83 (0.47–1.47) 

Figures 1 through 3 illustrate the annual recurrence rates for patients who had tumors with different histologic grades (Fig. 1), ER levels (Fig. 2), and numbers of involved axillary lymph nodes (Fig. 3). Women with grade 3 tumors had a ≥5% annual risk of recurrence during the first 5 years after diagnosis with a peak annual risk of 7.6% during the second year after diagnosis. The annual recurrence risk of for women with grade 1 or 2 tumors was relatively constant at approximately 2% per year beyond the first year after diagnosis. Women who had tumors that expressed low-positive ER levels also had a high early risk of recurrence that peaked during the second year after diagnosis. Pathologic involvement of from 4 to 9 axillary lymph nodes was associated with a high recurrence risk (≥5% per year), which persisted to 10 years after diagnosis. Women who had from 1 to 3 involved lymph nodes had a 3% to 4% per year risk of recurrence that persisted to 10 years. Women who had lymph node-negative tumors had a persistent approximately 2% per year recurrence risk.

Figure 1.

Annual breast cancer recurrence rates for women with grade 3 tumors (n = 1162) or with grade 1 or 2 tumors (n = 2464).

Figure 2.

Annual breast cancer recurrence rates for women with low tumor estrogen receptor (ER) positivity levels (n = 362) or with moderate/high tumor ER level (n = 2773). Fmol/mg indicates femtomoles per milligram cytosol protein; IHC, immunohistochemical.

Figure 3.

Annual breast cancer recurrence rates according to lymph node (N) classification for women with N0 tumors (n = 1819), N1 tumors (n = 1522), or N2 tumors (n = 405).

DISCUSSION

This study demonstrated that, among women with ER-positive, early-stage breast cancer who received treatment with definitive local therapy plus tamoxifen, lymph node-positive status, grade 3 tumors, or low-positive ER level tumors had a high absolute rate of breast cancer recurrence within 2.5 years after diagnosis. Lymph node positivity and low-positive ER levels also were associated significantly with the risk of early recurrence compared with delayed recurrence between 2.5 years and 5 years. The impact on timing of recurrence of locally advanced presentation was not assessed, because these patients were excluded from the study.

Analyses of annual breast cancer recurrence rates demonstrated the differing effects of grade, ER level, and lymph node status on the timing of recurrence. The risk of recurrence for high-grade and low-positive ER tumors peaked during the second year after diagnosis and declined thereafter. In contrast, a high positive lymph node count was associated with a high early recurrence risk that remained high throughout the follow-up interval.

Previous studies demonstrated that high grade is prognostic of poorer outcomes among women with ER-positive early breast cancers.18–20 Low ER levels, as determined by both biochemical and IHC assays21, 22 also reportedly are predictive of poorer outcomes among tamoxifen-treated patients. In a report from the Early Breast Cancer Trialists' Collaborative Group, tamoxifen-treated patients who had tumors with ER levels >100 fmol/mg cytosol of protein experienced fewer recurrences (49% ± 10) compared with patients who had tumors that expressed ER levels from 10 to 100 fmol/mg (36% ± 8).11

Although the benefits of therapy should be determined from prospective trials, the absolute 2.5-year risk estimates presented in Table 2 may allow the calculation of the theoretical absolute benefit of upfront AI in place of tamoxifen. For example, assuming a hazards ratio of 0.83 with upfront AI instead of tamoxifen,4 if a postmenopausal woman with from 4 to 9 positive lymph nodes and a 17.5% risk of recurrence within the first 2.5 years were treated with an AI instead of tamoxifen, then the absolute benefit potentially would be 2.9% at 2.5 years. If a woman with lymph node-negative breast cancer and a 3.9% risk of recurrence within the first 2.5 years were treated with an AI instead of tamoxifen, then her absolute benefit would be only 0.66%.

The risk-benefit calculations described above were based on retrospective data, and only randomized prospective trials can determine whether AI therapy is more beneficial than tamoxifen independent of the risk of early recurrence as identified in the current study. This appears to be the true: In a subset analysis of the Arimidex, Tamoxifen, Alone or in Combination (ATAC) trial, there was a trend toward a benefit for anastrozole in all subgroups based on grade; however, the benefit was significant only in the moderately differentiated subgroup.4, 9 In the lymph node-positive subset, there was a trend toward superiority for anastrozole; however, in the lymph node-negative subset, there was a significant benefit.4, 9 This pattern was reversed in the Breast Intergroup (BIG) 1–98 trial, in which letrozole was superior to tamoxifen in the lymph node-positive subsets but not in the lymph node-negative subsets,1 which the investigators believed was because of a lack of events in the early analysis. It should be noted that these subgroup analyses often were not predefined and were exploratory and under-powered.

Central subset analyses of tumor blocks of patients enrolled in the ATAC and BIG 1–98 trials23, 24 demonstrated that AIs were superior to tamoxifen irrespective of the degree of positivity of ER, HER-2 status, and progesterone receptor (PgR) status. In the neoadjuvant literature, Ellis et al. reported that letrozole was superior to tamoxifen at all levels of ER expression between Allred scores of 3 and 8.25 In the Immediate Preoperative Anastrozole, Tamoxifen, or Combined with Tamoxifen trial, anastrozole was equivalent to tamoxifen at high ER levels and was slightly better than tamoxifen at lower ER levels26; however, those differences were not statistically significant.

Because PgR and HER-2 levels were not available for the majority of patients in the current study, their prognostic value for early recurrence could not be assessed. Among 728 patients with known negative HER-2 status in the current study, there were 75 patients in Group 1 (10.3%), 69 patients in Group 2 (9.5%), and 584 patients in Group 3 (80.2%). This distribution did not differ significantly from distribution in the overall cohort reported herein (7.9%, 7.4%, and 84.7%, respectively). Because of small sample sizes, multivariate analyses of the residual effects of grade, ER level, and the number of involved lymph nodes could not be carried out in the HER-2 negative subset.

To our knowledge, this is the first study to quantify the risk of early recurrence among a geographically defined population of postmenopausal women with ER-positive breast cancer who were treated with tamoxifen. Until recently, the timing of recurrence has not been an important clinical factor in the selection of adjuvant hormone therapy. However, in the era of multiple effective adjuvant hormone strategies, the timing of recurrence may have clinical relevance. The current study illustrates that, among women with ER-positive, tamoxifen-treated, early-stage breast cancer, standard pathologic markers may be used to identify subgroups at higher risk of early recurrence. Current and planned randomized trials, including those evaluating molecular markers as selection criteria for adjuvant therapy, are awaited to better define the optimal short- and long-term hormone therapeutic strategy for patients with ER-positive breast cancer. ER-positive breast cancer undoubtedly is a heterogeneous disease; and, as our understanding of the various subgroups improves, a “1-size-fits-all” approach to hormone therapy may give way to individualized therapeutic strategies.

Ancillary