Abstract

Background

Indications for adjuvant endocrine treatment of breast cancer have gradually increased over the past several years. We aimed to define subgroups of patients who may or may not benefit from adjuvant endocrine therapy.

Methods

A population-based cohort of systemically untreated breast cancer patients (N = 3197) were identified within the registry of the Danish Breast Cancer Cooperative Group (DBCG). The patients were node negative and had estrogen receptor–positive and/or progesterone receptor–positive tumors (except medullary tumors) and were further characterized by the following risk factors: aged 35–74 years (grouped into 5-year categories) at surgery, tumor size (≤20 mm), and histopathology (grade 1 ductal carcinoma, grade 1 or 2 invasive lobular carcinoma, other or unknown histopathology). Standardized mortality ratios (SMRs) were calculated based on the mortality rate (observed number of deaths per 100 000 person-years) among patients relative to the mortality rate in the general population of women (expected number of deaths per 100 000 person-years). The association between standardized mortality ratio and risk factors were analyzed in univariate and multivariable Poisson regression models. All findings were validated in a subsequent DBCG cohort of breast cancer patients (N = 2710).

Results

The median follow-up after surgery was 14.8 years. In the study population there were 970 deaths compared with expected death of 737 women, which was an excess mortality of 233 deaths (SMR = 1.32, 95% CI = 1.24 to 1.40). Mortality rates were 2356 per 100 000 person-years in the study population and 1790 per 100 000 person-years in the general population of women. The mortality rate was associated with larger tumor size (11–20 mm tumors vs 1–10 mm tumors, SMR = 1.42, 95% confidence interval [CI] = 1.31 to 1.53 vs SMR = 1.12, 95% CI = 1.00 to 1.26). The mortality rate was also associated with age (35–59 years, SMR > 1) compared with that in the general population of age-matched women, except for a small subgroup of patients (aged 60–74 years, tumors ≤10 mm, grade 1 ductal carcinoma, and grade 1 or 2 lobular carcinoma: adjusted relative risk = 1.02, 95% CI = 0.89 to 1.16.).

Conclusions

A small subgroup of breast cancer patients who were 60 years or older and had hormone-responsive early-stage tumors up to 10 mm, and received no systemic adjuvant therapy, were not at increased risk of mortality compared with women in this age-group in the general population.

CONTEXT AND CAVEATS
Prior knowledge

The use of endocrine or hormone therapy has increased in breast cancer patients, but what patient subgroups would have the same prognosis as that of the general population (no breast cancer) without such adjuvant therapy is not well studied.

Study design

Standardized mortality ratios were estimated in a population-based Danish Breast Cancer Cooperative Group cohort of systemically untreated breast cancer patients, who had node-negative and hormone receptor–positive tumors, to compare the mortality rate in patients with the mortality rate in general population of Danish women.

Contribution

After a median follow-up of 14.8 years after surgery, mortality rate was increased in patients with larger tumors (11–20 mm) and aged 35–59 years. A small subgroup of women aged 60 years or older who had smaller (≤10 mm) and low histological grade tumors were not at increased risk of mortality compared with the general population.

Implication

The low-risk group of elderly patients (>60 years of age) with hormone-responsive small tumors of low histological grade would not benefit from adjuvant therapy.

Limitation

Lack of data on tumor immunohistochemical characteristics, gene profiling, and some biomarker expression did not allow a more precise analysis.

From the Editors

The meta-analyses performed by the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) have demonstrated a reduction in the annual recurrence rate by 31% in early-stage estrogen receptor–positive (ER + ) breast cancer patients who were treated with tamoxifen for 5 years ( 1 ). In a consensus statement from the 2009 St. Gallen International Breast Cancer Conference, the nodal status, tumor size, histological grade and type, peritumoral vascular invasion, and HER2 and hormone receptor status were categorized as the most useful markers in clinical patient management ( 2 ). The thresholds for the use of endocrine therapy, anti-HER2 therapy, and chemotherapy were defined separately, and targeted therapies against HER2 and ER were considered highly important ( 2 ).

In Denmark, a population-wide database of virtually all women diagnosed with breast cancer, who were treated according to standard national guidelines, was established by the Danish Breast Cancer Cooperative Group (DBCG) in 1977. The database includes information on multiple clinical and tumor characteristics permitting comparisons of mortality between groups. However, mortality reflects the sum of the patient’s baseline risk of dying from breast cancer and her risk of dying from other causes. Therefore, a comparison of the mortality rates of the breast cancer population with that of the general Danish female population will provide a better understanding of the associations between mortality rates and different patient and tumor characteristics. Because low-risk patients with node-negative and ER + tumors in Denmark hitherto solely receive locoregional treatment by surgery and radiotherapy following breast-conserving surgery, it is possible to evaluate the associations of various patient and tumor parameters with prognosis, when no adjuvant chemotherapy or endocrine treatment was given to the patients.

In this study, we aimed to evaluate the associations between risk factors (age at surgery, tumor size, and tumor histopathology) and mortality rate among low-risk patients with node-negative hormone receptor–positive tumors, relative to the mortality rate of the general Danish female population, to define subgroups of patients who, without systemic adjuvant therapy, may not experience a prognosis similar to the background population of women.

Methods

Study and Validation Cohorts

Since the establishment of the DBCG in 1977, virtually all diagnostic and treatment units in Denmark have applied DBCG guidelines for diagnostic procedures, surgery, radiotherapy, systemic therapy, and follow-up for early-stage breast cancer. Diagnostic, therapeutic, and follow-up data have been accumulated prospectively in the DBCG registry by the use of standardized forms ( 3 ).

According to the DBCG guidelines, treatment regimens have been recommended according to risk classifications made by standard algorithms. We selected a study cohort and a validation cohort among patients prospectively classified as low risk in the DBCG database. The low-risk classification was assigned to patients with favorable prognostic characteristics who received no systemic adjuvant treatment.

The study cohort (N = 3197) was selected within the DBCG low-risk program, which included patients with node-negative tumors up to 50 mm. These patients were not given adjuvant endocrine treatment of chemotherapy. The program enrolled patients from November 1, 1989, to April 30, 2001 ( 4 ). The selection criteria of the study cohort were age at surgery 35–74 years, tumor size 20 mm or less, ER + and/or progesterone receptor–positive (PR + ) (except medullary tumors), lymph node negative, four or more excised lymph nodes, surgery by mastectomy or breast-conserving surgery with additional radiotherapy, grade 1 ductal carcinoma, and grade 1 or 2 lobular carcinoma or carcinoma of other type or unknown diagnosis. Among premenopausal patients, only lobular histological grade 1 was included. Exclusion criteria were distant metastases of breast cancer, bilateral breast cancer, inflammatory breast cancer, prior and concurrent malignant disease apart from nonmelanoma skin tumors, in situ cancer of cervix uteri, or death, when observed within 60 days after surgery.

The validation cohort (N = 2710) was selected from subsequent DBCG low-risk programs implemented in 1999, 2001, and 2004 ( 3 ). Selection criteria were the same as the study cohort. The only modification was that patients in whom the axillary lymph node status was determined by sentinel node technique ( 5 ) could be included if the number of excised lymph nodes was one or more. For the validation cohort, the period of surgery was from July 1, 1998, to June 1, 2005. The exclusion criteria were identical to those applied to the study cohort.

Follow-up

DBCG data were linked with data on patient vital status and emigration from the Danish Civil Registration System. Survival time in the study cohort was determined from 60 days after surgery until the date of death, immigration, or end of follow-up (June 1, 2010), whichever came first. Data on mortality rate of the general Danish female population by age and calendar period were obtained from Statistics Denmark ( 6 ). In a previous project, data on disease events after surgery were obtained until April 1, 2007, by linkage to the Danish Hospital Discharge Register and validated by reviewing local medical reports ( 4 ).

Statistical Analysis

To estimate the relative risk of death among women in the study and validation cohorts compared with the general population of women, we calculated the standardized mortality ratios (SMRs), which is the ratio of the observed number of deaths among patients to the expected number of deaths in the general population. The expected number of deaths was estimated by multiplying the survival time accrued from the study and validation cohorts by the mortality rates of the general population of women matched by age (1-year groups) and calendar period (1-year groups). Estimates of standardized mortality ratio greater than unity indicate that breast cancer patients have a higher mortality rate than women of the general population matched on age and calendar year.

The survival time was subdivided according to risk factors (age at surgery, calendar period of surgery, follow-up time after surgery, tumor size, histopathology group, and number of excised lymph nodes), and standardized mortality ratios were calculated accordingly by multiplying survival time with the mortality rates of the general population of women. The age at surgery was categorized in 5-year intervals as follows: 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, and 70–74 years. The calendar period of surgery was subdivided into five intervals (November 1989 to December 1992, January 1993 to December 1995, January 1996 to May 1998, June 1998 to December 2001, and January 2002 to June 2005). The follow-up time began 60 days after surgery and was categorized into six intervals (0.164 to <1, 1 to <2, 2 to <3, 3 to <5, 5 to <10, and ≥10 years after surgery). Tumor size was categorized by maximum tumor size (1–5, 6–10, 11–15, and 16–20 mm). Histopathology group was categorized as grade 1 ductal carcinoma, grade 1 or 2 lobular carcinoma, and carcinoma of other types or unknown diagnoses. The number of excised lymph nodes was categorized as 1–3 (excision by sentinel node technique, only included in the validation cohort), 4–9, and 10 or more lymph nodes. In addition, menopausal status at surgery and type of surgery were included in the characteristics of the study and validation cohorts. The association between standardized mortality ratio and risk factors was analyzed in univariate and multivariable Poisson regression models. Risk factors were evaluated by two-sided likelihood ratio tests, and parameter estimates were reported with 95% likelihood ratio confidence intervals (CIs). P values less than .05 were considered statistically significant. Models were fitted to data, and maximum likelihood estimates of parameters were obtained by the PROC GENMOD of SAS version 9.1 (SAS Institute, Cary, NC). Follow-up time in the study cohort was quantified in terms of a Kaplan–Meier estimate of potential follow-up ( 7 ). The standardized mortality ratio of the study cohort was analyzed in three steps as follows: an initial analysis for reduction of the full model, a validation of the reduced model by comparison of risk factor estimates with estimates obtained in the validation cohort, and a final analysis for determination of threshold values of continuous risk factors.

Initial Analysis of the Full Model.

The full model included all risk factors. The observed and expected mortality rates per 100 000 person-years were reported together with the excess mortality rate and the standardized mortality ratio. The association between standardized mortality ratio and risk factors (age at surgery, tumor size, histopathology group, number of excised lymph nodes, calendar period of surgery, and follow-up time) was analyzed by univariate and multivariable models. In the multivariable models, the association between standardized mortality ratio and two-factor interactions between a subgroup of prognostic factors (age at surgery, tumor size, histopathology group, and number of excised lymph nodes) as well as interactions between prognostic factors and follow-up time was investigated in separate models. In these analyses, tumor size was grouped into 1–10 and 11–20 mm to increase the power of interaction tests. In all analyses, the interaction terms were tested for statistical significance ( P < .05). The risk factors that were non-statistically significantly ( P > .05) associated with standardized mortality ratio in the full model were removed to develop a reduced model.

Validation of the Reduced Model.

Validation of the reduced model was done by independent analysis of the validation cohort. Because of the shorter follow-up period of the validation group, the follow-up period in the validation analysis were restricted to 5 years after surgery in both the validation and study cohorts. Similar estimates of risk in the validation and the study cohorts for each risk factor were taken as confirmation of the risk factor association with standardized mortality ratio.

Final Analysis.

In the final analysis of the study cohort, we included age at surgery, tumor size, and histopathology. To determine threshold values for tumor size and age at surgery, these were classified by hierarchical indicator variables ( 8 ) (tumor size ≥6, ≥11, and ≥16 mm; and age at surgery ≥40, ≥45, ≥50, ≥55, ≥60, ≥65, and ≥70 years). Multistep backward selection was used to remove non-statistically significant indicator variables ( P > .05) from the model. The remaining variables were considered to identify subgroups with different standardized mortality ratios. The overall level of statistical significance of the risk factors was assessed by tests of global hypotheses of no difference among age-groups, tumor size groups, and histopathology groups, respectively.

The pattern of first events after surgery (locoregional and distant recurrence, contralateral breast cancer, second primary cancers, and death as first event) was analyzed by cumulative incidence functions ( 9 ). Cumulative incidences at 10 years after surgery were determined by the cmprsk-package of R-software (version 2.9; R Development Core Team, Vienna, Austria) and 95% pointwise confidence intervals based on the ln(−ln) transformation. The cumulative incidence functions were further analyzed by proportional hazards models for subdistribution ( 10 ), which were used to model age and tumor size effects directly on the cumulative incidence functions. The model assumption of proportional subdistribution hazards was investigated by plots of Schoenfeld type residuals vs time to events, and the hypotheses of no effect of age at surgery and no effect of tumor size, respectively, were tested by likelihood ratio tests. All statistical tests were two-sided, and all P values less than .05 were considered statistically significant.

Results

Characteristics of the Study and the Validation Cohorts

The characteristics of the study (N = 3197) and the validation cohorts (N = 2710) are shown in Table 1 . Age at surgery, menopausal status, histopathology group, and ER and/or PR status were similar in the two cohorts. Smaller tumors of 1–10 mm size were observed more often in patients in the validation cohort (41.1%) compared with patients in the study cohort (36.0%). Breast-conserving surgery was also noted more often in patients in the validation cohort (60.3%) compared with the study cohort (37.1%).

Table 1

Characteristics of low-risk patients in the Danish Breast Cancer Cooperative Group registry with diagnosis of primary invasive breast cancer *

 Study cohort Validation cohort 
Characteristic No. (%) No. (%) 
Total 3197 (100.0) 2710 (100.0) 
Calendar period of surgery   
    November 1989 to December 1992 726 (22.7) — 
    January 1993 to December 1995 1080 (33.8) — 
    January 1996 to May 1998 969 (30.3) — 
    June 1998 to December 2001 422 (13.2) 1208 (44.6) 
    January 2002 to June 2005 — 1502 (55.4) 
Age-group at surgery, y 
    35–39 77 (2.4) 54 (2.0) 
    40–44 229 (7.2) 175 (6.5) 
    45–49 458 (14.3) 331 (12.2) 
    50–54 520 (16.3) 444 (16.4) 
    55–59 510 (16.0) 500 (18.5) 
    60–64 553 (17.3) 488 (18.0) 
    65–69 479 (15.0) 398 (14.7) 
    70–74 371 (11.6) 320 (11.8) 
Menopausal status at surgery 
    Premenopausal 871 (27.2) 805 (29.7) 
    Postmenopausal 2326 (72.8) 1905 (70.3) 
Surgery 
    Mastectomy 2010 (62.9) 1077 (39.7) 
    Breast-conserving surgery 1187 (37.1) 1633 (60.3) 
Tumor size, mm 
    1–5 208 (6.5) 198 (7.3) 
    6–10 943 (29.5) 916 (33.8) 
    11–15 1160 (36.3) 1012 (37.3) 
    16–20 886 (27.7) 584 (21.5) 
Excised lymph nodes, No. 
    1–3 — 589 (21.7) 
    4–9 1029 (32.2) 403 (14.9) 
    ≥10 2168 (67.8) 1718 (63.4) 
Histopathology † 
    Grade 1 ductal carcinoma  2126 (66.5) ‡  1788 (66.0) § 
    Grade 1 or 2 lobular carcinoma  543 (17.0) ‖  513 (18.9) ¶ 
    Other type or unknown 528 (16.5) 409 (15.1) 
ER+ and/or PR+ # 
    No 71 (2.2) 2 (0.1) 
    Yes 3110 (97.3) 2707 (99.9) 
Not known 16 (0.5) 1 (0.0) 
 Study cohort Validation cohort 
Characteristic No. (%) No. (%) 
Total 3197 (100.0) 2710 (100.0) 
Calendar period of surgery   
    November 1989 to December 1992 726 (22.7) — 
    January 1993 to December 1995 1080 (33.8) — 
    January 1996 to May 1998 969 (30.3) — 
    June 1998 to December 2001 422 (13.2) 1208 (44.6) 
    January 2002 to June 2005 — 1502 (55.4) 
Age-group at surgery, y 
    35–39 77 (2.4) 54 (2.0) 
    40–44 229 (7.2) 175 (6.5) 
    45–49 458 (14.3) 331 (12.2) 
    50–54 520 (16.3) 444 (16.4) 
    55–59 510 (16.0) 500 (18.5) 
    60–64 553 (17.3) 488 (18.0) 
    65–69 479 (15.0) 398 (14.7) 
    70–74 371 (11.6) 320 (11.8) 
Menopausal status at surgery 
    Premenopausal 871 (27.2) 805 (29.7) 
    Postmenopausal 2326 (72.8) 1905 (70.3) 
Surgery 
    Mastectomy 2010 (62.9) 1077 (39.7) 
    Breast-conserving surgery 1187 (37.1) 1633 (60.3) 
Tumor size, mm 
    1–5 208 (6.5) 198 (7.3) 
    6–10 943 (29.5) 916 (33.8) 
    11–15 1160 (36.3) 1012 (37.3) 
    16–20 886 (27.7) 584 (21.5) 
Excised lymph nodes, No. 
    1–3 — 589 (21.7) 
    4–9 1029 (32.2) 403 (14.9) 
    ≥10 2168 (67.8) 1718 (63.4) 
Histopathology † 
    Grade 1 ductal carcinoma  2126 (66.5) ‡  1788 (66.0) § 
    Grade 1 or 2 lobular carcinoma  543 (17.0) ‖  513 (18.9) ¶ 
    Other type or unknown 528 (16.5) 409 (15.1) 
ER+ and/or PR+ # 
    No 71 (2.2) 2 (0.1) 
    Yes 3110 (97.3) 2707 (99.9) 
Not known 16 (0.5) 1 (0.0) 
*

Diagnosis based on the World Health Organization (WHO) classification of tumors ( 11 ). According to the Danish Breast Cancer Cooperative Group guidelines, patients were not treated with adjuvant medical treatment. Exclusion criteria were distant metastases of breast cancer, bilateral breast cancer, inflammatory breast cancer, prior and concurrent malignant disease apart from nonmelanoma skin tumors and in situ cancer of cervix uteri, or death, when observed within 60 days after surgery. ER+ = estrogen receptor–positive; PR+ = progesterone receptor–positive; — = not applicable.

Histological grading according to Elston and Ellis ( 12 ).

Grade unknown or not determinable included (n = 98 tumors).

§

Grade unknown or not determinable included (n = 49 tumors).

Grade unknown or not determinable included (n = 307 tumors).

Grade unknown or not determinable included (n = 263 tumors).

#

Medullary tumors with negative or not known hormone receptor status included.

Analysis of Mortality in the Study Cohort

The median potential follow-up time after surgery was 14.8 years (95% CI = 14.6 to 15.0 years), and 41 167 person-years were at risk of death. Because 970 women died in the study cohort compared with the expected death of 737 women in the age-matched general population of women, the excess mortality was 233 deaths (SMR = 1.32, 95% CI = 1.24 to 1.40). Mortality rates were 2356 per 100 000 person-years in the study population and 1790 per 100 000 person-years in the general population of women. The observed and expected mortality rates according to age at surgery, tumor size and histopathology are shown in Figure 1 . The excess mortality rate was highest for younger patients aged 35–39 years (1054 deaths per 100 000 person-years; SMR = 5.53, 95% CI = 3.11 to 8.95) and lowest for patients aged 60–64 years (301 deaths per 100 000 person-years; SMR = 1.14, 95% CI = 0.98 to 1.32) ( Table 2 ).

Table 2

Mortality in the study cohort compared with mortality in the general population of women *

Risk factor Person-years at risk  Mortality rate per 100 000 person-years
 
Relative mortality  Adjusted RR of mortality
 
Observed Expected Excess SMR (95% CI) RR (95% CI) P† 
Total 41 167 2356 1790 566 1.32 (1.24 to 1.40)   
Age at surgery, y       <.001 
    35–39 1088 1287 233 1054 5.53 (3.11 to 8.95) 4.88 (2.72 to 8.01)  
    40–44 3258 952 356 595 2.67 (1.84 to 3.72) 2.34 (1.59 to 3.31)  
    45–49 6425 1043 522 521 2.00 (1.56 to 2.52) 1.72 (1.31 to 2.21)  
    50–54 7060 1445 787 658 1.84 (1.50 to 2.22) 1.60 (1.28 to 1.98)  
    55–59 6709 1833 1291 542 1.42 (1.18 to 1.69) 1.26 (1.02 to 1.53)  
    60–64 7088 2469 2167 301 1.14 (0.98 to 1.32) 0.98 (0.82 to 1.18)  
    65–69 5593 4219 3386 833 1.25 (1.09 to 1.41) 1.08 (0.91 to 1.27)  
    70–74 3946 5626 5170 456 1.09 (0.95 to 1.24) 0.92 (0.77 to 1.10)  
Tumor size, mm       <.001 
    1–10 15 147 1888 1679 209 1.12 (1.00 to1.26) 1 (referent)  
    11–20 26 021 2629 1854 774 1.42 (1.31 to 1.53) 1.28 (1.12 to 1.48)  
Histopathology ‡       .0049 
    Grade 1 ductal carcinoma 27 491 2357 1743 614 1.35 (1.25 to 1.46) 1 (referent)  
    Grade 1 or 2 lobular carcinoma 6759 2796 1913 883 1.46 (1.26 to 1.68) 1.09 (0.93 to 1.28)  
    Other or unknown 6918 1923 1855 67 1.04 (0.87 to 1.22) 0.77 (0.64 to 0.93)  
Risk factor Person-years at risk  Mortality rate per 100 000 person-years
 
Relative mortality  Adjusted RR of mortality
 
Observed Expected Excess SMR (95% CI) RR (95% CI) P† 
Total 41 167 2356 1790 566 1.32 (1.24 to 1.40)   
Age at surgery, y       <.001 
    35–39 1088 1287 233 1054 5.53 (3.11 to 8.95) 4.88 (2.72 to 8.01)  
    40–44 3258 952 356 595 2.67 (1.84 to 3.72) 2.34 (1.59 to 3.31)  
    45–49 6425 1043 522 521 2.00 (1.56 to 2.52) 1.72 (1.31 to 2.21)  
    50–54 7060 1445 787 658 1.84 (1.50 to 2.22) 1.60 (1.28 to 1.98)  
    55–59 6709 1833 1291 542 1.42 (1.18 to 1.69) 1.26 (1.02 to 1.53)  
    60–64 7088 2469 2167 301 1.14 (0.98 to 1.32) 0.98 (0.82 to 1.18)  
    65–69 5593 4219 3386 833 1.25 (1.09 to 1.41) 1.08 (0.91 to 1.27)  
    70–74 3946 5626 5170 456 1.09 (0.95 to 1.24) 0.92 (0.77 to 1.10)  
Tumor size, mm       <.001 
    1–10 15 147 1888 1679 209 1.12 (1.00 to1.26) 1 (referent)  
    11–20 26 021 2629 1854 774 1.42 (1.31 to 1.53) 1.28 (1.12 to 1.48)  
Histopathology ‡       .0049 
    Grade 1 ductal carcinoma 27 491 2357 1743 614 1.35 (1.25 to 1.46) 1 (referent)  
    Grade 1 or 2 lobular carcinoma 6759 2796 1913 883 1.46 (1.26 to 1.68) 1.09 (0.93 to 1.28)  
    Other or unknown 6918 1923 1855 67 1.04 (0.87 to 1.22) 0.77 (0.64 to 0.93)  
*

Estimates from univariate and multivariable analyses of a reduced model where the number of excised lymph nodes, calendar period of surgery, and follow-up time after surgery were excluded. The exclusion of these factors did not affect the estimates of age, tumor size, and histopathology group. The RRs of different age-groups are shown for grade 1 ductal carcinoma tumors of 1–10 mm. CI = confidence interval; RR = adjusted relative risk of mortality; SMR = standardized mortality ratio.

P values were calculated using two-sided likelihood ratio tests.

Histological grading according to Elston and Ellis ( 12 ).

Figure 1

Observed and expected mortality rates per 100 000 person-years at risk for breast cancer patients. The study cohort of node-negative patients with estrogen receptor–positive and/or progesterone receptor–positive tumors who received no systemic adjuvant treatment were identified within the registry of the Danish Breast Cancer Cooperative Group and were further characterized by age at surgery, tumor size, and histopathology. The median potential follow-up was 14.8 years after surgery. The error bars represent 95% confidence intervals.

Figure 1

Observed and expected mortality rates per 100 000 person-years at risk for breast cancer patients. The study cohort of node-negative patients with estrogen receptor–positive and/or progesterone receptor–positive tumors who received no systemic adjuvant treatment were identified within the registry of the Danish Breast Cancer Cooperative Group and were further characterized by age at surgery, tumor size, and histopathology. The median potential follow-up was 14.8 years after surgery. The error bars represent 95% confidence intervals.

The association between relative mortality and age at surgery and tumor pathology in the study cohort was investigated by univariate analysis. The relative mortality was statistically significantly greater than unity for all age-groups below 60 years ( Table 2 ). The relative mortality was also greater than unity in patients with tumors of 11–20 mm size (SMR = 1.42, 95% CI = 1.31 to 1.53), whereas it did not differ from unity in patients with smaller tumors of 1–10 mm size (SMR = 1.12, 95% CI = 1.00 to 1.26). For patients with grade 1 ductal carcinoma and grade 1 or 2 lobular carcinoma tumors, the relative mortality was greater than unity, whereas it was not statistically different from unity in patients with other tumor types or unknown histology ( Table 2 ). The relative mortality did not differ statistically significantly from unity during the first 3 years after surgery (first year, SMR = 0.72, 95% CI = 0.45 to 1.07; second year, SMR = 0.89, 95% CI = 0.62 to 1.22; and third year, SMR = 1.01, 95% CI = 0.73 to 1.36), whereas an increased relative mortality compared with the general population was observed 3–5 years after surgery (SMR = 1.24, 95% CI = 1.01 to 1.49), 5–10 years after surgery (SMR = 1.39, 95% CI = 1.25 to 1.54), and 10 years or more after surgery (SMR = 1.43, 95% CI = 1.30 to 1.57) (data not shown in Table 2 ).

The association between relative mortality and risk factors was further investigated in multivariable analyses ( Table 2 ). A statistically significant association was detected between relative mortality and age at surgery ( P < .001), tumor size ( P < .001 for 11–20 vs 1–10 mm [referent]), and histopathology group ( P = .0049). No statistically significant association was observed with the period of surgery ( P = .51) and number of excised lymph nodes ( P = .33), and consequently these were excluded from the model. Follow-up period was also excluded from the model even though the association with relative mortality was statistically significant ( P = .0011), as this did not modify the estimates for age at surgery, tumor size, or histopathology group. According to the reduced model, patients with 11–20 mm tumors had an increased risk of mortality (relative risk [RR] = 1.28, 95% CI = 1.12 to 1.48) compared with patients with 1–10 mm tumors (referent). For patients with grade 1 ductal carcinoma (referent) and grade 1 or 2 lobular carcinoma, the risk of mortality was statistically significant higher (RR = 1.09, 95% CI = 0.93 to 1.28) than patients with other or unknown histopathology (RR = 0.77, 95% CI = 0.64 to 0.93) ( Table 2 ). The estimate of relative risk for age at surgery was for the age-group in question equivalent to the estimate of relative mortality for patients with grade I ductal carcinoma and tumor size 1–10 mm and grade 1 ductal carcinoma. Thus, for patients aged 35–59 years with small tumors and grade 1 ductal carcinoma, the relative mortalities were substantially larger than unity. As an example, the predicted mortality rate was 1.26 times higher than the mortality rate in the background population (RR = 1.26, 95% CI = 1.02 to 1.53) for patients aged 55–59 years. The two-way interactions among prognostic factors (age at surgery, tumor size, histopathology, and number of excised lymph nodes) as well as their interactions with follow-up time were investigated, and none was statistically significant.

Analysis of the Reduced Model for Relative Mortality in the Study and the Validation Cohorts

The validation analysis of relative mortality was done to confirm the findings in the analysis of the study cohort. As the follow-up in the validation cohort was limited to 5 years after surgery, the follow-up in the study cohort had to be reduced correspondingly. At follow-up until 5 years after surgery, the relative mortality in the validation cohort (SMR = 1.05, 95% CI = 0.89 to 1.24) was similar to that of the study cohort (SMR = 1.04, 95% CI = 0.90 to 1.19) (data not shown). A multivariable analysis of the reduced model, which included age at surgery, tumor size, and histopathology, showed that the estimates of age effects had a similar trend of decreasing relative risk of mortality with increasing age in the study and the validation cohorts. For tumor size, the relative risk of mortality was consistently higher for large tumors (11–20 mm) in both the study (RR = 1.32, 95% CI = 0.97 to 1.82) and the validation (RR = 1.21, 95% CI = 0.87 to 1.71) cohorts compared with small tumors (1–10 mm; referent). For histopathology group, the relative risk of mortality differed between the two populations (study cohort: grade 1 or 2 lobular carcinoma > other or unknown types of histology > grade 1 ductal carcinoma; validation cohort: grade 1 ductal carcinoma > other or unknown types of histology > grade 1 or 2 lobular carcinoma). The most important difference was a decrease in mortality risk in the validation cohort for the grade 1 or 2 lobular carcinoma group (adjusted RR = 0.64, 95% CI = 0.40 to 0.99) compared with the study cohort (adjusted RR = 1.17, 95% CI = 0.81 to 1.65) ( Table 3 ).

Table 3

Validation of the reduced multivariable model of mortality risk in the low-risk study cohort *

  Adjusted RR of mortality
 
  Study cohort †
 
Validation cohort
 
Risk factor RR (95% CI) P‡ RR (95% CI) P‡ 
Age at surgery, y  <.001  .009 
    35–39 7.73 (2.70 to 17.36)  3.65 (0.21 to 16.32)  
    40–44 2.98 (1.39 to 5.62)  3.82 (1.48 to 8.08)  
    45–49 1.06 (0.51 to 1.95)  1.66 (0.73 to 3.24)  
    50–54 0.73 (0.38 to 1.30)  1.01 (0.49 to 1.85)  
    55–59 0.84 (0.51 to 1.33)  1.42 (0.88 to 2.19)  
    60–64 0.71 (0.46 to 1.06)  1.36 (0.90 to 2.00)  
    65–69 0.79 (0.54 to 1.13)  0.85 (0.54 to 1.27)  
    70–74 0.77 (0.53 to 1.11)  0.77 (0.50 to 1.15)  
Tumor size, mm  .075  .26 
    1–10 1 (referent)  1 (referent)  
    11–20 1.32 (0.97 to 1.82)  1.21 (0.87 to 1.71)  
Histopathology §  .69  .075 
    Grade 1 ductal carcinoma 1 (referent)  1 (referent)  
    Grade 1 or 2 lobular carcinoma 1.17 (0.81 to 1.65)  0.64 (0.40 to 0.99)  
    Other or unknown 1.02 (0.69 to 1.48)  0.7 (0.41 to 1.14)  
  Adjusted RR of mortality
 
  Study cohort †
 
Validation cohort
 
Risk factor RR (95% CI) P‡ RR (95% CI) P‡ 
Age at surgery, y  <.001  .009 
    35–39 7.73 (2.70 to 17.36)  3.65 (0.21 to 16.32)  
    40–44 2.98 (1.39 to 5.62)  3.82 (1.48 to 8.08)  
    45–49 1.06 (0.51 to 1.95)  1.66 (0.73 to 3.24)  
    50–54 0.73 (0.38 to 1.30)  1.01 (0.49 to 1.85)  
    55–59 0.84 (0.51 to 1.33)  1.42 (0.88 to 2.19)  
    60–64 0.71 (0.46 to 1.06)  1.36 (0.90 to 2.00)  
    65–69 0.79 (0.54 to 1.13)  0.85 (0.54 to 1.27)  
    70–74 0.77 (0.53 to 1.11)  0.77 (0.50 to 1.15)  
Tumor size, mm  .075  .26 
    1–10 1 (referent)  1 (referent)  
    11–20 1.32 (0.97 to 1.82)  1.21 (0.87 to 1.71)  
Histopathology §  .69  .075 
    Grade 1 ductal carcinoma 1 (referent)  1 (referent)  
    Grade 1 or 2 lobular carcinoma 1.17 (0.81 to 1.65)  0.64 (0.40 to 0.99)  
    Other or unknown 1.02 (0.69 to 1.48)  0.7 (0.41 to 1.14)  
*

The RR of different age-groups is shown for grade 1 ductal carcinoma of 1–10 mm. CI = confidence interval; RR = adjusted relative risk of mortality.

Follow-up of 0.164–5 years after surgery.

P values were calculated using two-sided likelihood ratio tests.

§

Histological grading according to Elston and Ellis ( 12 ).

Final Analysis of Mortality in the Study Cohort After Validation

Final analysis was done to determine the cut points for the continuous risk factors age at surgery and tumor size. Histopathology group was maintained in the model, although the validation analysis showed inconsistent results at 5 years of follow-up. The multivariable analysis of the hierarchical indicator variables showed that increase in age at surgery was associated with decrease in risk of mortality (RR at age ≥40 years higher than age ≥55 years higher than age ≥60 years) and larger tumor size (≥11 mm) was associated with an increased risk of mortality. In Table 4 , relative risk values are presented for the corresponding age and tumor size categories. The categorization of histopathology group was reduced to two levels as follows: histological grade 1 ductal carcinoma and grade 1 and 2 lobular carcinoma (referent) vs other or unknown histology. Patients of other or unknown histopathology had a lower relative risk of mortality (RR = 0.76, 95% CI = 0.63 to 0.90) compared with patients of histological grade 1 ductal carcinoma and grade 1 and 2 lobular carcinoma (referent). For patients of referent histopathology, the relative risk according to age at surgery and tumor size is shown in Figure 2 . The adjusted relative risk of mortality was substantially larger than unity for younger patients aged 35–39 years and statistically significantly larger than unity for all patients, except those aged 60–74 years with small tumors of 1–10 mm size (RR = 1.02, 95% CI = 0.89 to 1.16).

Table 4

Final multivariable model of risk of mortality in the low-risk study cohort *

  Adjusted RR of mortality
 
Risk factor RR (95% CI) P† 
Age at surgery, y <.001 
    35–39 4.96 (2.77 to 8.14)  
    40–54 1.76 (1.48 to 2.08)  
    55–59 1.28 (1.04 to 1.56)  
    60–74 1.02 (0.89 to 1.16)  
Tumor size, mm <.001 
    1–10 1 (referent)  
    11–20 1.28 (1.11 to 1.47)  
Histopathology ‡  .002 
    Grade 1 ductal carcinoma or grade 1 or 2 lobular carcinoma 1 (referent)  
    Other or unknown 0.76 (0.63 to 0.90)  
  Adjusted RR of mortality
 
Risk factor RR (95% CI) P† 
Age at surgery, y <.001 
    35–39 4.96 (2.77 to 8.14)  
    40–54 1.76 (1.48 to 2.08)  
    55–59 1.28 (1.04 to 1.56)  
    60–74 1.02 (0.89 to 1.16)  
Tumor size, mm <.001 
    1–10 1 (referent)  
    11–20 1.28 (1.11 to 1.47)  
Histopathology ‡  .002 
    Grade 1 ductal carcinoma or grade 1 or 2 lobular carcinoma 1 (referent)  
    Other or unknown 0.76 (0.63 to 0.90)  
*

The final analysis for the association between the hierarchical indicator variables (age ≥ 40 years, ≥ 55 years, and ≥ 60 years; tumor size ≥ 11 mm) and the RR of mortality. Follow-up was until June 1, 2010. CI = confidence interval; RR = adjusted relative risk of mortality.

P values were calculated using two-sided likelihood ratio tests.

Histological grading according to Elston and Ellis ( 12 ).

Figure 2

Association between adjusted relative risk (RR) of mortality and age at surgery and tumor size in the final multivariable model. The referent histopathology is grade 1 ductal carcinoma and grade 1 or 2 lobular carcinoma. For patients of referent histopathology, the horizontal line (adjusted RR = 1) is equivalent of a mortality similar to the mortality of women of the general population of women matched by age and calendar year. The error bars represent 95% confidence intervals. The median potential follow-up was 14.8 years after surgery.

Figure 2

Association between adjusted relative risk (RR) of mortality and age at surgery and tumor size in the final multivariable model. The referent histopathology is grade 1 ductal carcinoma and grade 1 or 2 lobular carcinoma. For patients of referent histopathology, the horizontal line (adjusted RR = 1) is equivalent of a mortality similar to the mortality of women of the general population of women matched by age and calendar year. The error bars represent 95% confidence intervals. The median potential follow-up was 14.8 years after surgery.

The analyses of associations between relative mortality and risk factors were supplemented by an analysis of recurrence pattern. The aim was not to make a comprehensive analysis of recurrence pattern but to give information on disease progression, which is the likely cause of excess mortality in breast cancer patients. In consequence, the association between the first disease events after surgery was investigated only for the validated risk factors age at surgery and tumor size, which were categorized as in the final model of relative mortality. At 10 years after surgery, the risk of distant recurrences were statistically significantly higher ( P < .001) among patients with large tumors of 11–20 mm size (ranging from 6.6% to 13.6%) than among patients with small tumors of 1–10 mm size (ranging from 2.3% to 8.0%) ( Table 5 and Figure 3 ). Locoregional recurrences were statistically significantly more common for younger patients aged 35–54 years (likelihood ratio test, test statistic D = 21.2, df = 1, P < .001) than for older patients aged 55–74 years and also more common for patients with large tumors than for patients with small tumors ( P = .023). Regarding nonrecurrence events, contralateral breast cancer did not differ among patients based on age and tumor size, whereas second primary cancer ( P < .001) and death as first event ( P < .001) were more common in older patients compared with younger patients. In the interpretation of breast cancer recurrence rates, it should be noted that there is a pronounced difference in the sum of the rates of competing events (death, secondary primary cancer, and contralateral breast cancer) between older patients aged 60–74 years (29.1%) and younger patients aged 35–54 years (9.7%).

Table 5

Cumulative incidence of first disease events in the low-risk study cohort until 10 years after surgery by age at surgery and tumor size *

Age at surgery, y Tumor size, mm No. of patients  Locoregional recurrence †  Distant recurrence ‡  Total recurrence §  Contralateral breast cancer ‖  Second primary cancer ¶  Death as first event # 
% (95% CI) % (95% CI) % (95% CI) % (95% CI) % (95% CI) % (95% CI) 
35–39 1–10 25 21.5 (7.4 to 40.4) 8.0 (1.3 to 22.9) 29.5 (12.5 to 48.8) 8.2 (1.3 to 23.5) 4.2 (0.3 to 18.3) 0.0 (0.0 to 0.0) 
 11–20 52 9.6 (3.5 to 19.5) 13.6 (5.9 to 24.4) 23.2 (12.7 to 35.5) 6.0 (1.5 to 15.0) 1.9 (0.2 to 9.0) 1.9 (0.2 to 9.0) 
40–54 1–10 460 5.4 (3.5 to 7.8) 2.3 (1.2 to 4.1) 7.7 (5.5 to 10.5) 3.1 (1.8 to 5.0) 3.6 (2.1 to 5.8) 2.5 (1.3 to 4.3) 
 11–20 747 9.5 (7.5 to 11.7) 8.0 (6.1 to 10.1) 17.4 (14.7 to 20.3) 4.8 (3.4 to 6.5) 3.0 (1.9 to 4.4) 2.1 (1.2 to 3.3) 
55–59 1–10 208 2.9 (1.2 to 6.0) 7.0 (4.0 to 11.2) 10.0 (6.3 to 14.7) 8.6 (5.2 to 13.0) 5.9 (3.2 to 9.7) 4.7 (2.3 to 8.4) 
 11–20 302 5.0 (2.9 to 7.9) 6.6 (4.1 to 9.9) 11.6 (8.2 to 15.6) 5.8 (3.5 to 9.0) 7.9 (5.1 to 11.5) 6.0 (3.6 to 9.2) 
60–74 1–10 458 4.0 (2.5 to 6.1) 3.8 (2.3 to 5.9) 7.8 (5.6 to 10.6) 5.4 (3.5 to 7.7) 8.3 (5.9 to 11.2) 11.5 (8.7 to 14.7) 
 11–20 945 4.4 (3.2 to 5.8) 8.9 (7.1 to 10.8) 13.2 (11.1 to 15.5) 6.2 (4.8 to 7.9) 9.1 (7.4 to 11.2) 15.7 (13.4 to 18.2) 
Age at surgery, y Tumor size, mm No. of patients  Locoregional recurrence †  Distant recurrence ‡  Total recurrence §  Contralateral breast cancer ‖  Second primary cancer ¶  Death as first event # 
% (95% CI) % (95% CI) % (95% CI) % (95% CI) % (95% CI) % (95% CI) 
35–39 1–10 25 21.5 (7.4 to 40.4) 8.0 (1.3 to 22.9) 29.5 (12.5 to 48.8) 8.2 (1.3 to 23.5) 4.2 (0.3 to 18.3) 0.0 (0.0 to 0.0) 
 11–20 52 9.6 (3.5 to 19.5) 13.6 (5.9 to 24.4) 23.2 (12.7 to 35.5) 6.0 (1.5 to 15.0) 1.9 (0.2 to 9.0) 1.9 (0.2 to 9.0) 
40–54 1–10 460 5.4 (3.5 to 7.8) 2.3 (1.2 to 4.1) 7.7 (5.5 to 10.5) 3.1 (1.8 to 5.0) 3.6 (2.1 to 5.8) 2.5 (1.3 to 4.3) 
 11–20 747 9.5 (7.5 to 11.7) 8.0 (6.1 to 10.1) 17.4 (14.7 to 20.3) 4.8 (3.4 to 6.5) 3.0 (1.9 to 4.4) 2.1 (1.2 to 3.3) 
55–59 1–10 208 2.9 (1.2 to 6.0) 7.0 (4.0 to 11.2) 10.0 (6.3 to 14.7) 8.6 (5.2 to 13.0) 5.9 (3.2 to 9.7) 4.7 (2.3 to 8.4) 
 11–20 302 5.0 (2.9 to 7.9) 6.6 (4.1 to 9.9) 11.6 (8.2 to 15.6) 5.8 (3.5 to 9.0) 7.9 (5.1 to 11.5) 6.0 (3.6 to 9.2) 
60–74 1–10 458 4.0 (2.5 to 6.1) 3.8 (2.3 to 5.9) 7.8 (5.6 to 10.6) 5.4 (3.5 to 7.7) 8.3 (5.9 to 11.2) 11.5 (8.7 to 14.7) 
 11–20 945 4.4 (3.2 to 5.8) 8.9 (7.1 to 10.8) 13.2 (11.1 to 15.5) 6.2 (4.8 to 7.9) 9.1 (7.4 to 11.2) 15.7 (13.4 to 18.2) 
*

First disease event or follow-up until April 1, 2007, whichever came first. The cumulative incidence values were calculated by cumulative incidence functions ( 9 ) to analyze recurrence pattern and mortality. Associations of first disease events with age at surgery, tumor size, and age × size interaction were analyzed by proportional hazards models for subdistribution ( 10 ), and the hypotheses of no association were tested by likelihood ratio tests. CI = confidence interval.

Two-sided likelihood ratio test: age (3 df ): P < .001; size (1 df ): P = .023; and interaction age by size (3 df ): P = .17.

Two-sided likelihood ratio test: age (3 df ): P = .19; size (1 df ): P < .001; and interaction age by size (3 df ): P = .049.

§

Two-sided likelihood ratio test: age (3 df ): P = .005; size (1 df ): P < .001; and interaction age by size (3 df ): P = .10.

Two-sided likelihood ratio test: age (3 df ): P = .078; size (1 df ): P = .64; and interaction age by size (3 df ): P = .38.

Two-sided likelihood ratio test: age (3 df ): P < .001; size (1 df ): P = .68; and interaction age by size (3 df ): P = .82.

#

Two-sided likelihood ratio test: age (3 df ): P < .001; size (1 df ): P = .053; and interaction age by size (3 df ): P = .56.

Figure 3

Cumulative incidence of the first events 10 years after surgery. The first events were locoregional and distant recurrence, contralateral breast cancer, second primary cancers, and death as first event. Total recurrence is the sum of locoregional and distant recurrence.The error bars represent 95% confidence intervals.

Figure 3

Cumulative incidence of the first events 10 years after surgery. The first events were locoregional and distant recurrence, contralateral breast cancer, second primary cancers, and death as first event. Total recurrence is the sum of locoregional and distant recurrence.The error bars represent 95% confidence intervals.

Discussion

In this large population-based cohort study, we showed that subgroups of breast cancer patients with node-negative, hormone receptor–positive low-grade tumors up to 20 mm had mortality rates higher than the general population of women. Age less than 60 years at diagnosis and tumor size greater than 10 mm were independently associated with a worse prognosis. Younger age at surgery was associated with a progressively higher relative mortality, whereas the number of excised lymph nodes was not statistically significantly associated with relative mortality. Results regarding the estimated effects of age and tumor size are robust and were validated in an independent validation cohort, whereas the data of histopathologic subgroups in the study and validation cohorts were inconsistent. The worse prognosis for patients with large tumors compared with small tumors is in accordance with the finding of a larger rate of distant recurrences for large tumors compared with small tumors at 10 years after surgery. Age was not associated with distant recurrence in this study, however, the high relative mortality for younger patients is in accordance with a larger rate of locoregional recurrence and a low mortality rate of the age-matched general population of women.

Very young age at diagnosis of breast cancer, in particular, is associated with poor survival ( 13 , 14 ). This increased mortality is, to some extent, explained by presentation at a later stage and by known and more adverse negative prognostic factors among patients up to 40 years of age—number of positive lymph nodes, tumor size, negative hormone receptor status, overexpression of HER2, and markers of increased proliferation ( 15 ). However, systemic adjuvant treatment to some extent reduces the poor prognostic value of young age. From the population-based study by Kroman et al. ( 13 ), it was concluded that diagnosis of breast cancer at young age was associated with an increased risk of death, with women younger than 35 years at diagnosis having the worst prognosis of all age groups. They also found that the negative prognostic impact of age less than 45 years, to a large extent, was counteracted by adjuvant cytotoxic treatment. The data from that study ( 13 ) cannot be directly compared with the current study, as different analytical approaches were applied. Kroman et al. ( 13 ) estimated the relative excess mortality rates, whereas the current study estimated standardized mortality ratios. A biphasic relationship between age and mortality was described by Tai et al. ( 16 ) based on a large number of patients in the Surveillance, Epidemiology, and End Results (SEER) Program of the United States. In the current study, we took into account the mortality of the background population and showed that the increased mortality in elderly patients (>60 years of age) predominately reflects non-breast cancer mortality, as the relative risk of dying for these patients were not increased. Thus, the data strongly suggest that breast cancer–specific mortality in node-negative patients decreases with age. Interestingly, in a review article ( 17 ) on node-negative small tumors up to 10 mm (T1a,bN0M0), Hanrahan et al. ( 17 ) found that age less than 50 years at diagnosis was associated with poor prognosis. Other predictors were high tumor grade, ER tumor, high Ki67 index, HER2 augmentation, and larger tumors within the T1a,b subgroup.

In our study cohort, we found a similar relative risk in patients with invasive lobular carcinomas compared with invasive ductal carcinomas both after 5 years of follow-up (RR = 1.17 and 1.00, respectively) ( Table 3 ) and after 15 years of follow-up (RR = 1.09 and 1.00, respectively) ( Table 2 ). In contrast, patients with invasive lobular carcinomas in the validation cohort had a statistically significantly lower relative risk compared with those with invasive ductal carcinomas (RR = 0.64 and 1.00, respectively) ( Table 3 ). These are inconsistent results, and despite not being able to reproduce the findings on these subgroups in the validation group, we kept histopathologic subgroup as a variable in the final multivariable model. A previous finding by Pestalozzi et al. ( 18 ) indicates that the prognosis of ductal and lobular carcinoma develops differently over time. This study ( 18 ) reported that although prognosis in a large International Breast Cancer Study Group (IBCSG) study was better for invasive lobular carcinomas during the first years of follow-up, it became worse following long-term follow-up.

Tumor size is one of the strongest prognostic indicators ( 19 ). There is a statistically significant association between large size and positive lymph nodes ( 20 ), but the influence of tumor size on survival is not alone related to nodal status. Also in node-negative patients, increasing tumor size is statistically significantly negatively associated with survival ( 20 , 21 ). In 1989, Carter et al. ( 20 ) reported tumor size, lymph node status, and survival in almost 25 000 breast cancer patients from the American SEER database. The 5-year relative survival rates among node-negative patients were 99.2% and 98.3% for tumors less than 0.5 and 0.5–0.9 cm, respectively. For tumors 1 cm or larger, the survival rates declined from 95.8% for tumors 1.0–1.9 cm to 82.2% for tumors larger than 5 cm. Furthermore, according to another study based on SEER data, for smaller tumors up to 30 mm in size, a linear increase of mortality with tumor size was observed but for tumors larger than 30 mm, there were no further increases in mortality with increased tumor size ( 22 ). In addition, in four of nine studies reviewed by Mirza et al. ( 23 ), multivariable analysis showed that when age, tumor grade, proliferation and mitotic index, hormone receptor status, and treatment were included in the model, increased tumor size remained a statistically significant prognostic factor. These results are strongly supported by our current findings and underline the rationale of early detection of breast cancer by mammography screening programs as a mean of improving treatment results.

Several publications have focused on prognostic factors in node-negative patients ( 24 , 25 ). Tumor grade, overexpression of HER2, lymphovascular invasion, and percent S-phase were found to be associated with prognosis in the multivariable Cox proportional hazards analyses performed by Trudeau et al. ( 25 ). In the study by Hery et al. ( 24 ) on more than 2200 node-negative patients, grade and tumor size were the strongest prognostic variables in both univariate and multivariable analyses. Three prognostic groups could be constructed in that study. The low-risk group consisted of grade 1 tumors 10 mm or less, and the high-risk group consisted of grade 2 or 3 tumors larger than 20 mm size. The intermediate-risk group included tumors not belonging to low-risk or high-risk groups. The risk of metastasis at 10 years increased from 8% in the low-risk group to 15%–21% in the intermediate-risk group and 32%–38% in the high-risk group. In the overview by Mirza et al. ( 23 ), studies with more than 200 included patients and a median follow-up of at least 60 months were reviewed. Prognostic factors were defined as useful if a preponderance of studies indicated that the factor was statistically significantly associated with one or more survival outcome measures. The prognostic factors pinpointed by Trudeau et al. ( 25 ) and Hery et al. ( 24 ) were all found useful in this study. Interestingly, hormone receptors, HER2, tumor protein p53 (TP53; also known as p53), and DNA ploidy showed only limited association with survival. Because the current study is restricted to low-grade and hormone receptor–positive tumors, the occurrence of unfavorable tumor prognostic factors is likely to be minimized, but on the other hand, had Ki67 and HER2 data been available, further refinement of the prognostic groups could have been possible.

Gene expression profiling of breast cancer may add further prognostic information, although in multivariable analysis, nodal status and tumor size are still statistically significant independent prognostic factors ( 19 ). Two biologically distinct ER + subtypes have been identified: luminal A and luminal B ( 26 , 27 ). The luminal B subtype includes some HER2-enriched tumors but is otherwise characterized by expressing genes associated with high proliferation such as cyclin B1 (CCNB1), antigen identified by monoclonal antibody Ki-67 (MKI67), and v-myb myeloblastosis viral oncogene homolog (avian)-like 2 (MYBL2) ( 28 ). As a surrogate for DNA microarray in subtyping luminal breast cancers, immunohistochemical classification by ER, PR, and HER2 is used ( 29 , 30 ). By these parameters, the luminal B subclassification is connected with the hormone receptor–positive tumors with high proliferation rate and often HER2 augmented ( 28 ).

The present series of hormone receptor–positive tumors cannot be divided in the luminal A and luminal B subtypes, as neither HER2 nor Ki67 data are available, but a dominance of luminal A tumors is expected. In the study of node-negative hormone receptor–positive breast cancer patients by Cheang et al. ( 28 ), 66% tumors were luminal A and 34% were luminal B. Of the 318 luminal B tumors, only 55 (17%) were found to be HER2 enriched. Poorer relapse-free survival was observed for the HER2-normal and HER2-enriched luminal B tumors (hazard ratios 1.4 and 1.6, respectively). In the current study, we have restricted the study population to grade 1 ductal carcinoma and grade 1 or 2 invasive lobular carcinoma, which should minimize, but not eliminate, the luminal B proportion ( 31 ). Therefore, although the present classification by nodal status, hormone receptors, histological grade, and tumor size precisely identifies a subgroup of patients that has similar mortality as women of same age who did not receive systemic adjuvant treatment, additional immunohistochemical information most likely could help identifying those patients, who might or might not benefit from further adjuvant treatment. Also, gene profiling tools as the Oncotype DX recurrence score ( 32 ) are evaluated in more studies as both a prognostic and predictive factor in ER + early-stage breast cancer. Final results are not available yet, although the recurrence score seems to have impact on decision making in increasing number of breast cancer cases ( 33 ).

Patients in subgroups with mortality rates exceeding the background mortality may potentially benefit from systemic adjuvant treatment. The present investigation point at higher mortality rates among patients with tumor size 11–20 mm, and for patients aged 35–59 years independent of tumor size. However, we identified a low-risk group with mortality rates comparable to the background population. This low-risk group of endocrine-responsive tumors was restricted to patients of age 60 years or older, with small tumors up to 10 mm in size, and with low histopathology grade (grade 1 ductal carcinoma, grade 1 or 2 invasive lobular carcinoma, or of other or unknown type). Indeed, although the Panel at the 2009 St. Gallen International Breast Cancer Conference recommended adjuvant endocrine treatment in almost all patients whose tumors show evidence of endocrine responsiveness ( 2 ), the present results indicate that there is a small subgroup of node-negative patients, who might not benefit from such an approach.

The major strengths of this study are the population-based risk estimation of death, the large study population, and the long follow-up period with more than 41 000 person-years of risk. Furthermore, the clinical data and the tumor characteristics were collected and evaluated prospectively. It is a well-defined and uniform cohort of patients with early breast cancer, who were not receiving systemic treatment during prolonged follow-up. The results are therefore not confounded by different treatment modalities and provide the opportunity to investigate the effects of prognostic factors on the natural history of breast cancer.

This study is limited by the lack of immunohistochemical tumor characteristics and gene profiling. Data on HER2, topoisomerase (DNA) II alpha (TOP2A), and Ki67 expression would have allowed us to make a more exact evaluation of the data and to perform a more precise definition of a group of patients completely matching the current low-risk criteria.

In conclusion, our results indicate that there is a small subgroup of patients 60 years of age or older with small (≤10 mm) node-negative, grade 1 ductal carcinoma or grade 1 or 2 invasive lobular carcinoma, ER + and/or PR + tumors, who experience a prognosis similar to the background population of women even without systemic adjuvant therapy.

Funding

There was no funding for this study.

References

1.
Early Breast Cancer Trialists’ Collaborative Group
Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomised trials
Lancet.
 , 
2005
, vol. 
365
 
9472
(pg. 
1687
-
1717
)
2.
Goldhirsch
A
Ingle
JN
Gelber
RD
Coates
AS
Thurlimann
B
Senn
HJ
Thresholds for therapies: highlights of the St Gallen International Expert Consensus on the primary therapy of early breast cancer 2009
Ann Oncol.
 , 
2009
, vol. 
20
 
8
(pg. 
1319
-
1329
)
3.
Moller
S
Jensen
MB
Ejlertsen
B
, et al.  . 
The clinical database and the treatment guidelines of the Danish Breast Cancer Cooperative Group (DBCG); its 30-years experience and future promise
Acta Oncol.
 , 
2008
, vol. 
47
 
4
(pg. 
506
-
524
)
4.
Christiansen
P
Al-Suliman
N
Bjerre
K
Moller
S
Recurrence pattern and prognosis in low-risk breast cancer patients—data from the DBCG 89-A programme
Acta Oncol.
 , 
2008
, vol. 
47
 
4
(pg. 
691
-
703
)
5.
Friis
E
Galatius
H
Garne
JP
Organized nation-wide implementation of sentinel lymph node biopsy in Denmark
Acta Oncol.
 , 
2008
, vol. 
47
 
4
(pg. 
556
-
560
)
6.
Statistics Denmark
StatBank Denmark
  
http://www.statistikbanken.dk/ . Accessed August 4, 2010
7.
Schemper
M
Smith
TL
A note on quantifying follow-up in studies of failure time
Control Clin Trials.
 , 
1996
, vol. 
17
 
4
(pg. 
343
-
346
)
8.
Walter
SD
Feinstein
AR
Wells
CK
Coding ordinal independent variables in multiple regression analyses
Am J Epidemiol.
 , 
1987
, vol. 
125
 
2
(pg. 
319
-
323
)
9.
Gray
R
A class of K-sample tests for comparing the cumulative incidence of a competing risk
Ann Stat.
 , 
1988
, vol. 
16
 
3
(pg. 
1141
-
1154
)
10.
Fine
JP
Gray
RJ
A proportional hazards model for the subdistribution of a competing risk
J Am Stat Assoc.
 , 
1996
, vol. 
94
 
446
(pg. 
496
-
509
)
11.
Tavassoli
FA
Devilee
P
Tumors of the Breast and Female Genital Organs
 , 
2003
Lyon, France
ARC Press
12.
Elston
CW
Ellis
IO
Pathological prognostic factors in breast cancer. I. The value of histological grade in breast cancer: experience from a large study with long-term follow-up
Histopathology.
 , 
1991
, vol. 
19
 
5
(pg. 
403
-
410
)
13.
Kroman
N
Jensen
MB
Wohlfahrt
J
Mouridsen
HT
Andersen
PK
Melbye
M
Factors influencing the effect of age on prognosis in breast cancer: population based study
BMJ.
 , 
2000
, vol. 
320
 
7233
(pg. 
474
-
478
)
14.
Arriagada
R
Le
MG
Dunant
A
Tubiana
M
Contesso
G
Twenty-five years of follow-up in patients with operable breast carcinoma: correlation between clinicopathologic factors and the risk of death in each 5-year period
Cancer.
 , 
2006
, vol. 
106
 
4
(pg. 
743
-
750
)
15.
Hartley
MC
McKinley
BP
Rogers
EA
, et al.  . 
Differential expression of prognostic factors and effect on survival in young (< or =40) breast cancer patients: a case-control study
Am Surg.
 , 
2006
, vol. 
72
 
12
(pg. 
1189
-
1194
)
16.
Tai
P
Cserni
G
Van De Steene
J
, et al.  . 
Modeling the effect of age in T1-2 breast cancer using the SEER database
BMC Cancer.
 , 
2005
, vol. 
5
 pg. 
130
  
doi:10.1186/1471-2407-5-130
17.
Hanrahan
EO
Valero
V
Gonzalez-Angulo
AM
Hortobagyi
GN
Prognosis and management of patients with node-negative invasive breast carcinoma that is 1 cm or smaller in size (stage 1; T1a, bN0M0): a review of the literature
J Clin Oncol.
 , 
2006
, vol. 
24
 
13
(pg. 
2113
-
2122
)
18.
Pestalozzi
BC
Zahrieh
D
Mallon
E
, et al.  . 
Distinct clinical and prognostic features of infiltrating lobular carcinoma of the breast: combined results of 15 International Breast Cancer Study Group clinical trials
J Clin Oncol.
 , 
2008
, vol. 
26
 
18
(pg. 
3006
-
3014
)
19.
Soerjomataram
I
Louwman
MW
Ribot
JG
Roukema
JA
Coebergh
JW
An overview of prognostic factors for long-term survivors of breast cancer
Breast Cancer Res Treat.
 , 
2008
, vol. 
107
 
3
(pg. 
309
-
330
)
20.
Carter
CL
Allen
C
Henson
DE
Relation of tumor size, lymph node status, and survival in 24,740 breast cancer cases
Cancer.
 , 
1989
, vol. 
63
 
1
(pg. 
181
-
187
)
21.
Chia
SK
Speers
CH
Bryce
CJ
Hayes
MM
Olivotto
IA
Ten-year outcomes in a population-based cohort of node-negative, lymphatic, and vascular invasion-negative early breast cancers without adjuvant systemic therapies
J Clin Oncol.
 , 
2004
, vol. 
22
 
9
(pg. 
1630
-
1637
)
22.
Verschraegen
C
Vinh-Hung
V
Cserni
G
, et al.  . 
Modeling the effect of tumor size in early breast cancer
Ann Surg.
 , 
2005
, vol. 
241
 
2
(pg. 
309
-
318
)
23.
Mirza
AN
Mirza
NQ
Vlastos
G
Singletary
SE
Prognostic factors in node-negative breast cancer: a review of studies with sample size more than 200 and follow-up more than 5 years
Ann Surg.
 , 
2002
, vol. 
235
 
1
(pg. 
10
-
26
)
24.
Hery
M
Delozier
T
Ramaioli
A
, et al.  . 
Natural history of node-negative breast cancer: are conventional prognostic factors predictors of time to relapse?
Breast.
 , 
2002
, vol. 
11
 
5
(pg. 
442
-
448
)
25.
Trudeau
ME
Pritchard
KI
Chapman
JA
, et al.  . 
Prognostic factors affecting the natural history of node-negative breast cancer
Breast Cancer Res Treat.
 , 
2005
, vol. 
89
 
1
(pg. 
35
-
45
)
26.
Perou
CM
Sorlie
T
Eisen
MB
, et al.  . 
Molecular portraits of human breast tumours
Nature.
 , 
2000
, vol. 
406
 
6797
(pg. 
747
-
752
)
27.
Sorlie
T
Perou
CM
Tibshirani
R
, et al.  . 
Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications
Proc Natl Acad Sci U S A.
 , 
2001
, vol. 
98
 
19
(pg. 
10869
-
10874
)
28.
Cheang
MC
Chia
SK
Voduc
D
, et al.  . 
Ki67 index, HER2 status, and prognosis of patients with luminal B breast cancer
J Natl Cancer Inst.
 , 
2009
, vol. 
101
 
10
(pg. 
736
-
750
)
29.
Carey
LA
Perou
CM
Livasy
CA
, et al.  . 
Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study
JAMA.
 , 
2006
, vol. 
295
 
21
(pg. 
2492
-
2502
)
30.
Livasy
CA
Karaca
G
Nanda
R
, et al.  . 
Phenotypic evaluation of the basal-like subtype of invasive breast carcinoma
Mod Pathol.
 , 
2006
, vol. 
19
 
2
(pg. 
264
-
271
)
31.
Calza
S
Hall
P
Auer
G
, et al.  . 
Intrinsic molecular signature of breast cancer in a population-based cohort of 412 patients
Breast Cancer Res.
 , 
2006
, vol. 
8
 
4
pg. 
R34
 
32.
Paik
S
Shak
S
Tang
G
, et al.  . 
A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer
N Engl J Med.
 , 
2004
, vol. 
351
 
27
(pg. 
2817
-
2826
)
33.
Lo
SS
Mumby
PB
Norton
J
, et al.  . 
Prospective multicenter study of the impact of the 21-gene recurrence score assay on medical oncologist and patient adjuvant breast cancer treatment selection
J Clin Oncol.
 , 
2010
, vol. 
28
 
10
(pg. 
1671
-
1676
)
The authors are solely responsible for the study design, data collection, analysis and interpretation of the data, writing the article, and decision to submit the article for publication.