Assessment of 25-Year Survival of Women With Estrogen Receptor–Positive/ERBB2-Negative Breast Cancer Treated With and Without Tamoxifen Therapy

Key Points Question Are clinically used markers of breast cancer, such as tumor size, tumor grade, progesterone receptor status, and Ki-67 status, independently associated with 25-year survival and tamoxifen treatment benefit among patients with breast cancer? Findings In this secondary analysis of data from 565 postmenopausal women with lymph node–negative, estrogen receptor–positive, and ERBB2-negative breast cancer who participated in the Stockholm tamoxifen randomized clinical trial (STO-3), tumor size and tumor grade were significantly associated with long-term (25-year) survival. A significant tamoxifen treatment benefit was observed among patients with larger tumors, lower tumor grades, and progesterone receptor–positive tumors. Meaning This study’s findings suggest that tumor size and tumor grade is associated with long-term survival, and patients with larger tumors, lower tumor grades, and progesterone receptor–positive status experienced significant treatment benefit with receipt of tamoxifen therapy.


Introduction
Breast cancer is the most common female cancer in the Western world and one of the major causes of death among women. It is widely recognized as a heterogeneous disease with a long natural history. 1 Over the past decades, a gradual increase in survival has been observed because of early detection, precise diagnosis, and improved treatment. Among women with estrogen receptor (ER)positive tumors, treatment with adjuvant endocrine therapy is generally recommended to reduce the risk of recurrence and improve survival. 2 However, approximately one-half of patients with ER-positive disease do not benefit from endocrine therapy, and approximately 1 in 4 patients later develop distant metastasis and die of breast cancer. 3,4 Clinically used breast cancer markers are known to provide short-term survival estimates for up to 10 years after primary diagnosis. Numerous studies have reported that large tumors, 5 high tumor grades, [6][7][8][9] and high Ki-67 expression are associated with worse short-term survival. [10][11][12] Tumor size and tumor grade are also routinely used to make decisions about adjuvant treatment, whereas the association between Ki-67 status and treatment benefit from endocrine therapy remains unclear. [13][14][15] In addition, studies have indicated that progesterone receptor (PR) status might not provide independent information about prognosis in combination with other breast cancer markers, 16 and the predictive value of PR has been debated. 2,17 Patients with ER-positive tumors have a continuous long-term risk of distant recurrence and death compared with patients with ER-negative tumors. 3,4,[18][19][20] A study from the Early Breast Cancer Trialists' Collaborative Group found that the risk of distant recurrence continues steadily throughout the 5 to 20 years after primary diagnosis. 4 Among women with smaller tumors and lymph nodenegative disease (ie, T1N0), a cumulative risk of 13% for distant recurrence was reported. 4 The reasons for this long-term risk are unclear; however, it has been suggested that late fatal disease mechanisms may involve cancer cells remaining dormant over a long period. 21 Given the late onset of fatal disease among those with ER-positive breast cancer, it is challenging to estimate patients' longterm risk of fatal disease, and the ability of clinically used markers to independently estimate the long-term benefit of endocrine therapy has not been established.
Clinically used breast cancer markers are known to be associated with patient survival for up to 10 years after diagnosis. 16 However, the association of these markers with long-term survival has not been established, and there are few well-annotated clinical studies with long-term follow-up data    STO-3 clinical trial; of those, 565 patients had a diagnosis of ER-positive/ERBB2-negative breast cancer and were included in this secondary analysis ( Figure 1).

JAMA Network Open | Oncology
All residents in Sweden have a unique national registration number. This number allows automatic linkage with various personal records from national and regional registers, which provides high validity and essentially complete data coverage. Cancer registration is legally required in Sweden, and validation studies have reported that the Swedish Cancer Registry covers more than 96% of all cancer diagnoses in validation studies. 25 Information on metastatic disease was obtained from the Regional Stockholm Breast Cancer Quality Registry of the Regional Cancer Center Stockholm-Gotland in Stockholm. 26 Thus, through linkage with Swedish national and regional registers, complete long-term follow-up data from participants in the STO-3 clinical trial were available through December 31, 2016.

Immunohistochemistry
Immunohistochemical analyses and reannotation of ER, PR, ERBB2, and antigen Ki-67 among participants in the STO-3 study were performed in 2014. Breast

Tumor Grade and Size
Tumor grade (1-3) was assessed in 2014 by 1 pathologist according to the Nottingham system (also known as Elston-Ellis grading). 24 Tumor size was categorized into 3 groups based on clinical guidelines, with tumors of 10 mm or less classified as T1a and T1b (T1a/b), tumors of 11 mm to 20 mm classified as T1c, and tumors larger than 20 mm classified as T2.

Statistical Analysis
Survival Analysis

Multivariable Analysis of Long-term Survival
A multivariable Cox proportional hazards survival analysis for clinically used markers was performed to estimate long-term survival, adjusting for standard clinical patient and tumor characteristics.
Consistent with the Kaplan-Meier analysis, a statistically significant reduction in the long-term risk of distant recurrence was found among patients with smaller tumors (T1a/b and T1c) vs larger tumors  (Table). Consistent with the Kaplan-Meier analyses, a statistically significant difference in long-term DRFI by PR status and Ki-67 status was not found (Table). Similar results were observed for BCSS (eTable 2 in Supplement 1).

Multivariable Analysis of Long-term Treatment Benefit
A multivariable Cox proportional hazards survival analysis of clinically used markers by STO-3 clinical trial arm, adjusted for standard patient and tumor characteristics, was performed to estimate the long-term treatment benefit of tamoxifen therapy. A statistically significant reduction in long-term  Log-rank P = .02 The reported P values are based on a 2-sided log-rank test. PR indicates progesterone receptor. but not among those who had the smallest tumors (T1a/b) compared with patients who did not receive adjuvant treatment (Figure 3).

Recursive Partitioning Analysis
The recursive partitioning analysis first divided patients by tumor size, separating those with the smallest tumors (T1a/b) from those with larger tumors (T1c and T2) (Figure 4)  ( Figure 4). The first division, which comprised patients with the smallest tumors (T1a/b), had the best long-term survival, followed by the second division, which comprised patients with larger tumors (T1c and T2) from the tamoxifen treatment group. The third division, which comprised patients who did not receive adjuvant treatment and who were separated by tumor size (T1c and T2), had the worst survival. Similar results were observed in the analysis of long-term BCSS (eFigure 3 in Supplement 1).  No. at risk T1a and T1b  T1c and T2 treated  T1c untreated  T2 untreated   157  193  123  47   149  173  105  35   133  149  89  25   104  125  75  19   78  91  58  14   55  63  42  6   T1c and T2 treated   T1c untreated   T2 untreated   T1a and T1b Log-rank P < .001

Yes No
Received tamoxifen therapy?

Yes No
Tumor size >20 mm?
No Yes markers that were reannotated simultaneously and patients who received homogeneous treatment according to the STO-3 clinical trial arm.
The findings of this secondary analysis suggest that patients with larger tumors (T1c and T2) and lower tumor grades (1 and 2) receive substantial treatment benefit with tamoxifen therapy.
Furthermore, a long-term treatment benefit was observed among patients with PR-positive tumors and in patients who had low and medium to high Ki-67 expression, which is consistent with the results of a previous STO-3 study. 17

Limitations
This study has limitations. As with most long-term follow-up studies, clinical recommendations for disease management and treatment have changed since the initiation of the original clinical trial. The STO-3 clinical trial was performed before aromatase inhibitors became one of the recommended treatment options for patients with ER-positive breast cancer. In addition, when the STO-3 clinical trial was conducted, the duration of tamoxifen treatment was shorter, and the treatment dosage was higher than current recommendations. In the population-based STO-3 clinical trial cohort, approximately one-half of the patients had tumor samples available for molecular analysis. The present study therefore has limitations regarding the small number of patients available for subgroup analysis. We have, however, confirmed that patient and tumor characteristics in this secondary analysis were equally distributed and well balanced with those of the original STO-3 clinical trial cohort with regard to characteristics such as tumor size and ER status. 24 In addition, when performing immunohistochemical analysis, there is often some level of inaccuracy. However, in the present study, the clinically used markers were stained at a single medical laboratory in 2014 and assessed by experienced breast cancer pathologists who had been harmonized with regard to the scoring of immunohistochemical markers. 30

Conclusions
The findings of this study indicate that, among patients with lymph node-negative, ER-positive/ ERBB2-negative breast cancer from the STO-3 randomized clinical trial, tumor size followed by tumor grade were significantly associated with long-term risk of distant recurrence, as patients with larger tumors and higher tumor grades had significantly worse long-term survival compared to patients with smaller tumors and lower tumor grades. In contrast, PR status and Ki-67 status were not significantly associated with long-term survival in patients with lymph node-negative, ER-positive/ ERBB2-negative breast cancer. The findings further indicated that, among this selected subgroup, a