Radiation Therapy After Breast-Conserving Surgery in Women 70 Years of Age and Older: How Wisely Do We Choose?

Abstract

Background:

Despite data from randomized trials supporting omission of radiation therapy (RT) for women ≥ 70 years of age with T1, estrogen receptor positive (ER+) tumors undergoing breast-conserving therapy (BCT), RT usage remains high. We reviewed our institutional experience to determine if risk factors for local recurrence or co-morbidities influenced use.

Methods:

Women ≥ 70 years of age with T1, ER+, HER2 negative tumors undergoing BCT in 2010–2012 were identified from a prospectively maintained database. 10-year estimated mortality was calculated using the Suemoto index. The associations of clinicopathological features and mortality risk on receipt of RT were examined.

Results:

323 patients with 327 cancers were identified. Median age was 75 years, median tumor size was 1cm, and all were clinically node negative. 53.7% of patients received RT. RT usage decreased with age (73.6% age 70–74, 49.5% age 75–79, 33.3% age 80–84, 10.7% >85, p<0.001). Within age groups, estimated mortality did not impact RT usage. On multivariable analysis, only younger age and larger tumor size were associated with RT use. Recurrence-free survival was 98% versus 93% with and without RT (p=0.011), respectively. Those who received adjuvant radiation also had improved overall survival (92% versus 89%), though this effect did not reach statistical significance (p=0.051).

Conclusions:

Neither factors associated with risk of local recurrence nor estimated risk of death in 10 years were associated with use of adjuvant radiation in a large cohort of women ≥ 70 years of age with small ER+ breast cancers treated with breast-conserving surgery.

INTRODUCTION

A major goal in treating breast cancer is individualization of treatments to optimize local control and prevent recurrence while minimizing morbidity. The widespread adoption of sentinel node biopsy for axillary staging and the use of genomic testing to decrease chemotherapy utilization in patients with hormonally responsive tumors are examples of this approach.

Despite attempts to minimize morbidity, radiation therapy has remained the standard of care for patients undergoing breast-conservation therapy^1–3 based upon data from the Early Breast Cancer Trialists’ Cooperative Group meta-analyses demonstrating that the reduction in local recurrence associated with the use of radiation therapy is associated with an improvement in overall survival. However, this difference did not become apparent until 15 years of follow-up and only was found in women with a greater than 10% reduction in local recurrence at 5 years.⁴

In selected older women with small (<2cm), clinically node negative, estrogen receptor (ER) positive tumors undergoing breast-conserving surgery, the omission of adjuvant radiation therapy has been proven to be safe in randomized trials with long-term follow-up.^5–7 While there was a significant difference in local recurrence among patients who did and did not receive adjuvant radiation, this rate was low overall and there was no difference in overall survival between groups.

Despite these data, a change in the National Comprehensive Cancer Network guidelines⁸, and the “Choosing Wisely” campaign encouraging cost-effective, thoughtful cancer care⁹, the use of radiation in this population of women remains high nationwide.^10,11 We postulated that the decision to use radiation is influenced by both patient and tumor factors in this population of women at low risk of death from breast cancer.

METHODS

Following Institutional Review Board approval of this study, women age ≥70 years with T1 (<2cm), ER+, HER2-, clinically node-negative tumors undergoing breast-conserving surgery between 2010 and 2012 were identified from our prospectively maintained service database. ER positivity was defined at >1% of cells staining positive for estrogen receptor. Patients with pure in situ carcinoma or previous ipsilateral breast cancers were excluded. Bilateral breast cancers in the same patient were considered as separate events. Clinical, pathologic, treatment, and follow-up data were collected. All patients had an estimated 10-year mortality risk calculated using the Suemoto index.¹² This mortality prediction tool incorporates medical comorbidities, body mass index, lifestyle, functional status, and the patient’s assessment of her or his own health to provide an estimate of the risk of mortality in 10 years in patients age >60 years. The model is based on data from 16 different countries and 5 different cohorts. As this index is typically calculated prospectively, we utilized a standard institutional patient-completed questionnaire, which is completed at the time of initial surgical consultation to abstract and record information regarding functional status and the patient’s own assessment of her or his health.

The primary endpoint of the study was the use of adjuvant radiation. Secondary endpoints included locoregional and distant recurrence, and overall survival. Patient and treatment characteristics were summarized using the median and range for continuous variables, and the frequency and percentage for categorical variables. Univariable and multivariable associations with radiation were assessed using mixed effects models with a random surgeon effect to address the correlation among patients treated by the same surgeon. Factors significantly associated with radiation on univariable analysis were included in multivariable analysis. Kaplan-Meier methods were used to estimate overall survival (OS) and recurrence-free survival (RFS). OS is defined as the time to death from any cause and RFS is defined as the time to first recurrence. Follow-up time was calculated from the date of surgery, and patients without events at last follow-up date were censored. A landmark-analysis approach was utilized to evaluate the association between radiation, and overall and recurrence-free survival, with a landmark time of 12 weeks to account for the time from surgery to completion of radiation treatment since exact treatment dates were not always available. Univariable associations with survival and recurrence included a random surgeon effect to account for correlation between patients treated by the same surgeon. A p-value <0.05 was considered statistically significant. All statistical analyses were performed using R Software version 3.4.1 (R Core Development Team, Vienna, Austria).

RESULTS

There were 323 patients with 327 cancers who met the study inclusion criteria. The median age of the entire cohort was 75 years (range 70–100) at the time of surgery, and the median tumor size was 1 cm. The majority of patients (80.7%) had infiltrating ductal cancers. Invasive lobular carcinoma was present in 11.9%. Forty-eight percent of patients had poorly differentiated tumors, and 15.3% had evidence of lymphovascular invasion. Ten percent of patients had an extensive intraductal component. The median percentage of cells staining positive for the estrogen receptor was 95% (Table 1).

TABLE 1.

Patient and tumor characteristics

Characteristic	Overall (n = 327)	No radiation (n = 151)	Radiation (n = 176)	p-value
Median age at surgery	75 (70–100)	78 (70–100)	74 (70–88)	< 0.001
10-year estimate of mortality	43% (22–98%)	51% (22–98%)	37% (24–96%)	< 0.001
Tumor type				0.623*
Infiltrating Ductal	264 (80.7%)	120 (79.5%)	144 (81.8%)
Invasive Lobular	39 (11.9%)	18 (11.9%)	21 (11.9%)
Mixed ductal and lobular	4 (1.2%)	3 (2.0%)	1 (0.6%)
Pure mucinous/colloid/medullary/tubular/papillary	19 (5.8%)	9 (6.0%)	10 (5.7%)
Other	1 (0.3%)	1 (0.7%)	0 (0%)
Median tumor size (cm)	1 (0.1–2.0)	0.9 (0.1–1.9)	1.1 (0.1–2.0)	0.026
Extensive intraductal component	32 (9.8%)	14 (9.3%)	18 (10.2%)	0.751
Differentiation				0.674
Well differentiated	39 (11.9%)	20 (13.2%)	19 (10.8%)
Moderately differentiated	91 (27.8%)	43 (28.5%)	48 (27.3%)
Poorly differentiated	157 (48%)	68 (45%)	89 (50.6%)
Lymphovascular invasion	50 (15.3%)	19 (12.6%)	31 (17.6%)	0.23
Median % of cells positive for ER	95%	95%	95%
Oncotype	13 (0–40)	14 (3–33)	13 (0–40)	0.693

^*comparing infiltrating ductal to all others

ER, estrogen receptor

Axillary staging was performed in 286 patients (87.4%); 98.6% of patients who had an axillary evaluation underwent sentinel lymph node biopsy (Table 2). Two hundred and seventy-eight patients (96.9%) who underwent axillary staging were node negative, and 11 had isolated tumor cells (pN0(i+)). Most patients (84.7%) began adjuvant endocrine therapy. Of the patients who started endocrine therapy, 155 (56%) lived and were followed for 5 years or longer. Of these 155 patients, 113 (73%) completed at least 5 years of endocrine therapy.

TABLE 2.

Treatment characteristics

Treatment Characteristic	Overall (n = 327)	No radiation (n = 151)	Radiation (n = 176)	p-value
Axillary Evaluation				0.434
Axillary Lymph Node Dissection	7 (2.1%)	2 (1.3%)	5 (2.8%)
Sentinel Lymph Node Biopsy	279 (85.3%)	119 (78.8%)	160 (90.9%)
None	41 (12.5%)	30 (19.9%)	11 (6.2%)
Endocrine Therapy	277 (84.7%)	123 (81.5%)	154 (87.5%)	0.111
Completion of Endocrine Therapy	205 (74.0%)	88 (71.5%)	117 (76%)	0.375
Adjuvant Chemotherapy	31 (9.5%)	5 (3.3%)	26 (14.8%)	< 0.001
Re-excision	52 (15.9%)	15 (9.9%)	37 (21.0%)	0.002

Adjuvant radiation therapy was received by 176 (53.8%) patients. Those receiving radiation therapy were significantly younger than those who did not (74 versus 78 years of age, p<0.001). Nearly three-quarters of patients age 70–74 years (73.6%) were treated with adjuvant radiation compared with 49.5% of those age 75–79 years, 33.3% of those age 80–84 years, and 10.7% of those age 85 years and older (Fig. 1). Patients treated with adjuvant radiation had a lower estimate of their risk of mortality at 10 years (37% versus 51%, p<0.001)(Table 1).

Fig. 1.

Adjuvant radiation stratified by age group.

Patients receiving adjuvant radiation had larger tumors (1.1cm versus 0.9cm, p=0.026). There was no difference in the percentage of poorly differentiated tumors, lymphovascular invasion, or an extensive intraductal component between patients who did and did not receive radiation. Among patients with data available, there was no difference in Oncotype DX (Genomic Health, Redwood City, CA) score based on radiation use (Table 1). Treatment characteristics varied between patients receiving and not receiving RT, with those in the RT group being significantly more likely to undergo re-excision and to receive chemotherapy (Table 2). There were 24 patients (7.3%) who did not receive adjuvant radiation, chemotherapy, or endocrine therapy (Table 2).

The Suemoto index incorporates age in 5-year increments to predict 10-year mortality, and we analyzed age as a continuous variable; therefore, age and Suemoto index have a strong positive association (p<0.001). We evaluated the interaction between continuous age and 10-year risk of mortality on receipt of adjuvant radiation and found no significant effect (p=0.112), suggesting that patients of the same age have the same odds for receipt of radiation regardless of Suemoto index. On multivariable analysis, increased age was associated with decreased odds of receipt of radiation and increased tumor size was associated with increased odds of receipt of radiation, whereas the need for 1 or 2 re-excisions demonstrated a non-significant increase in odds of radiation (p=0.055) and Suemoto index was not independently associated with receipt of radiation (Table 3).

TABLE 3.

Multivariable analysis of factors associated with receipt of adjuvant radiation

Factor	OR	95% CI	p-value
Age at surgery	0.38	0.25–1.00	< .001
E-prognosis score	0.76	0.52–1.00	0.142
Tumor size	1.52	1.17–2.00	0.002
Number of re-excisions			0.055
0	Ref
1	2.44	1.14–5.00
2	2.02	0.11–37.00

After a median follow-up of 5.1 years, 36 patients died from any cause with an overall 5-year survival of 89% (95% confidence interval [CI] 85–93%). Among patients who received adjuvant radiation, 5-year overall survival was 92% (95% CI 88–97%) versus 85% in those who did not receive adjuvant radiation (95% CI 79–92%)(Fig. 2a), though this difference was not statistically significant (p=0.051). Fifteen patients have developed a recurrence of their initial cancer, a 5-year recurrence-free survival of 95% (95% CI 93–98%). Recurrence-free survival was significantly better among patients who received radiation (98% [95% CI 95–100%]) than among those who did not receive radiation (93% [95% CI 88–98%])(p=0.011)(Fig. 2b). Fourteen of fifteen (93.3%) recurrences were locoregional; there was one distant recurrence without locoregional recurrence. Among the patients who recurred locoregionally, 7 had an ipsilateral breast recurrence, 5 had an ipsilateral axillary recurrence, and 2 had an ipsilateral breast and axillary recurrence for a total of 9 ipsilateral breast tumor recurrences and 7 ipsilateral axillary recurrences. Among patients with an ipsilateral breast tumor recurrence, 3 had adjuvant radiation and 5 did not. Of patients with an ipsilateral axillary recurrence, 2 had adjuvant radiation and 6 did not. Of the 7 patients with ipsilateral axillary recurrence, 4 underwent axillary surgery (3 had SLNB and 1 had ALND) and 3 had no surgical axillary evaluation.

Fig. 2.

A) Overall survival and B) recurrence-free survival stratified by adjuvant radiation.

Of the 7 axillary recurrences, 3 occurred in patients who took no endocrine therapy. Two refused therapy and 1 patient was felt to have such minimal invasive disease that medication was felt to be optional. One recurrence was in association with a local recurrence. In total, the rate of isolated local recurrence was 3/323 (0.9%).

On univariable analysis, age, tumor size, a higher estimated 10-year mortality risk, and the presence of lymphovascular invasion were significantly associated with recurrence.

DISCUSSION

In our study of women ≥ 70 years of age with small ER+ tumors treated by lumpectomy at an academic medical center, the use of adjuvant radiation was 50% overall and nearly 75% in women age 70–75 years despite data from randomized controlled trials demonstrating no survival advantage of adjuvant radiation in this population. In the CALGB study in women treated with lumpectomy and tamoxifen, the addition of radiation did not change survival with over 10 years of follow-up (67% versus 66%) despite a significant difference in locoregional-recurrence-free survival (98% versus 90% at 10 years).^5,6 Similarly, the PRIME II trial compared patients with tumors up to 3 cm in size and showed no difference in overall survival at 5 years (93.9% in both groups) despite an absolute difference in ipsilateral breast tumor recurrence of about 3% between groups (1.3% versus 4.1%).⁷

Two retrospective studies did demonstrate improved survival when adjuvant radiation was given in conjunction with endocrine therapy in older women. However, the propensity matching performed does not eliminate the selection biases inherent in retrospective reviews, and their findings should not challenge the results of the prospective randomized trials.^13,14 In fact, a meta-analysis of over 2300 patients, including the 2 randomized studies, showed that while the addition of radiation to tamoxifen reduced local recurrence from 60 to 10 per 1000 patients at 10 years follow-up, and reduced axillary recurrence from 12 to 3 per 1000 at 5 years¹⁵, there was no difference in distant recurrence or overall survival.^3,5,7,16

Though adjuvant radiation was used in about half of our study population, use was notably less than reported in other large observational series. McCormick and colleagues evaluated changes in practice patterns after the publication of CALGB 9343.¹¹ In 2009, 5 years after the publication of CALGB 9343, 88% of over 1000 women 70–74 years of age who met criteria for inclusion in CALGB 9343 identified in the National Comprehensive Cancer Network (NCCN) database were treated with adjuvant radiation, a decrease from 94% in 2000. Usage in this same age group in our study was significantly less, with 75% undergoing radiation therapy. The majority of patients received a hypofractionated regimen, thus shortening the time required for treatment. It is unclear whether this weighed into the decision for treatment as opposed to observation. Palta et al evaluated over 40,000 women in the SEER database and confirmed a similarly small (68.65 to 61.7%) but statistically significant decline in the use of adjuvant radiation in women over 70 years of age after publication of CALGB 9343.¹⁰ Our decreased usage when compared to national rates is perhaps a reflection of the increased acceptance and adoption of the data supporting its omission in this patient population.

Similar to the findings of McCormick et al, we found the use of radiation therapy was strongly related to age. Around 75% of women 70–75 years of age received adjuvant radiation compared with fewer than 10% of women older than 85 years of age in our study. McCormick et al, using NCCN data, also showed that radiation use declined with age, decreasing from 80% in women 70–75 years of age to 41% for women over 80 years of age compared to the overall 6% decline.

Our data also indicate that use of radiation decreased as the Suemoto index increased, suggesting that mortality was considered in the decision-making process. In contrast, Soulos et al used a Surveillance, Epidemiology, and End Results (SEER) dataset to evaluate the use of RT among Medicare beneficiaries 3 years before and after publication of CALGB, and found a minimal and equivalent (between 3 and 4%) decrease in radiation between those with life expectancy greater than 10 years and those with life expectancy less than 5 years at the time of treatment.¹⁷ When the use of RT was evaluated by 5-year age groups in our study, the Suemoto index, used as a surrogate for life expectancy, was not associated with the use of adjuvant radiation. Thus, age alone may have been a larger consideration than predicted mortality.

It is possible that the nearly 50% rate of omission of radiation seen in our study does not reflect the broader community practice of breast surgeons. Shumway et al surveyed over 800 radiation oncologists and surgeons in 2015–2016. Despite the study time frame being even more recent than our study period, 40% of surgeons and 20% of radiation oncologists still responded that omission of radiation in this elderly population was not reasonable. Similarly, a significant proportion of respondents incorrectly identified the use of radiation in this group as improving patient survival, and overestimated the risk of local recurrence associated with radiation omission.¹⁸

Though radiation does not improve survival in this patient population, it does improve local control, and this risk reduction may be important to some women. In our study, recurrence was low, though, as expected, recurrence-free survival was better in patients who received adjuvant radiation compared to those who did not (98% versus 92%), but without a statistically significant difference in 5-year overall survival between groups. However, for some women, avoidance of local recurrence may have a significant impact on their quality of life, even though it may not improve their overall survival. Radiation is well tolerated with a with minimal toxicity and good cosmesis in older women¹⁹, and hypofractionated treatment minimizes inconvenience by significantly decreasing the duration of treatment. A 2014 publication comparing the cost effectiveness of no radiation, external beam radiation therapy (EBRT), and intensity-modulated radiation therapy (IMRT) in women eligible for the CALGB 9343 study found that while EBRT was cost-effective for older women with favorable breast cancer, the cost-effectiveness decreased with increasing age and with a decreased expected 10-year survival as well as with the use of IMRT or brachytherapy.²⁰ Thus, for women with a high likelihood of surviving 10 years, prevention of a local recurrence with adjuvant radiation may improve quality of life in a cost-effective manner. Despite this rationale, in our study on multivariable analysis, aside from tumor size, no factors known to increase risk of local recurrence, such as the presence of a poorly differentiated tumor or lymphovascular invasion, were associated with receipt of radiation.

Our data are limited in the ability to determine all of the factors used in decision making in this population due to their retrospective nature. Additionally, and not inconsequentially, we were not able to assess patient preference for the use of radiation and how this may have impacted its use.

It is clear that, while omission of radiation in women ≥ 70 years of age with small, ER+ tumors has no impact on overall survival, the use of radiation in this population remains high, even at an academic center. Aside from age and tumor size, no consistent features, including those known to influence local recurrence, were associated with the use of adjuvant radiation. Additionally, within age groups, a patient’s 10-year risk of mortality was not associated with the use of adjuvant radiation.

In this patient population, risk factors for recurrence, the patient’s risk of death from competing comorbidities, and, most importantly, patient preference regarding tolerance for a local recurrence and quality of life related to radiation therapy, must be carefully balanced. The development of a model which incorporates risk factors for recurrence and risk of death from other causes to facilitate individualized care in this patient population is the next step in optimizing care.

Synopsis:

Despite evidence that omission of RT in women with small estrogen receptor positive tumors is safe, use remains high. Neither factors consistent with risk of local recurrence or competing risk of death in 10 years predicted use of adjuvant radiation.