Has Hypofractionated Whole Breast Radiation Therapy Become Standard of Care in the United States? An Updated Report from National Cancer Database

Abstract

Introduction/Background:

We aimed to update the previous evaluation of hypofractionated whole breast irradiation (HF-WBI) use over time in the United States and factors related to its adoption for patients undergoing a lumpectomy from 2004–2016.

Materials and Methods:

Among the patients who underwent a lumpectomy, we identified 688,079 patients with early stage invasive breast cancer and 248,218 patients with ductal carcinoma in situ (DCIS) in the National Cancer Database from 2004–2016. We defined HF-WBI as 2.5–3.33 Gy/fraction to the breast, while conventional fractionated whole-breast irradiation (CF-WBI) as 1.8–2.0 Gy/fraction. We evaluated the trend of HF-WBI utilization and examined factors associated with HF-WBI utilization using logistic regression models.

Results:

Among invasive cancer patients, the usage of HF-WBI increased exponentially from 0.7% in 2004 to 15.6% in 2013 and then to 38.1% in 2016. Among DCIS patients, the usage of HF-WBI has increased significantly from 0.42% in 2004 to 13.4% in 2013 and then to 34.3% in 2016. Factors found to be associated with HF-WBI use included age, patient geographical location, race/ethnicity, tumor stage, grade, treating facility type and volume.

Conclusion:

HF-WBI utilization in the United States has more than doubled from 2013 to 2016. Although its use is close to that of CF-WBI, HF-WBI is still far from the preferred standard of care in the United States. We identified several patient and facility factors that can impact the uptake of HF-WBI treatment.

INTRODUCTION

Whole-breast radiation therapy can be categorized in two different ways: conventional fractionated whole-breast irradiation (CF-WBI) and hypofractionated whole-breast irradiation (HF-WBI). CF-WBI has been the traditional form of therapy, used as the standard early-stage breast cancer treatment method after breast conserving surgery (BCS)¹. It is typically delivered over the course of 5–7 weeks with a range of 1.8–2.0 Gy/fractions to the breast. Alternatively, HF-WBI involves administering 2.5–3.33 Gy/fractions to the breast over the course of 3 weeks². The increase in Gy/fractions over a shorter time period results in a more efficient form of therapy that spares patients weeks of their time.

Overwhelmingly results of randomized trials have demonstrated that HF-WBI has similar recurrence rates and disease-free survival as CF-WBI^2–6. The 10-year follow-up of these trials confirmed the efficacy of HF-WBI.^{4, 7} Additionally, HF-WBI reports comparable outcomes as CF-WBI and similar rates of toxicity^2–5. Not only are the regimens comparable in treatment results, it has been shown that HF-WBI is the more cost-effective route of treatment for breast cancer patients^{8, 9}. An analysis performed in 2017 demonstrated that HF-WBI resulted in higher quality-adjusted life years and lower cost when compared to CF-WBI⁹.

The American Society of Radiation Oncology (ASTRO) recommendations published in 2011 advised HF-WBI administration for patients older than 50 years, having a pT1–2N0 tumor, and not treated with chemotherapy¹⁰. After the guidelines were published, an analysis of National Cancer Database (NCDB) data reported that the usage of HF-WBI reached 15.6% among invasive cancer patients and 13.4% among DCIS patients in 2013¹¹, which were significant higher than the HF-WBI utilization rates reported before 2012^{12, 13}. Another study utilizing commercial claims data found that HF-WBI use reached 34.5% in the hypofractionation-endorsed patients and 21.2% in the hypofractionation-permitted patients in 2013⁸. While the usage during the time period of analysis saw a significant increase, the United States’ rates of HF-WBI use were remarkably lower than other countries at the time¹⁴. Most notably after the 2002 Ontario Clinical Oncology Group trial were published,¹⁵ outlining shorter radiation therapy schedules, Ontario, Canada observed over 70% of eligible patients receiving HF-WBI¹⁶. Similarly, an evaluation conducted in 2010 found that the United Kingdom used HF-WBI as the standard form of treatment¹⁷.

The NCDB has collected more recent data on the use of the varying forms of whole breast radiation since 2013. Currently, there are no publically available publications assessing the status of HF-WBI adoption in the United States apart from the previous analysis through 2013, how the adoption of HF-WBI in practices around the nation are being distributed amongst eligible patients, and the factors that may affect whether a patient receives HF-WBI.

In this study we have three aims. Our primary aim is to update the evaluation of HF-WBI usage over time and its current rates with the most updated data available. Secondly, we aim to identify factors related to the adoption of HF-WBI in the United States, and examine if radiotherapy treatments across different regions of the U.S. differ. Lastly, we aim to evaluate the adoption of HF-WBI across cancer treatment facilities.

MATERIALS and METHODS

Study Samples

In this study, we analyzed HF-WBI use trends over time and by several factors. We used data from the NCDB, the American College of Surgeons and the American Cancer Society’s database collecting hospital registry data covering approximately 80% of new breast cancer cases in the US.¹⁸ The database contains information on radiation treatment schedules, dosages, fractions, boost fields, as well as demographic and clinical information. With the data, we are able to analyze factors related to HF-WBI, the way HF-WBI is administered, and the trend of use compared to CF-WBI. No patient, provider, or hospital identifiers were examined in this study, no protected health information was reviewed, and the analysis was retrospective using de-identified data, so institutional review board review was waived for this study.

Using the NCDB data from 2004–2016, this study included women with the following criteria. Mirroring the study sample from the previous analysis from 2004–2013, women with a first cancer diagnosis of breast cancer, had finished or had started their first course of cancer treatment performed at the reporting facility, and received lumpectomy were included. Patients with ductal carcinoma in situ (DCIS) and patients with invasive breast cancer who had AJCC pT1–2N0–1 disease were included. Patients who underwent neoadjuvant therapy, had pT3–4N2–3 disease, and underwent a mastectomy were excluded. Furthermore, women ≥70 years of age with clinically lymph node-negative, estrogen receptor-positive, and T1 breast cancer were excluded because these patients do not always necessarily receive whole breast irradiation^{19, 20}. As a contrast, we also investigated the types of post-lumpectomy radiotherapy in women with pT3 or pN2–3 breast cancer.

Variables Studied

NCDB recorded total fractions of radiation, dose to the breast, boost dose, and the total days of undergoing radiotherapy. By separating the total fractions of radiotherapy into the fractions of initial radiotherapy to the breast and boost radiotherapy within the total days of therapy, we could calculate dose per fraction to the breast as well as the dose per fraction in the boost therapy (see details in Supplemental Materials). We defined HF-WBI as 2.5–3.33 Gy/fraction to the breast, while conventional therapy (CF-WBI) as 1.8–2.0 Gy/fraction^{2, 4}. Using this classification, almost all HF-WBI patients (99%) received 14–18 fractions and 98% of CF-WBI had 22–28 fractions. The distribution of common regimens of HF-WBI (at least 200 patients) are shown in Supplementary Table 1.

The distribution and categorization of demographic, clinical, and facility factors are listed in Table 1. Age was analyzed as a 10-group categorical variable. Tumor size was categorized as < 2.0 centimeters or 2.1–5.0 centimeters, while nodal stage was analyzed as a categorical variable. Comorbidity status was represented as Deyo/Charlson comorbidity index²¹. Facility location was recorded as the U.S. Census regions (www.census.gov) and states in each region can be found in the Table 1 footnotes. The median household income of each patient’s area of residence was derived from the 2012 American Community Survey data (www.census.gov/programs-surveys/acs). Facility volume was calculated as the average number of breast cancer patients per year at each reporting facility.

Table 1.

Characteristics in invasive breast cancer patients treated with HF-WBI vs. CF-WBI

	CF-WBI (n=433,152)	HF-WBI (n=81,864)	%HF-WBI	AOR (95% CI)a	Chi-square*
Age at diagnosis					4175.14
<40	16,045	921	5.43	1.0 (ref)
40–44	31,208	2,574	7.62	0.74 (0.67–0.82)
45–49	54,150	5,930	9.87	0.94 (0.86–1.04)
50–54	64,826	10,606	14.06	1.43 (1.31–1.57)
55–59	71,909	13,944	16.24	1.66 (1.51–1.82)
60–64	74,557	17,714	19.20	1.93 (1.76–2.12)
65–69	68,444	19,681	22.33	2.25 (2.04–2.47)
70–74	23,398	4,053	14.76	2.53 (2.28–3.69)
75–79	16,611	3,065	15.58	3.32 (2.98–3.69)
80+	12,004	3,376	21.95	5.96 (5.35–6.65)
Race/ethnicity					355.84
White	346,741	65,473	15.88	1.0 (ref)
Black	46,907	6,838	12.72	0.85 (0.82–0.88)
Hispanic	19,738	4,098	17.19	1.05 (1.01–1.10)
Asian	13,041	4,017	23.55	1.48 (1.41–1.55)
Other/unknown	6,725	1,438	17.62	1.08 (1.00–1.17)
Insurance status					40.70
Not Insured	7,556	1,033	12.03	1.04 (0.96–1.13)
Private Insurance	278,025	48,487	14.85	1.0 (ref)
Medicaid	24,711	3,945	13.77	0.91 (0.87–0.95)
Medicare	112,990	26,681	19.10	0.93 (0.90–0.96)
Other Government	4,407	820	15.69	0.88 (0.80–0.96)
Insurance Status Unknown	5,463	898	14.12	1.03 (0.94–1.13)
Charlson/Deyo comorbidity index					48.79
0	377,767	70,130	15.70	1.0 (ref)
1	46,102	9,441	17.00	0.92 (0.89–0.95)
2	7,372	1,698	18.72	0.88 (0.82–0.94)
≥3	1,911	595	23.74	0.86 (0.76–0.96)
N stage in invasive cancer					5902.16
pN0	335,194	75,205	18.32	1.0 (ref)
pN1	91,714	5,372	5.53	0.26 (0.25–0.26)
T stage in invasive cancer					285.38
No Tumor	3,033	498	14.10	0.53 (0.47–0.59)
≤2.0 cm	329,214	67,925	17.10	1.0 (ref)
2.1–5.0 cm	99,573	13,242	11.74	0.80 (0.78–0.82)
Laterality			4		54.71
Right	213,775	41,715	16.33	1.0 (ref)
Left	219,081	40,122	15.48	0.93 (0.91–0.95)
Histology in invasive cancer
Ductal	355,692	65,660	15.58
Lobular	31,315	7,277	18.86
Ductal & lobular	21,755	4,403	16.83
Others	24,390	4,524	15.65
Grade					324.77
1	105,244	26,505	20.12	1.0 (ref)
2	178,282	35,781	16.72	0.86 (0.84–0.88)
3	125,906	15,729	11.11	0.75 (0.73–0.78)
Chemotherapy					1999.11
No	236,904	63,183	21.05	1.0 (ref)
Yes	186,779	17,173	8.42	0.54 (0.53–0.56)
Estrogen receptor					25.48
Negative	78,284	9,584	10.91	1.0 (ref)
Positive	348,102	71,874	17.11	1.09 (1.06–1.13)
Surgical margin					48.82
Negative	413,598	79,254	16.08	1.0 (ref)
Positive	16,512	2,241	11.95	0.81 (0.77–0.86)
Facility locationb					3142.51
New England	33,612	5,919	14.97	1.0 (ref)
Middle Atlantic	66,798	15,097	18.43	1.03 (0.98–1.07)
South Atlantic	88,220	15,263	14.75	1.16 (1.11–1.21)
East North Central	80,690	15,251	15.90	1.31 (1.25–1.36)
East South Central	22,798	3,700	13.96	1.15 (1.09–1.22)
West North Central	32,231	5,884	15.44	1.21 (1.15–1.28)
West South Central	23,911	2,798	10.48	0.63 (0.59–0.67)
Mountain	16,335	5,540	25.33	2.99 (2.84–3.16)
Pacific	52,512	11,491	17.95	1.45 (1.39–1.52)
Facility volume, number of breast cancers per year					1452.74
<120	92,402	11,705	11.24	1.0 (ref)
120–207	109,787	17,031	13.43	1.26 (1.23–1.30)
208–342	117,859	21,541	15.45	1.36 (1.31–1.40)
343+	110,329	28,549	20.56	1.84 (1.78–1.90)
Facility type					1204.00
Community Cancer Program	242,626	36,930	13.21	1.0 (ref)
Integrated Network Cancer Program	60,892	12,192	16.68	1.30 (1.26–1.34)
Academic/Research Program	113,589	31,821	21.88	1.54 (1.50–1.58)
Median household income of residential zip code					272.95
<$38,000	59,921	9,117	13.21	1.0 (ref)
$38,000-$47,999	90,593	14,684	13.95	1.05 (1.02–1.09)
$48,000-$62,999	116,182	21,156	15.40	1.14 (1.10–1.18)
≥$63,000	164,431	36,739	18.26	1.29 (1.24–1.33)
Distance to facility					99.23
< 5.0 miles	139,232	24,646	15.04	1.0 (ref)
5.0–9.9 miles	110,744	19,972	15.28	0.92 (0.89–0.94)
10.0–19.9 miles	97,973	19,175	16.37	0.95 (0.92–0.97)
20.0–49.9 miles	63,105	12,743	16.80	0.98 (0.95–1.01)
>=50.0 miles	20,432	5,202	20.29	1.14 (1.09–1.20)
Urban/Rural					330.98
Large metropolis (pop. ≥1 million)	225,462	47,891	17.52	1.0 (ref)
Metropolis (pop. <1 million)	139,480	22,226	13.74	0.83 (0.81–0.85)
Urban (pop. ≥20000)	23,886	3,774	13.64	0.92 (0.88–0.97)
Urban (pop. 2500–19999)	27,159	4,624	14.55	1.04 (0.99–1.09)
Rural	5,839	1,149	16.44	1.20 (1.10–1.30)

^aCalculated from multivariable logistic regression, adjusting for year of diagnosis and all other variables with adjusted odds ratios in the table. Variables without adjusted odds ratios were not included in the final logistic regression model.

^bFacility location: New England (CT, MA, ME, NH, RI, VT), Middle Atlantic (NJ, NY, PA), South Atlantic (DC, DE, FL, GA, MD, NC, SC, VA, WV), East North Central (IL, IN, MI, OH, WI), East South Central (AL, KY, MS, TN), West North Central (IA, KS, MN, MO, ND, NE, SD), West South Central (AR, LA, OK, TX), Mountain (AZ, CO, ID, MT, NM, NV, UT, WY), Pacific (AK, CA, HI, OR, WA).)

^*All p-values <0.001

Abbreviations: HF-WBI, hypofractionated whole breast irradiation; CF-WBI, conventional fractionated whole breast irradiation; pop., population; AOR, adjusted odds ratio; CI, confidence interval

Statistical Analysis

We started our analysis by examining the trend of radiation therapy overall from 2004–2016 using a generalized linear model with log link and binomial distribution. First, we examined the use of any radiotherapy, HF-WBI, CF-WBI, and accelerated partial breast irradiation (APBI) among three subgroups of patients, including patients with invasive cancer, DCIS, and patients who were 50 years or older, had pT1–2N0 disease and did not receive chemotherapy. Patients in the last subgroup had been recommended for HF-WBI by ASTRO in 2011¹⁰. Second, we examined the trend of HF-WBI use over time by clinical and facility factors to show which group of patients adopted HF-WBI earlier. Third, we examined clinical and facility factors that are related to HF-WBI use using multivariable logistic regressions in patients with invasive breast cancer and DCIS separately. Stepwise backward selection approach was used and variables with p<0.001 were kept. Adjusted odds ratios (OR) and 95% confidence intervals (CI) were estimated from these logistic regression models. In addition, the adjusted trend of HF-WBI use (versus CF-WBI) was also plotted for selected factors using predicted probabilities from the multivariable logistic regression. A sensitivity analysis was done by restricting only to cases whose all treatments were received at the reporting facilities. Lastly, we conducted facility-level analysis by calculating the percentage of HF-WBI use within each facility over time among facilities that treated at least 10 breast cancer patients each year. Statistical analyses were conducted using the STATA 15 software package (StataCorp, College Station, TX).

RESULTS

HF-WBI use over time

This study included 936,297 post-lumpectomy breast cancer patients, analyzing 688,079 invasive breast cancer patients and 248,218 DCIS patients over the years 2004 to 2016. We examined radiotherapy use over the years and separated by radiotherapy type. Figure 1 shows the trends of radiotherapy use over time among the breast cancer patients who had a lumpectomy. Among invasive cancer patients, the usage of HF-WBI increased significantly from 0.7% in 2004 to 15.6% in 2013 and then to 38.1% in 2016 (p<0.001, Figure 1A). Though a substantial decrease in CF-WBI accompanied the considerable increase in HF-WBI use, overall radiotherapy use demonstrated a slight increase (83.3% in 2004 and 88.6% in 2016, p<0.001). Similarly to invasive patients, usage of HF-WBI among DCIS patients has increased significantly from 0.4% in 2004 to 13.4% in 2013 and then to 34.3% in 2016 (p<0.001, Figure 1B). Figure 1C displays the overall trend in radiotherapy use among invasive breast cancer patients who are 50 years or older, had a pathologic T1–2N0 disease, and did not use chemotherapy, a group of patients recommended to receive HF-WBI according to the 2011 ASTRO Guidelines. Particularly for this population, HF-WBI use increased dramatically over the years from 0.9% in 2004 to 49.4% in 2016.

Figure 1.

Trend of post-lumpectomy radiation therapy over time in patients with invasive breast cancer who received lumpectomy (A), patients with ductal carcinoma in situ (DCIS) (B), and patients aged 50 years or older with pT1–2N0 breast cancer, without prior chemotherapy (C). APBI, accelerated partial breast irradiation; CF-WBI, conventional fractionated whole breast irradiation; HF-WBI, hypofractionated whole breast irradiation.

With interest, we also examined post-lumpectomy HF-WBI use in patients with pT3 or pN2–3 breast cancer (n=25,191). Prior to 2014, few patients (0.8%, 153 of 20,195) received HF-WBI in this group, while in 2014–2016, a small percentage of the patients (4.0%, 198 of 4996) received HF-WBI. This mainly confined to patients with pT3N0 disease diagnosed in 2014–2016 (21.8%, 153 of 570).

Factors related to the use of HF-WBI and trend of HF-WBI use over time

Table 1 showed characteristics of 81,864 invasive breast cancer patients received HF-WBI versus 433,152 received CF-WBI. In the multivariable analysis of factors related to the use of HF-WBI (column 4), we found that age at diagnosis, nodal status, receipt of chemotherapy, and census region of facility located were strongly correlated with HF-WBI utilization. There was monotonic increasing trend that the older the patients, the more likely they received HF-WBI. Node positive patients were less likely to receive HF-WBI than node negative patients, and patients receiving chemotherapy were less likey to recive HF-WBI. Patients in Mountain census region were more likely to receive HF-WBI, while patients in West South Central census region were less likely than other regions. We also found several other factors moderately and statistically significantly associated with the use of HF-WBI, including race, insurance, Charlson comorbidity index, T stage, laterality, histological grade, estrogen receptor status, surgical margin, facility type and volume, median income of communities, distance to facility, and urban-rural continuum. In particular, compared to White women, African American women were less likely and Asian American women were more likely to receive HF-WBI. As the volume of a facility (number of patients treated per facility) increased, the more likely the facility was to administer HF-WBI. We also observed that academic cancer programs were most likely to administer HF-WBI in comparison to community programs in the multivariable analysis. In the sensitivity analysis restricting to patients whose all treatments were all received at the reporting facilities, the results are similar to the primary analsyis, except that the associations for facility type and facility volume were stronger (Supplementary Table 2).

Based on the multivariable model in Table 1, we estimated the adjusted percentage of invasive breast cancer patients who received HF-WBI by six demographic and clinical factors. Each factor’s graph illustrates the adjusted trend of HF-WBI use over time from 2004 to 2016 (Figure 2). There is a clear and gradual increase in the rate of HF-WBI use in all the subgroups defined by the selected factors, most notably after 2011–2012. However, the rates of HF-WBI use have yet to reach 60% (signifying the “majority” percentage benchmark) in any of the subgroups. Adjusting for all other factors, older patients adopted HF-WBI earlier; for example, use of HF-WBI reached 20% in patients ≥70 years old in 2011 but patients <40 years old did not reach 20% utilization until 2016. Similarly, while adjusting for all other factors, Asian Americans adopted HF-WBI the earliest, while African American patients adopted it at a slower rate. There was a large difference between patients according to tumor stage; approximately 50% of patients with node negative T1 or T2 tumors had used HF-WBI in year 2016, but <30% in patients with node positive T1 or T2 diseases in the same year. Only slight differences were seen between histological grade, estrogen receptor status, or insurance type in the adoption of HF-WBI.

Figure 2.

Uptake of Hypofractionated Radiotherapy over Time in Patients with Invasive Breast Cancer, by demographic and clinical factors. Adjusted Percents of patients receiving hypofractionated radiotherapy are presented. Variables in Table 1 were adjusted for in the model.

The uptake of HF-WBI also varied by several facility and location factors (Figure 3). Each factor’s graph illustrates the adjusted trend of HF-WBI use over time. Most notably is the difference in usage rate across census regions, with the Mountain region (AZ, CO, ID, MT, NM, NV, UT, WY) starting the earliest and West South Central region (AR, LA, OK, TX) the latest. Academic or high volume cancer programs led the adoption of HF-WBI than other cancer programs. Small differences were observed according to median income of communities, or the distance to facility (defined as the distance to the reporting facility). Across the urban-rural continuum, we do see a slight lead in HF-WBI uptake in rural communities.

Figure 3.

Uptake of Hypofractionated Radiotherapy over Time in Patients with Invasive Breast Cancer, by facility and locational factors. Adjusted Percents of patients receiving hypofractionated radiotherapy are presented. Variables in Table 1 were adjusted for in the model.

Among DCIS patients, 126,658 received CF-WBI and 26,216 received HF-WBI (Table 2). The results for DCIS patients and the factors that affected HF-WBI use were similar to that of the invasive breast cancer patients. The most significant factors found were age and facility location.

Table 2.

Characteristics in DCIS patients treated with HF-WBI vs. CF-WBI

	CF-WBI (n=126,658)	HF-WBI (n=26,216)	%HF-WBI	AOR (95% CI) a	Chi-square*
Age at diagnosis					1430.69
<40	2,597	190	6.82	1.0 (ref)
40–44	10,863	1,037	8.71	0.67 (0.52–0.86)
45–49	17,313	2,082	10.73	0.93 (0.73–1.18
50–54	19,450	3,379	14.80	1.34 (1.06–1.71)
55–59	20,450	4,116	16.75	1.62 (1.28–2.05)
60–64	18,737	4,486	19.32	1.91 (1.50–2.42)
65–69	16,716	4,510	21.25	2.17 (1.70–2.76)
70–74	10,859	3,289	23.25	2.92 (2.28–3.73)
75–79	6,567	1,960	22.99	3.92 (3.05–3.73)
80+	3,106	1,167	27.31	6.84 (5.29–8.86)
Race/ethnicity					121.84
White	97,695	19,836	16.88	1.0 (ref)
Black	15,813	2,867	15.35	0.79 (0.74–0.85)
Hispanic	6,225	1,393	18.29	0.91 (0.82–1.00)
Asian	4,923	1,697	25.63	1.44 (1.32–1.58)
Other/unknown	2,002	423	17.44	0.95 (0.80–1.12)
Insurance status					25.44
Not Insured	1,940	371	16.05	1.49 (1.27–1.75)
Private Insurance	81,418	14,508	15.12	1.0 (ref)
Medicaid	5,936	1,198	16.79	0.97 (0.88–1.07)
Medicare	34,392	9,603	21.83	1.02 (0.96–1.09)
Other Government	1,388	262	15.88	0.94 (0.77–1.14)
Insurance Status Unknown	1,584	274	14.75	0.99 (0.81–1.22)
Charlson/Deyo comorbidity index					17.41
0	111,343	22,412	16.76	1.0 (ref)
1	12,943	3,093	19.29	0.88 (0.83–0.94)
2	1,911	529	21.68	0.89 (0.77–1.03)
≥3	461	182	28.30	0.92 (0.69–1.23)
Tumor size in DCIS					105.61
≤ 1.0 cm	48,101	8,392	14.85	1.0 (ref)
1.1–2.0 cm	23,284	4,218	15.34	0.85 (0.81–0.89)
2.1–3.0 cm	7,292	1,427	16.37	0.81 (0.76–0.87)
>3.0 cm	6,515	1,195	15.50	0.72 (0.67–0.78)
Laterality					22.03
Right	61,535	13,156	17.61	1.0 (ref)
Left	65,072	13,055	16.71	0.91 (0.87–0.95)
Histology in DCIS					53.71
Ductal	95,208	20,004	17.36	1.0 (ref)
Ductal & lobular	3,675	688	15.77	0.90 (0.79–1.03)
Comedocarcinoma	12,446	2,079	14.31	0.82 (0.77–0.88)
Cribriform	11,347	2,660	18.99	1.13 (1.06–1.21)
Papillary	2,952	628	17.54	1.11 (0.98–1.26)
Paget’s disease	315	77	19.64	0.90 (0.60–1.34)
Others	715	80	10.06	0.78 (0.55–1.10)
Grade
1	15,372	2,979	16.23
2	41,909	9,783	18.93
3	44,630	9,813	18.02
Estrogen receptor
Negative	16,418	3,548	17.77
Positive	95,599	21,842	18.60
Surgical margin					12.46
Negative	121,253	25,412	17.33	1.0 (ref)
Positive	4,383	671	13.28	0.80 (0.71–0.91)
Facility locationb					829.62
New England	10,963	1,740	13.70	1.0 (ref)
Middle Atlantic	20,629	4,716	18.61	1.05 (0.95–1.16)
South Atlantic	25,891	5,200	16.73	1.29 (1.18–1.42)
East North Central	25,097	5,336	17.53	1.35 (1.23–1.48)
East South Central	6,413	1,159	15.31	1.02 (0.90–1.16)
West North Central	9,802	1,926	16.42	1.34 (1.20–1.50)
West South Central	6,930	951	12.07	0.82 (0.72–0.94)
Mountain	4,102	1,595	28.00	3.72 (3.31–4.19)
Pacific	14,234	3,403	19.29	1.49 (1.35–1.65)
Facility volume, number of breast cancers per year					460.90
<120	27,201	3,658	11.85	1.0 (ref)
120–207	32,772	5,722	14.86	1.40 (1.31–1.49)
208–342	35,404	7,061	16.63	1.46 (1.37–1.57)
343+	30,774	9,120	22.86	2.08 (1.94–2.23)
Facility type					280.67
Community Cancer Program	72,248	11,971	14.21	1.0 (ref)
Integrated Network Cancer Program	18,251	4,145	18.51	1.31 (1.24–1.40)
Academic/Research Program	33,562	9,910	22.80	1.54 (1.46–1.62)
Median household income of residential zip code					61.52
<$38,000	17,349	3,053	14.96	1.0 (ref)
$38,000-$47,999	25,658	4,758	15.64	1.07 (1.00–1.16)
$48,000-$62,999	33,701	6,731	16.65	1.15 (1.07–1.24)
≥$63,000	49,360	11,614	19.05	1.29 (1.20–1.39)
Distance to facility
< 5.0 miles	43,105	8,156	15.91
5.0–9.9 miles	33,707	6,758	16.70
10.0–19.9 miles	28,361	6,186	17.91
20.0–49.9 miles	16,465	3,751	18.55
≥ 50.0 miles	4,545	1,321	22.52
Urban/Rural					182.33
Large metropolis (pop. ≥1 million)	67,695	15,769	18.89	1.0 (ref)
Metropolis (pop. <1 million)	40,507	6,931	14.61	0.78 (0.75–0.82)
Urban (pop. ≥20000)	6,515	1,087	14.30	0.87 (0.79–0.97)
Urban (pop. 2500–19999)	7,084	1,360	16.11	1.17 (1.06–1.28)
Rural	1,533	380	19.86	1.64 (1.39–1.94)

^aCalculated from multivariable logistic regression, adjusting for year of diagnosis and all other variables with adjusted odds ratios in the table. Variables without adjusted odds ratios were not included in the final logistic regression model.

^bFacility location: New England (CT, MA, ME, NH, RI, VT), Middle Atlantic (NJ, NY, PA), South Atlantic (DC, DE, FL, GA, MD, NC, SC, VA, WV), East North Central (IL, IN, MI, OH, WI), East South Central (AL, KY, MS, TN), West North Central (IA, KS, MN, MO, ND, NE, SD), West South Central (AR, LA, OK, TX), Mountain (AZ, CO, ID, MT, NM, NV, UT, WY), Pacific (AK, CA, HI, OR, WA).

^*All p-values <0.001

Abbreviations: HF-WBI, hypofractionated whole breast irradiation; CF-WBI, conventional fractionated whole breast irradiation; pop., population; AOR, adjusted odds ratio; CI, confidence interval

Facility-level Analysis on the Diffusion of HF-WBI

A total of 1253 cancer programs/centers were in the facility-level analysis, including 841 community programs, 189 integrated network programs, and 223 academic programs. The diffusion of HF-WBI across cancer programs in the United States was a slow, gradual process (Supplementary Figure 1 and Video Clip). In 2004, 30 facilities started to give HF-WBI to at least 5% of their breast cancer patients, and the use of HF-WBI started mainly at large academic cancer programs (n=13). Until 2007, there were only two academic cancer programs that gave HF-WBI to more than 50% of their patients. By 2011, only 4.0% of cancer programs gave HF-WBI to more than 50% of their patients. After 2012, most cancer programs had started to use HF-WBI but there was a large variation in the proportion of patients receiving HF-WBI across programs; some centers treated few patients, while others treated a higher percentage of patients (Figure 4). By 2016, the vast majority of cancer programs were delivering HF-WBI; 45.2% of all centers gave HF-WBI to more than 50% of their patients and only few centers gave HF-WBI to all of their patients.

Figure 4.

Scatter plot of percent of patients receiving hypofractionated radiotherapy within each treating facility against numbers of patients treated in each facility in years of 2013–2016. Each dot represents a cancer treatment facility that treated at least 20 patients.

DISCUSSION

HF-WBI utilization in the United States has increased exponentially in the last 13 years and has more than doubled from 2013 to 2016. Although its use is close to that of conventional WBI, HF-WBI is still far from the preferred standard of care in the United States. More specifically, many patients who are preferred candidates for HF-WBI based on the 2011 ASTRO guidelines are still receiving conventional WBI in 2016. This is in contrast to other countries including United Kingdom and Canada, where HF-WBI are the standard of care for post-lumpectomy breast cancer ^{14, 16}.

We identified several patient and facility factors that might impact the uptake of HF-WBI treatment in both invasive breast cancer patients and DCIS patients. The most noteworthy factors related to the uptake of HF-WBI were age of patient, race/ethnicity, nodal status, tumor size, and facility location, volume, and types. While there were multiple factors that are related to HF-WBI use, the percentage of HF-WBI use was well below 60% in many of the categories defined by these factors. Nevertheless, these factors may help us to understand the underlying reasons for slow uptake of HF-WBI. For example, we see a substantial difference in utilization rate based on geographical location. Patients in the Mountain regions are more likely to receive HF-WBI treatment than those of other areas. Less frequency traveling to hospital might be one reason for patients in Mountain states to adopt HF-WBI earlier. Interestingly, patients with left breast cancer were slightly less likely to receive HF-WBI than patients with right breast cancer, suggesting that some doctors may still concern about the cardiotoxicity associated with high dose per fraction in HF-WBI. Our findings may also inform clinicians and treatment facilities about the types of patients currently benefiting from HF-WBI use and those potentially underutilizing this treatment option.

The diffusion of HF-WBI across treating facilities in the U.S. was a gradual and slow process and the adoption of HF-WBI within a facility was not uniform among patients. Some academic cancer centers apparently lead the way of HF-WBI adoption, followed by cancer programs in integrated networks, then by community centers. As illustrated in Figure 4 and supplemental figure 1, there was no overwhelming percentage of eligible patients receiving the treatment, suggesting that within a cancer center, doctors may offer this therapy regimen on a case-by-case basis that may not adhere to ASTRO recommendations. Additionally, the lower uptake for hypofractionated therapy use in DCIS patients may be influenced by the lack of randomized data. The recently published Danish Breast Cancer Group (DBCG) HYPO trial (NCT00909818) found that moderately hypofractionated breast irradiation of node-negative breast cancer or DCIS patients did not result in more breast induration or locoregional recurrence compared with standard fractionated therapy.²² These findings have the potential of increasing the uptake HF-WBI utilization in the DCIS patient population moving forward.

Although HF-WBI is a cost-effective option for breast cancer patients⁹, it is still not a standard of care in the U.S. The extent to which financial incentives may partially contribute to the slow uptake of HF-WBI is currently unknown. In an average US hospital, it is estimated that there would be an annual reduction of over $300,000 in revenue if they were to utilize HF-WBI at a rate of over 70%¹⁴. These decreases in revenue from radiation oncology reflect broader shifts in the health care system to less volume-based care and more value-based care²³. The temporal trend of increased use of hypofractionated whole breast radiotherapy coincides with a growing cost-consciousness in cancer care that has motivated seeking greater efficiency in health care delivery for patients at lower costs²⁴. Policy changes and the development of new reimbursement models to support high-quality radiation oncology care may reduce the disincentives of the current payment system for using highly cost-effective treatments like HF-WBI ²⁵. The current study has no detailed data on insurance policy and financial incentives, but future studies should compare insurance benefit plans, especially commercial insurance policies, which allow different co-payments and deductibles, as well as reimbursement structure for providers to investigate financial reasons for slow uptake of HF-WBI.

In 2018, ASTRO published new guidelines, stating that HF-WBI is the preferred standard of care in most women with early-stage breast cancer²⁶, which expanded the suitable patient popualtions than the 2011 guidelines. This new recommendation may speed up the adoption of HF-WBI and make it the standard of care in the U.S. The uptake of this therapy has increased elsewhere in recent years after similar guidelines were made. In a retrospective cohort study in Ontario, Canada, the rate of HF-WBI use in 2015 was noted to have increased to 75% from 64% in 2009 in patients with invasive breast cancer and DCIS.²⁷ However, continued monitoring of radiotherapy utilization trends is necessary, as the 2011 ASTRO recommendation has not caused a noticeable change to the uptake curve of HF-WBI.

This study had two notable strengths. One was the large national sample with detailed collection of information about breast cancer patient demographics, diagnosis, treatment, and multiple facility factors up to 2016. The second strength was our precise definition of HF-WBI. We created a list of all the combinations of dose to the breast, boost dose, and number of and days on radiotherapy to avoid misclassification in radiation types. Therefore, the current study can provide an accurate estimate of the utilization trend of HF-WBI in the U.S. This study also had several limitations. Treating facilities that participate in the NCDB are Commission on Cancer (CoC) accredited centers, which are more likely to be large academic centers and less likely to be rural centers; the NCDB is not a population-based dataset²⁸. Nonetheless, the NCDB captured over 70% of all diagnosed cancers in the US²⁸, so the conclusions may still be valid even if the sample is not representative. Additionally, the NCDB database does not include detailed information on insurance policies of patients, so we were not able to explore whether financial incentives play a role on the adoption of HF-WBI.

CONCLUSION

In summary, the utilization of HF-WBI in the United States has exponentially increased from 2004 to 2016. Although its use is close to that of conventional WBI, HF-WBI is still far from the preferred standard of care in the United States. In the era of COVID-19 pandemic, it is more urgent to adopt HF-WBI in order to reduce unnecessary clinical visits and minimize the exposure to the dangerous coronavirus. With the publication of 2018 ASTRO guidelines on HF-WBI and COVID-19 pandemic, future studies can continuously monitor the trend of HF-WBI utilization, and further investigate financial incentive factors on the use of HF-WBI.