Race-related Subsequent Breast Events Following Ductal Carcinoma in Situ: A Surveillance, Epidemiology, and End Results–based analysis

Abstract

BACKGROUND

The risk of subsequent ductal carcinoma in situ (DCIS) or invasive breast cancer (IBC) has been evaluated in either breast after a DCIS diagnosis; we modeled competing risks of ipsilateral and contralateral DCIS or IBC subtypes by self-reported race.

METHODS

A cohort of 198,827 women diagnosed with primary unilateral DCIS between 2000 and 2022 was identified from the U.S. SEER tumor registries. Competing subdistributional hazard (sHR) models were used to estimate DCIS or IBC laterality-associated risks overall and for the estrogen receptor (ER+) or progesterone receptor (PR+) overexpressing (ER+/PR+) and ER−PR− expressing tumor subtypes. Cox models were used for sub-analysis to estimate the overall risk of a second event (DCIS or IBC).

RESULTS

Within 10 (±6.1) average years of follow-up after initial DCIS, 16,148 women had a subsequent event (25.7% DCIS and 74.3% IBC). Overall, compared with White women, Black women had an elevated risk for IBC for either tumor subtype whereas Asian and Hispanic women had an elevated risk for ER−PR− IBC. For tumor aggressiveness by laterality, Black women had an elevated IBC risk in either breast for ER−PR− (sHR = 1.82, 0.95% CI 1.53–2.17 in the ipsilateral and 1.512, 0.95% CI 1.24–1.86 in the contralateral breast), as did Hispanic women and Asian women in the ipsilateral breast only; with stronger association vs ER+/PR+, p _het = 0.0001.

CONCLUSION

These contemporary data reflect treatment patterns since 2000, showing an elevated risk of subsequent breast tumors in either breast among Black women after DCIS.

Precis:

Black women continue to face an elevated risk of aggressive IBC, possibly driven by distinct biological pathways. Further evidence of risk in both breasts after DCIS supports additional research to understand how to identify these women with high-risk to provide appropriate therapy and surveillance.

Introduction

Ductal carcinoma in situ (DCIS) of the breast is a preinvasive form of breast cancer characterized by the proliferation of atypical epithelial cells within the breast ducts.¹ In 2025, approximately 59,080 new cases of DCIS are expected to be diagnosed in the U.S.² Since the widespread adoption of mammography screening in 2000,³ DCIS incidence rates have remained stable for White women,^4,5 but have increased annually by 1.6% for Black women and 1.0% for women of other races.⁶

DCIS is often considered a precursor to invasive breast cancer. In the Surveillance, Epidemiology, and End Results (SEER) registries for women diagnosed with DCIS through 2009, more than 70% of cases were subsequent invasive breast cancers.⁷ Additionally, risk following DCIS varies by race and ethnicity.^7,8 Hispanic and Black women with DCIS have a higher risk of subsequent breast events compared with White women.^7–9 Therefore, accurate race-specific risk estimates are essential to guide treatment planning and management for DCIS patients. Risk of subsequent breast outcomes following DCIS whether in the ipsilateral,^10,11 contralateral,^12,13 or either breast^14–16 may be overestimated due to exclusion or censoring of women who develop breast events in the opposite breast (denominator) while the events of interest remain constant.

This study provides an update on the risk of subsequent breast events after DCIS – reflecting recent practice, and contemporary surveillance and diagnosis, while improving risk estimation accuracy through competing risk methodology, thereby mitigating potential risk estimation bias. Leveraging data from the U.S. SEER 2000–2022 cancer registries, the primary objective of this study was to estimate the risk of subsequent breast events—either subsequent DCIS or IBC, as well as IBC alone, following a primary DCIS diagnosis—by laterality of ER subtypes, stratified by self-reported race and ethnicity. The overall race-associated risks of subsequent breast events after DCIS were also assessed.

Methods

Data Source and Patient Selection

All women diagnosed with primary in situ diagnosis from 2000 to 2022 (n = 296,300) were identified from the NCI’s SEER 18 registries (November 2023 Submission). The SEER cancer registries collect and publish population-level data on cancer incidence and survival, representing ~48% of the U.S. population.^17,18 We first excluded women diagnosed with lobular carcinoma in situ (n = 49,564). Of the remaining 246,736 DCIS only cases, eligible women had to be ≥20 years old at their initial DCIS diagnosis (no exclusions), with unilateral DCIS as the only primary or first of ≥2 primary cancer diagnoses (n = 33,386 excluded); have known race or ethnicity information (n = 2,425 unknown race excluded and 227 non-Hispanic American Indian/Alaskan Native women excluded); and have at least six months of follow-up from the index DCIS (n = 11,871 total excluded, 5 without follow-up information, 7,323 with IBC and 4,543 with DCIS within 6-months). Due to insufficient sample size for meaningful analysis, 227 non-Hispanic American Indian/Alaska Native women were excluded from the analysis. Following all exclusion criteria, 198,827 women were eligible for analysis of the risk of subsequent breast outcomes (DCIS or IBC) after the index DCIS.

Exposure Measures

SEER included the following information of the index DCIS cases: age at diagnosis; year at diagnosis; geographic region of the 16 central SEER registries (18 geographic areas, with California and Georgia as whole states); SEER-aggregated race categories based on self-report included “Black”, “White”, “Hispanic (all races)”, and “Asian/Pacific Islander”, which we subsequently referred to as non-Hispanic White, non-Hispanic Asian, non-Hispanic Black, and Hispanic. Histopathological features available in SEER included histological subtypes of the index DCIS and subsequent breast events, tumor grade and size, breast quadrants, estrogen receptor (ER), progesterone receptor (PR), surgery type of the index DCIS site, and radiotherapy receipt. Human epidermal growth factor receptor 2 (HER2) was available since 2009. The following specific histologic subtype ICDO-3 (International Classification of Diseases for Oncology, Third Edition) codes were identified as DCIS: 8050, 8201, 8230, 8500, 8501, 8503, 8504, 8507, 8509, and 8523.¹⁹ IBC was identified with ICD (C50) diagnosis codes (Supplemental Table I).

Outcome Measures

The main outcome of interest was subsequent breast events (DCIS or IBC) diagnosed at least six months after the index DCIS, allowing sufficient time for the outcome to occur.²⁰ Specifically, laterality-associated risks of subsequent DCIS or IBC overall and by ER subtypes (ER+ or PR+ overexpressing and ER− and PR− expressing tumors) were assessed by race and ethnicity. The laterality of the original DCIS was used to classify subsequent breast events into ipsilateral and contralateral events. Ipsilateral subsequent events occurred in the same breast as the original DCIS, while contralateral subsequent events occurred in the opposite breast from the original DCIS. Women who underwent mastectomy (bilateral or unilateral) or had unknown information about surgical procedure for the index DCIS were excluded (n = 31,784) from the laterality analysis. Except, women who did not undergo the removal of the unaffected contralateral breast and were not removed from the assessment of contralateral events. The risk of subsequent DCIS or IBC was also evaluated independent of laterality (non-competing model). The risk of subsequent IBC alone following DCIS was assessed as part of a sub-analysis. As with the main analysis, IBC laterality-associated risks overall and by ER tumors were assessed by race and ethnicity. The risks of subsequent invasive breast cancers based on ER, PR, and HER2 (available only after 2009) status were assessed as additional analysis. Women in this cohort were followed from at least 6 months after the index DCIS diagnosis until either the first occurrence of a subsequent breast event (DCIS or IBC), death, loss to follow-up, or the study endpoint of December 31, 2022 (follow-up time was calculated by SEER).

Statistical Analysis

Age and age-standardized baseline differences were compared with the Pearson chi-square (χ²) for categorical patient characteristics. To assess the risk of subsequent DCIS or IBC by laterality of the original DCIS, the Fine and Gray proportional hazards model was used to estimate adjusted subdistribution hazard ratios (sHRs) and 95%CIs (confidence intervals)—accounting for competing breast outcomes from ipsilateral and contralateral breast,²¹ reflecting current clinical practice.²² Clinically, censoring is informative, given that subsequent breast events are a result of the underlying DCIS diagnosis and are not independent of each other. Therefore, we modeled rather than censored competing (contralateral) breast outcomes. To assess the risk of subsequent DCIS or IBC after DCIS, overall and by ER− and PR− (ER−PR−) and ER+ or PR+ (ER+/PR+), an adjusted Cox proportional hazards model was used to estimate hazard ratios (HRs) and 95%CIs by race and ethnicity. All models were adjusted for the available demographic and index DCIS clinical characteristics on the relationship between race and subsequent breast events after the index DCIS. The proportionality assumption of the hazard models was assessed with the interaction effect of time with each covariate in the model and with Schoenfeld-type residual plots for the Cox²³ and Fine and Gray²⁴ models.

The Cochran’s heterogeneity test was used to assess differences in HRs and sHRs between the tumor subtypes for each race group (the differences in estimates for ER−PR− and ER+/PR+ overall and by laterality of the original DCIS). All statistical analyses were based on α < 0.05 (2-sided). Statistical analyses were performed in SAS 9.4 (RRID: SCR_008567, Institute Inc., Cary, NC). This research was prepared in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.²⁵ The use of de-identified and publicly available SEER data allows this research to be exempt from the Washington University in Saint Louis Institutional Review Board (IRB) and the patient-informed consent requirement to be waived.

Results

Of the 198,827 eligible women with DCIS in SEER between 2000–2022, 65.8% were White, 11.9% were Asian, 11.4% were Hispanic (all races), and 10.9% were Black (Table I). Overall, the median duration of follow-up was 113 months (range: 6–269). By race and ethnicity, the median duration of follow-up was 106 months for Hispanic women, 109 for Black women, 111 for Asian women, and 124 for White women. All age-standardized clinical characteristics, including year at DCIS diagnosis and follow-up time, differed significantly between race groups (Table I). Asian and Hispanic women were diagnosed with their initial DCIS (baseline) at a younger age (Table I).

TABLE I.

Age and age-standardized demographic and clinical characteristics of 198,827 women with DCIS in SEER (2000–2022) ^a by race and ethnicity

	White	Asian	Hispanic	Black
	130,761	23,704	22,627	21,735	P-value d
Age at DCIS, (y): b μ±sd	60±13	57±21	56±64	59±01
20 – 49	(21.8)	(29.7)	(31.1)	(23.4)
50 – 59	(27.6)	(29.3)	(30.0)	(28.8)
60 – 69	(27.0)	(24.7)	(23.6)	(28.2)
≥70	(23.6)	(16.3)	(15.3)	(19.6)	<.0001
Year at DCIS
2000 – 2007	(30.1)	(20.7)	(20.1)	(22.4)
2008 – 2015	(34.0)	(30.7)	(30.5)	(31.8)
2016 – 2022	(35.9)	(48.6)	(49.3)	(45.8)	<.0001
Maximum-survival time, (y): μ±sd	10±4	9±3	8±9	9±1
0.5 – 6	(33.4)	(42.4)	(45.3)	(42.1)
7 – 12	(26.3)	(26.4)	(26.1)	(27.6)
13 – 18	(26.0)	(21.0)	(20.2)	(21.2)
19 – 22	(14.3)	(10.2)	(8.4)	(9.1)	<.0001
DCIS hormone receptor status
ER−PR−	(9.5)	(9.6)	(9.1)	(8.4)
ER+/PR+	(65.6)	(71.3)	(70.4)	(73.1)	<.0001
Unknown or borderline c	(24.9)	(19.1)	(20.4)	(18.5)
DCIS histologic subtype
Intraductal, NOS	(72.8)	(73.9)	(75.2)	(74.4)
Comedo	(8.6)	(7.5)	(7.3)	(7.3)
Cribriform	(7.9)	(8.1)	(7.7)	(7.1)
Solid	(5.8)	(5.1)	(4.6)	(4.4)
Papillary	(4.9)	(5.4)	(5.2)	(6.5)	<.0001
DCIS tumor size, cm
<2.0	(58.6)	(60.4)	(58.7)	(55.1)
2.0 – 4.9	(13.0)	(19.3)	(15.4)	(14.4)
≥5.0	(5.1)	(5.8)	(6.3)	(7.3)	<.0001
Unknown or missing	(23.3)	(14.5)	(19.6)	(23.2)
DCIS tumor grade
Well-differentiated (I)	(11.6)	(12.8)	(12.6)	(12.9)
Moderately differentiated (II)	(35.3)	(40.4)	(38.9)	(37.8)
Poorly differentiated (III)	(32.6)	(30.8)	(29.7)	(31.1)
Undifferentiated (IV)	(7.4)	(6.5)	(6.5)	(4.0)	<.0001
Unknown or missing	(13.1)	(9.7)	(12.2)	(14.2)
DCIS treatment modality
No surgery	(2.1)	(2.9)	(3.3)	(3.7)
BCS only	(34.2)	(29.9)	(34.0)	(31.4)
BCS and radiotherapy	(46.3)	(44.8)	(44.6)	(44.5)
Mastectomy	(16.3)	(21.2)	(16.7)	(18.9)	<.0001
Unknown or missing	(1.1)	(1.2)	(1.4)	(2.6)

Abbreviations: DCIS, ductal carcinoma in situ; SEER, Surveillance, Epidemiology, and End Results; ER, estrogen-receptor; PR, progesterone-receptor; BCS, breast-conserving (preserving) surgery.

^aAge-standardized % reported (^b estimates for age at DCIS diagnosis not standardized).

^cBorderline ER/PR values are defined by SEER as “Borderline; undetermined whether positive or negative”.

^dχ² p-value on the comparison between Non-Hispanic White (White), Non-Hispanic Asian (Asian), Non-Hispanic Black (Black), and Hispanic (all races) women with DCIS.

A total of 16,148 women had subsequent breast events during the mean of 9.94 (±6.05) years of follow-up. Among these events, 4,150 (25.7%) were DCIS and 11,998 (74.3%) were IBC. Overall, compared with White women, Black women had an elevated risk of subsequent breast events (HR = 1.52, 95%CI, 1.44, 1.59), as did Asian women (HR = 1.18, 95% CI 1.12, 1.24) and Hispanic women (HR 1.09, 95%CI 1.03, 1.15, Table II). When assessed by initial DCIS tumor subtypes, the magnitude of risks for subsequent breast events differed across race groups; with the strongest risk observed for Black women (HR = 1.79, 95%CI 1.58, 2.04 for ER−PR− and HR = 1.44, 95%CI 1.36, 1.52 for ER+/PR+, p _het < 0.0001; Table II, Supplemental Figure I shows number of the overall breast events overtime by race).

TABLE II.

Adjusted hazard ratios (HR) for development of subsequent DCIS and IBC and IBC alone overall and by receptor status among 198,827 women with DCIS diagnosis in SEER (2000–2022), presented by self-reported race and ethnicity

	Overall			ER+ or PR+			ER− & PR−
	Cases No.	HR a	(95%CI)	Cases No.	HR a	(95%CI)	Cases No.	HR a	(95%CI)
DCIS and IBC
White	10,554		1.00	8,269		1.00	1,388		1.00
Asian	1,965	1.18	(1.12, 1.24)	1,500	1.09	(1.02, 1.16)	283	1.33	(1.15, 1.53)
Hispanic	1,587	1.09	(1.03, 1.15) c	1,198	1.01	(0.95, 1.08)	230	1.20	(1.04, 1.39)
Black	2,042	1.52	(1.44, 1.59)	1,557	1.44	(1.36, 1.52)	319	1.79	(1.58, 2.04)
Total	16,148			P d het <0.0001
IBC alone
White	7,947		1.00	6,446		1.00	1,092		1.00
Asian	1,336	1.02	b (0.96, 1.09)	1,040	0.98	(0.91, 1.06)	212	1.28	(1.09, 1.51)
Hispanic	1,198	1.05	b (1.00, 1.12)	925	1.00	(0.93, 1.07)	192	1.28	(1.09, 1.50)
Black	1,517	1.44	b (1.36, 1.52)	1,138	1.36	(1.28, 1.46)	279	1.96	(1.70, 2.25)
Total	11,894			P d het <0.0001

Abbreviations: DCIS, ductal carcinoma in-situ; IBC, invasive breast cancer; SEER, Surveillance, Epidemiology, and End Results; CI, confidence interval; ER, estrogen-receptor; PR, progesterone-receptor.

^aModels were adjusted for the SEER registry, age at initial DCIS, year at initial DCIS, histologic subtype of initial DCIS, tumor grade and size of initial DCIS, hormone receptor status of initial DCIS, and index DCIS treatment modality.

^bEstimates reported from Cox model stratified on the covariates in violation of proportionality hazards assumption.

^cChance in inference with complete case analysis for the overall DCIS/IBC among Hispanic women (HR = 1.06, 95%CI, 0.98, 1.15).

^dP-values were calculated using Cochran’s heterogeneity test to evaluate differences in the risk estimates between tumor subtypes.

Note: There were 662 (5.5%) IBC ER and 746 (6.2%) IBC PR missing outcome values out of the total number of subsequent IBC cases (n=11,894) not included in the analysis. Similarly, there were 681 DCIS ER (16.4%) and 1,077 DCIS PR (26.0%) missing outcome values not included in the analysis out of the total subsequent DCIS cases (n = 4,150).

For the main analysis of DCIS or IBC laterality of the original DCIS (n = 14,712), when compared with White women, Black women had an elevated risk in each breast (sHR = 1.59, 95%CI 1.47, 1.71 in the ipsilateral and 1.14, 95%CI 1.05, 1.23 in the contralateral breast, Table III). This elevated risk was observed overall and by initial DCIS receptor subtype. The observed subsequent DCIS or IBC tumor subtype associated risks were stronger among Black women for ER−PR− than for ER+/PR+ (sHR = 1.82, 95%CI 1.53, 2.17 for ER−PR− and 1.51, 95%CI 1.39, 1.65 for ER+/PR+, p _het = 0.0001 in the ipsilateral breast and sHR = 1.52 for ER−PR− and 1.11 for ER+/PR+ in the contralateral breast; Table III). Subtype associated risks were stronger among Black women for subsequent triple-negative (HR−/HER2−) breast cancers (HR = 2.01, 95% CI, 1.63–2.47), and Asian, Hispanic, and Black women with DCIS had an elevated risk of subsequent HR−/HER2+ breast cancers (Supplemental Table II). The risk estimates of women who underwent surgery in the sensitivity analysis were no longer significant for Hispanic women for DCIS/IBC overall or by tumor subtype in the ipsilateral breast.

TABLE III.

Adjusted subdistributional hazard ratios (sHR) for development of subsequent DCIS and IBC and IBC alone overall and by receptor status among 167,043 women with DCIS diagnosis in SEER (2000–2022), presented by laterality across self-reported race/ethnicity

	Overall			ER+ or PR+			ER− & PR−
	No.	sHR b	(95%CI)	No.	sHRb	(95%CI)	No.	sHR b	(95%CI)
DCIS and IBC
Ipsilateral a
White	4,695		1.00	3,493		1.00	780		1.00
Asian	855	1.10	c (1.02, 1.19)	625	1.02	(0.92, 1.12)	146	1.30	c (1.07, 1.57)
Hispanic	780	1.11	c (1.02, 1.21)	576	1.12	(1.02, 1.23) d	134	1.21	c (1.01, 1.48)
Black	1,000	1.59	c (1.47, 1.71)	741	1.51	(1.39, 1.65)	183	1.82	c (1.53, 2.17)
				P e het < 0.0001
Contralateral a
White	4,910		1.00	3,950		1.00	619		1.00
Asian	930	1.07	(1.00, 1.16)	726	1.05	(0.96, 1.15)	136	1.33	(1.08, 1.63) d
Hispanic	653	0.92	(0.84, 1.00)	506	0.88	(0.80, 0.97) d	88	1.02	(0.81, 1.30)
Black	822	1.14	(1.05, 1.23)	637	1.11	(1.01, 1.21)	132	1.52	(1.24, 1.86)
				P e het < 0.0001
IBC alone
Ipsilateral a
White	3,621		1.00	2,801		1.00	604		1.00
Asian	672	1.12	(1.02, 1.23)	505	1.07	(0.97, 1.20)	114	1.25	(1.01, 1.55)
Hispanic	627	1.16	(1.06, 1.27)	473	1.14	(1.03, 1.27)	114	1.30	(1.05, 1.60)
Black	787	1.53	(1.41, 1.67)	579	1.49	(1.35, 1.64)	154	1.82	(1.50, 2.20)
				P e het < 0.0001
Contralateral a
White	3,561		1.00	3,024		1.00	402		1.00
Asian	534	0.83	c (0.76, 0.92)	437	0.82	(0.74, 0.92) d	78	1.15	(0.88, 1.50)
Hispanic	449	0.88	c (0.79, 0.98)	369	0.85	(0.76, 0.95) d	56	1.02	(0.76, 1.37)
Black	556	1.16	c (1.05, 1.27)	430	1.00	(0.90, 1.12)	99	1.68	(1.32, 2.13)
				P e het < 0.0001

Abbreviations: DCIS, ductal carcinoma in-situ; IBC, invasive breast cancer; SEER, Surveillance, Epidemiology, and End Results; CI, confidence interval; ER, estrogen-receptor; PR, progesterone-receptor; het, heterogeneity.

^aWomen who had mastectomy (except for those who did not undergo removal of unaffected contralateral breast as indicated by primary site codes 41, 51, and 71) or unknown information of surgical procedure for the initial DCIS were excluded from the laterality analysis. Subsequent breast events that occur in the same breast as the original DCIS are referred to as ipsilateral and those that occur in the opposite breast from the original DCIS as contralateral.

^bModels were adjusted for the SEER registry, age at initial DCIS diagnosis, year at initial DCIS diagnosis, histologic subtype of initial DCIS, tumor grade and size of initial DCIS, hormone receptor status of initial DCIS, and initial DCIS treatment modality.

^cEstimates reported from Cox model stratified on the covariates in violation of proportionality hazards assumption.

^dChance in inference with complete case analysis were observed among Hispanic women for DCIS/IBC ER+ or PR+ (sHR = 1.14, 95%CI, 0.99, 1.31 in the ipsilateral and 0.90, 95%CI 0.78, 1.02 in the contralateral breast); among Asian women for DCIS/IBC ER−PR− (sHR = 1.17, 0.87, 1.55 in the contralateral breast); the overall IBC ER+/PR+ in the contralateral breast for Hispanic (sHR = 0.86, 0.74, 1.00) and Asian women (sHR = 0.87, 0.74, 1.02); among Asian women for IBC ER+/PR+ in the contralateral breast (sHR = 0.90, 95%CI 0.77, 1.04).

^eP-values were calculated using Cochran’s heterogeneity test to evaluate differences in the risk estimates between tumor subtypes.

We additionally considered IBC risks alone in women following the index DCIS. The results were largely unchanged. Overall, Black women also had an elevated risk of subsequent IBC after initial DCIS regardless of tumor subtype (Table III). In the sensitivity analysis restricted to women who underwent surgery, the risk estimates for ipsilateral IBC were attenuated among Asian women but remained significant for Hispanic women. Analysis by laterality for subsequent IBC (Table III) again showed results largely consistent with the combined breast event presentation.

Discussion

We evaluated the contemporary risk of subsequent breast events (DCIS or IBC) among women with unilateral primary DCIS diagnosed and reported in SEER from 2000–2022, by race and ethnicity. In this extended follow-up, Black women continue to have an elevated risk of subsequent breast events, irrespective of tumor subtype or laterality. Notably, Black women continue to have an elevated risk of subsequent IBC aggressive tumors (ER−PR−)—aligning with a recent molecular profile of DCIS that reported biological differences related to cancer initiation, recurrence, and invasive progression in Black compared with White women.²⁶

Compared with our prior SEER findings among women with DCIS diagnosed before 2015,^7,8 with an inclusion of more recent breast events, we confirmed the overall increase in IBC risk for ER−PR− for all women. Our current findings also confirmed Black women have an increased IBC risk in both the ipsilateral and contralateral breast, irrespective of tumor subtype. Molecular mechanisms driving ipsilateral breast events differ between Black and White women.²⁶ We also observed that the proportion of both IBC subtypes following DCIS was higher for Black than for White, Asian, or Hispanic women. It is possible that this disparity may reflect a higher baseline risk of breast cancer in Black women, yet prior population-based data, a combination of SEER and National Program of Cancer Registries²⁷ show that Black women have similar level of risk of triple-negative breast cancer, at similar levels to those observed in our DCIS cohort. This suggests that other contributors, such as genetics, socioeconomic and behavioral influences play a role in determining whether and how DCIS progresses. Nevertheless, the ongoing rise in more aggressive breast tumors in Black women, in both breasts underscores the need for continued study and early detection improvement in this high-risk group. The higher risk of the triple-negative subtype of IBC in Black compared with White women,^28,29 also contributes to worsened prognosis (vs ER+ tumors) and is the main mortality driver between Black and White women, accounting for 56% of the difference in deaths.³⁰ This disparity is further exacerbated by the lack of currently available targeted therapies for these cancers. The mortality rate by ER+ in Black women is also higher than in White women,³¹ further highlighting the multifactorial nature of racial disparities in breast cancer outcomes.

Hispanic and Asian women have an increased risk of aggressive or ER−PR− tumors in the ipsilateral breast. It is unclear whether the observed higher risk in the ipsilateral breast of Hispanic women in this cohort is partly due to untreated DCIS cases, as seen in a recent SEER-based analysis among Asian women,³² if it is attributed to delays in timely follow-up/diagnosis or influenced by genetic factors. In this cohort, Hispanic women were diagnosed with DCIS at a younger age, which is noteworthy given that younger age at diagnosis is associated with an increased risk of IBC and even death.³³ In the sensitivity analysis restricted to women who underwent surgery, the elevated risk of ipsilateral DCIS/IBC events among Hispanic women was attenuated, suggesting that the increased risk in this group of women may in part be attributed to the lack of surgery. However, Hispanic women continued to demonstrate a significantly elevated risk of ipsilateral IBC alone, in contrast to Asian women, whose risk was not elevated for the assessment of IBC alone. Additionally, as expected, women who received radiation therapy had lower rates of ipsilateral breast events. These rates did not differ significantly across racial groups, suggesting that receipt of radiation therapy may mitigate racial risk differences in local recurrence. Still, we acknowledge that SEER only captures first course of treatment, which may affect the risk of subsequent ipsilateral breast events and limits our ability to fully assess racial disparities in these outcomes.

In this extended follow-up, unlike in our prior findings, Asian women had an elevated risk for ER−PR− IBC in the ipsilateral but not in the contralateral breast for ER−PR− tumors.⁸ A recent SEER study reported lower overall incidence rates of IBC and IBC for ER−PR− tumors among Asian women in all age groups,³⁴ but, these population-level incidence rates reflect general trends and do not capture specific risk dynamics following DCIS. Additionally, delays in timely follow-up after an abnormal mammogram reported among Asian women³⁵ could contribute to late-stage detection or underestimation of true risk, particularly for aggressive ER−PR− tumors. Risk estimates among the subgroups of Asian women (i.e. Japanese, Chinese, Korean) for developing subsequent breast cancers differ,³² thus it is likely that these aggregated findings are masking important associations that warrant further exploration. Asian women in this study were more likely to have an increased risk of a second DCIS in the contralateral breast, both overall (HR = 1.61) and for either receptor subtype (HR = 1.55 for ER+/PR+ and 2.01 for ER−PR−), but the number of cases was small.

NCCN recommended post-DCIS surveillance includes annual imaging and clinical exams every 6 months for 5 years, then annually.³⁶ While lower mammogram screening uptake has been reported among minority women in general,³⁴ data on post-DCIS care remain limited but crucial for understanding racial disparities in DCIS outcomes. Lower surveillance could not account for excess incidence of invasive breast cancer, however one study of women diagnosed with DCIS between 2008 and 2014 observed that Black and Hispanic women had lower surveillance uptake compared to White women.³⁷ DCIS is generally detected by mammogram, and sensitivity of detecting breast events is lower for women with dense (62–68% detection sensitivity) than women with fatty breasts (86–89%). For the majority of White women, mammogram sensitivity for detecting DCIS was reported between 75% and 85%.³⁸ Breast density varies by race and ethnicity, yet data are limited on its impact on DCIS detection.^39,40

Several limitations of our study should be considered. Missing data on tumor characteristics that were adjusted for in the analysis, include the initial DCIS tumor size (21.9%), DCIS grade (12.7%), DCIS ER status (23%), and DCIS PR status (30%). The total number of missing values for each tumor characteristic by race is presented in Supplemental Table III. These tumor characteristics of the initial DCIS cases are potentially predictive of subsequent breast events, and may not be missing at random, imputing them could introduce bias. Instead, a missingness indicator⁴¹ was used in the analysis to retain all available cases. A complete case analysis was also conducted as a sensitivity analysis, and the overall inference did not change meaningfully. Where changes in inference were observed, they are primarily due to exclusion of women with the outcome who had missing the tumor characteristics in the complete case analysis. Invasive breast cancer is associated with adiposity,⁴² hormone replacement therapy,⁴³ or family history of breast cancer.⁴⁴ SEER did not have this information for us to assess whether the direction or magnitude of the observed risks are modified by appropriate adjustments for these factors associated with breast cancer. Specifically, the lack of adjustment in our models for hormone therapy may particularly influence the estimated risk of subsequent receptor-positive cancers. Last, we may have limited power to detect an association for certain race- and laterality-stratified ERPR tumor groups, especially ER−PR− tumors, which constitute 10–25% of all breast cancers.^45,46

The major strength of this study is the availability of a nationally representative contemporary sample of all women with DCIS diagnosed and reported in the U.S. SEER registries from 2000–2022, with an average of ~10 years of follow-up. Thus, the large sample size preserves the generalizability of risk estimates of subsequent breast outcomes for women of different races. SEER is the only comprehensive source of population-based cancer incidence data in the U.S.⁴⁷ Risk of ipsilateral subsequent breast events is often a primary focus of studies,^4,11 however, subsequent breast events in the contralateral breast after DCIS remain important²⁰ and are clinically relevant in counseling women about their treatment decisions. Clinically, evaluating competing breast events depends on treatment to the same (ipsilateral) breast, which can affect the risk in the opposite (contralateral) breast and help avoid overestimating the risk of future breast events.⁴⁸

In conclusion, the observed heterogeneity of subsequent breast events by tumor subtype continues to reveal insights into cancer biology across self-reported races among women with DCIS—adding to the complexity of racial disparity in breast cancer outcomes, such as tumor aggressiveness, stage at diagnosis or survival rates. A comprehensive understanding of screening patterns and adherence to post-DCIS follow-up care can help in part explain the observed race-associated differences in second breast tumors among women with DCIS. Emerging evidence on different pathways among Black women and elevated aggressive IBC is consistent with overall IBC patterns among these women.²⁶ Evidence of risk in both breasts after DCIS supports additional research to understand how to identify these women with high-risk to provide appropriate therapy and surveillance.

Data Sharing Statement:

Data used for this study are publicly available and can be accessed through https://seer.cancer.gov/.