Trends in breast cancer death rates from ductal carcinoma in situ versus invasive cancer in the United States

Abstract

Background

Breast ductal carcinoma in situ (DCIS), considered a non-obligate precursor to invasive breast cancer, has been widely studied; however, trends in death rate from DCIS remain undocumented.

Methods

To examine the trend in death rate from DCIS among women (age at death, 40 + years) in the US, we quantified annual percent changes (APCs) and average APCs in age-standardized death rates using the Surveillance, Epidemiology, and End Results Program (SEER) 8 incidence-based mortality database and Joinpoint regression. The trend was quantified from 2000 to 2021, capturing deaths among women diagnosed between 1975 and 2021 and allowing for a 25-year follow-back period before each death year. Additionally, we employed the SEER 12 to assess disparities in death rates from DCIS by race and ethnicity during 2017–2021, based on diagnoses from 1992–2021. Trends in death rates from invasive breast cancer were analysed similarly.

Results

Death rate from DCIS increased by 2.43% annually (95% confidence interval [CI] = 1.34–3.54) from 2000–2015, and then plateaued, leading to an overall statistically non-significant increase from 2000–2021 (1.01% per year; 95%CI = -0.20–2.24). This trend was in stark contrast with an average of 1.61% annual decrease in death rates for invasive breast cancer (95%CI = -1.95 to -1.27). From 2016–2021, death rate from DCIS was 87% higher among non-Hispanic Black women than non-Hispanic White women (2.98 vs 1.60 per 100,000), a relatively larger disparity than the 39% estimated for death rate from invasive cancer (55.96 vs 40.28 per 100,000).

Conclusions

Distinct epidemiologic patterns in death rates between DCIS and invasive breast cancer highlight important implications for understanding the natural history and management of these malignancies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13058-025-02068-9.

Background

Breast ductal carcinoma in situ (DCIS), considered a non-obligate precursor to invasive breast cancer, has been widely studied for its epidemiology, clinicopathologic features, treatment, and outcomes. In the United States, the incidence rate of DCIS surged from the early 1980s, coinciding with the widespread adoption of mammography screening, and stabilized at approximately 40 per 100,000 women (age at diagnosis, 20 + years) by around 2010 [1, 2]. Currently, an estimated 56,000 women are diagnosed with DCIS each year [1], accounting for 1 in 5 breast cancers diagnosed via mammography screening [3]. Trends in death rates from DCIS, however, remain undocumented in the United States and elsewhere, primarily because mortality databases—typically collected by national civil registration and vital statistics systems—lack information associated with cancer diagnosis. Leveraging population-based cancer registry data from the Surveillance, Epidemiology, and End Results Program (SEER) linked to national mortality data, this study examines the trend in death rate from DCIS in the United States and compares the pattern with that observed for invasive breast cancer.

Materials and methods

Data source and study participants

To examine the trend in death rate from DCIS among women, we used the SEER 8 incidence-based mortality database (Atlanta [Metropolitan], Connecticut, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, and Utah), covering 8.3% of the U.S. population [4]. The SEER 8 incidence-based mortality database includes information on cases diagnosed since 1975, with their records linked to death certificate data. This historic database allows for secular trend analyses of cancer death rates by various tumor characteristics over the longest period in the United States. To enable assessment by race and Hispanic ethnicity, we also used the SEER 12 incidence-based mortality database that includes cases diagnosed since 1992 and race and ethnicity information [5]. Expanding on SEER 8, SEER 12 adds Alaska, Los Angeles, Rural Georgia, and San Jose-Monterey, covering 12.2% of the U.S. population.

From both SEER 8 and SEER12, we identified women diagnosed with DCIS as their primary incident cancer and who died from breast cancer (age at death, 40 + years). DCIS was defined per the International Classification of Disease for Oncology, 3rd Edition (ICD-O-3) (topography, C500-C509; histology 8201, 8230, 8500–8507, 8523; behavior, 2) [6] and the cause of death was derived from death certificates in the mortality database from the National Center for Health Statistics. DCIS-related deaths included breast cancer deaths among women diagnosed with DCIS as either first or higher-order primary cancer (e.g., second primary DCIS diagnosis following first primary lung cancer). To account for a favorable prognosis of DCIS and minimize differential influence of follow-back time across death years, we considered a 25-year follow-back (i.e., burn-in) period of incidence before each death year (incidence, 1975–2021 and mortality, 2000–2021 for SEER 8; incidence, 1992–2021 and mortality, 2017–2021 for SEER 12) [7]. Using a shorter burn-in period, such as 20 years, allowed for a longer trend analysis of death rates (starting from 1995 instead of 2000 for SEER 8), but resulted in fewer deaths captured (potential underestimation; Supplementary Fig. 1). In contrast, extending the burn-in period to 30 years did not substantially increase the number of deaths, yet it shortened the timeframe for trend analysis (starting from 2005). Therefore, a 25-year burn-in period was selected as a balanced approach (Supplementary Fig. 1). Similarly, women diagnosed with primary incident invasive breast cancer (topography, C500-C509; histology, all histologies excluding 9050–9055, 9140, 9590–9993; behavior, malignant, 4) and who died from breast cancer (age at death, 40 + years) were identified, incorporating the same 25-year follow-back period. Multiple primary invasive breast cancers were included, except for higher-order invasive breast cancers that were preceded by a DCIS diagnosis. We excluded autopsy or death certificate-only cases from the analysis.

Statistical analysis

Based on the SEER 8, long-term trends in age-standardized death rates were quantified between 2000 and 2021 for DCIS and invasive breast cancer, separately, using Joinpoint regression. Joinpoint uses a segmented line regression model to fit a series of straight lines on a logarithmic scale and estimates annual percent changes (APCs) and corresponding 95% confidence intervals (CIs) (Supplementary Method). Average APCs were also calculated to summarize the trend throughout the study period. Using the SEER 12, racial and ethnic differences in death rates were measured from 2017 to 2021 by calculating rate ratios with corresponding CIs determined by the Tiwari method [8]. Calculations of annual age-standardized death rates and rate ratios were performed using SEER*Stat 8.4.4. A two-sided P-value of < 0.05 was considered statistically significant.

Results

From 2000–2021, there were 2,175 breast cancer deaths among women diagnosed with DCIS in 8 SEER areas (1.53 per 100,000), substantially fewer than among women diagnosed with invasive breast cancer (61,386 deaths, 43.83 per 100,000). Figure 1  shows trends in death rates from DCIS and those from invasive breast cancer (not preceded by a history of DCIS diagnosis), separately. From 2000 to 2015, death rates from DCIS increased by 2.43% annually (95%CI = 1.34–3.54), and then decreased by 2.46% per year until 2021, although the decrease was not statistically significant (95%CI = -5.98 to 1.19). These trends led to a statistically non-significant annual increase of 1.01% from 2000 to 2021 (95%CI = -0.20 to 2.24). In contrast, death rate from invasive cancer continuously decreased by an average of 1.61% annually (95%CI = -1.95 to -1.27) (Fig. 1).

Fig. 1.

Trends in death rates from breast ductal carcinoma in situ versus invasive breast cancer from 2000 to 2021 among women aged 40 and above in 8 SEER registries. Note: Abbreviation, APC, annual percent change. Trends were quantified using Joinpoint version 5.3.0, incorporating a 25-year burn-in period of incidence before each death year (Supplementary Method). Grid Search and weighted Bayesian Information Criterion methods were used for model fitting. Parametric method was used to determine confidence intervals for APCs

From 2017–2021, death rates from DCIS in 12 SEER registries differed by race and ethnicity (Table 1); non-Hispanic Black women had the highest death rate (2.98 per 100,000), 87% higher (95% CI = 1.52–2.27) than non-Hispanic White women (1.60 per 100,000), followed by Asian or Pacific Islander women (1.26 per 100,000) and Hispanic women (1.20 per 100,000). For invasive breast cancer, non-Hispanic Black women had a 39% higher rate (95%CI = 1.33–1.45) than non-Hispanic White women (55.96 vs 40.28 per 100,000), followed by Hispanic (32.66 per 100,000) and Asian or Pacific Islander (27.09 per 100,000) women.

Table 1.

Incidence-based mortality rates from breast ductal carcinoma in situ versus invasive breast cancer during 2017–2021 among women aged 40 and above by race and ethnicity in 12 SEER registries

	Ductal carcinoma in situ			Invasive breast cancer
Race and ethnicitya	Deaths, N	Death rate per 100,000 person-years (95% confidence interval)	Rate ratio (95% confidence interval)	Deaths, N	Death rate per 100,000 person-years (95% confidence interval)	Rate ratio (95% confidence interval)
Totalb	888	1.61 (1.50–1.72)	-	20,802	38.65 (38.11–39.19)	-
White	531	1.60 (1.46–1.74)	1 (reference)	12,787	40.28 (39.55–41.01)	1 (reference)
Black	132	2.98 (2.49–3.55)	1.87 (1.52–2.27)	2,536	55.96 (53.75–58.23)	1.39 (1.33–1.45)
Asian American or Pacific Islander	114	1.26 (1.04–1.52)	0.79 (0.64–0.97)	2,342	27.09 (25.98–28.23)	0.67 (0.64–0.70)
Hispanic	97	1.20 (0.97–1.47)	0.75 (0.60–0.94)	2,887	32.66 (31.45–33.90)	0.81 (0.78–0.85)

^aRace and ethnicity information in SEER are abstracted from medical records. Rates for White, Black, and Asian American or Pacific Islander women are exclusive of Hispanic ethnicity. American Indian or Alaska Native women were not analyzed separately due to limited numbers

^bIncludes American Indian or Alaska Native women and women of unknown race and ethnicity

Data were from the Surveillance, Epidemiology, and End Results (SEER) 12 registries (Alaska, Atlanta [Metropolitan], Connecticut, Hawaii, Iowa, Los Angeles, New Mexico, Rural Georgia, San Francisco-Oakland, San Jose-Monterey, Seattle-Puget Sound, and Utah), which cover 12.2% of the U.S. population. Incidence-based mortality from breast cancer was estimated among women diagnosed with ductal carcinoma in situ from 1997 to 2021 and within 25 years (burn-in period) before death. Cases identified from only autopsy records or death certificates were excluded (Supplementary Method)

Discussion

Based on population-based cancer registry data linked to mortality database, the death rate from DCIS has not decreased over the past 22 years through 2021, contrasting sharply with the approximately 2% annual decrease observed in death rates from invasive breast cancer. The absence of a declining death rate from DCIS may indicate a steady occurrence of fatal DCIS and suggest that advancements in DCIS treatment have not been sufficient to reduce mortality. The effectiveness of DCIS treatment has been evaluated solely on trials using invasive breast cancer or DCIS recurrence as endpoints, with evidence lacking demonstrating its impact on reducing breast cancer-specific mortality [9, 10]. To reduce breast cancer mortality following DCIS diagnosis, identifying factors associated with fatal DCIS remains essential to optimizing the clinical management of DCIS [11], while also continuing efforts to address concerns about the overtreatment of low-risk DCIS [12].

Although the absolute number of deaths and death rates were substantially lower for DCIS compared with invasive breast cancer, racial disparity was more pronounced for death rates from DCIS, especially between non-Hispanic Black women and non-Hispanic White women (rate ratio, 1.87 vs 1.39). The racial gap cannot be attributed to the difference in incidence rate of DCIS, as it was lower among Black than White women until the late 1990s, after which the rates became relatively comparable [2, 10, 13]. This finding aligns with previous population-based studies showing that Black women diagnosed with DCIS have a higher likelihood than White women of developing second DCIS or invasive breast cancer (relative ratios, 1.21 to 1.46) [14] and dying from breast cancer (hazard ratios, 2.55) with the associations being persistent even after adjusting for age, tumor characteristics, treatment receipt, and household income [10]. However, a study of Missouri Cancer Registry showed that Black women were more likely than White women to underutilize or delay radiation therapy after breast-conserving surgery [15], indicative of residual confounding in prior studies. Additionally, tissue-based studies identified distinct molecular signatures associated with self-reported race [16, 17], pointing to molecular mechanisms that may intersect with systematic and social determinants of health.

Limitations of this study are inherent to the use of historic SEER data with relatively lower coverage, underrepresentation of diverse populations and reliance on aggregated data. Although invasive breast cancer was intended to be classified as not preceded by DCIS, DCIS identification relied solely on diagnoses recorded in the cancer registry during the study period. As a result, women with DCIS diagnosed before the study period or with undiagnosed DCIS—possibly due to lack of screening—who were later diagnosed with invasive cancer may have been misclassified as having no prior DCIS. Cause of death information, derived from death certificates, is subject to misclassification and potential variation in accuracy between women with DCIS and those with invasive cancer is unknown. Additionally, it is unclear whether the 25-year follow-up is sufficient to capture all breast cancer deaths after DCIS, given the uncertainty of survival time. However, this follow-up period is probably adequate to characterize mortality trends from most lifetime deaths related to DCIS, if any, particularly among older women, who represent the majority of deaths (age at death, 50 + years; 94% of all deaths), given the presence of numerous competing risks in later life.

In conclusion, based on high-quality population-based incidence-based mortality data with a 25-year follow-back from each year of breast cancer death, the death rate from DCIS in the United States has not decreased from 2000 to 2021. The trend contrasts with a consistent decline in death rates from invasive breast cancer and highlights important implications for understanding the natural history and management of these malignancies.

Authors’ contributions

Sung: Conceptualization, Validation, Writing – original draft, Writing – review & editing Jiang: Conceptualization, Data curation, Formal analysis, Visualization, Writing – review & editing Jatoi: Conceptualization, Writing – review & editing Jemal: Conceptualization, Writing – review & editing.

Funding

This work was supported by the Intramural Research Department of the American Cancer Society (no grant number applies). The American Cancer Society had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The findings and conclusions in this article are those of the authors and do not necessarily represent the official positions of the American Cancer Society.

Data availability

The data are publicly available and can be obtained from the National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) Program (https://seer.cancer.gov/data/access.html).

Declarations

Ethical approval and consent to participate

The study used de-identified publicly available data, which is considered non-human participants research under the Common Rule (US federal regulation 45 CFR §46); thus, institutional review board approval and informed consent were not needed.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.