Socioeconomic status impacts tumor biology, treatment, and outcomes in over 200,000 patients with invasive lobular carcinoma of the breast: an analysis of the National Cancer Database

TL;DR

This study shows that socioeconomic status affects tumor biology, treatment, and survival in over 200,000 breast cancer patients with invasive lobular carcinoma.

Contribution

The first study to evaluate how different components of socioeconomic status influence invasive lobular carcinoma specifically.

Findings

Patients with lower socioeconomic status had larger tumors and higher-grade tumors.

Low SES patients had a 24% higher risk of death within 5 years compared to high SES patients.

Lower SES was associated with less endocrine therapy use in hormone receptor-positive tumors.

Abstract

While the impact of socioeconomic factors on breast cancer diagnosis, treatment, and outcomes are well-documented, few studies have focused on invasive lobular carcinoma (ILC), the second most common type of breast cancer. We evaluated the relationships between race and socioeconomic status (SES) with clinicopathological characteristics and outcomes in patients with stage I-III ILC using the National Cancer Database (NCDB).

We used the NCDB, a national oncology database, to evaluate insurance status, a composite measure of SES (education and income), clinicopathological characteristics, and outcomes in patients with stage I-III ILC. Clinicopathologic variables included tumor size, presence of lymphovascular invasion (LVI), and tumor receptor subtype (hormone receptor, HR), and tumor grade. Overall survival was analyzed with multivariable Cox proportional hazards models.

We identified 269,657 patients with stage I-III ILC. Patients in the Medicaid/no insurance group and those with lower SES had larger tumors, more positive lymph nodes, fewer HR+ tumors, and higher-grade tumors. Those in the low SES group had higher rates of chemotherapy use and, in those with HR+ tumors, lower rates of endocrine therapy use. In a multivariable model adjusting for SES, self-identified race/ethnicity, age, stage, receptor subtype, grade, treatment, and Charlson–Deyo score, patients with low SES had a 24% higher risk of death by 5 years compared to patients with high SES (HR 1.24, 95% CI 1.19–1.30, p < 0.001).

While our study confirms several known disparities in the presentation, outcomes, and treatment of breast cancer, this is the first evaluation to assess how different components of SES influence ILC specifically.

Full text

Introduction

Breast cancer is a leading cause of cancer-related deaths in women across the globe [1]. Invasive lobular carcinoma (ILC) is the second most common type of breast cancer, accounting for up to 15% of all breast cancer cases [2]. Compared to the more prevalent invasive ductal carcinoma (IDC), ILC is characteristically more hormonally driven and takes on a diffuse, infiltrative pattern of growth due to the lack of E-cadherin [2]. This contributes to its distinct appearance on imaging, response to systemic therapy, and pattern of recurrence, all of which distinguish ILC from IDC [3, 4]. Recent investigations have shown strong correlations between differences in socioeconomic status (SES) and variations in breast tumor features, treatment, and survival, indicating a potentially critical role of SES in breast cancer outcomes, but have not focused on differences by tumor histology [5–8].

Socioeconomic status (SES) is a composite measure of an individual’s economic or social position relative to others, typically including measures like education and income, and has well-established associations with various health outcomes, including cancer [9–14]. In a single institution analysis of patients with ILC, we previously established that lower SES, as measured by the Area Deprivation Index (ADI), is linked to larger tumors, increased rates of lymphovascular invasion (LVI), and higher grade tumors [15]. Additionally, we found lower SES to be associated with higher likelihood of mastectomy and lower receipt of endocrine therapy, highlighting discrepancies in treatment related to variations in SES. Other studies evaluating health disparities in patients with ILC have shown higher rates of diagnosis and worse outcomes for those with non-White race [16, 17]. To date, these investigations examining the relationship between socioeconomic factors and ILC remain limited.

While associations between SES and breast cancer in general have been well-established, these findings may not fully translate to ILC due to its distinct biological and clinical characteristics. To address this, we queried the National Cancer Database (NCDB) to assess the relationships between a composite measure of SES, using income and education, and the clinicopathological features, treatment, and outcomes of a large cohort of patients with stage I-III ILC. Additionally, we evaluated whether prior associations between race and outcomes in those with ILC would persist when considering composite measures of SES.

Methods

This was a retrospective cohort study using the 2010–2016 National Cancer Database (NCDB) participant user files (PUFs). The National Cancer Database is a comprehensive database containing data from over 1,500 sites across the United States, capturing approximately 70% of all newly diagnosed cancer cases in the United States [18]. We collected socioeconomic status, clinicopathologic characteristics, and survival outcomes for patients with stage I-III ILC treated at all NCDB-participating institutions. Patients with stage IV or metastatic disease were excluded.

Covariates were selected based on clinical relevance and their potential role as confounders in the relationship between SES and ILC presentation, treatment, and outcomes. Race and ethnicity data were collected and grouped as White-identifying, Black-identifying, East Asian-identifying (including Chinese, Japanese, and Korean), Spanish/Hispanic origin-identifying (as defined by the NCDB), or Other. Tumor receptor subtype was classified by estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor 2 (HER2) status, and grouped as HR+HER2−, HR−HER2− (triple negative), or HER2+. To capture a broader measure of SES, we created a composite SES score using annual income and education quartiles. For income, the NCDB grouped annual income into the following categories: < $40,227,$40,227–$50,353,$50,354–$63,332, and ≥$63,333. The NCDB’s education quartiles were defined by the following percentages of people over the age of 25 in the patient’s zip code without a high school degree: ≥ 17.6%, 10.9%–17.5%, 6.3%–10.8%, and < 6.3%. These quartile assignments (1, 2, 3, 4) were then added together to create a composite measure of SES, which was grouped as follows: low (2–3), mid-low (4–5), mid-high (6–7), and high (8) [19]. Insurance status, as a categorical variable, was collected and analyzed separately. Categories were private insurance, Medicare, and Medicaid/no insurance. Given the small proportion of uninsured patients (< 1%), and the expansion of Medicaid coverage to uninsured patients under the Breast and Cervical Cancer Prevention and Treatment Act, we combined the Medicaid and no insurance. We excluded the < 3% of subjects with “other government insurance” and unknown insurance. Comorbidities were assessed using the Charlson–Deyo Comorbidity Index, ranging from 0 to 6 with 0 representing no comorbid health conditions, 1 and 2 representing mild comorbid health conditions, and ≥ 3 representing moderate to severe comorbidities [20].

Statistical analysis

Data were analyzed in Stata 16.1. Hypothesis tests were two-sided, and the significance threshold was set to 0.05. Chi-squared tests were used to compare categorical variables and analysis of variance (ANOVA) to compare continuous variables across groups. We compared tumor characteristics (size, number of positive lymph nodes, receptor subtype, grade, and LVI) and treatment type (receipt of chemotherapy, endocrine therapy, and type of surgery) by insurance type and composite SES (based on income and education quartiles) in patients with stage I-III ILC. We also used logistic regression to evaluate the association between socioeconomic status and surgery type, and we used multinomial logistic regression to evaluate the association between socioeconomic status and receptor subtype while controlling for race/ethnicity.

We evaluated overall survival using a Cox proportional hazards model for insurance type and composite SES, controlling for self-identified race/ethnicity, age at diagnosis, stage, receptor subtype, tumor grade, receipt of chemotherapy, receipt of endocrine therapy, and Charlson–Deyo comorbidity score. We performed a sensitivity analysis adding distance from hospital and urban/metro vs. rural (2013 classification) to the model. Administrative censoring was applied at five years. We analyzed the unknown/missing category for variables that had > 5% missing data and used complete case analysis for all other variables.

Results

Study cohort

We identified 269,657 cases of patients with stage I-III ILC in the NCDB. Of these, 226,868 identified as White (84%), 20,948 as Black (7.8%), 11,848 as Spanish/Hispanic (4.4%), 1,817 as East Asian (0.7%), and 5,889 as other (2.2%). Overall, most tumors (92.4%) were HR+HER2−, consistent with prior studies of ILC. Most tumors were grade 2, and LVI was present in 15%. Among those with HR+ tumors (92.4%), 88% received endocrine therapy. Chemotherapy was utilized in 38% of the entire cohort. For surgical treatment, 51% underwent mastectomy and 49% underwent lumpectomy (see Table 1).Table 1ILC characteristics in overall study cohortVariableStudy cohort (N = 269,657)Age, years (mean ± SD)62.9 ± 12.4Self-identified race/ethnicity White84% Black7.8% East Asian < 1% Spanish/Hispanic4.4% Other2.2% [unknown/missing, n][2287]Income  < $40,22713$40,227–$50,35319$50,354–$63,33223$63,33345% [unknown/missing, n][3816]Education (no HS degree)  ≥ 17.6%15% 10.9%–17.5%22% 6.3%–10.8%29%  < 6.3%33% [unknown/missing, n][3383]Tumor size, mm (mean ± SD)^a^25.8 ± 35.3Positive lymph nodes, n (mean ± SD)^b^1.64 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 4.14Receptor Subtype HR+/HER2−92.4% HR−/HER2−1.8% HER2+ [unknown/missing, n]5.8%[117,577]Tumor grade 126% 262% 3 [unknown/missing, n]12%[26,968]Lymphovascular invasion present [not applicable, n] [unknown/missing, n]15%[6][21,569]Endocrine therapy* [unknown/missing, n]88%[3215]Chemotherapy [unknown/missing, n]38%[8181]Surgery Mastectomy51% Lumpectomy [unknown/missing, n]49%[1150]Data reported from complete case analyses. SD, standard deviation; HR, hormone receptor; HER2, human epidermal growth factor receptor 2; +  receptor status positive; − receptor status negative. *Endocrine therapy in HR+ patients only (n = 140,585)^a^n = 267,109^b^n = 257,993

Insurance type

Most patients had private insurance (52%), followed by Medicare (42%) and Medicaid/no insurance (5.8%). Patients with Medicare were older than patients with private insurance and Medicaid/no insurance (73.2 ± 8.0 years compared to 55.7 ± 9.4 years for private insurance and 54.7 ± 10.1 years for those with Medicaid/no insurance). Insurance differed significantly by self-identified race/ethnicity (p < 0.001), with a smaller proportion of Black-identifying and Spanish/Hispanic-identifying patients having private insurance (48% and 48%, respectively, versus 52% for White, 65% for East Asian, and 61% for Other). Those with Medicaid/no insurance had larger tumors, more positive lymph nodes, more triple negative tumors, more grade 3 tumors, and more LVI (Table 2). Receipt of chemotherapy and endocrine therapy differed significantly by insurance type (both p < 0.001). Those with Medicaid/no insurance received chemotherapy more often than those with private insurance or Medicare (53% versus 48% in private insurance and 23% in Medicare insurance groups). Patients with private insurance received endocrine therapy more commonly (84% versus 81% in Medicaid/no insurance group and 78% in Medicare group). There were no clinically significant differences in receipt of surgery by insurance status (99.7% vs. 99.7% vs. 99.5% for private, Medicare, and Medicaid/no insurance, respectively), however, the type of surgery received differed (p < 0.001). Patients with Medicare underwent lumpectomy more frequently (52% in Medicare insurance group versus 46% in private insurance group and 42% in Medicaid/no insurance group).Table 2ILC characteristics by insuranceCharacteristicInsurance (n = 262,463)Medicaid/no insurance (n = 15,303)Medicare (n = 110,831)Private insurance (n = 136,329)p-valueAge, years (mean ± SD)54.7 ± 10.173.2 ± 8.055.7 ± 9.4 < 0.001Tumor size, mm (mean ± SD)^a^30.0 ± 38.524.2 ± 26.925.0 ± 30.7 < 0.001Positive lymph Nodes, n (mean ± SD^)b^2.4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 5.01.5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 4.11.7 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 4.1 < 0.001Receptor subtype < 0.001 ER+/PR+HER2−49%46%46% ER+/PR−/HER2−5.7%8.1%5.0% HER2+4.3%3.1%3.3% Triple negative1.3%1.1% < 1% Unknown/missing40%42%45%Tumor grade < 0.001 121%25%23% 255%55%55% 3 Unknown/missing15%10%10%9.9%11%10%Lymphovascular invasion Not presentUnknown/missing [not applicable, n]18%68%15%[0]12%75%13%[5]14%73%13%[0] < 0.001Data reported from complete case analyses. Unknown/missing category analyzed if data missing for > 5%. SD, standard deviation; HR, hormone receptor; HER2, human epidermal growth factor receptor 2; +  receptor status positive; − receptor status negative^a^n = 260,089^b^n = 251,629

Composite socioeconomic status

When grouped by composite SES, 43,394 (16%) patients were categorized as having low SES, 65,472 (25%) had mid-low SES, 84,908 (32%) had mid-high SES, and 72,067 (27%) had high SES. Composite SES category differed significantly by self-identified race (p < 0.001), with Black-identifying patients having the smallest proportion of patients with high SES and the greatest proportion with low SES (8.9% with high SES versus 29% for White, 38% for East Asian, 12% for Spanish/Hispanic, and 30% for Other; 45% with low SES versus 13% for White, 9.5% for East Asian, 36% Spanish/Hispanic, and 12% for Other). Similar to patients in the Medicaid/no insurance group, those with lower SES had larger tumors, more positive lymph nodes, fewer HR positive tumors, more triple negative tumors, more grade 3 tumors, and more LVI (Table 3). While triple negative and HER2 + tumors were rare in this dataset, we found a significant association between receptor subtype and composite SES category (p < 0.001). Higher rates of triple negative ILC in those with lower composite SES (1.3% triple negative in low SES group and < 1% in high SES group). This association persisted when controlling for race/ethnicity, with patients with low SES more likely to have triple negative subtype. Compared to the high SES group, those with low SES had 1.35 times the odds of triple negative subtype (OR 1.35, 95% CI 1.20–1.53, p < 0.001). There were small but statistically significant differences in systemic therapy by composite SES category. Those in the low SES group had the highest proportion receiving chemotherapy and, for those with HR+ tumors, the lowest proportion receiving endocrine therapy (33% chemotherapy in low SES versus 30% for high SES; 88% endocrine therapy in the low SES group versus 89% for high SES when restricted to HR+ subtypes). There were no clinically significant differences in the receipt of surgery overall by SES (99.4% vs. 99.6% vs. 99.6% vs. 99.6% for low, mid-low, mid-high, and high SES, respectively). However, the odds of mastectomy increased by 7% as for each category decrease in SES (OR 1.07, 95% CI 1.06–1.08, p < 0.001: 50% had mastectomy in the high SES group, 51% in the mid-high group, 53% in the mid-low group, and 55% in the low SES group).Table 3ILC characteristics by composite SESComposite SES (n = 265,841)CharacteristicLow (n = 43,394)Mid-low (n = 65,472)Mid-high (n = 84,908)High (n = 72,067)p-valueTumor size, mm (mean ± SD)^a^27.3 ± 36.726.6 ± 36.125.4 ± 34.324.6 ± 34.8 < 0.001Positive lymph Nodes, n (mean ± SD)^b^1.9 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 4.41.8 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 4.31.6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 4.11.5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 3.9 < 0.001Receptor subtype < 0.001 ER+/PR+/HER2-45%45%46%47% ER+/PR−/HER2-6.5%6.3%6.4%6.3% HER2+3.6%3.4%3.2%3.0% Triple negative1.3%1.0% < 1% < 1% Unknown/missing44%44%43%43%Tumor grade < 0.001 123%24%23%24% 254%55%56%56% 3 Unknown/missing12%11%11%10%11%9.9%10%9.5%Lymphovascular invasion Not presentUnknown/missing [not applicable, n]14%70%15%[1][1]13%72%14%[3][3]13%73%13%[1][1]13%75%12%[1][1] < 0.001Data reported from complete case analyses. Unknown/missing category analyzed if data missing for > 5%. SD, standard deviation; HR, hormone receptor; HER2, human epidermal growth factor receptor 2; +  receptor status positive; − receptor status negative^a^n = 263,337^b^n = 254,341

Overall survival

In a multivariable model adjusted for insurance type, composite SES, age at diagnosis, stage, receptor subtype, tumor grade, receipt of chemotherapy, receipt of endocrine therapy, and Charlson–Deyo comorbidity score, Black-identifying participants had a 15% higher risk of death at 5 years after diagnosis compared to White-identifying participants (Table 4). In contrast, East Asian-identifying, Spanish/Hispanic-identifying, and Other participants had a 21%, 27%, and 28% lower risk of death at 5 years compared to White-identifying patients, respectively. Other factors associated with improved overall survival in this model included receipt of chemotherapy and endocrine therapy. Medicare or Medicaid/no insurance was associated with shorter overall survival compared to private insurance. Lower SES was associated with shorter overall survival compared to high SES, with the low SES group having 24% increased risk of death. Higher stage disease, ER+/PR-/HER2- or triple negative subtype compared to ER+/PR+/HER2− subtype, higher grade tumors, and higher Charlson-Deyo comorbidity score were also associated with shorter overall survival.Table 4. Multivariable cox proportional hazards model for death by five years with insurance status, composite socioeconomic status, self-identified race/ethnicity, age at diagnosis, stage, grade, receptor subtype, receipt of chemotherapy, receipt of endocrine therapy, and Charlson–Deyo comorbidity index in patients with ILCVariablesHazard ratio95% confidence intervalp-valuePrivate insurance (reference)Medicare insurance1.201.15–1.24 < 0.001Medicaid/no insuranceHigh SES (reference)1.751.65–1.86 < 0.001Low SES1.241.19–1.30 < 0.001Mid low SES1.231.19–1.28 < 0.001Mid high SES1.111.07–1.15 < 0.001White (reference)East Asian0.790.64–0.990.04Black1.151.10–1.21 < 0.001Spanish/Hispanic0.730.68–0.79 < 0.001Other0.720.63–0.82 < 0.001Age at Diagnosis (per year)1.051.05–1.05 < 0.001Stage 1 (reference)Stage 21.661.60–1.72 < 0.001Stage 34.384.22–4.55 < 0.001Grade 1(reference)Grade 21.141.10–1.18 < 0.001Grade 3Unknown/missing1.691.181.61–1.761.12–1.24 < 0.001 < 0.001ER+/PR+/HER2− (reference)ER+/PR−/HER2−1.431.35–1.51 < 0.001Triple negative1.681.53–1.85 < 0.001HER2+Unknown/missing1.111.111.01–1.211.07–1,140.02 < 0.001Receipt of Chemotherapy0.830.80–0.86 < 0.001Receipt of Endocrine Therapy0.490.47–0.50 < 0.001Charlson Deyo Score 0 (reference)Charlson Deyo Score 11.471.42–1.52 < 0.001Charlson Deyo Score 22.252.13–2.39 < 0.001Charlson Deyo Score \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ge$$\end{document} 33.072.80–3.36 < 0.001Model included insurance, composite SES, race/ethnicity, age, stage, receptor subtype, grade, receipt of chemotherapy, receipt of endocrine therapy and Charlson–Deyo comorbidity index score, ER estrogen receptor, PR progesterone receptor, HER2 human epidermal growth factor receptor 2; +  receptor status positive; − receptor status negative. Data available in n = 219,316

When adding distance from hospital and urban/metro versus rural classification to the above multivariable model, urban/metro versus rural classification was not significantly associated with overall survival (HR 1.01, 95% CI 0.91–1.13, p = 0.82), and all other associations reported from the main model remained almost identical in magnitude, direction, and significance.

Discussion

Our study demonstrates that known disparities in breast cancer treatment and outcomes extend to those with ILC, the second most common type of breast cancer. Although a majority of ILC tumors were HR+/HER2− and low to intermediate grade, we found that low SES was associated with more aggressive tumor features, including higher grade and increased LVI. These findings are consistent with our prior single-institution work where we found that greater area deprivation index was associated with larger ILC tumor size and greater LVI [15]. We also observed similar differences in treatment in both this national analysis and our institutional cohort, with lower SES being associated with higher mastectomy rates and less endocrine therapy use in those with HR+ disease. The consistency between our single-institution findings and this national cohort strengthens the evidence that socioeconomic factors impact ILC presentation and outcomes.

These findings are also in line with recent advancements in our understanding of ILC. Although ILC has historically been viewed as a homogenous tumor type, recent studies show heterogeneity in the molecular subtypes of ILC [21–23]. The presence of high grade and LVI in ILC is atypical, however we found that these factors were significantly more common in those with lower SES suggesting that tumor heterogeneity in ILC may be influenced by SES factors. The underlying mechanism driving these aggressive features in a typically low proliferative tumor type is unknown and remains an important area for future research.

Recent studies have shown that outcomes disparities by self-identified race appear to impact those with HR positive breast cancers in particular. Black-identifying patients with ILC experienced worse overall survival compared to other race groups, despite the generally favorable prognosis associated with HR+ tumors. This trend is supported by existing literature that has identified similarly worse survival and outcomes for Black-identifying patients with HR+ tumors [24–26]. Proposed explanations for these disparities include differences in tumor biology, treatment delays, and barriers to consistent healthcare [27, 28]. Additionally, the impact of an individual’s lived environment may have significant impact on tumor biology. Persistent social, political, and economic marginalization of Black-identifying women may contribute to higher comorbidities, changes in allostatic load, and poorer outcomes [29–31]. This concept, coined the weathering hypothesis, has been supported by several studies as a possible explanation for how environmental challenges may contribute to epigenetic changes in Black-identifying patients [32–34]. Investigators have shown that chronic stress from neighborhood disadvantage is associated with the upregulation of pro-inflammatory pathways, such as the Conserved Transcriptional Response to Adversity (CTRA) gene expression profile in leukocytes [35–38]. The CTRA response has been shown to be increased in both low-income and Black-identifying patients, offering a potential biologic mechanism through which structural inequities influence clinical outcomes [39–42].

In our analysis, the observation of high stage at presentation could be partially explained by limited access to screening mammography in those with lower SES [43]. Educational levels and health literacy, for example, may also impact patients’ understandings of breast cancer symptoms and screening guidelines, influencing how they engage with the healthcare system [44, 45]. Geographic barriers, such as residence in rural or medically underserved areas, can also limit access to timely care, especially when coupled with transportation challenges and caregiving or work responsibilities [46, 47]. Additionally, the use of hormone replacement therapy—which has been associated with the development of ILC—is more common in those with higher SES [48, 49]. In one analysis, those taking combined estrogen and progestin hormone replacement therapy for menopausal symptoms had significantly increased risk of developing ILC compared to non-users [50].

In summary, our analysis of the NCDB validates our previous single institution findings, showing that socioeconomic factors influence not only tumor biology but also treatment and outcomes in patients with ILC. Limitations include the high prevalence of missing data in the NCDB, and the lack of data on important factors such as surveillance, duration of endocrine therapy, and adherence to endocrine therapy [51]. Our study did not capture breast cancer-specific survival, limiting our ability to distinguish between deaths caused by tumor biology and other causes, including factors associated with SES. Additionally, the diagnosis of ILC was not centrally confirmed, raising the possibility that some histologic diagnoses were incorrectly classified as the diagnostic challenges of ILC are well-documented [52]. Finally, because of the lack of more granular SES data in the NCDB and the inability to use more complex measures of SES, such as the Area Deprivation Index, our reliance on education and income as SES indicators may not fully capture other relevant aspects of SES that could impact outcomes.

These findings highlight the potential for heterogeneity in ILC, and the importance of understanding how socioeconomic factors influence tumor development and outcomes. Our composite measure of SES made use of income and education, variables that have independently been associated with cancer outcomes and are readily available in the NCDB, to better model cumulative socioeconomic disadvantage. Future studies should investigate the mechanisms driving the development of aggressive ILC features in those with low SES, with a focus on the roles of genetics, environment, and access to treatment.