Associations of alcohol use with expression of stromal markers in benign breast biopsy samples

Abstract

Background

We explored the associations of alcohol consumption with expression of α-Smooth Muscle Actin (α-SMA), Tenascin-C (TNC), Fibroblast Activation Protein (FAP), Matrix Metalloproteinase-14 (MMP14), and Calcyclin (s100a6) stromal markers in benign breast biopsy samples.

Methods

The study included 683 cancer-free women from the Nurses’ Health Study II who had biopsy-confirmed benign breast disease (BBD). Alcohol consumption was assessed with semi-quantitative Food Frequency Questionnaires. The data on breast cancer (BCa) risk factors were obtained from biennial questionnaires. Immunofluorescence for stromal markers was performed on tissue microarrays. For each core, the % positivity was quantified by inForm v2.6.0. Generalised linear regression was used to examine associations between alcohol consumption (recent [at biopsy date] and cumulative average from all available questionnaires before the biopsy date) and log-transformed expression of each marker, adjusting for BCa risk factors and BBD subtype.

Results

In multivariate analysis, we observed a suggestive positive association of cumulative average alcohol with TNC (β per 11 g/day = 0.59, 95% CI −0.03,1.22, p = 0.06). Alcohol consumption was not associated with α-SMA, FAP, s100a6, and MMP14.

Conclusion

Alcohol consumption may be associated with an increased normal stromal fibroblast activation as measured by TNC expression in histologically normal breast tissue. Future studies are warranted to confirm our findings.

Introduction

Alcohol consumption is a well-established, modifiable breast cancer risk factor with numerous studies reporting positive associations [1–3]. Women who drink 15–30 grams of alcohol per day (about 1–2 drinks) are at a 10–40% increased risk of breast cancer compared to non-drinkers, and each additional drink per day increases the risk by about 7% [4–7]. In addition, research shows that although alcohol consumption increases breast cancer risk among pre- and postmenopausal women, postmenopausal women may face a slightly greater risk [3]. Importantly, the recent report from the US Surgeon General emphasises that the associations of alcohol with breast cancer are causal [8].

Various mechanisms may explain the influence of alcohol consumption on breast cancer risk. As part of the oxidative metabolism pathway, ethanol is converted to acetaldehyde, which forms adducts with DNA, leading to disruption of DNA synthesis and repair and inducing mutations [9, 10]. Importantly, compared to the liver, breast tissue significantly differs in its ability to detoxify acetaldehyde, resulting in its accumulation and increasing toxicity in this environment [2, 10]. Other mechanisms by which alcohol can influence breast cancer risk include oxidative stress and inflammation, as well as changes in metabolism of oestrogens [9, 10].

In addition to breast cancer risk, in some previous studies, alcohol consumption was also positively associated with mammographic breast density, one of the strongest risk factors for breast cancer [11–13]. Mammographic breast density is reflective of the proportion of stromal, epithelial, and adipose tissues in the breast. The stroma contributes greatly (approximately 30%) to the variation in breast density vs. only 4–7% explained by epithelium [14, 15]. Importantly, emerging evidence has highlighted the active role of stroma in tumorigenesis via various mechanisms. Normal (resident) fibroblasts are the most abundant stromal cells in the tumour microenvironment (TME), primarily responsible for extracellular matrix synthesis and remodelling and creating and maintaining connective tissue [16]. In addition to their involvement in important regulatory pathways, normal breast fibroblasts can give rise to cancer-associated fibroblasts (CAFs) [17, 18]. Whether alcohol can influence the activity of normal fibroblasts is unknown. To fill these gaps, we examined the associations of alcohol consumption with α-Smooth Muscle Actin (α-SMA), Tenascin-C (TNC), Fibroblast Activation Protein (FAP), Matrix Metalloproteinase-14 (or Metallo-peptidase, MMP14), and s100a6 (Calcyclin), all of which are considered to be markers of fibroblast activation. While there have been no studies of stromal tissue markers in normal tissue of cancer-free women, these stromal markers have been studied in breast tumours and adjacent normal tissue and have been previously linked to more unfavourable tumour characteristics and patient outcomes [19–26].

Methods

Study population

This study included cancer-free women with incident biopsy-confirmed benign breast disease (BBD) identified within the Nurses’ Health Study II (NHSII) cohort. The NHSII is an ongoing prospective cohort study established in 1989, which enrolled 116,430 female registered nurses aged 25–42 years from across the United States [27]. Information on breast cancer risk factors was collected at the baseline and updated biennially. At baseline and on subsequent questionnaires, participants were asked whether they had ever received a diagnosis of BBD and, if so, whether it had been confirmed by fine-needle aspiration or biopsy. Our study included 683 cancer-free women with biopsy-confirmed BBD with available data on stromal markers, alcohol consumption, and important covariates. The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health, and those of participating registries as required. Consent was obtained or implied by return of questionnaires.

Benign breast biopsy confirmation and tissue microarray construction

Women who reported a biopsy-confirmed BBD diagnosis were contacted to obtain permission to retrieve their pathology specimens for review [27]. Hematoxylin and eosin (H&E)-stained breast tissue slides were centrally reviewed by blinded pathologists and classified into one of three histologic subtypes: nonproliferative, proliferative disease without atypia, and proliferative disease with atypia. A study pathologist re-evaluated H&E sections to annotate normal terminal duct-lobular units (TDLUs), which were then used for tissue microarray (TMA) construction at the Dana-Farber/Harvard Cancer Center Tissue Microarray core facility. Up to three 0.6 mm cores of normal TDLU were included in the TMA blocks [28–30].

Immunofluorescence for stromal markers and image analysis

A 5 µm paraffin section was cut from each TMA block and stained for stromal markers using an automated multiplex immunofluorescence (IF) technique at the University of Florida Pathology Core Lab using the Leica Bond Autostainer according to a previously established protocol (αSMA: Abcam, Cambridge, MA, Cat# ab5694, RRID: AB_2223021, 1:400 dilution; FAP: Abcam, Cambridge, MA, Cat# ab207178, RRID: AB_2864720, 1:50 dilution; MMP14, Abcam, Cambridge, MA, Cat# ab51074, RRID: AB_881234, 1:150 dilution; Tenascin-C, Abcam, Cambridge, MA, Cat# ab108930; RRID: AB_10865908, 1:200 dilution; s100a6, Abcam, Cambridge, MA, Cat# ab181975, RRID: AB_3697229, 1:300 dilution). To distinguish stromal from epithelial compartments, epithelial regions were identified using the cytokeratin marker AE-1/AE-3, enabling accurate segmentation and quantification of stromal staining (panCK AE1/AE3, Novus Bio, Centennial, CO, Cat# NBP2-29429, RRID: AB_3068002, 1:500 dilution).

Using optimised imaging parameters (DAPI 5.12 ms, Opal 480 9.39 ms, Opal 520 52.82 ms, Opal 570 36.79 ms, Opal 620 36.81 ms, Opal 690 30.33 ms, and auto fluorescence 94.89 ms), TMA sections were digitised at ×20 magnification using the Phenolmager (Akoya Biosciences, Marlborough, MA). IF quantification was performed using in InForm v2.6.0 (Akyoa Biosciences). The experienced operator selected seven representative cores in which at least one of the five stromal markers was highly expressed to train tissue segmentation into stromal and epithelial regions based on pan-cytokeratin (Pan-CK) and αSMA expression as well as determine minimum threshold for each marker. Thresholds for marker positivity were set as: PanCK $>$0.1, FAP $>$0.25, TNC > 0.25, MMP14 $>$0.3, αSMA $>$0.17, and s100a6 $>$1.

Cell-level data for all detected cores were exported as a .CSV file, and subsequent data processing was performed in R. PanCK-positive cells were excluded from the analysis to eliminate epithelial and myoepithelial populations. SMA-positive cells in epithelial regions were also excluded as these were most likely endothelial cells. The total number of the remaining cells (in the stromal regions) was used as the denominator for downstream quantification.

For each TMA core, marker expression within the stromal compartment was quantified on a continuous scale as the percent of stromal cells staining positive for a given marker, regardless of staining intensity.

Assessment of alcohol consumption

In NHSII, alcohol consumption was assessed using semi-quantitative Food Frequency Questionnaires (FFQ) in 1989, 1991, and then every 4 years thereafter. Study participants reported their alcohol consumption over the previous year, specifying intake of wine, beer and liquor. Standard drink definitions include: 1 can or bottle of beer, a 4-oz glass of wine, or a single drink/shot of liquor. Participants indicated their drinking frequency using eight categories: 1–3 per month, 1 per week, 2–4 per week, 5–6 per week, 1 per day, 2–3 per day, 4–6 per day, or $\ge$6 per day. Total alcohol consumption in grams was calculated as the sum of daily drinks multiplied by average ethanol content per drink (12.8 g for regular beer, 11.3 g for light beer, 11.0 g for wine, and 14.0 g for liquor). Recent alcohol consumption was determined based on the questionnaire cycle closest to the date of biopsy and was modelled both as a continuous (per 11 g/day, equivalent to 1 drink) and categorical variable (0 g/day [reference group], <11 g/day, 11–22 g/day, and $\ge$22 g/day as well as 0 g/day [reference group], <11 g/day, and ≥11 g/day). Cumulative average alcohol consumption was computed by averaging all of a participant’s reported intakes from baseline through biopsy date. It was also modelled as a continuous as well as categorical variable with the above-mentioned approaches. In our study sample, based on the biopsy year, for the majority of participants (75.6%) the cumulative average was calculated based on 4 or more FFQs; for 7.0% it was based on 3 FFQs, and for 17.4% it was based on 2 FFQs.

Covariate information

Data on breast cancer risk factors were obtained from the biennial questionnaires at the time of the biopsy. Women were considered to be postmenopausal if they reported having no menstrual periods within the previous 12 months (with natural menopause), or having both ovaries removed, or having a hysterectomy, and being 54 years or older for ever smokers or 56 years or older for never smokers [31, 32].

Statistical analysis

Associations between alcohol consumption and log-transformed expression of five stromal markers (αSMA, TNC, FAP, s100a6, and MMP14) were evaluated using multivariate linear regression. The risk estimates were adjusted for a priori selected breast cancer risk factors: age (continuous, years), age at menarche ( < 12, 12, 13, >13, unknown), BMI (continuous, kg/m²), combined parity/age at first birth (parous with first birth before age 25, parous with first birth after age 25, nulliparous, unknown), family history of breast cancer (yes/no), BBD subtype (nonproliferative, proliferative disease without atypia, and proliferative with atypia), and menopausal status/postmenopausal hormone use (premenopausal, postmenopausal/no hormone use, postmenopausal/past hormone use, postmenopausal/current hormone use, and postmenopausal/unknown hormone use status). We additionally examined age and BMI-adjusted models as well as full models without adjustment for BBD subtype. Also, since tenascin had a large number of observations with 0% staining positivity, we additionally used probit transformation to confirm our findings. Median intake values within each alcohol intake category were used for the test of trend. Finally, even though we could not assess the differences in the associations between pre- and postmenopausal women due to the small number of postmenopausal women in our sample, in exploratory analysis we examined associations separately in premenopausal women. All analyses were conducted using SAS software (version 9.4, SAS Institute, Cary, NC, USA). Significance was evaluated at the 0.05 level using two-sided tests.

Results

Among 683 cancer-free women with incident BBD, the average age at biopsy was 44 years (range of 27–63). The majority of women had proliferative disease without atypia (65.3%), followed by non-proliferative disease (26.2%) and proliferative disease with atypia (8.5%). The majority of women were premenopausal at biopsy (81.3%). Table 1 summarises age-adjusted characteristics of women in the study sample by the level of cumulative average alcohol consumption. In our sample, 150 (22%) women were non-drinkers, 477 (70%) women consumed less than one drink per day ( < 11 g/day), and 56 (8%) consumed at least one drink per day ( > 11 g/day). Distribution of the stromal markers’ expression in our study sample and their correlations with each other are presented in Supplementary Tables 1 and 2, respectively. Distribution of markers’ expression by the level of cumulative average alcohol consumption is presented in Fig. 1. Since tenascin had large number of observations with 0% staining positivity, we additionally present distribution of categorical expression (0, <1, 1–< 10, and ≥10%) (Supplementary Fig. 1) as well as distribution of log-transformed tenascin (Supplementary Fig. 2) by the level of cumulative average alcohol.

Table 1.

Age-adjusted characteristics of women with biopsy-confirmed incident benign breast disease in the Nurses’ Health Study II, by cumulative alcohol use.

Characteristic	Cumulative average alcohol consumption
	None n = 150	<11 g/day n = 478	≥11 g/day n = 56
Mean (SD)
Age at BBD biopsy (years)a	44.95 (6.20)	44.21 (6.02)	44.68 (5.84)
Age at menarche (years)	12.44 (1.38)	12.55 (1.36)	12.72 (1.34)
Body Mass Index (kg/m2)	26.30 (6.00)	25.50 (5.46)	24.00 (5.29)
αSMA (%)	9.54 (5.32)	9.77 (5.62)	9.68 (4.72)
TNC (%)	1.14 (4.63)	0.66 (3.72)	0.68 (2.26)
FAP (%)	7.05 (9.54)	7.52 (9.49)	9.24 (9.62)
s100a6 (%)	21.27 (15.00)	21.27 (14.18)	24.41 (18.75)
MMP14 (%)	8.64 (10.51)	10.20 (12.33)	14.36 (17.04)
Percentages
Parity/age at first birth
Nulliparous	15	13	15
Parous, age <25 years	34	29	25
Parous, age ≥25 years	51	58	59
Family history of BBD	8	10	8
Benign breast disease
Non-proliferative	24	27	21
Proliferative without atypia	67	65	65
Proliferative with atypia	9	8	14
Smoking status
Never smoked	77	64	34
Past smoker	17	27	40
Current smoker	7	9	26
Menopausal status/postmenopausal hormone useb
Premenopausal	79	83	81
Postmenopausal/never used	4	4	7
Postmenopausal/past use	2	2	1
Postmenopausal/current use	8	8	2

αSMA alpha-smooth muscle actin, BBD benign breast disease, CI confidence interval, FAP fibroblast activation protein, MHT menopausal hormone therapy, MMP14 matrix metalloproteinase, NA not applicable, s100a6 calcyclin, SD standard deviation, TNC tenascin-C.

^aValue is not age-adjusted^..

^bThe table does not include participants with unknown menopausal status.

Fig. 1. Distribution of stromal markers by cumulative average drinking in the study sample.

Each figure represents respective medians, quartiles, and outliers for each stromal marker by the level of drinking (None, <11g/day, ≥11g/day).

Results from the full multivariate and reduced age- and BMI-adjusted models are presented in Table 2 and Supplementary Table 3, respectively. In the multivariate analysis, continuous recent and cumulative alcohol consumption were not associated with the expression of αSMA, FAP, s100a6, and MMP14 (for recent: β per 11 g or one drink = 0.03, 95% CI −0.06, 0.13; β = 0.16, 95% CI −0.18, 0.49; β = −0.05, 95% CI −0.25, 0.14; and β = 0.14, 95% CI −0.22, 0.50, respectively; for cumulative: β per 11 g or one drink = 0.01, 95% CI −0.11, 0.12; β = 0.09, 95% CI −0.29, 0.48; β = −0.06, 95% CI −0.29, 0.16; and β = 0.06, 95% CI −0.35, 0.48, respectively). Continuous cumulative average alcohol consumption was associated with a borderline significant increase in TNC expression (β per 11 g or one drink = 0.59, 95% CI −0.03, 1.22, p = 0.06), which was more apparent in the reduced models (β = 0.69, 95% CI 0.07, 1.32; Supplementary Table 3). When modelled as categorical (drinks per day), neither recent nor cumulative average alcohol consumption was associated with any of the markers (p for trend for all >0.05). However, though within a small strata of heavy drinking in our study sample, heavy cumulative average alcohol consumption ( ≥ 22 g/day [≥2 drinks/day]), was associated with higher expression of TNC (β = 2.50, 95% CI 0.20, 4.80) (Table 2), with a similar significant association observed in the reduced model (β = 2.65, 95% CI 0.35, 4.95; Supplementary Table 3). The findings remained similar in the models without adjustment for BBD subtype (Supplementary Table 4). The results were also similar in the additional models with probit-transformation for tenascin (data not shown). No associations were found in the secondary analysis among 555 premenopausal women (Supplementary Table 5); the association of continuous cumulative average alcohol with TNC was no longer significant (p = 0.23) likely due to a smaller number of premenopausal women.

Table 2.

Association of alcohol use with expression of stromal markers (log-transformed) in benign breast biopsy samples (beta coefficients and 95% Confidence Intervals)^a.

Alcohol use	N	αSMA	TNC	FAP	s100a6	MMP14
Recent alcohol (at BBD)
Continuous per 11 g or 1 drink	664	0.03 (−0.06, 0.13)	0.36 (−0.18, 0.89)	0.16 (−0.18, 0.49)	−0.05 (−0.25, 0.14)	0.14 (−0.22, 0.50)
Drinks per day
Non-drinker	239	Reference	Reference	Reference	Reference	Reference
<1	356	−0.03 (−0.15, 0.09)	0.32 (−0.34, 0.98)	0.30 (−0.11, 0.71)	0.01 (−0.23, 0.25)	0.18 (−0.27, 0.62)
1–< 2	52	0.03 (−0.20, 0.25)	0.02 (−1.20, 1.24)	0.44 (−0.32, 1.19)	−0.12 (−0.57, 0.32)	0.08 (−0.73, 0.90)
≥2	17	0.17 (−0.20, 0.54)	1.59 (−0.41, 3.59)	0.62 (−0.63, 1.86)	−0.09 (−0.82, 0.65)	0.67 (−0.66, 2.00)
p-trend	664	0.36	0.26	0.21	0.61	0.43
Drinks per day
Non-drinker	239	Reference	Reference	Reference	Reference	Reference
<1	356	−0.03 (−0.15, 0.09)	0.32 (−0.34, 0.98)	0.30 (−0.11, 0.71)	0.01 (−0.23, 0.25)	0.17 (−0.27, 0.62)
≥1	69	0.06 (−0.14, 0.26)	0.40 (−0.69, 1.50)	0.48 (−0.20, 1.16)	−0.11 (−0.51, 0.29)	0.23 (−0.50, 0.95)
p-trend	664	0.51	0.50	0.19	0.56	0.57
Cumulative average
Continuous per 11 g or 1 drink	683	0.01 (−0.11, 0.12)	0.59 (−0.03, 1.22)	0.09 (−0.29, 0.48)	−0.06 (−0.29, 0.16)	0.06 (−0.35, 0.48)
Drinks per day
Non-drinker	150	Reference	Reference	Reference	Reference	Reference
<1	477	0.10 (−0.04, 0.23)	0.23 (−0.51, 0.97)	0.24 (−0.22, 0.70)	0.07 (−0.20, 0.34)	0.36 (−0.13, 0.85)
1–< 2	43	0.17 (−0.08, 0.42)	0.27 (−1.11, 1.64)	0.80 (−0.05, 1.65)	0.11 (−0.39, 0.62)	0.35 (−0.55, 1.26)
≥2	13	0.12 (−0.30, 0.54)	2.50 (0.20, 4.80)	0.09 (−1.32, 1.50)	−0.45 (−1.29, 0.39)	0.32 (−1.20, 1.83)
p-trend	683	0.33	0.11	0.24	0.58	0.64
Drinks per day
Non-drinker	150	Reference	Reference	Reference	Reference	Reference
<1	477	0.10 (−0.04, 0.23)	0.23 (−0.51, 0.98)	0.24 (−0.22, 0.70)	0.07 (−0.20, 0.34)	0.36 (−0.13, 0.85)
≥1	56	0.16 (−0.07, 0.39)	0.79 (−0.46, 2.03)	0.64 (−0.13, 1.40)	−0.02 (−0.48, 0.44)	0.35 (−0.48, 1.17)
p-trend	683	0.28	0.24	0.14	0.82	0.62

αSMA alpha-smooth muscle actin, BBD benign breast disease, CI confidence interval, FAP fibroblast activation protein, MMP14 matrix metalloproteinase, s100a6 calcyclin, TNC tenascin-C.

^aAdjusted for age (continuous), BMI (continuous), a family history of breast cancer (Yes/No), menopausal status/postmenopausal hormone use (premenopausal, postmenopausal/no hormones, postmenopausal/past hormones, postmenopausal/current hormones, postmenopausal/unknown hormone use status), age at menarche ( < 12, 12, 13, >13, unknown), combined parity/age at first birth (parous with first birth before age 25, parous with first birth at or after age 25, nulliparous, unknown), and BBD category (non-proliferative, proliferative without atypia, and proliferative with atypia).

Discussion

In this study of 683 cancer-free women with biopsy-confirmed incident BBD in the NHS II cohort, we examined the associations between recent and cumulative average alcohol consumption and the expression of stromal markers αSMA, TNC, FAP, s100a6, and MMP14. We found suggestive positive associations of cumulative average alcohol consumption with TNC expression.

Alcohol consumption is a well-established risk factor for breast cancer, with numerous studies showing a positive, dose-dependent association [1–3]. The International Agency for Research on Cancer (IARC) classifies alcohol as a group 1 carcinogen, with causal associations for breast cancer reported by the U.S. Surgeon General [8]. Previously suggested mechanisms underlying these associations include DNA damage caused by acetaldehyde, oxidative stress and promotion of inflammation [9, 10]. Additionally, alcohol consumption may elevate oestrogens (both circulating and in the breast tissue) via increases in aromatase activity [2, 9]. A recent study also suggested that alcohol may influence breast tissue composition [33]. Our findings of positive associations of cumulative average alcohol consumption with TNC expression suggest that alcohol can potentially influence normal breast fibroblasts. While all the selected markers for this study are associated with fibroblast activation, ECM remodelling, and tumour progression, unlike the rest of the markers that are localised in the cellular plasma, membrane or nucleus, TNC is secreted into ECM [34]. Previous studies in the breast tumours suggest that TNC contributes to carcinogenesis by promoting epithelial–mesenchymal transition (EMT), stem-cell signalling, and immune-suppressive environment and, as the result, enhancing cell motility, angiogenesis, invasion, and metastatic spread [35–37]. TNC expression in breast tumours has been linked to poorer patient outcomes and more aggressive tumour characteristics [38]. Prior reports showed expression of TNC in breast tumour and, to a much lesser degree, in adjacent normal tissue [36, 38]. Our findings now demonstrate that TNC may also be expressed in histologically normal breast tissue of cancer-free women, suggesting its potential involvement in the early stages of neoplastic transformation. If associations between alcohol and TNC are confirmed in subsequent studies, results may suggest that the earliest effects of alcohol on breast carcinogenesis may involve fibroblast activation in normal breast tissue. Given the complexity of the stromal environment, future studies should incorporate more relevant markers to comprehensively examine this complex relationship.

This is the first study to evaluate associations of alcohol consumption and the expression of several stromal markers in cancer-free women. The study used the data and breast biopsy samples from the Nurses’ Health Study II, a large, prospective cohort study with nearly 4 decades of follow-up and detailed information on BBD status and established breast cancer risk factors. Despite these strengths, several limitations should be noted. Biopsy samples were retrieved from a single region of breast tissue, which may limit the generalisability of findings to the entire breast. However, previous NHS studies have demonstrated that this sampling technique allows identification of markers associated with BCa risk, as well as investigating their associations with various breast cancer risk factors [39–46]. Although alcohol intake was assessed prospectively, self-reported consumption may be subject to misclassification. Nonetheless, prior validation studies have demonstrated high reproducibility and validity of alcohol data collected by the FFQ, with strong correlations between reported alcohol intake on FFQ and multiple-week diet records (r = 0.90), as well as repeated FFQs administered 4 years apart (r = 0.84) [47, 48]. Therefore, the FFQ remains a reliable tool for measuring long-term alcohol consumption. It should be noted that in our sample, only 18.7% of women consumed ≥11 g (≥1 drink) of alcohol per day at the time of the biopsy, and only 18.2% reported an average cumulative alcohol consumption ≥11 g (≥1 drink), limiting our ability to examine the effect of heavy drinking definitively. Finally, timing of exposure may also play a role, as adolescence is considered a window of susceptibility for breast carcinogenesis and early-life exposures can have long-lasting effects on subsequent breast cancer risk [2]. Some previous studies have demonstrated a positive association of adolescent alcohol consumption with the risk of proliferative BBD [49, 50], higher estradiol levels [51] and breast cancer [52, 53]. Previous studies that examined associations of early-lifer alcohol consumption as well as other breast cancer risk factors with breast cancer risk had varying definitions of the adolescent period (ranging between ages 10–20 years). In the context of breast cancer risk, adolescence could be defined as the period from the onset of puberty to the completion of breast development (around the age of 17–18 years). Additionally, the time between menarche and the first full-time pregnancy is of particular interest as the breast tissue remains undifferentiated and more susceptible to various influences during this period. In our study sample, data on adolescent alcohol consumption (as reported for the ages between 15 and 17 in NHSII) were available only for 444 (65%) participants, and among those, 415 reported no alcohol consumption during adolescence; thus, this association could not be evaluated.

In conclusion, our findings provide the first evidence of suggestive associations of alcohol consumption with TNC expression in histologically normal breast tissue. Future studies are warranted to confirm these findings and expand the set of relevant stromal markers to better capture the potential complexity of alcohol’s influence on stromal-epithelial interactions. Additionally, investigations into the associations of alcohol consumption during adolescence, as well as before the first full-term pregnancy, with stromal marker expression may help to better understand the biological mechanisms linking early-life alcohol consumption to breast cancer risk.

Author contributions

LY and RT conceived of and designed the study, directed statistical analyses, interpreted results, substantially revised initial drafts of the paper and provided final review and approval. AA, MR, and LY performed statistical analyses. YH and BS assessed IF results. AA wrote the first draft of the manuscript, which was revised with contributions from LY, YH, MR, BR and RT. LY supervised the overall study progress. All authors read and approved the final manuscript.

Funding

This work was supported by the National Cancer Institute at the National Institutes of Health [CA277817, CA240341 to LY, CA131332, CA175080, P01 CA087969 to RMT, UM1 CA186107 to MS, U01 CA176726 to WW], Avon Foundation for Women, Susan G. Komen for the Cure®, and Breast Cancer Research Foundation.

Data availability

The data that support the findings of this study are available from the Nurses’ Health Studies; they are not publicly available. Investigators interested in using the data can request access, and feasibility will be discussed at an investigator’s meeting. Limits are not placed on scientific questions or methods, and there is no requirement for co-authorship. Additional data sharing information and policy details can be accessed at http://www.nurseshealthstudy.org/researchers.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

This study was performed in accordance with the Declaration of Helsinki. The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health, those of participating registries as required, and the University of Florida Institutional Review Boards. Informed consent was obtained or implied by return of questionnaires. Women provided written informed consent for obtaining their benign biopsy samples.

Supplementary information

The online version contains supplementary material available at 10.1038/s41416-026-03481-3.