Overview of study designs and smoking-related variables in included breast cancer subtype analyses.
| Study (author, year) | Country/Setting | Design, recruitment & follow-up | Population & eligibility | Sample/Cases | Exposure assessment (smoking) | Smoking variables used in analyses | Timing of exposure assessment | Tumor subtype definition & markers | Covariates in multivariable models | Main smoking-breast cancer associations (vs. never smokers) |
|---|---|---|---|---|---|---|---|---|---|---|
| Kabat et al., 2011 [22] | USA; multi-centre WHI cohort (clinical trials + OS; 40 centres) | Prospective cohort; recruitment 1993–1998; median follow-up 8.0 years; close-out 12 Sept 2005 | Postmenopausal women aged 50–79 years enrolled in WHI CT or OS; excluded if prior breast cancer or mastectomy, or missing key exposure/outcome data | Cohort analysed: 148,030 women. Cases: TNBC 300; ER+ (HER2 status known) 2,479. Exclusions: prior breast cancer/mastectomy 8,735; missing outcome 690; breast cancer without definite ER/PR/HER2 2,263; missing smoking 1,773; missing alcohol 318 | Baseline self-administered questionnaires (health habits/lifestyle) at WHI entry; smoking was entirely self-reported | Collected: ever smoked ≥ 100 cigarettes; age at initiation; current/former status; age at quitting; cigarettes/day; years smoking. Analysed as: (1) smoking status (never/former/current); (2) cigarettes/day (0–4, 5–14, 15–24, ≥ 25); (3) age at start (< 20, ≥ 20 years); (4) duration (< 20, 20–29, ≥ 30 years); (5) pack-years (< 20, 20–40, ≥ 40), all vs. never | Single baseline assessment at enrolment (1993–1998), prior to diagnosis; no repeated updates of smoking were used in analyses | Breast cancer cases self-reported then centrally adjudicated. TNBC: ER−/PR−/HER2− (absence of ER and PR expression and no HER2 over-expression). ER+ breast cancer: ER positive with known HER2 status. Markers: ER, PR, HER2; Ki-67 not reported. IHC cut-offs not reported; no PAM50 or St Gallen algorithms | Base model: age; age at menarche; age at first full-term pregnancy; parity; age at menopause; BMI; waist circumference; oral contraceptive use; hormone therapy (never, estrogen only, estrogen + progestin, both); history of breast biopsy; family history of breast cancer in first-degree relative; mammogram in past 2 years; physical activity (MET-h/week); education; ethnicity; WHI trial arm or OS component. Smoking models additionally adjusted for alcohol intake. Alcohol models additionally adjusted for pack-years (0, < 20, 20–40, ≥ 40) | Smoking status—TNBC (n = 300): former vs. never HR 0.91 (95% CI 0.70–1.16); current vs. never HR 1.09 (0.69–1.72) → no clear association. Smoking status—ER+ (n = 2,479): former vs. never HR 1.14 (1.05–1.24); current vs. never HR 1.05 (0.88–1.25) → modest increase in ER+ risk among former smokers. Dose metrics (ER+): positive trends for cigarettes/day (p = 0.02), duration (p = 0.03), and pack-years (p = 0.01; ≥ 40 pack-years HR 1.24, 95% CI 1.06–1.44), while no significant trends for TNBC (all p-trend > 0.3). Baseline distribution by smoking: among TNBC cases, 51.5% never, 41.1% former, 7.4% current; among ER+ cases, 47.8% never, 46.1% former, 6.1% current |
| Kawai et al., 2014 [12] | USA; three-county Seattle-Puget Sound metropolitan area (King, Pierce, Snohomish counties); population-based | Population-based case-control study of women aged 20–44 years; cases diagnosed 2004–2010, identified via SEER Cancer Surveillance System; controls selected by random-digit dialing; interviews generally within ~2 years of reference date | Women 20–44 years of age, resident in the three-county area; cases: incident invasive breast cancer with no prior in situ or invasive breast cancer; controls: cancer-free women frequency matched to cases on 5-year age groups. Exclusions from analysis: ER−/HER2+ (n = 60) due to small numbers; missing ER/PR/HER2 (n = 28); missing smoking data (5 controls, 8 cases) | Initially, 1,359 eligible cases; 1,056 (78%) interviewed. After exclusions, analytic case set: 960 invasive breast cancer cases (778 ER+, 182 TN). Controls: 1,489 eligible; 943 (63%) interviewed; after excluding 5 with missing smoking data, 938 controls. Final analytic dataset: 938 controls, 778 ER+ cases, 182 ER−/PR−/HER2− (TN) cases | In-person structured interview; detailed lifetime smoking history up to reference date (diagnosis date for cases, assigned reference date for controls). Smoking self-reported. Ever/never based on ≥ 100 cigarettes lifetime; collected ages started/stopped for each period, intensity (cigarettes/day), and recency | Derived variables: (1) ever smoked (never/ever ≥ 100 cigarettes); (2) recency (never/current-recent/former; current-recent = smoking within 2 years of reference date, former = quit ≥ 2 years before reference date); (3) total years smoked (never/< 5.0/5.0–9.9/10.0–14.9/≥ 15.0); (4) age at initiation (never/≤ 14/15–17/≥ 18); (5) pack-years (never/< 2.5/2.5–4.9/5.0–9.9/10.0–14.9/≥ 15.0); (6) years since quitting for former smokers (never/ < 5/5–9.9/≥ 10); (7) initiation before menarche (no/yes); (8) initiation before first birth among parous women (no/yes) | Exposure history was restricted to the period before the reference date; interviews were conducted on average 18 months (cases) and 20 months (controls) after the reference date (medians 16 and 19 months, respectively). Smoking history recalled retrospectively but over a relatively recent time window | ER and PR positivity is defined as ≥ 1% positive staining of tumour cells; negativity is defined as 0–1% positive staining. HER2 positivity is defined as IHC 3+ and/or FISH-positive; HER2 negativity is defined as IHC 0/1+ and/or FISH-negative. Tumours with HER2 IHC 2+ and no FISH result are classified as HER2 unknown. Subtypes: ER+ (all ER+ regardless of PR/HER2); triple-negative (ER−/PR−/HER2−). ER−/HER2+ group excluded due to small numbers | Polytomous logistic regression comparing ER+ and TN cases separately to the common control group. All models adjusted for age (5-year categories) and reference year (continuous). Potential confounders evaluated: education, income, race/ethnicity, oral contraceptive use, mammography history, first-degree family history, BMI, age at menarche, parity, number of full-term pregnancies, age at first live birth, alcohol use, and physical activity. Only age at first live birth changed the risk estimates > 10%, so the final models adjusted for age, reference year, and age at first live birth. No significant effect modification detected | Ever vs. never smoking: overall breast cancer OR 1.3 (95% CI 1.1–1.7); ER+ OR 1.4 (1.1–1.8); TN OR 1.1 (0.7–1.6)—elevation confined to ER+. Current/recent vs. never: ER+ OR 1.4 (1.0–2.0); TN OR 1.2 (0.7–2.1). Former vs. never: ER+ OR 1.4 (1.0–1.8); TN OR 0.9 (0.6–1.5). Pack-years (ever smokers, overall): < 2.5 PY ER+ OR 1.5 (1.1–2.1); ≥ 15 PY ER+ OR 1.7 (1.1–2.5); TN estimates near null, no dose-response. Among current/recent smokers: ≥ 15 years smoking ER+ OR 1.5 (1.1–2.1); ≥ 10 pack-years ER+ OR 1.6 (1.1–2.4); no corresponding increase for TN (OR ≈ 1.0). Years since quitting among former smokers: ER+ risk appeared to return toward baseline ≥ 10 years after cessation. Overall pattern: modest increased risk for ER+, no clear association for TN in young women 20–44. |
| Butler et al., 2016 [21] | USA; 24 adjoining counties in central and eastern North Carolina; Carolina Breast Cancer Study (CBCS) phases I & II | Population-based case-control study. Cases: first primary invasive breast cancer diagnosed 1 May 1993–30 Sept 1995 (Phase I) or 1 May 1996–30 Sept 2001 (Phase II), identified via rapid case ascertainment through the NC Central Cancer Registry. Controls: incidence-density sample, frequency-matched to cases by 5-year age, race, county; controls 20–64 from DMV, ≥ 65 from Medicare. Case response rate 76%, control response 55%. | Female residents of 24-county region, aged 20–74 years, with first diagnosis of invasive breast cancer (cases) or cancer-free at selection (controls). Oversampling of black and younger women to improve power for subtype/race analyses. | Overall analytic set: 1,808 invasive cases and 1,564 controls. Subtype-specific analyses: Luminal cases ≈ 737 (369 never, 368 ever smokers); Basal-like cases ≈ 205 (114 never, 91 ever smokers); controls 1,564 (840 never, 724 ever). Race-stratified counts were reported separately (black: 788 cases/718 controls; white or non-black: 1,020 cases/846 controls). | In-person nurse-administered interview using a standardized questionnaire, typically ~6 months after case ascertainment. Smoking self-reported: lifetime history, age at initiation, age at cessation, number of packs per day. Active smokers were defined as women who had smoked ≥ 100 cigarettes in their lifetime. | Derived variables: (1) ever smoking (never/ever ≥ 100 cigarettes); (2) smoking status (never/former/current)-current = still smoking at interview or quit at same age as case/control selection, former = quit before selection; (3) smoking dose (packs/day: never, < ½, ½–1, > 1); (4) duration (never, ≤ 10, 11–20, > 20 years); duration separately for current and former smokers; (5) years since quitting in former smokers (never, < 5, 5–10, 11–20, > 20); (6) age at initiation (never, ≤ 15, 16–20, > 20 years); (7) initiation relative to menarche and first full-term pregnancy (never; ≤ menarche; after menarche ≥ 11 years before FFTP; after menarche < 11 years before FFTP). | Smoking history up to age at case/control selection; interviews on average ~6 months after diagnosis/selection. Classification of current vs. former explicitly referenced to age at case/control selection to limit reverse causation due to post-diagnosis quitting. | ER and PR positivity was obtained from medical records. Tumour blocks were centrally stained for HER2, HER1, and CK5/6 by IHC in the UNC core lab. Subtypes: luminal = ER+ and/or PR+, regardless of HER2; basal-like = ER−, PR−, HER2−, and HER1+ and/or CK5/6+. Assay procedures and cut-offs for positivity were previously described in CBCS methodological papers. | Unconditional logistic regression with polytomous outcomes (luminal and basal-like vs. common control group). All ORs adjusted for age, race, first-degree family history of breast cancer, alcohol use, menopausal status, hormone replacement therapy use, oral contraceptive use, parity, age at first birth, age at first breastfeeding, age at menarche, BMI, and for randomized recruitment probabilities via offset term. The same adjustment set was applied in overall and subtype-specific models. | Overall breast cancer: Ever vs. never OR 1.07 (95% CI 0.92–1.25); current vs. never OR 1.02 (0.84–1.24); former vs. never OR 1.11 (0.93–1.33). Duration > 20 vs. never OR 1.33 (1.09–1.61); among former smokers, > 20 yrs vs. never OR 1.54 (1.15–2.07); years since quitting 5–10 vs. never OR 1.39 (1.01–1.93). Subtype-specific: ever vs. never—luminal OR 1.12 (0.92–1.36), basal-like OR 0.96 (0.69–1.32). Current vs. never—luminal OR 1.10 (0.86–1.41), basal-like OR 0.82 (0.54–1.24). Duration > 20 yrs vs. never—luminal OR 1.51 (1.19–1.93), basal-like OR 0.90 (0.57–1.43). Dose > 1 pack/day vs. never—luminal OR 1.08 (0.81–1.44), basal-like OR 0.47 (0.25–0.89) (inverse). Tests of heterogeneity showed statistically different ORs by subtype for dose (p = 0.02) and duration (p < 0.01). Race-stratified: among black women, ever vs. never and especially long duration are more strongly associated with Luminal cancer [current vs. never OR 1.53 (1.04–2.26); duration > 20 yrs vs. never OR 2.06 (1.38–3.06)], with former smoking associated with increased basal-like risk [OR 1.71 (1.02–2.86)]. Among white women, associations are weaker or null for luminal; basal-like shows an inverse association with high dose [> 1 pack/day vs. never OR 0.38 (0.16–0.90)]. Overall pattern: long-term smoking is positively associated with luminal breast cancer, especially in black women, with no clear increase and some inverse signals for basal-like disease. |
| Park et al., 2016 [23] | USA; pooled data from 4 studies of African American women: Carolina Breast Cancer Study (CBCS), Women’s Circle of Health Study (WCHS), Black Women’s Health Study (BWHS), Multiethnic Cohort (MEC) | Pooled case-control analysis within the AMBER Consortium. CBCS & WCHS: population-based case-control; BWHS & MEC: prospective cohorts contributing nested case-control data. Diagnosis/enrolment periods: CBCS 1993–2014 (20–74 y); WCHS 2002–2013 (20–75 y); BWHS cohort initiated 1995 (21–69 y); MEC 1993–1996 (45–75 y). | African American women. Cases: first diagnosis of invasive breast cancer or DCIS with available smoking and receptor data. Controls: African American women without breast cancer, selected within each parent study (population controls in CBCS/WCHS; ~4 matched controls per case by birth year and questionnaire cycle in BWHS/MEC). | Eligible: 5,819 cases, 17,453 controls. After excluding 105 with missing smoking: 5,791 breast cancer cases and 17,376 controls. Subtype-specific: ER+ cases 3,099; ER− cases 1,511; triple-negative (TNBC) cases 694 (ER−/PR−/HER2−). | Smoking information was collected in each study via self-administered questionnaires or in-person interviews; all self-reported. Lifetime active smoking history up to index date (age at initiation, cessation, quantity). Variables were harmonised across studies for pooled analysis. | Harmonised active-smoking variables: (1) smoking status (never/former/current); (2) age at initiation (≤ 14, 15–17, 18–20, ≥ 21 y); (3) cigarettes/day (< 5, 5–14, 15–24, ≥ 25); (4) duration (< 10, 10–19, ≥ 20 years); (5) pack-years (< 10, 10–19, ≥ 20); (6) among parous women, years smoked before first birth (never smokers; smoked only after first birth; 1–5, 6–9, ≥ 10 years before first birth). Never-smokers are the reference. Passive smoking was also assessed, where available, but it is secondary. | Smoking history ascertained for the period prior to the index date: diagnosis date for cases; corresponding reference date for controls (matched by study-specific procedures). Analyses use smoking exposure up to that date only. | Pathology data from hospital records and/or cancer registries. Tumours classified by ER, PR, and HER2. Subtypes used in analysis: ER+ (any ER-positive), ER−, and TNBC defined as ER−/PR−/HER2−. HER2 is not available for some earlier cases; those with missing receptor data were excluded from subtype-specific models. Assays and cut-offs followed routine clinical practice in each centre; specific % thresholds were not detailed. | Multivariable unconditional logistic regression. Basic adjustment: age, study, calendar year of interview, geographic region. Fully adjusted model: additionally, education, age at menarche, age at first birth, parity, age at menopause, oral contraceptive use, estrogen-only therapy, combined estrogen + progestin therapy, BMI, family history of breast cancer, and alcohol use. All categorical with “unknown” levels as specified in the paper. Same adjustment set for the overall and subtype-specific models. | Overall breast cancer: in premenopausal women, both former and current smokers had modestly lower risk vs. never (OR ≈ 0.8, no dose-response by duration or pack-years). In postmenopausal women, long duration (≥ 20 y) and higher pack-years (≥ 20) were associated with modestly higher risk (OR ≈ 1.14–1.16). By subtype: for ER+ disease, long-term smoking (≥ 20 y) showed a small positive association (duration ≥ 20 y vs. never OR ≈ 1.11, p-trend ~0.03), while associations for ER− and TNBC were close to null (ORs around 1.0 for former/current, duration and pack-years, with no clear trends). Overall pattern: smoking has little to no association with ER− or TNBC, with only a modest increase in risk for long-term smoking in ER+ and postmenopausal African American women. |
| Ellingjord-Dale et al., 2017 [13] | Norway; nationwide Norwegian Breast Cancer Screening Program | Nested case-control within a population-based mammography screening cohort, 2006–2014. Women 50–69 invited every 2 years for two-view mammography; attendance ≈ 75%. Cases were identified through the Cancer Registry of Norway; for each case, 5 controls matched on year of birth (± 3 y) and year of last screening (± 3 y). | Women aged 50–69 years who attended screening 2006–2014, completed risk-factor questionnaires, and had no history of invasive cancer (except non-melanoma skin cancer) or DCIS before 1 Jan 2006. Cases: first invasive breast cancer (ICD-10 C50) with ER, PR, HER2 data. Controls: cancer-free, alive, resident in Norway at case diagnosis, matched as above. | Screening cohort: 344,348 women eligible. After excluding missing covariates, 4,952 breast cancer cases remained. For controls, after exclusions, 197,854 women; from these, 24,760 controls matched (5 per case). Subtype classification possible for 4,402 cases: luminal A-like 2,761; luminal B-like HER2− 709; luminal B-like HER2+ 367; HER2+ 204; triple-negative 361. | Self-administered questionnaires completed at the last screening before diagnosis for cases and corresponding screening round for controls (if missing, previous round used; ≈ 16.5% from earlier questionnaire). Smoking history self-reported. | Smoking status: never/past/current. Intensity: number of cigarettes per day (current): never, 1–4, 5–9, 10–19, ≥ 20. Cumulative exposure: pack-years = (avg cigarettes/day ÷ 20) × years smoked, categorised as < 2.5, 2.5–4.9, 5.0–9.9, 10.0–14.9, 15.0–19.9, ≥ 20. Smoking was also analysed at ages 30–39 and 40–49 years in supplementary tables, but the main subtype table uses current status, current intensity, and lifetime pack-years. | Exposures (including smoking) were taken from the last questionnaire before diagnosis (or prior round if missing), i.e., current status and intensity at/near the time of screening, always prior to diagnosis for cases and corresponding date for controls. | ER, PR, HER2 obtained from pathology reports submitted to the Cancer Registry. ER+: ≥ 10% nuclear staining 2006–Jan 2012; ≥ 1% from Feb 2012 onwards. PR+: ≥ 10% throughout. HER2: IHC 0/1+ = negative; 3+ = positive; 2+ confirmed by in situ hybridization. If IHC 2+ and ISH positive (or ISH positive with missing IHC) → HER2+; if IHC 2+ and ISH negative → HER2−. Subtypes (St Gallen-based): luminal A-like (ER+PR+HER2−); luminal B-like HER2− (ER+PR−HER2−); luminal B-like HER2+ (ER+PR±HER2+); HER2+ (ER−PR−HER2+); triple-negative (ER−PR−HER2−). | Conditional logistic regression matched on birth year & screening year. All ORs mutually adjusted for BMI (≤ 22, 23–25, 26–28, > 28 kg/m2), education (primary, high school, bachelor/master), age at menarche (9–12, 13, 14, 15–18 y), number of pregnancies ≥ 6 months (0, 1, 2, 3, ≥ 4), and menopausal status (pre, peri, post). For smoking models, additionally adjusted for alcohol intake (never, 1, 2, 3–4, ≥ 5 glasses/week) and physical activity (0, 1, 2–3, 4–5, ≥ 6 hours/week). | Overall breast cancer: smoking status: past vs. never OR 1.06 (95% CI 0.98–1.15); current vs. never OR 1.13 (1.03–1.23), p-trend = 0.006. Cigarettes/day (current): 10–19 vs. never OR 1.22 (1.06–1.39); ≥ 20 vs. never OR 1.41 (1.06–1.89), p-trend = 0.001. Pack-years: ≥ 20 vs. < 2.5 pack-years OR 1.26 (1.08–1.46), p-trend = 0.004. By subtype: for luminal A-like cancers, past vs. never OR 1.12 (1.01–1.23), current vs. never OR 1.18 (1.05–1.32), p-trend = 0.003. Current 10+ cigarettes/day vs. never OR 1.27 (1.07–1.50), and pack-years ≥ 20 vs. < 2.5 OR 1.27 (1.04–1.54), p-trend = 0.01. For luminal B-like HER2−, current smoking OR 1.22 (0.98–1.52); 10+ cigarettes/day vs. never OR 1.38 (1.00–1.89); pack-years ≥ 20 vs. < 2.5 OR 1.62 (1.09–2.40), p-trend = 0.03. For luminal B-like HER2+, HER2+ (ER−PR−HER2+) and triple-negative cancers, smoking status, intensity and pack-years showed no clear associations (ORs around 1.0, non-significant trends). Overall interpretation: smoking increases risk of luminal A-like and luminal B-like HER2− breast cancers, particularly at ≥ 10 cigarettes/day or ≥ 20 pack-years, with no detectable association for HER2+ or triple-negative disease. |
| Gomes et al., 2022 [24] | Brazil; state of Paraíba, Northeast Brazil; two breast-cancer reference centres (FAP, Campina Grande; HNL, João Pessoa) and public/rural primary care centres | Hospital-based case-control study. Cases: invasive operable breast cancer diagnosed and treated 2017–2020 at FAP or HNL. Controls: healthy women recruited in the same period from the same hospitals and three rural public health-care centres. Participants interviewed March 2017–March 2020; median time from diagnosis to interview ≈ 9 months. | Cases: women ≥ 18 years with invasive BC diagnosed ≤ 36 months before recruitment; no in situ tumours, BC recurrence, or previous other cancers. Controls: women without any cancer or chronic disease (e.g., diabetes, heart disease), age-matched to cases (± 5 years), only one control per family. | Total sample: 313 invasive BC cases and 321 controls. Of cases, 224 (71.6%) were postmenopausal. For molecular subtype analysis, 12 cases were excluded for missing IHC data, leaving 301 cases: luminal A 54 (17.9%), luminal B 175 (58.1%), HER2 29 (9.7%), TNBC 43 (14.3%). | Structured questionnaire administered face-to-face by study authors. Cases interviewed in chemotherapy/radiotherapy units; controls interviewed in waiting rooms of health-care centres. Information collected on: family history of cancer in first-degree relatives, alcohol consumption (ever/never and frequency), smoking (ever/never), oral contraceptive use (ever/never), age at menarche, parity, reproductive phase, etc. Height and weight for BMI from medical records (measured before treatment). Smoking, alcohol, and contraceptive use were all self-reported. | Main risk factors analysed: family history (yes/no); BMI categories (normal 18.5–24.99, overweight 25–29.99, obesity ≥ 30.0 kg/m2); alcohol consumption (ever/never, plus monthly frequency categories: never, 1–2, 3–7, ≥ 8 times/month); ever smoked (yes/no); oral contraceptive use (yes/no); menopausal status (pre/post); age at menarche (< 12 vs. ≥ 12 years); nulliparity (yes/no). For subtype models, these were entered into polytomous logistic regression with healthy controls as the reference. | Questionnaire responses and BMI measurements reflect pre-diagnosis lifetime history up to diagnosis/interview; cases were included if diagnosed within the previous 36 months (median 9 months before interview), so exposures predominantly pre-diagnostic but some potential for post-diagnosis change/recall bias. | ER, PR, and HER2 status were obtained from pathology reports. Subtypes are defined as: luminal A: ER+ and/or PR+, HER2−, Ki-67 < 14%; luminal B: ER+ and/or PR+ and HER2+, or ER+ and/or PR+, HER2−, Ki-67 ≥ 14%; HER2 subtype: HER2+, ER−, PR−; triple-negative (TNBC): ER−, PR−, HER2−. All tumours invasive. | Overall case-control models: logistic regression with backward selection. Final model adjusted for menopause status, age at menarche, smoking (ever/never), and age (categorical); exposures retained in the final model: family history, BMI, alcohol consumption, and contraceptive use. Subtype models: polytomous logistic regression with healthy controls as reference; final model adjusted for menopause status, nulliparity, smoking, age at menarche, and age (categorical). | Overall BC risk (all women, cases vs. controls, multivariable model): family history yes vs. no: OR 1.78 (95% CI 1.22–2.59). Obesity vs. normal BMI: OR 1.69 (1.08–2.63); overweight vs. normal: OR 1.37 (0.92–2.04). Alcohol consumption (ever vs. never): OR 2.21 (1.44–3.39). Contraceptive use (ever vs. never): OR 2.99 (2.09–4.28). Ever smoked (yes vs. no) was associated with BC in age-adjusted analysis (OR 1.51, 95% CI 1.07–2.12) but was not retained in the final multivariable model. Stratified by menopausal status: among postmenopausal women, obesity alone increased BC risk: OR 2.02 (1.22–3.37); alcohol consumption increased risk: OR 4.15 (2.13–8.11). Among premenopausal women, obesity was not associated; nulliparity increased BC risk: OR 4.19 (1.65–10.49), whereas among postmenopausal women nulliparity was protective (OR 0.36, 0.18–0.70). Alcohol dose-response: vs. never, 1–2 times/month OR 1.27 (0.71–2.27); 3–7 times/month OR 3.82 (1.96–7.43); ≥ 8 times/month OR 5.00 (2.00–12.51). Subtype-specific risks vs. controls (polytomous logistic regression, adjusted model): Family history increased risk of luminal A (OR 3.78, 1.90–7.52) and TNBC (OR 2.58, 1.27–5.23); association for luminal B attenuated to borderline (OR 1.49, 0.96–2.31). Obesity (vs. normal) increased risk mainly for TNBC (OR 4.06, 1.58–10.42) and luminal B (OR 1.87, 1.13–3.11); no clear effect for luminal A or HER2. Overweight showed non-significant positive trends for luminal A and TNBC. Alcohol consumption increased the risk of luminal A strongly: OR 7.08 (3.40–14.73), and luminal B more modestly: OR 1.77 (1.07–2.92); no clear association for HER2 or TNBC. Contraceptive use increased the risk of luminal A (OR 4.48, 2.09–9.58), luminal B (OR 3.08, 2.02–4.69), and HER2 (OR 4.89, 1.92–12.44), but not clearly TNBC (OR 1.57, 0.77–3.22). Early menarche (< 12 years) particularly increased TNBC risk (age-adjusted OR 4.54, 2.15–9.58). Overall pattern: obesity and alcohol are strongly linked to TNBC and luminal subtypes, respectively, while smoking shows only a modest crude association with overall BC and no clear independent subtype-specific effect after adjustment. |
| Ihenacho et al., 2022 [25] | USA; population-based in Los Angeles County (CA) and Metropolitan Detroit (Oakland, Wayne, Macomb counties), via SEER registries | Population-based case-control study of young-onset breast cancer (YOBC). Recruitment 2010–2015. Cases ascertained by rapid case ascertainment through LA and Detroit SEER; controls selected by area-based sampling from Census postal addresses, frequency-matched to cases on race (NHB/NHW), region, and 5-year age group. | US-born women, self-identified female, non-Hispanic Black (NHB) or non-Hispanic White (NHW), aged 20–49 years at reference date, residing in LA County or Metropolitan Detroit. Cases: incident, invasive, primary breast cancer diagnosed 2010–2015, confirmed histologically. Controls: cancer-free women meeting the same demographic and residence criteria. | In total 1,812 invasive YOBC cases (1,130 NHW, 682 NHB) and 1,381 controls (716 NHW, 665 NHB) completed interviews. Smoking status was missing for 18 (14 cases, 4 controls), leaving 1,798 cases and 1,377 controls for main smoking analyses. Tumour subtype data are missing for 130 cases, leaving 1,670 cases for subtype analyses. Subtypes: luminal A, luminal B, HER2-type, triple-negative (numbers not all given in text but all included in polytomous models). | In-person structured interview using a life-history calendar to enhance recall. Detailed lifetime personal cigarette smoking history was obtained: smoking status, age at start, periods of cessation, average cigarettes per day (CPD), total years smoked, timing of initiation relative to first full-term pregnancy (FFTP). Smoking is self-reported. | Ever smoking: ≥ 1 cigarette/day for ≥ 6 months (yes/no). Smoking status: never/formerly smoked /currently smoke; women who quit ≤ 1 year before reference date were classified as current smokers. CPD: < 5, 5–19, ≥ 20. Pack-years: (< 5, 5–19, ≥ 20), calculated as (CPD/20) × years smoked. Age at initiation: < 18, 18–24, ≥ 25 years. Time since initiation: < 20, 20–29, ≥ 30 years. Time since quitting among former smokers: 1–10, ≥ 10 years (vs. never). Timing relative to FFTP (parous women): initiated after FFTP vs. initiated before FFTP vs. never smoked. All smoking exposures use “never smoked” as reference. | Exposure history was defined up to the reference date: the date of invasive BC diagnosis for cases and the date 4 months before interview for controls. Lifetime smoking variables (status, CPD, pack-years, age at initiation, time since initiation, timing vs. FFTP) are constructed using data up to that reference date; thus pre-diagnostic for cases. | Tumour subtypes were derived from SEER pathology data (hospital/registry) based on ER, PR, HER2 and tumour grade. Categorised as: luminal A (ER/PR+, HER2−, grade 1/2); luminal B (ER/PR+, HER2+, any grade, or ER/PR+, HER2−, grade ≥ 3); HER2-type (ER−, PR−, HER2+); triple-negative (TNBC) (ER−, PR−, HER2−). | Multivariable logistic regression for overall YOBC and polytomous logistic regression for subtypes. All models sample-weighted. Adjustment set (final models): study site (LA/Detroit), age (20–29, 30–39, 40–49 years), household poverty level (HHP ≥ 200% vs. < 200% of federal poverty level), first-degree family history of BC (no/yes/unknown), BMI 12 months before reference date (underweight, normal, overweight, obese), lifetime cumulative alcohol intake (5 categories including abstainers), joint parity/age at FFTP (combined categories), and menopausal status (premenopausal vs. peri/post). Same covariates used for overall and subtype-specific models; BMI was also explored in sensitivity analyses (with and without adjustment, as potential mediator). | Overall YOBC (all subtypes combined): ever vs. never smoking was associated with increased YOBC risk: aOR 1.20 (95% CI 1.00–1.44). By subtype (ever vs. never): strong heterogeneity (p = 0.01). Increased risk for luminal A aOR 1.34 (1.06–1.68) and HER2-type aOR 1.97 (1.23–3.16); no association for luminal B aOR 1.04 (0.78–1.39) or TNBC aOR 0.92 (0.68–1.25). Smoking status: current vs. never significantly increased luminal A risk aOR 1.36 (1.02–1.81); former vs. never aOR 1.33 (0.98–1.71). For HER2-type, former vs. never aOR 2.41 (1.45–4.01), current vs. never aOR 1.58 (0.84–2.99). Dose/intensity: risk of HER2-type YOBC increased with higher CPD and pack-years; e.g., for HER2-type, higher categories of CPD and pack-years show monotonic rises in aORs. Age at initiation: ≥ 25 years vs. never was associated with increased YOBC overall aOR 1.91 (1.24–2.96), luminal A aOR 2.25 (1.32–3.84), and TNBC aOR 1.94 (1.03–3.64); while < 18 years vs. never increased HER2-type risk aOR 2.36 (1.36–4.09). Time since initiation: ≥ 30 years since initiation vs. never increased luminal A risk aOR 1.55 (1.07–2.26) and HER2-type aOR 2.77 (1.32–5.79). Timing vs. FFTP (parous): initiation before FFTP vs. never increased YOBC overall aOR 1.25 (1.02–1.54) and luminal A aOR 1.45 (1.11–1.89); HER2-type aOR 1.79 (0.99–3.25, not statistically significant). Little evidence of interaction by race or SEP; ever smoking increased overall YOBC, particularly among NHW women (aOR 1.39, 1.06–1.82) but not NHB (aOR 0.96, 0.70–1.31). Overall conclusion: lifetime smoking is positively associated with young-onset luminal A and HER2-type breast cancer, with no clear association for Luminal B or TNBC. |
| Peñalver-Argüeso et al., 2023 [14] | Spain; population-based multi-case-control study (MCC-Spain) in 12 provinces, with 22 collaborating hospitals for cases and population controls from general practitioner (GP) lists in the same catchment areas | Population-based case-control study. Recruitment 2008–2013. Incident, histologically confirmed invasive breast cancer cases identified soon after diagnosis from pathology/oncology services; controls randomly sampled from GP lists during the same period. No follow-up (point-in-time case-control). | Women aged 20–85 years, resident ≥ 6 months in the recruitment area, able to complete interview. Cases: incident invasive breast cancer, no previous breast cancer. Controls: cancer-free women from the same catchment population, with no history of breast cancer (or the specific tumour under study). | Total: 1,733 invasive breast cancer cases and 1,903 controls. Pathological subtype information available for 1,578 cases (91.1%): HR+ (ER or PR+, HER2−): 1,144; HER2+: 300; TN (ER−/PR−/HER2−): 134. Analyses stratified by menopausal status: pre/peri-menopausal 610 cases/547 controls; post-menopausal 1,122 cases/1,352 controls. | Standardised face-to-face interviewer-administered questionnaire. Detailed lifetime tobacco history plus sociodemographic, anthropometric (self-reported weight and height 1 year before interview), reproductive, family history and lifestyle variables. Women reporting < 100 cigarettes over their lifetime were classified as never smokers. All smoking information is self-reported. | Main smoking variables: smoking status 1 year before interview (never; former ≥ 10 years since quitting; former < 10 years; current); age at initiation (never, ≥ 18, < 18 years); duration (< 20, 20–30, > 30 years); intensity (< 15 vs. ≥ 15 cigarettes/day); pack-years (< 10, 10–25, > 25); among parous women, years smoking before first birth (< 10 or ≥ 10) and cigarettes/day before first birth (< 15 or ≥ 15). All contrasts use never smokers as reference. | All smoking variables refer to exposure up to 1 year before diagnosis (cases) or reference date (controls), to minimise reverse causation and allow a minimum latency. | ER, PR, HER2 data from hospital pathology/cancer registries. Subtypes grouped as: HR+: ER+ and/or PR+, without HER2 overexpression; HER2+: HER2 overexpressed, any ER/PR; TN: ER−, PR−, HER2−. Ki-67 not used. IHC cut-offs not explicitly detailed; classification reflects routine diagnostic practice. | Overall breast cancer models: unconditional logistic regression adjusted for age and province, then additionally for education, age at first birth, number of children, menopausal status, previous breast biopsies, family history of breast cancer and alcohol consumption. Pre/peri-menopausal models were further adjusted for oral contraceptive use; post-menopausal models were further adjusted for BMI, hormone replacement therapy and age at menopause. Subtype analyses: multinomial logistic regression (controls as reference) with the same covariate set (including menopausal/BMI/HRT terms). | All women (overall BC): smoking status 1 year before interview showed no clear association: vs. never, former ≥ 10 y OR ≈ 0.92; former < 10 y OR ≈ 0.87; current OR ≈ 1.02 (all CIs include 1; p-trend ≈ 0.97). Pre-/peri-menopausal: smoking tended to increase risk – starting smoking at ≥ 18 y OR ≈ 1.5; duration > 30 y OR ≈ 1.8; similar positive (mostly non-significant) trends for higher pack-years. Post-menopausal: smoking appeared protective, especially at low to moderate exposure and in overweight/obese women: duration > 30 y OR ≈ 0.69; intensity < 15 cig/day OR ≈ 0.70; pack-years < 10 OR ≈ 0.68; 10–25 OR ≈ 0.62 (no reduction at > 25 pack-years). In post-menopausal women, inverse associations were concentrated in BMI ≥ 25 kg/m2. By subtype: overall (all ages), no strong heterogeneity; in post-menopausal women, long-term former smoking (quit ≥ 10 y) was associated with reduced HER2+ risk (RRR ≈ 0.28) and low-intensity smoking with reduced TN risk (RRR ≈ 0.28), but numbers were small, and most other subtype-specific estimates were close to null. |
BC: breast cancer; BMI: body mass index; CBCS: Carolina Breast Cancer Study; CI: confidence interval; CPD: cigarettes per day; CT: Clinical Trial; DCIS: ductal carcinoma in situ; DMV: Department of Motor Vehicles; ER: estrogen receptor; FAP: Fundação Assistencial da Paraíba; FFTP: first full-term pregnancy; FISH: fluorescence in situ hybridisation; GP: general practitioner; HER1: human epidermal growth factor receptor 1; HER2: human epidermal growth factor receptor 2; HNL: Hospital Napoleão Laureano; HR: hazard ratio; IHC: immunohistochemistry; ISH: in situ hybridisation; LA: Los Angeles; MEC: Multiethnic Cohort; MET-h/week: metabolic equivalent task hours per week; NC: North Carolina; NHB: Non-Hispanic Black; NHW: Non-Hispanic White; OR: odds ratio; OS: observational study; PR: progesterone receptor; PY: pack-years; SEER: Surveillance, Epidemiology, and End Results Program; TN: triple-negative; TNBC: triple-negative breast cancer; UNC: University of North Carolina; USA: United States of America; WHI: Women’s Health Initiative; YWHHS: Young Women’s Health History Study.