Abstract
Background
Metabolism-related metrics have been widely investigated for their relationship with breast cancer risk but are mostly based on single values. Weight gain during adulthood has been related to an increased risk of breast cancer, while the relationship with changes in glucose and lipids remain largely unknown.
Methods
Women aged 20–80 were included from the general population-based Lifelines cohort when they had two assessments: 2007–2013 and 2014–2017. The following metrics were measured once at each of the two periods: body mass index (BMI), waist to height ratio (WHtR), hemoglobin A1c (HbA1c), HDL cholesterol (HDL-C), total cholesterol (TC), and triglyceride (TG). Women with a history of cancer, follow up less than 12 months, or who were pregnant during assessments were excluded. Mean annual changes (mean ACs) were calculated for each metric between the two periods, and further categorized into three groups - decrease, no change, and increase. Cox proportional hazards models were conducted to estimate their associations with breast cancer occurrence, reporting adjusted hazard ratios (aHR) with 95% confidence intervals (CI).
Results
During a median of 92.3 months follow-up, 1,202 of 58,785 women were diagnosed with breast cancer. Among women with a baseline BMI < 25 kg/m2, a negative association between BMI decrease and breast cancer risk was observed in contrast to their counterparts with no change (aHR: 0.75, 95% CI: 0.56–0.99). In addition, relative to the no change group, breast cancer risk was positively associated with reductions in HbA1c (aHR: 1.23, 95% CI: 1.07–1.40) and TG (aHR: 1.26, 95% CI: 1.06–1.49).
Conclusions
Changes over time in this real-world dataset from the general population highlight the benefits of weight loss and the harms of decreased glucose and TG in relation to breast cancer risk. These longitudinal patterns are affected by age, BMI, and initial values, emphasizing the importance of personalized metabolic health management.
Supplementary Information
The online version contains supplementary material available at 10.1186/s13058-025-02105-7.
Keywords: Breast cancer risk, Longitudinal changes, BMI, HbA1c, Triglyceride
Introduction
Metabolic dysfunction has emerged as a serious worldwide health concern due to lifestyle factors such as diet, alcohol, smoking and physical inactivity, and biological factors such as central obesity, and/or to non-modifiable factors like genetics and age. Rather than a definitive diagnosis, metabolic dysfunction represents a continuum of metabolic disturbances, encompassing hyperglycemia, dyslipidemia, insulin resistance, and systemic inflammation [1, 2]. Obesity is often regarded as a precursor to the emergence of metabolic dysfunction [3], and its link with breast cancer, the most common malignancy among women in the world, has been extensively studied and is well recognized [4, 5]. Not only baseline values, but also weight changes throughout adulthood were related to postmenopausal breast cancer risk [6–8]. However, body mass index (BMI) is criticized as an imperfect metric that fails to identify all metabolically unhealthy individuals. It is estimated that nearly one-third of obese individuals are metabolically healthy, usually referring to normal glucose and lipid metabolism parameters in addition to the absence of hypertension [9, 10]. Simultaneously, there was an estimated quarter of normal-weight adults to be metabolically unhealthy, typically characterized by predisposing factors such as hyperinsulinemia, hypertriglyceridemia, abdominal fat distribution and elevated blood pressure; these factors can increase the risk to subsequent development of type 2 diabetes mellitus (T2DM) and coronary artery disease [11]. Metabolic dysfunction is increasingly recognized as a significant risk factor for obesity-related cancers, independent of obesity status [2]. Thus, incorporating additional indicators of metabolic dysfunction could enhance cancer risk estimation.
Dyslipidemia has been linked to cancer progression and metastasis in animal and laboratory studies [12]. One of the constituents in a lipid panel, cholesterol, is a fundamental cell membrane component and hormones precursor [13], with inconsistent epidemiological evidence that mainly relied on single assessments. Some studies suggested a negative relation between total cholesterol (TC) and breast cancer risk [14], perhaps due to the anti-oxidative and anti-inflammatory effects of high-density lipoprotein cholesterol (HDL-C) [15–17]. Others however reported a U-shaped association between TC/HDL-C and breast cancer risk in women with T2DM [18]. Regarding triglycerides (TG), meta-analyses offer less definitive evidence, with some suggesting an inverse relationship between TG levels and breast cancer risk [19–21]. With respect to hemoglobin A1c (HbA1c), meta-analyses have failed to establish a significant link between HbA1c levels and breast cancer risk in both diabetic and nondiabetic women [22, 23]. However, a recent longitudinal study suggested a potential association between low HbA1c levels (< 5%, 5.0–5.4% versus 5.5–5.9%) and breast cancer incidence in nondiabetic individuals [24].
This study assesses associations between longitudinal changes in BMI, waist-to-hip ratio (WHtR), HbA1c, and lipid profiles and breast cancer risk in a general population-based cohort, while adjusting for variations in baseline values, BMI and menopausal status.
Methods
Lifelines and Palga
A cohort study was conducted by use of the Lifelines database, for which the design and rationale have been described in detail elsewhere [25–27]. In brief, Lifelines is a multi-disciplinary prospective population-based cohort study examining, in a unique three-generation design, the health and health-related behaviours of 167,729 persons living in the North of the Netherlands. Questionnaire data, measurements, and biological samples have been collected from the entire cohort in general assessment rounds. Up to now, there data from the baseline (2007–2013), second (2014–2017) and third (2019–2023) assessment round are available. This study is conducted according to the Declaration of Helsinki and approved by the medical ethics committee of the Universitair Medical Center Groningen (no. 2007/152) and is ISO certified (9001:2008 Healthcare). A written informed consent was collected from all participants. Since 1971, Palga has played an indispensable role as the sole data expert in the Netherlands responsible for collecting, safeguarding, and providing access to pathology data, and achieved full nationwide coverage in 1991 (https://www.palga.nl/, accessed on 4th July, 2025). The integration of Lifelines with Palga further enriched this data by linking demographics, blood/anthropometric metrics, and cancer diagnoses.
Inclusion and exclusion criteria
Women aged 20–80 with measurements of either BMI, WHtR, HbA1c, HDL-C, TC, or TG at baseline and second assessment rounds available in Lifelines were included. Woman with a history of cancer except non-melanoma skin cancer at baseline, or a follow up less than 12 months after the second assessment round, or who were pregnant at the time of either assessment round, were excluded. To account for potential differences in measurement dates, follow-up periods were calculated separately for anthropometric measures (e.g., weight, height, waist circumference) and blood measures (e.g., HbA1c, HDL-C, TC, TG). Analyses for each metric of interest included only participants with complete data at two time points. For instance, when working on the mean annual change (mean AC) in BMI, only women with BMI measured at both two time points were included in the analysis.
Baseline definition
The date of the baseline examination for anthropometric metrics (i.e., BMI and WHtR), glucose (i.e., HbA1c), or blood lipids (i.e., TC, HDL-C and TG), was defined as the index date at baseline.
Metrics of interest
The metrics of interest are defined as the mean AC in the following metabolism-related metrics: BMI, WHtR, HbA1c, HDL-C, TC and TG. In detail, the mean AC in years was calculated as: mean AC = (X2 – X1)/(t2 – t1), where X1 and X2 separately stands for the baseline and follow-up values obtained in the first and second examinations, and t1 and t2 are the dates of the first and second physical examinations. To clarify, t1 and t2 may show slight variations between anthropometric and blood metrics, and we used the exact dates corresponding to each specific exposure. The mean AC for each metric was further categorized into three groups—decrease, no change, and increase—based on standard deviation (SD). Specifically, mean AC values below − 0.5 × SD were classified as the decrease, those equal to or above 0.5 × SD as the increase, and values between these thresholds as the no change. As we focused on variations within individuals, with an expectation of a high correlation between the two measures, we selected 0.5 × SD instead of 1 × SD to achieve a balanced sample size across groups.
Outcome of interest
The outcome of interest was time to breast cancer diagnosis, including invasive and in situ carcinomas. Participants were followed from the time at the second physical examination till the diagnosis of breast cancer or death or the last date of follow up: March 16th, 2023. Mortality data were obtained from the Municipal Personal Records Database (Basisregistratie Personen, BRP), which contains personal information on all residents of the Netherlands. Death certificates are available for residents. Breast cancer diagnoses were retrieved from the Palga database, where daily pathology results are automatically submitted to. Given the reliability of both the Municipal BRP database and the Palga database, loss to follow-up is not anticipated.
Potential confounders
The following variables at baseline were collected and regarded as potential confounders. Several were categorized into three groups in tertiles, namely age in years (T1: <41; T2: 41–49; T3: ≥50), BMI in kg/m2 (T1: <23.3, T2: 23.3–26.7, T3:≥26.8), WHtR (T1: <0.47, T2: 0.47–0.52, T3: ≥0.53), HbA1c in % (T1: <5.4, T2: 5.4–5.5, T3: ≥5.6%), HDL-C in mmol/L (T1: <1.4, T2: 1.4–1.7, T3:≥1.8), TC in mmol/L (T1: <4.5, T2: 4.5–5.3, T3: ≥5.4), and TG in mmol/L (T1: <0.72, T2: 0.72–1.05, T3: ≥1.06). Education (low, middle, high) and smoking history (non-smokers, < 10, ≥10 packyears) were separately categorized in three levels; the others were treated as binary variables, namely age at menarche (≤ 13, > 13 years), menopause (yes/no), having biological children (yes/no), alcohol use (≤ 1, > 1 glasses per day), hormonal contraception (e.g., the pill, injection, or Mirena IUD; yes/no), diabetes (yes/no), hypertension (yes/no), and history of high cholesterol diagnosis (yes/no).
Statistical analyses
All data analyses were performed using R version 4.3.2. In the cohort prior to selection based on follow-up periods and complete metric data, several baseline variables had missing values. The missing rates were as follows: age at menarche (4.6%), menopausal status (19.4%), alcohol use (6.3%), smoking history (5.1%), education (1.8%), hormonal contraception use (12.5%), HDL-C (0.9%), TC (0.9%), TG (0.9%), HbA1c (1.4%), BMI (0.1%), and WHtR (0.1%). For variables with a missing rate < 10%, multiple imputations with a chained equation were used. If a missing rate was ≥ 10%, a separate category was designated for the missing values. But for menopausal status, baseline age at 50 was used as a proxy to determine the status for missing values [28]. Those aged < 50 were regarded as not stopping menstruation, while those aged ≥ 50 were regarded as stopping menstruation.
Descriptive statistics were reported as means with standard deviations (SD) or as counts with percentages. Baseline characteristics were compared between women with and without breast cancer and between the decrease/increase and no-change groups using Standardized Mean Difference (SMD) scores, excluding missing values. SMD quantifies the magnitude of baseline differences between two groups independent of sample size [29]. A notable difference between groups was defined as a positive SMD with a 95% confidence intervals (CIs) lower limit exceeding 0.1 or a negative SMD with an upper limit below − 0.1. This ensures a 95% likelihood that the effect size exceeds 0.1. SMD is used for descriptive purposes to indicate the effect size.
Cox proportional hazards models were conducted separately to test the associations of the mean ACs in BMI, WHtR, HbA1c, HDL-C, and TG with breast cancer occurrence. One model was for one metric of interest, and participants were further restricted to those who did not have missing values for that specific exposure at two assessments time points. We presented the results as adjusted hazard ratios (aHRs) with 95% CIs. The mean AC was included in models as a categorical variable with three groups: decrease, no change, and increase, with ‘no change’ being a reference category.
To understand the role of baseline values in the relation between mean AC and breast cancer risk, Model 1 only included age and baseline values of the metric of interest. Model 2 further included the mean AC of the metric. Model 3 was further adjusted for baseline values of other metabolism-related variables, in addition to Model 2. Model 4 (final model) was adjusted for diabetes, hypertension, dyslipidemia, age at menarche, menopausal status, biological children, alcohol, packyears, education, contraception in addition to Model 3. Since WHtR and BMI were both calculated with height, they were not adjusted simultaneously in the same model, and baseline BMI was used as the covariate in these models. The proportional hazards assumption was examined by the score test which revealed that age in tertiles and alcohol went against the assumption. Therefore, these two variables were treated as stratification factors in models.
To further understand the role of age, BMI and baseline values in the relation between mean ACs and breast cancer risk, subgroup analyses were conducted for these characteristics. Participants were grouped into two groups according to age at second measure at 50 years old, according to BMI at 25 kg/m2, and according to baseline values at medians. The medians for BMI, WHtR, HbA1c, HDL-C, TC and TG were 24.9 kg/m2, 0.5, 5.5%, 1.6, 5.0 and 0.87 mmol/L, respectively. Model 4 was run again in these subgroups, with baseline values of the exposure included as a continuous variable.
Invasive breast cancer exhibits more aggressive pathological features and carries a poorer prognosis compared to breast cancer in situ. To better understand how changes in these metrics are associated with invasive breast cancer specifically, a sensitivity analysis was conducted, restricting the outcome of interest to invasive cases only.
Results
Cohort characteristics
The patient selection procedure is shown in Fig. 1. At baseline, the Lifelines database contained a total of 94,511 females. In the cohort selected based on follow-up periods for anthropometric measurements, 58,785 women were included before applying criteria for complete metric data. The median time interval between two measures was 3.84 years (interquartile range: 3.09–4.59 years). After up to 111.6 months of follow-up (mean: 90.4 months; median: 92.3 months), 1,202 women had been diagnosed with breast cancer. Women with breast cancer were more likely to be older (49.70 vs. 45.54 years), post-menopausal (46.92% vs. 36.60%), having biological children (76.37% vs. 68.62%), and more frequently received lower education (36.52% vs. 28.26%, Table 1) compared to those without breast cancer. All the blood and anthropometric measures remained comparable between both groups, as indicated by the absolute lower/upper limit of the 95% CI for the SMD being > 0.1.
Fig. 1.
Flow chart for patients selection. Follow-up periods were calculated separately for anthropometric measures (e.g., weight, height, waist circumference) and blood measures (e.g., HbA1c, HDL-C, TC, TG) due to potential differences in measurement dates. Analyses for each exposure of interest included only complete data for two measurements at two time points
Table 1.
Distribution of baseline characteristics between women with and without breast cancer
| Baseline characteristics | mean (SD) / number (proportion, %) | SMD | |||
|---|---|---|---|---|---|
| non-BC (n = 57,583) | BC (n = 1,202) | ||||
| age (continuous, years) | 45.54 (12.11) | 49.70 (10.51) | -0.37 (-0.42, -0.31) | ||
| age (categorical, years) | |||||
| < 41 | 19,958 (34.66) | 227 (18.89) | 0.37 (0.32, 0.43) | ||
| 41–49 | 18,730 (32.53) | 435 (36.19) | |||
| ≥ 50 | 18,895 (32.81) | 540 (44.93) | |||
| age at menarche (continuous, years) | 13.10 (1.46) | 13.02 (1.41) | 0.06 (-0.00, 0.11) | ||
| age at menarche (categorical, years) | |||||
| < 13 | 34,865 (60.55) | 768 (63.89) | 0.06 (0.00, 0.12) | ||
| ≥ 13 | 20,080 (34.87) | 388 (32.28) | |||
| missing | 2638 (4.58) | 46 (3.83) | |||
| menopause | |||||
| no | 36,505 (63.40) | 638 (53.08) | 0.21 (0.15, 0.27) | ||
| yes | 21,078 (36.60) | 564 (46.92) | |||
| biological children | |||||
| no | 18,067 (31.38) | 284 (23.63) | -0.17 (-0.23, -0.12) | ||
| yes | 39,516 (68.62) | 918 (76.37) | |||
| diabetes | |||||
| no | 56,247 (97.68) | 1173 (97.59) | -0.01 (-0.06, 0.05) | ||
| yes | 1336 (2.32) | 29 (2.41) | |||
| hypertension | |||||
| no | 44,684 (77.60) | 896 (74.54) | -0.07 (-0.13, -0.01) | ||
| yes | 12,899 (22.40) | 306 (25.46) | |||
| cardiovascular disease | |||||
| no | 56,582 (98.26) | 1173 (97.59) | -0.05 (-0.10, 0.01) | ||
| yes | 1001 (1.74) | 29 (2.41) | |||
| high cholesterol diagnosis | |||||
| no | 51,582 (89.58) | 1052 (87.52) | -0.06 (-0.12, -0.01) | ||
| yes | 6001 (10.42) | 150 (12.48) | |||
| alcohol (glasses per day) | |||||
| ≤ 1 | 26,078 (45.29) | 554 (46.09) | 0.01 (-0.05, 0.07) | ||
| > 1 | 27,893 (48.44) | 577 (48.00) | |||
| missing | 3612 (6.27) | 71 (5.91) | |||
| smoking (packyears) | |||||
| 0 | 27,563 (47.87) | 510 (42.43) | 0.13 (0.07, 0.19) | ||
| ≥ 0, < 10 | 16,643 (28.90) | 353 (29.37) | |||
| ≥ 10 | 10,455 (18.16) | 269 (22.38) | |||
| missing | 2922 (5.07) | 70 (5.82) | |||
| education | |||||
| low | 16,275 (28.26) | 439 (36.52) | 0.18 (0.13, 0.24) | ||
| middle | 23,064 (40.05) | 414 (34.44) | |||
| high | 17,215 (29.90) | 322 (26.79) | |||
| missing | 1029 (1.79) | 27 (2.25) | |||
| contraception | |||||
| no | 4093 (7.11) | 90 (7.49) | 0.09 (0.03, 0.14) | ||
| yes | 46,343 (80.48) | 928 (77.20) | |||
| missing | 7147 (12.41) | 184 (15.31) | |||
| BMI (continuous, kg/m2) | 25.75 (4.60) | 26.27 (4.48) | -0.12 (-0.17, -0.06) | ||
| BMI (categorical, kg/m2) | |||||
| < 23.3 | 19,071 (33.12) | 349 (29.03) | 0.11 (0.05, 0.17) | ||
| 23.3–26.7 | 19,431 (33.74) | 397 (33.03) | |||
| ≥ 26.8 | 19,049 (33.08) | 456 (37.94) | |||
| missing | 32 (0.06) | 0 (0.00) | |||
| WHtR (continuous) | 0.51 (0.07) | 0.52 (0.07) | -0.16 (-0.22, -0.10) | ||
| WHtR (categorical) | |||||
| < 0.47 | 18,191 (31.59) | 306 (25.46) | 0.14 (0.08, 0.20) | ||
| 0.47–0.52 | 18,903 (32.83) | 415 (34.53) | |||
| ≥ 0.53 | 20,455 (35.52) | 481 (40.02) | |||
| missing | 34 (0.06) | 0 (0.00) | |||
| HbA1c (continuous, %) | 5.53 (0.42) | 5.60 (0.45) | -0.15 (-0.21, -0.09) | ||
| HbA1c (categorical, %) | |||||
| < 5.4 | 17,548 (30.47) | 307 (25.54) | 0.16 (0.10, 0.22) | ||
| 5.4–5.5 | 14,557 (25.28) | 274 (22.80) | |||
| ≥ 5.6 | 24,639 (42.79) | 610 (50.75) | |||
| missing | 839 (1.46) | 11 (0.92) | |||
| HDL-C (continuous, mmol/L) | 1.63 (0.40) | 1.62 (0.40) | 0.02 (-0.03, 0.08) | ||
| HDL-C (categorical, mmol/L) | |||||
| < 1.4 | 14,384 (24.98) | 325 (27.04) | 0.05 (-0.01, 0.10) | ||
| 1.4–1.7 | 23,257 (40.39) | 478 (39.77) | |||
| ≥ 1.8 | 19,415 (33.72) | 393 (32.70) | |||
| missing | 527 (0.92) | 6 (0.50) | |||
| TC (continuous, mmol/L) | 5.04 (1.00) | 5.18 (1.05) | -0.13 (-0.19, -0.07) | ||
| TC (categorical, mmol/L) | |||||
| < 4.6 | 19,017 (33.03) | 342 (28.45) | 0.11 (0.05, 0.16) | ||
| 4.6–5.3 | 18,094 (31.42) | 394 (32.78) | |||
| ≥ 5.4 | 19,945 (34.64) | 460 (38.27) | |||
| missing | 527 (0.92) | 6 (0.50) | |||
| TG (continuous, mmol/L) | 1.01 (0.56) | 1.09 (0.65) | -0.13 (-0.19, -0.07) | ||
| TG (categorical, mmol/L) | |||||
| < 0.72 | 18,385 (31.93) | 351 (29.20) | 0.08 (0.02, 0.14) | ||
| 0.72–1.05 | 19,286 (33.49) | 395 (32.86) | |||
| ≥ 1.06 | 19,385 (33.66) | 450 (37.44) | |||
| missing | 527 (0.92) | 6 (0.50) | |||
Abbreviation: BMI: body mass index; BC: breast cancer; HbA1c: hemoglobin A1c; HDL-C: HDL cholesterol; SMD: standardized mean difference; TC: total cholesterol; TG: triglyceride; and WHtR: waist to height ratio
Associations between mean ACs and breast cancer occurrence
The mean AC for BMI, WHtR, HbA1c, HDL-C, TC and TG were divided into three categories - decrease, no-change, and increase, respectively. The number of sample sizes in each category and the comparisons of baseline characteristics between the increase/decrease and the no change group are described in Supplementary Materials, and summarized in Table S1-S6 and Fig. S1.
BMI and WHtR
After adjustment in the Cox models, BMI AC was not significantly associated with breast cancer risk overall (aHR: 0.89, 95% CI: 0.76–1.04, Fig. 2and Table 2). Estimates remained similar in sensitivity analyses (Table S7). However, in the subgroup analyses, there was a positive association between BMI decrease and breast cancer risk in women with a baseline BMI < 25 kg/m2 (aHR: 0.75, 95% CI: 0.56–0.99, Fig. 3 and Table S9), indicating a possible protective effect of weight loss. Regarding WHtR, no significant associations were found neither in the whole cohort or in the subgroup analyses.
Fig. 2.
Bar plot and forest plot based on the primary Cox PH analyses to investigate the relation between longitudinal changes of metabolic-related metrics and breast cancer risk. meanAC values were divided into three categories based on ± 0.5 * SD, assuming minimal change for most participants. Baseline values were grouped into tertiles. The bar plot in middle displays the mean of meanAC values and baseline values in each category, with units as follows: kg/m² for BMI, % for HbA1c, and mmol/L for HDL-C, TC, and TG (no unit for WHtR). The forest plot on the right represents fully adjusted Cox PH models, accounting for age, baseline values, age at menarche, menopausal status, biological children, diabetes, hypertension, high cholesterol diagnosis, alcohol, smoking, education, and hormonal contraception. For WHtR, baseline values used for adjustment include WHtR, HbA1c, HDL-C, TC, and TG, while for other exposures, they include BMI, HbA1c, HDL-C, TC, and TG. Abbreviation: BMI: body mass index; BC: breast cancer; Expo.: exposures; HR: hazard ratio; HbA1c: hemoglobin A1c; HDL-C: HDL cholesterol; meanAC: mean annual change; Ref.: reference; T1 – T3: the first, second and third tertile of baseline values; TC: total cholesterol; TG: triglyceride; and WHtR: waist to height ratio
Table 2.
Cox proportional hazards models in the whole dataset
| Characteristics | univariate analysis: meanAC | Model 1: baseline + age |
Model 2 | Model 3 | Model 4 | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Sample size (comparison / reference) |
HR (95% CI) | P | HR (95% CI) | P | HR (95% CI) | P | HR (95% CI) | P | HR (95% CI) | P | ||||||
| BMI | ||||||||||||||||
| meanAC | ||||||||||||||||
| decrease | 11,330 (221) / 31,586 (681) | 0.90 (0.78, 1.05) | 0.190 | - | - | 0.88 (0.75, 1.03) | 0.105 | 0.88 (0.75, 1.02) | 0.097 | 0.89 (0.76, 1.04) | 0.143 | |||||
| increase | 14,634 (300) / 31,586 (681) | 0.95 (0.83, 1.08) | 0.422 | - | - | 0.99 (0.87, 1.14) | 0.937 | 0.99 (0.87, 1.14) | 0.936 | 1.00 (0.87, 1.14) | 0.954 | |||||
| baseline | ||||||||||||||||
| T2 | - | - | - | 1.00 (0.86, 1.15) | 0.949 | 1.00 (0.87, 1.16) | 0.967 | 0.99 (0.85, 1.15) | 0.891 | 0.97 (0.84, 1.13) | 0.714 | |||||
| T3 | - | - | - | 1.13 (0.98, 1.30) | 0.101 | 1.15 (1.00, 1.33) | 0.058 | 1.10 (0.94, 1.28) | 0.242 | 1.08 (0.92, 1.26) | 0.346 | |||||
| WHtR | ||||||||||||||||
| meanAC | ||||||||||||||||
| decrease | 15,685 (332) / 25,375 (522) | 1.01 (0.88, 1.16) | 0.903 | - | - | 1.01 (0.87, 1.16) | 0.940 | 1.01 (0.88, 1.16) | 0.872 | 1.03 (0.89, 1.18) | 0.711 | |||||
| increase | 16,466 (346) / 25,375 (522) | 1.03 (0.90, 1.18) | 0.674 | - | - | 1.03 (0.90, 1.18) | 0.636 | 1.03 (0.90, 1.18) | 0.680 | 1.03 (0.90, 1.19) | 0.631 | |||||
| baseline | ||||||||||||||||
| T2 | - | - | - | 1.12 (0.96, 1.30) | 0.139 | 1.12 (0.96, 1.30) | 0.135 | 1.10 (0.94, 1.28) | 0.230 | 1.07 (0.92, 1.25) | 0.398 | |||||
| T3 | - | - | - | 1.09 (0.94, 1.26) | 0.272 | 1.09 (0.94, 1.27) | 0.264 | 1.02 (0.87, 1.21) | 0.774 | 0.99 (0.84, 1.18) | 0.936 | |||||
| HbA1c | ||||||||||||||||
| meanAC | ||||||||||||||||
| decrease | 16,620 (391) / 28,408 (561) | 1.15 (1.01, 1.31) | 0.030 | 1.20 (1.06, 1.37) | 0.006 | 1.20 (1.05, 1.37) | 0.007 | 1.23 (1.07, 1.40) | 0.003 | |||||||
| increase | 8139 (172) / 28,408 (561) | 1.04 (0.87, 1.23) | 0.690 | 0.97 (0.81, 1.15) | 0.713 | 0.95 (0.80, 1.13) | 0.563 | 0.97 (0.81, 1.15) | 0.731 | |||||||
| baseline | ||||||||||||||||
| T2 | - | - | - | 0.94 (0.79, 1.12) | 0.482 | 0.92 (0.77, 1.09) | 0.323 | 0.92 (0.77, 1.09) | 0.318 | 0.91 (0.77, 1.08) | 0.300 | |||||
| T3 | - | - | - | 1.06 (0.91, 1.23) | 0.452 | 1.00 (0.86, 1.17) | 0.979 | 0.99 (0.84, 1.16) | 0.855 | 0.99 (0.85, 1.17) | 0.929 | |||||
| HDL-C | ||||||||||||||||
| meanAC | ||||||||||||||||
| decrease | 13,060 (262) / 24,669 (507) | 1.00 (0.86, 1.17) | 0.962 | - | - | 1.03 (0.88, 1.20) | 0.727 | 1.03 (0.89, 1.20) | 0.697 | 1.04 (0.90, 1.22) | 0.574 | |||||
| increase | 15,863 (360) / 24,669 (507) | 1.05 (0.92, 1.21) | 0.448 | - | - | 1.06 (0.93, 1.21) | 0.391 | 1.06 (0.93, 1.21) | 0.399 | 1.06 (0.93, 1.22) | 0.373 | |||||
| baseline | ||||||||||||||||
| T2 | - | - | - | 0.88 (0.76, 1.02) | 0.089 | 0.88 (0.76, 1.02) | 0.091 | 0.89 (0.77, 1.04) | 0.151 | 0.89 (0.76, 1.03) | 0.122 | |||||
| T3 | - | - | - | 0.82 (0.70, 0.95) | 0.009 | 0.82 (0.70, 0.95) | 0.011 | 0.84 (0.70, 1.00) | 0.051 | 0.83 (0.69, 0.99) | 0.039 | |||||
| TC | ||||||||||||||||
| meanAC | ||||||||||||||||
| decrease | 12,627 (259) / 26,583 (575) | 0.95 (0.82, 1.10) | 0.514 | - | - | 0.92 (0.79, 1.07) | 0.296 | 0.91 (0.78, 1.06) | 0.244 | 0.93 (0.80, 1.09) | 0.362 | |||||
| increase | 14,382 (295) / 26,583 (575) | 0.94 (0.81, 1.08) | 0.356 | - | - | 0.88 (0.76, 1.01) | 0.079 | 0.88 (0.76, 1.01) | 0.070 | 0.88 (0.76, 1.01) | 0.073 | |||||
| baseline | ||||||||||||||||
| T2 | - | - | - | 1.02 (0.87, 1.19) | 0.818 | 1.01 (0.87, 1.18) | 0.852 | 1.04 (0.89, 1.22) | 0.623 | 1.03 (0.88, 1.20) | 0.752 | |||||
| T3 | - | - | - | 0.90 (0.77, 1.06) | 0.205 | 0.90 (0.76, 1.06) | 0.209 | 0.93 (0.78, 1.11) | 0.412 | 0.90 (0.76, 1.08) | 0.262 | |||||
| TG | ||||||||||||||||
| meanAC | ||||||||||||||||
| decrease | 9110 (231) / 32,012 (640) | 1.24 (1.06, 1.44) | 0.006 | - | - | 1.23 (1.04, 1.46) | 0.014 | 1.23 (1.04, 1.46) | 0.015 | 1.26 (1.06, 1.49) | 0.008 | |||||
| increase | 12,470 (258) / 32,012 (640) | 1.05 (0.90, 1.21) | 0.550 | - | - | 0.98 (0.85, 1.14) | 0.816 | 0.97 (0.83, 1.12) | 0.643 | 0.97 (0.84, 1.13) | 0.719 | |||||
| baseline | ||||||||||||||||
| T2 | - | - | - | 0.95 (0.82, 1.10) | 0.476 | 0.93 (0.80, 1.08) | 0.326 | 0.91 (0.78, 1.06) | 0.217 | 0.91 (0.78, 1.06) | 0.220 | |||||
| T3 | - | - | - | 1.03 (0.89, 1.19) | 0.712 | 0.95 (0.80, 1.11) | 0.494 | 0.88 (0.73, 1.06) | 0.175 | 0.88 (0.73, 1.06) | 0.185 | |||||
Model 1: only age and baseline values of the metric of interest; Model 2: Model 1 + the mean AC of the metric; Model 3: baseline values of other metabolism-related variables, in addition to Model 2. Model 4 (final model): diabetes, hypertension, dyslipidemia, age at menarche, menopausal status, biological children, alcohol, packyears, education, contraception in addition to Model 3
Abbreviation: BMI: body mass index; HbA1c: hemoglobin A1c; HDL-C: HDL cholesterol; HR: hazard ratio; meanAC: mean annual change; SD: standard deviation; SMD: standardized mean difference; T1 – T3: the first, second and third tertile of baseline values; TC: total cholesterol; TG: triglyceride; and WHtR: waist to height ratio
Fig. 3.
Subgroup Cox PH analyses to investigate the relation between longitudinal changes of metabolic-related metrics and breast cancer risk. Participants were divided into subgroups by age at second measurement (50 years) or BMI (25 kg/m²), and fully adjusted Cox PH models were performed separately within each subgroup. Adjustments included age, baseline values, age at menarche, menopausal status, biological children, diabetes, hypertension, high cholesterol diagnosis, alcohol, smoking, education, and hormonal contraception. For WHtR, baseline values for adjustment included WHtR, HbA1c, HDL-C, TC, and TG; for other exposures, baseline values included BMI, HbA1c, HDL-C, TC, and TG. In BMI subgroups, baseline BMI was included as a continuous rather than categorical variable in the model. Abbreviation: BMI: body mass index; BC: breast cancer; Expo.: exposures; HR: hazard ratio; HbA1c: hemoglobin A1c; HDL-C: HDL cholesterol; MAC: mean annual change; Ref.: reference; T1 – T3: the first, second and third tertile of baseline values; TC: total cholesterol; TG: triglyceride; and WHtR: waist to height ratio
HbA1c, HDL-C, TC, and TG
As shown in Fig. 2, and Table 2, the whole cohort analysis indicated an association between reduction in HbA1c levels and an increased breast cancer risk (aHR: 1.23, 95% CI: 1.07–1.40). This association was particularly evident in a subgroup of women aged ≥ 50 years (aHR: 1.23, 95% CI: 1.04–1.47, Fig. 3 and Table S8), those with a baseline BMI < 25 kg/m² (aHR: 1.33, 95% CI: 1.09–1.62, Fig. 3 and Table S9), and women with baseline HbA1c levels < 5.5% (aHR: 1.38, 95% CI: 1.08–1.76, Fig. S2 and Table S10). No association was observed for HDL-C mean AC and TC mean AC. Regarding TG, a reduction was associated with a higher breast cancer risk (aHR: 1.26, 95% CI: 1.06–1.49, Fig. 2and Table 2), particularly among older women (aHR: 1.29, 95% CI: 1.05–1.60, Fig. 3 and Table S8), women with a baseline BMI ≥ 25 kg/m² (aHR: 1.34, 95% CI: 1.08–1.67, Fig. 3 and Table S9), or women with baseline TG ≥ 0.87 mmol/L (aHR: 1.22, 95% CI: 1.01–1.48, Fig. S2 and Table S10). Estimates in the primary analyses remained similar in sensitivity analyses (Table S7).
Discussion
Summary
This study confirmed that, within a general population-based cohort, weight loss was associated with a reduced risk of breast cancer in women with a BMI < 25 kg/m2. Additionally, reductions in HbA1c and TG were linked to an increased risk of breast cancer.
Weight gain is widely acknowledged as a risk factor for breast cancer across diverse ethnic groups, with extensive evidence highlighting its long-term effects that manifest years after the weight gain [7, 8]. However, this association was not observed in our study, likely due to differences in weight change definitions and timing. This study used BMI instead of weight to estimate annual changes; assuming heights of 160–180 cm, BMI increases corresponded to annual weight gains of 1.6–2.0 kg, while BMI decrease reflected weight losses of 1.7–2.1 kg. In comparison, the EPIC [8] and HUNT studies [30] found significant associations between weight gain, i.e., 5–10 kg or > 10 kg in EPIC, and 5.0–7.49 kg or ≥ 7.5 kg per 10 years in HUNT, and breast cancer risk, relative to stable weight (± 2.5 kg). Notably, these studies involved longer intervals between measurements (> 20 years), whereas the Lifelines cohort focused on changes over a shorter period (mean interval: 4 years). Timing also differed: EPIC assessed BMI changes from age 20, HUNT from an average age of ~ 40, and Lifelines from later adulthood, with a mean baseline age of 46 years. The HUNT study further reported that weight gain before or around menopause increased risk, while later weight gain showed no clear association [30]. In our study, BMI increase was not significantly associated with breast cancer risk, even in subgroup analyses by age 50. Although we did not differentiate between pre- and postmenopausal breast cancer, most patients in our cohort were diagnosed after age 50 (78%), indicating our findings primarily reflected postmenopausal cases. Interestingly, baseline BMI showed an age-dependent relationship with breast cancer risk, that is consistent with prior studies [31, 32], highlighting the complex interplay between BMI, weight gain timing, and breast cancer risk.
In this study, an inverse relationship between weight loss and breast cancer risk was observed, particularly in women with a BMI < 25, a finding not observed in the EPIC or HUNT studies. However, supporting evidence comes from the Pooling Project of Prospective Studies of Diet and Cancer, where sustained weight loss, even modest (> 2–4.5, > 4.5–<9, or ≥ 9 kg) compared to stable weight (± 2 kg) was associated with reduced breast cancer risk in women aged 50 years and older [33]. Similarly, the WHI Observational Study included women aged 50 years and older, and found that women who lost ≥ 5% of their weight had a lower breast cancer risk compared with those with stable weight [34]. Notably, as we did not observe an association between BMI decrease and cancer risk in women ≥ 25 kg/m2, the Cancer Prevention Study-II (CPS-II) Nutrition Cohort also revealed no association between weight loss and postmenopausal breast cancer risk among overweight and obese women [35]. Moreover, one early study found that for women who reached their highest weight before age 45, weight loss after that age was associated with a reduced risk of postmenopausal breast cancer. In contrast, weight loss in women who reached their highest weight after age 45 showed no association with risk [36]. This highlighted the timing of weight loss [37], while these patterns were not evident in subgroup analyses based on age 50. Clinical trials have consistently shown that weight loss additionally resulted in maintenance of lean mass, greater fitness, greater fat loss and favorable effects on some sex hormones [38, 39]. Our findings emphasized the importance of weight control in middle adulthood for breast cancer prevention. Future research should explore the role of sustained weight loss in breast cancer prevention, simultaneously considering the timing, baseline BMI, menopausal status.
Decrease in blood metrics associated with an increased breast cancer risk
HbA1C
HbA1c reflects an average of three-month blood sugar level. In contrast to high levels, less attention has been paid to low levels of HbA1c. We observed that a reduction in HbA1c is related to an increased risk of breast cancer, particularly in women aged ≥ 50 years old, < 25 kg/m2, or those with baseline HbA1c levels < 5.5%. This finding is in disagreement with in vitro evidence suggesting hyperglycemia is associated with carcinogenic effects [40] and previous epidemiological findings of no clear relation between baseline HbA1c and breast cancer risk among women with diabetes [18]. This may be partially due to the different populations studied: the previous study focused on individuals with diabetes with mean HbA1c around 7% [18], while this study examined the general population with median Hba1c of 5.5%. In a large Japanese cohort of nondiabetic participants, repeated HbA1c values were collected. An increased breast cancer incidence in those with HbA1c values of 5.0–5.4% and below 5.0% was found, compared to a reference range of 5.5–5.9%, suggesting that low HbA1c may be associated with an increased breast cancer risk [24]. In addition, a high risk of all cancers, especially liver cancer, was also seen in the group with lower HbA1c (levels < 5.0%), compared with individuals without known diabetes and HbA1c levels of 5.0-5.4% [41]. By focusing on changes between two sequential measurements, this study further added to the evidence that a reduction in already normal HbA1c might be related to an increased breast cancer risk.
While high glucose levels are associated with cancer risk, the potential link between low HbA1c and cancer is not well understood. Low HbA1c levels might reflect compliance with diet, underlying health issues such as malnutrition, anemia, or organ dysfunction [42, 43]. Impaired hepatic function, for instance, which can increase red cell turnover and destruction of red cells via hypersplenism, leading to (falsely) lowering of HbA1c levels relative to their blood glucose [41]. Another potential mechanism may involve insulin-like growth factor I (IGF-I), which enhances insulin sensitivity and lowers blood glucose in humans [44]. However, high serum IGF-I levels are also associated with an increased risk of breast cancer [45]. While causality cannot be established, these findings suggest that a decrease in HbA1c might (in)directly indicate an increased risk of an occult neoplasm or a future malignancy diagnosis.
Triglyceride
The relationship between TG and cancer remains debated in both the general population and individuals with T2DM. In this study, the association between decreasing TG levels and an increased risk of breast cancer was observed. In agreement with this, a Chinese cohort reported that TG levels below 1.70 mmol/L were associated with an increased cancer risk [46]. Furthermore, a meta-analysis and a recent two-sample Mendelian randomization study indicated that serum TG levels or genetically elevated TG levels might be inversely associated with breast cancer risk [47, 48].
Possible explanations include higher baseline TG levels and concomitantly lower baseline HDL-C levels in the TG decrease group compared to the other groups. Although models were adjusted for these variables, residual confounding cannot be entirely ruled out. Additionally, TG may have protective effects by countering the toxic accumulation of saturated lipids through the release of the unsaturated fatty acid oleate from lipid droplets into phospholipid pools [49]. Considering the glycolytic pathway enzymes are effectively an extension of lipogenesis, and TG storage in adipose tissue is crucial for energy homeostasis [50], a significant reduction in triglyceride levels may indicate malnutrition or linked to low glucose and insulin levels, that could increase breast cancer risk.
Strengths and limitations
Lifelines is a large general population cohort that provides extensive and detailed information on both physical and mental health, and Palga is a nationwide network and registry of histo- and cytopathology in the Netherlands that covers nearly 100% of pathologically confirmed cancers in the Netherlands. The linkage of these two databases enables the study of longitudinal changes associated with the development of cancer. In addition, a comprehensive set of laboratory confounders was included, with baseline values categorized to minimize the impact of extreme values, enhance clinical interpretability, and address potential violations of the PH assumption. Despite these strengths, several limitations warrant careful consideration.
First, the time intervals between two measures varied among individuals. The proportion of the time intervals between two blood measures ranging from three to five years measures was 61%. To account for this, mean ACs were calculated. Thus, interpretations should be cautious and not extend beyond this time interval. Second, the latency period used may not have been long enough to accurately detect incident breast cancer [51], and reverse causality bias cannot be entirely ruled out. To account for that, we excluded patients who got breast cancer diagnoses within 12 months after the second measurement. Nevertheless, this may not fully exclude the possibility that these observed longitudinal changes might be related to the progression of existing undetected cancer in some individuals. Third, information on the use of glucose-lowering medications and statins was not available. However, given the low prevalence of diabetes and cardiovascular disease (~ 2%) in this cohort, the absence of medication data likely did not significantly impact our findings. Fourth, as estrogen receptor status is not structurally collected, we do not have these data for all breast cancers. So, we are unable to further explore the potential modification of estrogen receptor in the associations between changes of these metrics (i.e., BMI) and breast cancer risk [52]. Finally, residual confounding cannot be excluded, even though our analyses were adjusted for a wide range of confounders.
Conclusions
Weight loss appears to be linked to a reduced risk of breast cancer in women with a BMI of < 25 kg/m2. Reductions in HbA1c, especially in women with an age ≥ 50 and a BMI of < 25 kg/m2 and those who had baseline HbA1c values of < 5.5%, were related to an increased breast cancer risk, as were reductions in TG, particularly in women with an age of ≥ 50 and a BMI of > 25 kg/m2. These findings highlight the benefits of weight loss, and the complex regulation of glucose and lipid metabolism, and emphasize the importance of maintaining balanced and stable HbA1c and TG levels to decrease breast cancer risk.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
The authors wish to acknowledge the services of the Lifelines Cohort Study, the contributing research centers delivering data to Lifelines, and all the study participants.
Abbreviations
- aHR
Adjusted hazard ratios
- BMI
Body mass index
- 95% CI
95% confidence intervals
- mean ACs
Mean annual changes
- HbA1c
Hemoglobin A1c
- HDL-C
High-density lipoprotein cholesterol
- IGF-I
Insulin-like growth factor I
- SD
Standard deviation
- SMD
Standardized mean difference
- TC
Total cholesterol
- T2DM
Type 2 diabetes mellitus
- TG
Triglyceride
- WHtR
Waist to height ratio
Author contributions
F.Z. contributed in Data curation, Methodology, Formal Analysis, and Writing – original draft. G.S. contributed in Conceptualization, Methodology, Supervision and Writing – review and editing. G.W.L. contributed in Methodology, Supervision and Writing – review and editing. G.H.dB., B.vdV. and Q.Z. contributed in Supervision and Writing – review and editing. All authors read and approved the final manuscript.
Funding
Fan Zhang was supported by the Graduate School of Medical Sciences, University of Groningen, University Medical Center Groningen: a PhD position grant; and the Special Fund for Science and Technology of Guangdong Province in 2020 [grant number 200110115891683].
Data availability
Data may be obtained from a third party and are not publicly available. Researchers can apply to use the Lifelines data used in this study. More information about how to request Lifelines data and the conditions of use can be found on their website (https://www.lifelines-biobank.com/researchers/working-with-us/step-1-prepare-and-submit-your-application).
Declarations
Ethics approval and consent to participate
This study is conducted according to the Declaration of Helsinki and approved by the medical ethics committee of the Universitair Medical Center Groningen (no. 2007/152) and is ISO certified (9001:2008 Healthcare). A written informed consent was collected from all participants.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Wildman RP, Muntner P, Reynolds K, McGinn AP, Rajpathak S, Wylie-Rosett J, et al. The obese without cardiometabolic risk factor clustering and the normal weight with cardiometabolic risk factor clustering: prevalence and correlates of 2 phenotypes among the US population (NHANES 1999–2004). Arch Intern Med. 2008;168:1617–24. [DOI] [PubMed] [Google Scholar]
- 2.Karra P, Winn M, Pauleck S, Bulsiewicz-Jacobsen A, Peterson L, Coletta A, et al. Metabolic dysfunction and obesity-related cancer: beyond obesity and metabolic syndrome. Obes (Silver Spring). 2022;30:1323–34. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.O’Neill S, O’Driscoll L. Metabolic syndrome: a closer look at the growing epidemic and its associated pathologies. Obes Rev. 2015;16:1–12. [DOI] [PubMed] [Google Scholar]
- 4.Chen J. Multiple signal pathways in obesity-associated cancer. Obes Rev. 2011;12:1063–70. [DOI] [PubMed] [Google Scholar]
- 5.Chiappetta S, Bottino V. Obesity-associated cancer prevention. Lancet Healthy Longev. 2023;4:e520–1. [DOI] [PubMed] [Google Scholar]
- 6.Friedenreich CM. Review of anthropometric factors and breast cancer risk. Eur J Cancer Prev. 2001;10:15–32. [DOI] [PubMed] [Google Scholar]
- 7.Liu Y, Warren Andersen S, Wen W, Gao YT, Lan Q, Rothman N, et al. Prospective cohort study of general and central obesity, weight change trajectory and risk of major cancers among Chinese women. Int J Cancer. 2016;139:1461–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Ellingjord-Dale M, Christakoudi S, Weiderpass E, Panico S, Dossus L, Olsen A, et al. Long-term weight change and risk of breast cancer in the European prospective investigation into cancer and nutrition (EPIC) study. Int J Epidemiol. 2022;50:1914–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lin H, Zhang L, Zheng R, Zheng Y. The prevalence, metabolic risk and effects of lifestyle intervention for metabolically healthy obesity: a systematic review and meta-analysis: A PRISMA-compliant Article. Med (Baltim). 2017;96:e8838. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Bluher M. Metabolically healthy obesity. Endocr Rev. 2020;41:bnaa004. [DOI] [PMC free article] [PubMed]
- 11.Mathew H, Farr OM, Mantzoros CS. Metabolic health and weight: Understanding metabolically unhealthy normal weight or metabolically healthy obese patients. Metabolism. 2016;65:73–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Ben Hassen C, Goupille C, Vigor C, Durand T, Gueraud F, Silvente-Poirot S, et al. Is cholesterol a risk factor for breast cancer incidence and outcome? J Steroid Biochem Mol Biol. 2023;232:106346. [DOI] [PubMed] [Google Scholar]
- 13.Cui J, Shen Y, Li R. Estrogen synthesis and signaling pathways during aging: from periphery to brain. Trends Mol Med. 2013;19:197–209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Murdock DJ, Sanchez RJ, Mohammadi KA, Fazio S, Geba GP. Serum cholesterol and the risk of developing hormonally driven cancers: A narrative review. Cancer Med. 2023;12:6722–67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Jafri H, Alsheikh-Ali AA, Karas RH. Baseline and on-treatment high-density lipoprotein cholesterol and the risk of cancer in randomized controlled trials of lipid-altering therapy. J Am Coll Cardiol. 2010;55:2846–54. [DOI] [PubMed] [Google Scholar]
- 16.Pirro M, Ricciuti B, Rader DJ, Catapano AL, Sahebkar A, Banach M. High density lipoprotein cholesterol and cancer: marker or causative? Prog Lipid Res. 2018;71:54–69. [DOI] [PubMed] [Google Scholar]
- 17.Mazzuferi G, Bacchetti T, Islam MO, Ferretti G. High density lipoproteins and oxidative stress in breast cancer. Lipids Health Dis. 2021;20:143. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Zhang F, de Bock GH, Denig P, Landman GW, Zhang Q, Sidorenkov G. Role of serum lipids, blood glucose and blood pressure in breast cancer risk for women with type 2 diabetes mellitus. Clin Epidemiol. 2023;15:109–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Ma HQ, Cui LH, Li CC, Yu Z, Piao JM. Effects of serum triglycerides on prostate cancer and breast cancer risk: A Meta-Analysis of prospective studies. Nutr Cancer. 2016;68:1073–82. [DOI] [PubMed] [Google Scholar]
- 20.Nouri M, Mohsenpour MA, Katsiki N, Ghobadi S, Jafari A, Faghih S et al. Effect of serum lipid profile on the risk of breast cancer: systematic review and Meta-Analysis of 1,628,871 women. J Clin Med. 2022;11:4503. [DOI] [PMC free article] [PubMed]
- 21.Amerizadeh A, Vaseghi G, Farajzadegan Z, Asgary S. An updated systematic review and Meta-Analysis on association of serum lipid profile with risk of breast cancer incidence. Int J Prev Med. 2022;13:142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Boyle P, Koechlin A, Pizot C, Boniol M, Robertson C, Mullie P, et al. Blood glucose concentrations and breast cancer risk in women without diabetes: a meta-analysis. Eur J Nutr. 2013;52:1533–40. [DOI] [PubMed] [Google Scholar]
- 23.Lin C, Cai X, Yang W, Lv F, Nie L, Ji L. Glycemic control and the incidence of neoplasm in patients with type 2 diabetes: a meta-analysis of randomized controlled trials. Endocrine. 2020;70:232–42. [DOI] [PubMed] [Google Scholar]
- 24.Kobayashi D, Kuriyama N, Noto H, Aida A, Takahashi O, Shimbo T. Development of malignancies and changes in time-dependent hemoglobin A1c among a nondiabetic population: longitudinal analysis. Acta Diabetol. 2020;57:189–96. [DOI] [PubMed] [Google Scholar]
- 25.Stolk RP, Rosmalen JG, Postma DS, de Boer RA, Navis G, Slaets JP, et al. Universal risk factors for multifactorial diseases: lifelines: a three-generation population-based study. Eur J Epidemiol. 2008;23:67–74. [DOI] [PubMed] [Google Scholar]
- 26.Scholtens S, Smidt N, Swertz MA, Bakker SJ, Dotinga A, Vonk JM, et al. Cohort profile: lifelines, a three-generation cohort study and biobank. Int J Epidemiol. 2015;44:1172–80. [DOI] [PubMed] [Google Scholar]
- 27.Sijtsma A, Rienks J, van der Harst P, Navis G, Rosmalen JGM, Dotinga A. Cohort profile update: lifelines, a three-generation cohort study and biobank. Int J Epidemiol. 2022;51:e295–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.de Kat AC, Dam V, Onland-Moret NC, Eijkemans MJ, Broekmans FJ, van der Schouw YT. Unraveling the associations of age and menopause with cardiovascular risk factors in a large population-based study. BMC Med. 2017;15:2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates. 1988. [Google Scholar]
- 30.Alsaker MD, Janszky I, Opdahl S, Vatten LJ, Romundstad PR. Weight change in adulthood and risk of postmenopausal breast cancer: the HUNT study of Norway. Br J Cancer. 2013;109:1310–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Baer HJ, Tworoger SS, Hankinson SE, Willett WC. Body fatness at young ages and risk of breast cancer throughout life. Am J Epidemiol. 2010;171:1183–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Collaborative PBC, Schoemaker G, Nichols MJ, Wright HB, Brook LB, Jones MN. Association of body mass index and age with subsequent breast cancer risk in premenopausal women. JAMA Oncol. 2018;4:e181771. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Teras LR, Patel AV, Wang M, Yaun SS, Anderson K, Brathwaite R, et al. Sustained weight loss and risk of breast cancer in women 50 years and older: A pooled analysis of prospective data. J Natl Cancer Inst. 2020;112:929–37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Chlebowski RT, Luo J, Anderson GL, Barrington W, Reding K, Simon MS, et al. Weight loss and breast cancer incidence in postmenopausal women. Cancer. 2019;125:205–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Teras LR, Goodman M, Patel AV, Diver WR, Flanders WD, Feigelson HS. Weight loss and postmenopausal breast cancer in a prospective cohort of overweight and obese US women. Cancer Causes Control. 2011;22:573–9. [DOI] [PubMed] [Google Scholar]
- 36.Trentham-Dietz A, Newcomb PA, Egan KM, Titus-Ernstoff L, Baron JA, Storer BE, et al. Weight change and risk of postmenopausal breast cancer (United States). Cancer Causes Control. 2000;11:533–42. [DOI] [PubMed] [Google Scholar]
- 37.Ligibel JA, Basen-Engquist K, Bea JW. Weight management and physical activity for breast cancer prevention and control. Am Soc Clin Oncol Educ Book. 2019;39:e22–33. [DOI] [PubMed] [Google Scholar]
- 38.van Gemert WA, Schuit AJ, van der Palen J, May AM, Iestra JA, Wittink H, et al. Effect of weight loss, with or without exercise, on body composition and sex hormones in postmenopausal women: the SHAPE-2 trial. Breast Cancer Res. 2015;17:120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Duggan C, Tapsoba JD, Stanczyk F, Wang CY, Schubert KF, McTiernan A. Long-term weight loss maintenance, sex steroid hormones, and sex hormone-binding Globulin. Menopause. 2019;26:417–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Warburg O. On the origin of cancer cells. Science. 1956;123:309–14. [DOI] [PubMed] [Google Scholar]
- 41.Goto A, Noda M, Sawada N, Kato M, Hidaka A, Mizoue T, et al. High hemoglobin A1c levels within the non-diabetic range are associated with the risk of all cancers. Int J Cancer. 2016;138:1741–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Schovanek J, Cibickova L, Ctvrtlik F, Tudos Z, Karasek D, Iacobone M, et al. Hypoglycemia as a symptom of neoplastic disease, with a focus on Insulin-like growth factors producing tumors. J Cancer. 2019;10:6475–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Aggarwal V, Schneider AL, Selvin E. Low hemoglobin A(1c) in nondiabetic adults: an elevated risk state? Diabetes Care. 2012;35:2055–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Clemmons DR. Involvement of insulin-like growth factor-I in the control of glucose homeostasis. Curr Opin Pharmacol. 2006;6:620–5. [DOI] [PubMed] [Google Scholar]
- 45.Furstenberger G, Senn HJ. Insulin-like growth factors and cancer. Lancet Oncol. 2002;3:298–302. [DOI] [PubMed] [Google Scholar]
- 46.Yang X, Ma RC, So WY, Yu LW, Kong AP, Ko GT, et al. Low triglyceride and nonuse of Statins is associated with cancer in type 2 diabetes mellitus: the Hong Kong diabetes registry. Cancer. 2011;117:862–71. [DOI] [PubMed] [Google Scholar]
- 47.Ni H, Liu H, Gao R. Serum lipids and breast cancer risk: A Meta-Analysis of prospective cohort studies. PLoS ONE. 2015;10:e0142669. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Chen K, Li J, Ouyang Y, Liu G, Xie Y, Xu G, et al. Blood lipid metabolic profiles and causal links to Site-Specific cancer risks: A Mendelian randomization study. Nutr Cancer. 2024;76:175–86. [DOI] [PubMed] [Google Scholar]
- 49.Ackerman D, Tumanov S, Qiu B, Michalopoulou E, Spata M, Azzam A, et al. Triglycerides promote lipid homeostasis during hypoxic stress by balancing fatty acid saturation. Cell Rep. 2018;24:2596–e26052595. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Ferre P, Foufelle F. SREBP-1c transcription factor and lipid homeostasis: clinical perspective. Horm Res. 2007;68:72–82. [DOI] [PubMed] [Google Scholar]
- 51.Nadler DL, Zurbenko IG. Estimating cancer latency times using a Weibull model. Adv Epidemiol. 2014;1–8.
- 52.Nguyen HL, Geukens T, Maetens M, Aparicio S, Bassez A, Borg A, et al. Obesity-associated changes in molecular biology of primary breast cancer. Nat Commun. 2023;14:4418. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data may be obtained from a third party and are not publicly available. Researchers can apply to use the Lifelines data used in this study. More information about how to request Lifelines data and the conditions of use can be found on their website (https://www.lifelines-biobank.com/researchers/working-with-us/step-1-prepare-and-submit-your-application).