Abstract
Purpose.
As obesity and type 2 diabetes (T2D) have been increasing worldwide, we investigated their association with breast cancer incidence in the Reykjavik Study.
Methods.
During 1968–1996, approximately 10,000 women (mean age=53±9 years) completed questionnaires and donated blood samples. T2D status was classified according to self-report (N=140) and glucose levels (N=154) at cohort entry. A linkage with the Icelandic Cancer Registry provided breast cancer incidence through 2015. Cox regression with age as time metric and adjusted for known confounders was applied to obtain hazard ratios (HR) and 95% confidence intervals (CI).
Results.
Of 9,606 participants, 294 (3.1%) were classified as T2D cases at cohort entry while 728 (7.8%) women were diagnosed with breast cancer during 28.4±11.6 years of follow-up. No significant association of T2D (HR=0.95; 95%CI, 0.56–1.53) with breast cancer incidence was detected except among the small number of women with advanced breast cancer (HR=3.30; 95%CI, 1.13–9.62). Breast cancer incidence was elevated among overweight/obese women without (HR=1.18; 95%CI, 1.01–1.37) and with T2D (HR=1.35; 95%CI, 0.79–2.31). Height also predicted higher breast cancer incidence (HR=1.03; 95%CI, 1.02–1.05). All findings were confirmed in women of the AGES-Reykjavik sub-cohort (N=3,103) who returned for an exam during 2002–2006. With a 10% T2D prevalence and 93 incident breast cancer cases, the HR for T2D was 1.18 (95%CI, 0.62–2.27).
Conclusions.
These findings in a population with low T2D incidence suggest that the presence of T2D does not confer additional breast cancer risk and confirm the importance of height and excess body weight as breast cancer risk factors.
Keywords: Breast cancer, type 2 diabetes, obesity, anthropometry, prospective cohort, Cox regression
Introduction
The rising incidence of obesity and type 2 diabetes (T2D) affects populations around the world but at different rates across ethnic groups and geographic locations [1, 2]. The important role of obesity in the development of postmenopausal breast cancer has been studied extensively [3–6] while the possible influence of T2D appears to be weaker and results are less consistent [7, 8]. Due to the strong influence of body mass index (BMI) on T2D as well as breast cancer development the exploration of the combined impact of T2D and obesity on breast cancer incidence is of great interest. A worldwide modeling effort estimated attributable risks of 6.9% and 2.2% for high BMI (≥25 k/m2) and T2D, respectively, with a combined risk of 8.9% [9]. Two meta-analyses reported 20% [10] and 23% [7] higher breast cancer risk for women with T2D. However, separation by adjustment for weight status reduced the estimated elevated breast cancer risk associated with T2D to 16% as compared to 33% in studies that did not include BMI in their models [7]. Within the Multiethnic Cohort (MEC) of African American, Native Hawaiian, Japanese American, Latino, and white participants [11], T2D was significantly associated with breast cancer (HR=1.15; 95% CI, 1.07–1.23), but adjusting for BMI lowered the relative risk to 1.08 (95% CI, 1.00–1.16). As many of the previous studies relied on self-reported T2D diagnoses, individuals with elevated fasting glucose levels may have been missed resulting in biased estimates for the T2D-breast cancer association. Prediabetes, either impaired fasting glucose (IFG) or impaired glucose tolerance (IGT), may also increase the risk to develop breast cancer [12, 13]. T2D incidence in Iceland is relatively low compared with other Western countries today [1] and was even lower in the Reykjavik Study initiated in 1968 when detailed information on lifestyle and anthropometric measures was collected. At the time, 12% of the female participants were in the obese category [14]. Using population data from this cohort offers a unique opportunity to examine the association of undiagnosed T2D and IFG with breast cancer risk. Therefore, the current analysis aimed to determine the independent contributions of obesity, T2D, and IFG at baseline to develop breast cancer within the Reykjavik Study and its sub-study, the Age, Gene/Environment Susceptibility (AGES)-Reykjavik Study initiated 34 years later.
Methods
Study Population.
The current analysis builds on a prospective cohort, the Reykjavik Study, initiated in 1967 for male participants and in 1968 for female participants by the Icelandic Heart Association. Participants were all current residents of the capital area in Iceland in 1966 and born between 1908 and 1935, and thus aged from 31 to 88 years [15, 16]. The enrollment occurred in six waves (1968–69, 1971–72, 1977–79, 1981–84, 1987–91, and 1994–96) when 10,049 women agreed to be part of the investigation (71% response rate). At cohort entry, information on age, year of birth, education, height, weight, parity, regular sports activity since age 20, and medical history was collected.
Detailed reproductive information was obtained through a linkage of 91% participants of the Reykjavik Study with the Cancer Detection Clinic (CDC) cohort, an investigation established in 1964 that builds on the nationwide, centralized cervical- and breast cancer screening programs [17]. It includes all women in Iceland who attend annual cervical cancer (20–69 years) and breast cancer screening (40–69 years). Information closest to study entry was obtained on age at menarche (continuous) and age at first birth (none, <25, ≥25 years) but not a detailed mammography screening history. To reduce the substantial number of missing values for age at first birth, women who reported no children when entering the Reykjavik Study were coded as “no birth”. Women classified as nulliparous in the CDC cohort who reported a child in the Reykjavik Study were categorized as “25 and older” given that all participants were at least 33 years old at cohort entry.
AGES-Reykjavik Study.
During 2002–06, the AGES-Reykjavik Study with a focus on body composition and metabolism re-examined 5,764 participants (57% women) aged 66–96 years [15]. As described elsewhere [18], 8,030 of the 11,549 surviving cohort members of the original Reykjavik Study were randomly chosen and invited for another exam with a 71.8% response rate. Respondents had significantly lower triglycerides, fasting blood glucose, and BMI and included a smaller percentage of smokers [15]. At the exam, updated information on chronic conditions, including T2D status, as well blood samples for glucose and insulin were collected. In addition to the same questions as at the baseline exam in the Reykjavik Study, the AGES-Reykjavik Study also asked about family history of breast cancer, reproductive history, and alcohol intake allowing better adjustment for known breast cancer risk factors. In total, 3,103 women had sufficient information for the current analysis within the AGES-Reykjavik cohort. Both studies were approved by the National Bioethics Committee in Iceland (VSN-17–189). Informed consent was obtained from all participants.
Diabetes Status.
At entry to both the Reykjavik and the AGES-Reykjavik Study, T2D status was based on self-reports (”Do you have diabetes now or earlier?”) plus age at diagnosis, diabetes diet, and diabetes medication as categories of insulin vs. tablets without specific information. Glucose levels at fasting and 90 minutes after a 50 g glucose load were obtained in both studies to detect undiagnosed disease [14, 19]. For both cohorts, glucose levels in a capillary blood sample were estimated by the Hoffman-ferri-cyanide method, adapted to the Technicon-Method N-9a [15, 16]. Those with fasting glucose levels ≥126 mg/dL and 90-minute levels >200 mg/dL were classified as T2D and those ranging 100–125 mg/dL as IFG [20]. As oral glucose tolerance tests (OGTT) with 2-hour glucose levels were not available, IGT could not be assessed [21]. A summary variable indicated abnormal glucose status, i.e., IFG, T2D, or insulin-dependent diabetes. Cases diagnosed at an early age and insulin use were classified as type 1 diabetes and not considered in the current analysis.
Diagnosis of Breast Cancer.
For both cohorts, follow-up for first diagnosis of invasive breast cancer was done by record linkage with the population-based Icelandic Cancer Registry through 2015 [22], using the unique Icelandic identification numbers. The 118 cases of invasive breast cancer diagnosed before the baseline exam in the Reykjavik study were excluded. After removing 35 women without information on height and body weight, the data set for analysis had 9906 women with 728 incident breast cancer cases diagnosed in the period 1969–2015. In the AGES-Reykjavik Study, 3103 women remained in the dataset after exclusions; 93 incident breast cancer cases were detected by 2015.
Statistical analysis.
The associations of T2D (self-reported and based on glucose testing), IFG, and glucose levels with breast cancer incidence were examined using Cox regression models with age as time metric and adjusted for birth cohort (1908–14, 1915–19, 1920–24, 1925–29, 1930–36), BMI calculated from weight and height measurements at entry (<18.5, 18.5-<25, 25-<30, ≥30 kg/m2), height (continuous), education (primary, secondary, college), regular sports since age 20 (yes/no), smoking status (never, past, current), age at menarche (9–12, 13, 14, 15–21 years), parity (none, 1–2, ≥3, missing), and age at first childbirth (none, <25, ≥25 years, missing). We also included location of early life residence as determined by classifying each woman’s place of residence from birth using three categories: capital area, coastal villages, and rural areas as a lower risk of breast cancer had been observed in women who lived in coastal villages than in the capital area during early life [23]. Missing covariate information was included into the models as a separate category except for BMI, for which women with missing values had been excluded. Information on hormone therapy and diabetes medication was not available for the cohort.
An analysis restricted to postmenopausal breast cancers was performed after removing 22 cases diagnosed in women ≤50 years. Separate models by BMI status, as well as models with different outcomes according to stage, estrogen (ER) and progesterone (PR) receptor status were examined for effect modification. To explore if obesity and T2D are independent predictors of breast cancer risk, models with and without BMI were compared. For a closer examination of possible interactions [24], an interaction test was performed and a combined variable incorporating BMI and T2D (BMI <25 kg/m2/no T2D, <25 kg/m2/T2D, ≥25 kg/m2/no T2D, ≥25 kg/m2/T2D) was created and modeled with BMI <25 kg/m2/no T2D as the reference category using the same covariates as above.
Separate analyses for models with T2D, IFG, and combined T2D/IFG as exposure were conducted for the 3,103 women in the AGES-Reykjavik Study, which included information from the AGES questionnaire for age at menarche, smoking, and physical activity and, in addition, breastfeeding (yes, no), hormone therapy (never, ever, missing), and alcohol intake (<1 or ≥1 drink/week). Parity and age at first live birth were combined into one variable (no children, parous <25 or ≥25 years). Missing values were coded as a separate category for all covariates.
Results
Of the 9,606 women in the Reykjavik Study, 294 (3.1%) were classified as having T2D at study entry, based on self-report (N=140) or on glucose levels (N=154). For 9,587 participants, a fasting blood sample was available. Another 134 (1.4%) women were classified with IFG according to fasting glucose levels. During 28.4±11.6 years of follow-up, 728 (7.8%) breast cancer cases were diagnosed resulting in 13,826 person-years for breast cancer cases as opposed to 258,935 person-years for women without breast cancer. Of all breast tumors, 51% were diagnosed at TNM stage 1, 36% at stage 2, and 6% at stages 3/4 and 7% did not have stage information. The respective mean ages at cohort entry and at breast cancer diagnosis were 53.2±9.0 and 69.7±11.2 years. Within the study population, 2.4% were underweight (<18.5 kg/m2), 52.6% normal weight, 33.3% overweight (25-≤30 kg/m2), and 11.7% obese (≥30 kg/m2). Women with a higher BMI were more likely to have TD2 (Table 1): the respective proportions for low, normal, overweight, and obese BMI with T2D were 2.2, 1.4, 3.5, and 9.6%. An inverse trend of birth cohort with the presence of T2D was seen; the percentage decreased from 6.7 to 1.8% from the oldest to the youngest birth cohort. The proportion of women with T2D was higher for those with early life residence in the capital (3.4%) than for those with coastal (3.0%) and rural (2.7%) residence.
Table 1.
Characteristics of the Study Population by Diabetes Status, Reykjavik Study, 1968–2015
| Reykjavik Studya | AGES Studyb | ||||||
|---|---|---|---|---|---|---|---|
| Characteristic | Category | No T2D | T2D | % T2D | All women | Number | % T2D |
| Participants | Number | 9,312 | 294 | 3.1 | 9,606 | 3,103 | 10.1 |
| Breast cancer cases | Number | 712 | 16 | 2.2 | 728 | 93 | 11.88 |
| Deaths | Number | 6,544 | 255 | 3.8 | 6,799 | 1,414 | 12.6 |
| Person-years | Number | 266,738 | 6,023 | 2.2 | 272,761 | 30,769 | 9.5 |
| Follow-up time | Mean ± std | 28.6±11.5 | 20.5±10.9 | - | 28.4±11.6 | 9.9±3.4 | - |
| Age at cohort entry | Mean ± std | 53.0±8.9 | 60.4±9.2 | - | 53.2±9.0 | 76.8±5.9 | - |
| Age at breast cancer | Mean ± std | 69.7±11.2 | 73.1±9.7 | - | 69.7±11.2 | 82.0±6.3 | - |
| Age at diabetes | Mean ± std | NA | 56.8±11.7 | - | 56.8±11.7 | 75.5±9.4 | - |
| Height (cm) | Mean ± std | 163.2±5.7 | 161.8±6.0 | 163.1±5.7 | 160.6±5.8 | - | |
| Birth cohort | 1907–1914 | 1,348 | 97 | 6.7 | 1,445 | 72 | 12.3 |
| 1915–1919 | 1,636 | 53 | 3.1 | 1,689 | 298 | 10.5 | |
| 1920–1924 | 2,055 | 56 | 2.6 | 2,111 | 642 | 12.2 | |
| 1925–1929 | 2,058 | 48 | 2.3 | 2,106 | 850 | 10.7 | |
| 1930–1936 | 2,215 | 40 | 1.8 | 2,254 | 928 | 7.7 | |
| Location of early life residence | Reykjavik | 3,282 | 115 | 3.4 | 3,397 | 1,142 | 12.1 |
| Coastal area | 3,084 | 95 | 3.0 | 3,179 | 1,009 | 9.6 | |
| Rural | 2,445 | 69 | 2.7 | 2,514 | 796 | 8.0 | |
| Missing | 501 | 15 | 2.9 | 516 | 156 | 9.0 | |
| Education, yrs | Primary | 5,034 | 190 | 3.6 | 5,224 | 1,367 | 11.1 |
| Secondary | 3,495 | 88 | 2.4 | 3,583 | 1,396 | 9.2 | |
| College | 783 | 16 | 2.0 | 799 | 340 | 10.0 | |
| BMI, kg/m2 | <18.5 | 223 | 5 | 2.2 | 228 | 56 | 5.4 |
| ≥18.5 to <25 | 4,992 | 69 | 1.4 | 5,061 | 1,012 | 5.6 | |
| ≥25 to <30 | 3,083 | 113 | 3.5 | 3,196 | 1,258 | 9.6 | |
| ≥30 | 1,014 | 107 | 9.6 | 1,121 | 777 | 17.0 | |
| Smoking status | Never | 4,099 | 134 | 3.2 | 4,223 | 1,624 | 10.8 |
| Past | 1,418 | 70 | 4.7 | 1,488 | 993 | 10.0 | |
| Current | 3,795 | 90 | 2.3 | 3,885 | 377 | 8.8 | |
| Regular sports | No | 7,249 | 239 | 3.2 | 7,488 | 1,882 | 10.6 |
| Yes | 2,063 | 55 | 2.6 | 2,118 | 775 | 7.5 | |
| Age at menarche, yrs | ≤12 | 1,578 | 61 | 3.7 | 1,639 | 572 | 11.5 |
| 13 | 2,417 | 82 | 3.3 | 2,499 | 753 | 8.9 | |
| 14 | 2,710 | 49 | 1.8 | 2,759 | 972 | 9.4 | |
| ≥15 | 1,848 | 51 | 2.7 | 1,899 | 640 | 8.6 | |
| Missing | 759 | 51 | 6.3 | 810 | 166 | 20.5 | |
| Number of children | 0 | 937 | 40 | 4.2 | 977 | 218 | 8.3 |
| 1–2 | 2,452 | 53 | 2.1 | 2,505 | 661 | 3.3 | |
| ≥3 | 5,874 | 201 | 3.3 | 6,075 | 2,060 | 11.6 | |
| Missing | 49 | 0 | 0 | 49 | 164 | 20.7 | |
| Age at first live birth, | <25 | 4,654 | 137 | 2.9 | 4,792 | 1,788 | 8.3 |
| yrs | ≥25 | 3,059 | 77 | 2.3 | 3,136 | 893 | 9.5 |
| Missing | 662 | 40 | 6.5 | 702 | 204 | 9.5 | |
T2D diagnosis based on self-report (N=140) or elevated glucose levels (N=154).
Values are for exams conducted among AGES participants in 2002–20006.
No significant association of T2D with breast cancer was detected in a model adjusted for age only (HR, 1.01; 95% CI, 0.62–1.67) and in fully adjusted models without and with BMI included as confounder (Table 2). The respective HRs were 0.99 (95% CI, 0.60–1.64) and 0.95 (95% CI, 0.42–1.89). In stratified analyses by self-reported vs. glucose-based T2D, similar HRs were obtained (HR 0.89; 95% CI, 0.57–1.56) and (HR 1.01; 95% CI, 0.52–1.96). Restricting the BMI-adjusted model to the 706 breast cancer cases diagnosed >50 years showed similar results. With IFG, IFG/T2D, or fasting glucose as exposure, the respective HRs in BMI-adjusted models were 0.88 (95% CI, 0.44–1.78), 0.95 (95% CI, 0.64–1.42), and 1.02 (95% CI, 0.96–1.08) for each increase of 10 mg/dL fasting glucose.
Table 2.
Type 2 Diabetes and Breast Cancer Risk, Reykjavik and AGES Study, 1968–2015
| Characteristic | Exposure category | Number of cases | Person-years | HRa | 95% CI | |||
|---|---|---|---|---|---|---|---|---|
| Exposed | Unexposed | Exposed | Unexposed | |||||
| Reykjavik Study | T2D (model without BMI) | 16 | 712 | 6,023 | 266,738 | 0.99 | 0.60, 1.64 | |
| (N=9,606) | T2D (model with BMI) | 16 | 712 | 6,023 | 266,738 | 0.95 | 0.57, 1.56 | |
| T2D (postmenopausal) | 16 | 690 | 6,023 | 266,480 | 0.95 | 0.58, 1.57 | ||
| IFGb | 8 | 720 | 3,165 | 269,596 | 0.88 | 0.44, 1.78 | ||
| T2D and IFGc | 26 | 702 | 9,606 | 269,596 | 0.95 | 0.64, 1.42 | ||
| Fasting glucose (per 10 mg/dL) | 16 | 712 | 6,023 | 266,738 | 1.02 | 0.96, 1.08 | ||
| AGES Studyd | T2D (model without BMI) | 11 | 82 | 2,912 | 27,857 | 1.24 | 0.66, 2.36 | |
| (N=3,103) | T2D (model with BMI) | 11 | 82 | 2,912 | 27,857 | 1.18 | 0.62, 2.27 | |
| IFG | 35 | 58 | 10,243 | 20,526 | 1.19 | 0.78, 1.82 | ||
| T2D and IFG | 46 | 47 | 13,155 | 17,614 | 1.26 | 0.83, 1.92 | ||
| Fasting glucose (per 10 mg/dL) | 11 | 82 | 2,912 | 27,857 | 0.97 | 0.87, 1.09 | ||
| Stratified Models for Reykjavik Study | ||||||||
| Hormone receptorsd | ER+/PR+ | 4 | 264 | 5,915 | 259,410 | 0.69 | 0.25, 1.87 | |
| ER-/PR- | 3 | 95 | 5,894 | 254,971 | 1.22 | 0.38, 3.93 | ||
| ER+/PR- and ER-/PR+ | 1 | 133 | 5,877 | 255,751 | 0.28 | 0.04, 2.03 | ||
| Tumor stagee | TNM 1 | 9 | 363 | 5,966 | 259,869 | 1.05 | 0.54, 2.05 | |
| TNM 2 | 2 | 258 | 5,880 | 258,439 | 0.31 | 0.08, 1.24 | ||
| TNM 3/4 | 4 | 43 | 5,893 | 253,918 | 3.30 | 1.13, 9.62 | ||
Hazard ratios (HR) and 95% confidence intervals (CI) obtained by Cox regression with age as time metric and adjusted for BMI (except when noted), height, birth cohort, education, place of residence, smoking status, physical activity, age at menarche, parity, and age at first live birth
Presence of impaired fasting glucose (100–125 mg/dL)
Presence of impaired fasting glucose (100–125 mg/dL) or T2D based on self-report of fasting glucose
Women in AGES study (N=3,103); models also include alcohol intake, hormone treatment, and breast feeding. Parity and age at first live birth were combined into one variable (no children, parous <25 or ≥25 years).
Values were missing on hormone receptors for 228 cases and on tumor stage for 49 cases
Abbreviations: ER=estrogen receptor; PR=progesterone receptor; TNM=tumor, nodes, metastasis classification; BMI=body mass index.
In the subset of 3,103 women who participated in the AGES-Reykjavik Study, the mean age at study exam was 76.9±5.9 years. In this group, 313 (10%) women were classified as T2D and 1,019 (33%) as IFG. After entry into AGES, 93 (3%) women were diagnosed with incident breast cancer. With additional adjustment for alcohol intake, menopausal hormone therapy, and breast feeding (Table 2), the HRs for T2D at baseline of AGES were 1.24 (95% CI, 0.66–2.36) and 1.18 (95% CI, 0.62–2.27) before and after adjustment for BMI. Although the risk estimate was elevated when the women with T2D and IFG were combined and adjusted for BMI (HR 1.26; 95% CI, 0.83–1.92), it did not reach statistical significance. The HR per 10 mg/dL fasting glucose was 0.97 (95% CI, 0.87–1.09).
Within the Reykjavik Study cohort, stratification by ER/PR status showed slightly elevated HRs for ER-/PR- tumors but no statistically significant association. On the other hand, findings differed by stage at diagnosis. Whereas the incidence of tumors diagnosed at stage I and II was not associated with T2D status, the risk estimate for TNM stages 3/4 tumors among women with T2D was significantly elevated (HR=3.30; 95% CI, 1.13–9.62) although the number of breast cancer cases with and without T2D was small (n=4 and 43).
Of the covariates, height and BMI were the most significant predictors of breast cancer after adjustment for all the covariates in the overall model (Table 3). Every 1 cm of additional height predicted a 3% (95% CI, 1.02–1.05) or every 10 cm a 37% (95% CI, 1.19–1.57) higher breast cancer risk. Compared to normal weight women, the respective HRs of underweight, overweight, and obese women were 0.86 (95% CI, 0.48–1.53), 1.15 (95% CI, 0.97–1.36), and 1.38 (95% CI, 1.09–1.75). After removing the 22 cases diagnosed <50 years, these HRs strengthened to 0.82 (95% CI, 0.45–1.49), 1.18 (95% CI, 1.00–1.39), and 1.41 (95% CI, 1.10–1.79).
Table 3.
Anthropometric Risk Factors and Breast Cancer Risk, Reykjavik and AGES Study, 1968–2015
| Study | Risk factor | Categories | Number of cases | Person-years | HRa | 95% CI | ||
|---|---|---|---|---|---|---|---|---|
| Exposed | Unexposed | Exposed | Unexposed | |||||
| Reykjavik Study | Height | Per cm | 16 | 712 | 6,023 | 266,738 | 1.03 | 1.02, 1.05 |
| (N=9,606) | BMI | <18.5 | 0 | 12 | 74 | 5,604 | 0.86 | 0.48, 1.53 |
| ≥18.5 - <25 | 2 | 380 | 1,595 | 148,077 | 1.00 | Ref | ||
| ≥25 - <30 | 6 | 238 | 2,307 | 86,525 | 1.15 | 0.97, 1.36 | ||
| ≥30 | 8 | 82 | 2,046 | 26,531 | 1.38 | 1.09, 1.75 | ||
| AGES Studyb | Height | Per cm | 11 | 82 | 2,912 | 27,857 | 1.02 | 0.98, 1.06 |
| (N=3,103) | BMI | <18.5 | 0 | 1 | 18 | 432 | 0.82 | 0.11, 6.13 |
| ≥18.5 - <25 | 0 | 26 | 532 | 9,168 | 1.00 | Ref | ||
| ≥25 - <30 | 5 | 33 | 1,130 | 11,602 | 1.06 | 0.64, 1.76 | ||
| ≥30 | 6 | 22 | 1,231 | 6,654 | 1.28 | 0.73, 2.23 | ||
Hazard ratios (HR) and 95% confidence intervals (CI) obtained by Cox regression with age as time metric and adjusted for birth cohort, education, place of residence, smoking status, physical activity, age at menarche, parity, age at first live birth, and BMI or height.
Women in AGES study; models also include alcohol intake, hormone treatment, and breast feeding. Parity and age at first live birth were combined into one variable (no children, parous <25 or ≥25 years).
Abbreviations: T2D=type 2 diabetes; BMI=body mass index.
A formal interaction test between BMI and T2D was not significant (p=0.74). Results from the analysis of the combined BMI and T2D variable, using low BMI and no TD2 as reference (Figure 1), showed that having normal BMI and T2D did not predict higher breast cancer incidence (HR=0.46; 95% CI, 0.11–1.83). However, breast cancer incidence was significantly elevated for women with high BMI and no T2D (HR=1.18; 95% CI, 1.01–1.37). Participants with high BMI (≥25 kg/m2) and T2D had a non-significant 35% increased risk of breast cancer (95% CI 0.71–2.58). When using the combined BMI/T2D variable within the AGES-Reykjavik Study, no breast cancers were diagnosed in the low BMI/no T2D group while the respective risk estimates for the high BMI group without (HR=1.06; 95% CI, 0.66–1.69) and with T2D (HR=1.67; 95% CI, 0.82–3.40) were not significant although the HR for the high BMI women with T2D was higher than for those without.
Figure 1. Breast Cancer by Type 2 Diabetes and BMI Statusa, Reykjavik and AGES-Reykjavik Study, 1968–2015.
Abbreviations: Low BMI = <25 kg/m2
High BMI = ≥25 kg/m2 and no type 2 diabetes
T2D = Presence of type 2 diabetes 2
aHazard ratios (HR) and 95% confidence intervals (CI) obtained by Cox regression with age as time metric and adjusted for birth cohort, height, education, place of residence, smoking status, physical activity, parity, age at menarche and first live birth.
Women in AGES study (N=3,103); models also include alcohol intake, hormone treatment, and breast feeding. Parity and age at first live birth were combined into one variable (no children, parous <25 or ≥25 years).
Discussion
Over a follow-up time of close to 30 years, women with T2D and/or IFG who participated in the Reykjavik Study did not experience a higher breast cancer risk than those without T2D, previously known or undiagnosed cases based on abnormal glucose values, but excess body weight and height were associated with breast cancer incidence independent of T2D status. The higher incidence of postmenopausal breast cancer among participants with excess body weight [25] and height [26] was comparable to reports from other white populations. In a secondary analysis stratified by stage at diagnosis, the risk for advanced breast cancer was elevated in women with T2D, but this may be a chance finding due to the small number of women in this group. Despite the smaller sample size of the AGES-Reykjavik Study, its confirmatory results were important because of the adjustment for additional known risk factors.
The current results agree with a worldwide study showing that BMI contributes more to the burden of breast cancer than T2D [9] but disagrees with several meta-analyses showing a 20% higher risk [7, 10]. It is worth noting that despite the overall significant summary risk estimate in a large meta-analysis with 40 studies (SRR=1.27, 95% CI, 1.12–1.39), nearly half (19 vs. 21) of the studies reported null findings. Except for the importance of including BMI as a covariate, the divergent results are not well understood. The null finding for IFG parallels a summary report indicating a small risk of breast cancer associated with pre-diagnostic elevated glucose levels [27].
One possibility for the conflicting results is that the metabolic consequences of elevated BMI and T2D vary across populations. The concept of metabolically healthy obesity hypothesizes that some individuals with excess body weight remain free of metabolic consequences, such as insulin resistance or chronic inflammation as indicated by C-reactive protein [28]. Possible reasons may be differences in the distribution of body fat and the relative proportion of visceral fat, individual variation in coping with a positive energy balance, or a higher quality diet. As a result of these observations, it was proposed that the risk associated with cardiovascular disease might be lower among metabolically healthy obese individuals than for obese individuals with metabolic disturbances [29, 30]. This idea is supported by an evaluation of T2D incidence within the AGES-Reykjavik Study showing no difference in cardiovascular mortality rate by T2D status [31]. Based on information from the original Reykjavik Study, high BMI, family history of T2D, elevated triglycerides, and high systolic blood pressure at cohort entry were identified as predictors of T2D in later in life but a large proportion of participants with overweight/obesity had not developed T2D or other cardiovascular conditions [19]. Another interesting finding from the Reykjavik Study and the AGES-Reykjavik Study showed that the decline in all-cause mortality observed from 1993 to 2004 was 50% greater among participants with T2D than those without T2D [31]. This observation supports the hypothesis that Icelanders with T2D experience less serious consequences than individuals in some other geographic locations, possibly due to better treatment and control in a country with equal access to health care or to an overall healthier metabolic status [32].
The most serious limitations of the current analysis are the lack of information on T2D cases diagnosed after cohort entry and the small sample size of only 16 women with T2D and breast cancer resulting in low statistical power. Self-reports can be a problem but the availability of fasting glucose levels allowed the detection of undiagnosed cases and IFG despite the known low sensitivity (mean: 0.25; range: 0.19 to 0.32) and specificity (mean: 0.94; range: 0.92 to 0.96) of one fasting glucose measurement [33]. Although not likely, misclassification of type 1 diabetes as T2D may have added to the problem. Thus some non-differential misclassification with respect to exposure status can be expected, which is likely to move the hazard ratios towards unity. Using prevalent T2D as exposure adds to the problem that management and drug treatment of the disease may have affected BMI levels. As no OGTTs were available, only IFG and not IGT could be assessed; the two conditions do not always overlap [20]. Missing values for known confounders and the fact that information on alcohol intake, hormone therapy, and breast feeding was not available for the Reykjavik Study, may have biased the risk estimates. Also, the lack of detailed information on diabetes medication may have affected the results [34]. Despite the confirmation of our results within the AGES-Reykjavik Study with updated glucose levels and T2D reports, the null findings may not be conclusive given the limited sample sizes and power of the entire cohort and its sub-study.
Strengths of the current analysis include the long follow-up period, the glucose measurements at cohort entry, low rates of loss-to-follow-up, and the linkage with the Icelandic Cancer Registry with complete and countrywide follow-up [22]. The fact that the elevated breast cancer risk associated with obesity in the Reykjavik Study is comparable to published reports [25] also indicates the general validity of the study. The analysis is unique in that it is based on a population with relatively low T2D incidence, partly due to low prevalence of obesity at the time of recruitment for the Reykjavik Study [14, 35]. Even today, the age-adjusted prevalence of T2D in Iceland in 2017 was only 5.3% as compared to 6.8% for all of Europe, 10.8% in the Middle East, and 11.0% In North America according to the International Diabetes Federation [1].
In conclusion, our results from women with low obesity rates and T2D incidence suggest that a diagnosis of T2D may only confer a very small, if any, additional breast cancer risk beyond the elevated risk associated with obesity and height among participants of the Reykjavik Study. Possible reasons for the differences in comparison with other populations include speculations about better metabolic health and equal access to health care and treatment among Icelanders.
Acknowledgments
The AGES-Reykjavik Study was funded by NIH contract N01-AG-12100, the Intramural Research Program of the National Institute on Aging, the Icelandic Heart Association, and the Icelandic Parliament. This work was also supported by the Icelandic Centre for Research, RANNIS grant number: 152495051, http://en.rannis.is/ (to AH), and the Public Health Fund of the Icelandic Directorate of Health (to AH). GM was funded by a Fulbright Global Scholar Award (2017–2018).
References
- 1.International Diabetes Federation (2017) IDF Diabetes Atlas 2017. http://www.diabetesatlas.org/. Accessed on 6-21-0018. [PubMed]
- 2.Maskarinec G, Grandinetti A, Matsuura G et al. (2009) Diabetes prevalence and body mass index differ by ethnicity: the Multiethnic Cohort. Ethn Dis 19: 49–55 [PMC free article] [PubMed] [Google Scholar]
- 3.Barnes BB, Steindorf K, Hein R, Flesch-Janys D, Chang-Claude J (2011) Population attributable risk of invasive postmenopausal breast cancer and breast cancer subtypes for modifiable and non-modifiable risk factors. Cancer Epidemiol 35: 345–352 [DOI] [PubMed] [Google Scholar]
- 4.Parkin DM, Boyd L, Walker LC (2011) 16. The fraction of cancer attributable to lifestyle and environmental factors in the UK in 2010. Br J Cancer 105 Suppl 2: S77–S81 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Sprague BL, Trentham-Dietz A, Egan KM, Titus-Ernstoff L, Hampton JM, Newcomb PA (2008) Proportion of invasive breast cancer attributable to risk factors modifiable after menopause. Am J Epidemiol 168: 404–411 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bandera EV, Maskarinec G, Romieu I, John EM (2015) Racial and Ethnic Disparities in the Impact of Obesity on Breast Cancer Risk and Survival: A Global Perspective. Adv Nutr 6: 803–819 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Boyle P, Boniol M, Koechlin A et al. (2012) Diabetes and breast cancer risk: a meta-analysis. Br J Cancer 107: 1608–1617 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Hsu WC, Boyko EJ, Fujimoto WY et al. (2012) Pathophysiologic differences among Asians, native Hawaiians, and other Pacific Islanders and treatment implications. Diabetes Care 35: 1189–1198 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Pearson-Stuttard J, Zhou B, Kontis V, Bentham J, Gunter MJ, Ezzati M (2018) Worldwide burden of cancer attributable to diabetes and high body-mass index: a comparative risk assessment. Lancet Diabetes Endocrinol 6: e6–e15 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Larsson SC, Mantzoros CS, Wolk A (2007) Diabetes mellitus and risk of breast cancer: a meta-analysis. Int J Cancer 121: 856–862 [DOI] [PubMed] [Google Scholar]
- 11.Maskarinec G, Jacobs S, Park SY et al. (2017) Type II Diabetes, Obesity, and Breast Cancer Risk: The Multiethnic Cohort. Cancer Epidemiol Biomarkers Prev 26: 854–861 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gunter MJ, Xie X, Xue X et al. (2015) Breast cancer risk in metabolically healthy but overweight postmenopausal women. Cancer Res 75: 270–274 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Cust AE, Stocks T, Lukanova A et al. (2009) The influence of overweight and insulin resistance on breast cancer risk and tumour stage at diagnosis: a prospective study. Breast Cancer Res Treat 113: 567–576 [DOI] [PubMed] [Google Scholar]
- 14.Vilbergsson S, Sigurdsson G, Sigvaldason H, Hreidarsson AB, Sigfusson N (1997) Prevalence and incidence of NIDDM in Iceland: evidence for stable incidence among males and females 1967–1991--the Reykjavik Study. Diabet Med 14: 491–498 [DOI] [PubMed] [Google Scholar]
- 15.Harris TB, Launer LJ, Eiriksdottir G et al. (2007) Age, Gene/Environment Susceptibility-Reykjavik Study: multidisciplinary applied phenomics. Am J Epidemiol 165: 1076–1087 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Jonsdottir LS, Sigfusson N, Gudnason V, Sigvaldason H, Thorgeirsson G (2002) Do lipids, blood pressure, diabetes, and smoking confer equal risk of myocardial infarction in women as in men? The Reykjavik Study. J Cardiovasc Risk 9: 67–76 [PubMed] [Google Scholar]
- 17.Thorbjarnardottir T, Olafsdottir EJ, Valdimarsdottir UA, Olafsson O, Tryggvadottir L (2014) Oral contraceptives, hormone replacement therapy and breast cancer risk: a cohort study of 16 928 women 48 years and older. Acta Oncol 53: 752–758 [DOI] [PubMed] [Google Scholar]
- 18.Steingrimsdottir L, Halldorsson TI, Siggeirsdottir K et al. (2014) Hip fractures and bone mineral density in the elderly--importance of serum 25-hydroxyvitamin D. PLoS ONE 9: e91122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Olafsdottir E, Aspelund T, Sigurdsson G et al. (2009) Unfavourable risk factors for type 2 diabetes mellitus are already apparent more than a decade before onset in a population-based study of older persons: from the Age, Gene/Environment Susceptibility-Reykjavik Study (AGES-Reykjavik). Eur J Epidemiol 24: 307–314 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Genuth S, Alberti KG, Bennett P et al. (2003) Follow-up report on the diagnosis of diabetes mellitus. Diabetes Care 26: 3160–3167 [DOI] [PubMed] [Google Scholar]
- 21.World Health Organization (1999) Definition, Diagnosis and Classification of Diabetes Mellitus and its Complications. http://apps.who.int/iris/bitstream/10665/66040/1/WHO_NCD_NCS_99.2.pdf. Accessed on 3-6-2018.
- 22.Sigurdardottir LG, Jonasson JG, Stefansdottir S et al. (2012) Data quality at the Icelandic Cancer Registry: comparability, validity, timeliness and completeness. Acta Oncol 51: 880–889 [DOI] [PubMed] [Google Scholar]
- 23.Haraldsdottir A, Steingrimsdottir L, Valdimarsdottir UA et al. (2017) Early Life Residence, Fish Consumption, and Risk of Breast Cancer. Cancer Epidemiol Biomarkers Prev 26: 346–354 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Knol MJ, VanderWeele TJ (2012) Recommendations for presenting analyses of effect modification and interaction. Int J Epidemiol 41: 514–520 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Renehan AG, Tyson M, Egger M, Heller RF, Zwahlen M (2008) Body-mass index and incidence of cancer: a systematic review and meta-analysis of prospective observational studies. Lancet 371: 569–578 [DOI] [PubMed] [Google Scholar]
- 26.van den Brandt PA, Spiegelman D, Yaun SS et al. (2000) Pooled analysis of prospective cohort studies on height, weight, and breast cancer risk. Am J Epidemiol 152: 514–527 [DOI] [PubMed] [Google Scholar]
- 27.Boyle P, Koechlin A, Pizot C et al. (2013) Blood glucose concentrations and breast cancer risk in women without diabetes: a meta-analysis. Eur J Nutr 52: 1533–1540 [DOI] [PubMed] [Google Scholar]
- 28.Phillips CM (2017) Metabolically healthy obesity across the life course: epidemiology, determinants, and implications. Ann N Y Acad Sci 1391: 85–100 [DOI] [PubMed] [Google Scholar]
- 29.Stefan N, Haring HU, Schulze MB (2018) Metabolically healthy obesity: the low-hanging fruit in obesity treatment? Lancet Diabetes Endocrinol 6: 249–258 [DOI] [PubMed] [Google Scholar]
- 30.Neeland IJ, Poirier P, Despres JP (2018) Cardiovascular and Metabolic Heterogeneity of Obesity: Clinical Challenges and Implications for Management. Circulation 137: 1391–1406 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Olafsdottir E, Aspelund T, Sigurdsson G et al. (2013) Similar decline in mortality rate of older persons with and without type 2 diabetes between 1993 and 2004 the Icelandic population-based Reykjavik and AGES-Reykjavik cohort studies. BMC Public Health 13: 36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Healthcare Access and Quality Index based on mortality from causes amenable to personal health care in 195 countries and territories, 1990–2015: a novel analysis from the Global Burden of Disease Study 2015 (2017) Lancet 390: 231–266 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Barry E, Roberts S, Oke J, Vijayaraghavan S, Normansell R, Greenhalgh T (2017) Efficacy and effectiveness of screen and treat policies in prevention of type 2 diabetes: systematic review and meta-analysis of screening tests and interventions. BMJ 356: i6538. [DOI] [PubMed] [Google Scholar]
- 34.Walker JJ, Johnson JA, Wild SH (2013) Diabetes treatments and cancer risk: the importance of considering aspects of drug exposure. Lancet Diabetes Endocrinol 1: 132–139 [DOI] [PubMed] [Google Scholar]
- 35.Bergsveinsson J, Aspelund T, Gudnason V, Benediktsson R (2007) [Prevalence of type 2 diabetes mellitus in Iceland 1967–2002]. Laeknabladid 93: 397–402 [PubMed] [Google Scholar]