Abstract
Background:
Many studies have reported a positive association between diabetes and kidney cancer. However, it is unclear whether diabetes is a risk factor for kidney cancer independent of other risk factors, such as obesity and hypertension. We comprehensively examined the association of diabetes and its duration with incident kidney cancer in the prospective cohort Iowa Women’s Health Study (1986–2011).
Methods:
Diabetes status was self-reported at baseline (1986) and on five follow-up questionnaires. Cox proportional hazards regression was used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations of baseline and time-dependent diabetes with the risk of incident kidney cancer.
Results:
During the 25 years of follow-up, 245 cases of kidney cancer occurred among 36,975 post-menopausal women. In an age-adjusted model, there was a significant association between time-dependent diabetes and the risk of kidney cancer [HR (95% CI) = 1.76 (1.26, 1.45)]; the association was attenuated after multivariable adjustment for age, body mass index (BMI), waist-to-hip ratio (WHR), hypertension, physical activity, diuretic use, pack-years of smoking, alcohol intake, and total caloric intake [HR = 1.35 (0.94, 1.94)]. However, among non-obese women or women with a waist circumference less than 34.6 in., diabetes was significantly associated with kidney cancer risk: for time-dependent diabetes, HRs (95% CIs) were 1.82 (1.10, 3.00) among those with BMI < 30 kg/m2 and 2.18 (1.08, 4.38) among those with a waist circumference < 34.6 in..
Conclusions:
Our results suggest that diabetes is associated with kidney cancer risk among non-obese post-menopausal women.
Keywords: Diabetes, Kidney cancer, Post-menopausal woman, Obesity, Prospective cohort
1. Introduction
Kidney cancer has been on the rise over the last five decades representing 3–5% of new cancer cases in the United States [1], with an estimate of 73,750 new cases in 2020 [2]. The reason for the increase is still unclear. Thus, it is critical to identify risk factors for kidney cancer to design prevention strategies and aid in early detection and treatment of the disease.
Factors related to weight, such as body mass index (BMI), waist-to-hip ratio (WHR) and waist circumference, along with hypertension and smoking, have been consistently associated with the onset of kidney cancer [3–5]. Previous epidemiologic findings from the Iowa Women’s Health Study (IWHS) have also shown evidence of an association between weight-related measures and kidney cancer risk [BMI (Q4 vs. Q1): RR(95% CI) = 2.49 (2.39, 4.44)] [6]. Dietary factors may also play a role in this cancer development. Diet rich in calcium, vitamin C, vitamin E, and fiber may reduce the likelihood of developing this disease, although not all findings have been consistent [6–13].
Many studies have linked a history of diabetes to the increased risk of kidney cancer. The latest meta-analysis by Bao et al. showed that diabetes is associated with a 40% increased risk of kidney cancer [95% CI (1.16, 1.69)] [14]. However, previous findings are inconsistent regarding whether the diabetes-kidney cancer association is independent of obesity and other risk factors [15–19]. Thus, we aimed to examine the risk of kidney cancer associated with diabetes after adjusting for weight measures and other risk factors, in the IWHS cohort, a population-based prospective cohort of post-menopausal women. We also examined whether or not this association varied by weight measures and by duration of diabetes. The main strengths of our study are that we assessed diabetes at several time points, accounted for diabetes duration, and evaluated interactions between diabetes and well-established risk factors for kidney cancer.
2. Methods
2.1. Study population and data collection
The IWHS was initiated in 1986 when a questionnaire was mailed to 99,826 women who were randomly sampled from a driver’s license list provided by the Iowa Department of Transportation. A total of 41,836 women (approximately 42%) who responded to the baseline questionnaire constituted the IWHS cohort. To update risk factor information, subsequent questionnaires were mailed, with response rates indicated, in the following years: 1987 (91%), 1989 (89%), 1992 (83%), 1997 (79%), and 2004 (70%).
The baseline questionnaire asked about demographic and anthropometric characteristics, smoking, alcohol intake, diet, reproductive history, hormone therapy use, and medical history. At baseline, diabetes status was determined for each participant by one of two ways: 1) Subjects answered yes to the question: “Have you ever been told by a doctor that you have sugar diabetes (diabetes mellitus)?” or 2) Subjects indicated use of either insulin or pills for diabetes. Incident diabetes diagnosed by a physician was also reported by each participant on each follow-up questionnaire. Participants reporting a diagnosis of diabetes also reported their age at diagnosis by a physician. For this study, participants who reported diabetes onset before age 30 years (presumably, type 1 diabetes) were excluded; thus, the marjority of the diabetes cases were presumptive type 2 diabetes.
BMI was calculated from self-reported weight and height at baseline as weight in kilograms divided by height in meters squared. To calculate WHR, a tape measure was included with the baseline questionnaire; subjects were asked to have someone measure their waist, one inch above the umbilicus and their hips at the widest point. Previously, these measures have been determined to have both a high degree of accuracy and reliability [20]. Women also completed a baseline 127-item Willett food frequency questionnaire (FFQ) that queried about usual food and beverage intake over the previous year [21].
2.2. Kidney cancer ascertainment
To determine the number of incident kidney cancer cases from 1986 to 2011, the IWHS was linked to the State Health Registry of Iowa, part of the National Cancer Institute’s Surveillance, Epidemiology, and End Results Program; 257 cases of kidney cancer were identified (ICD-O-3 code: C64). Cancer stage was also provided by the registry. For each woman, total person-time was determined from the time of baseline until diagnosis of kidney cancer, death, leaving the state of Iowa, or the end of follow-up on December 31st, 2011, whichever occurred first.
2.3. Statistical analysis
All analyses were performed at a significant level of p = 0.05 using SAS statistical software (version 9.4, SAS Institute Inc, Cary, NC). We excluded women who reported a history of cancer (n = 3830) and women who were not considered post-menopausal at baseline (n = 547). Additionally, we excluded women who did not have a complete FFQ (i.e. number of missing items greater than 30 or had total caloric intake less than 600 or greater than or equal to 5000 kcal/day, n = 484) resulting in 36,975 women in the analytic cohort. Baseline dietary and lifestyle/medical characteristics were examined across diabetes status as mean ± standard deviation (SD) or percentage (%). We used Cox proportional hazards regression to calculate hazard ratios (HRs) for time to kidney cancer and 95% confidence intervals (CIs) in relation to the baseline diabetes, anthropometric measures, dietary factors, and lifestyle/medical characteristics. To account for diabetes that occurred during follow-up, we modeled diabetes as a time-dependent variable, which is defined as a variable whose value for a given subject may change over time [22]. Thus, women with time-dependent diabetes are those who had diabetes at baseline or developed diabetes over follow-up. We used an extended Cox proportional hazards regression to investigate the association between time-dependent diabetes and time to kidney cancer.
The associations between diabetes (baseline and time-dependent) and kidney cancer risk were examined using three models: adjusted for age (continuous) (Model 1), adjusted for age and BMI (< 25, 25−29.9, ≥30 kg/m2) (Model 2), and multivariable adjusted (Model 3). Demographic characteristics, lifestyle/medical factors, anthropometrics and dietary factors associated with diabetes and kidney cancer risk were examined as potential confounders and included in the final model if the change between the crude and adjusted parameter estimates was more than 10%. Thus, age (continuous), BMI (< 25, 25−29.9, ≥30 kg/m2), WHR (continuous), physical activity (high or moderate/low), pack-years of smoking (continuous), total caloric intake (continuous), hypertension (yes/no), diuretic use (yes/no) and alcohol intake (yes/no) were included in the final multivariable-adjusted model. A Pearson correlation coefficient (r = 0.40) between BMI and WHR indicated that these variables were not highly correlated and could be incorporated into the same model. We did not incorportate calcium intake into the model because adjustment for this variable did not change the association.
To test if adding diabetes to the multivariable model improves prediction of kidney cancer risk, we used logistic regression to compare c-index in several models: base model (adjusted for age, pack-years of smoking, hypertension, alcohol intake, diuretic use, physical activity, fiber intake, and total caloric intake); base model additionally adjusted for time-dependent diabetes and diabetes duration; base model adjusted for weight measures (BMI and waist-hip-ratio), and the fully adjusted model (base model + weight measures + time-dependent diabetes + diabetes duration). We also tested whether adding diabetes to the multivariable model improves model fit. We used Cox proportional hazard regression to compare Akaike’s Information Criterion (AIC) values for the base model; base model additionally adjusted for time-dependent diabetes; base model additionally adjusted for weight measures; and the fully adjusted model (base model + time-dependent diabetes + weight measures).
In addition, we examined whether BMI, WHR, waist circumference, or hypertension modified the association between time-dependent diabetes and kidney cancer since these risk factors are strongly related to the main exposure of interest [23–25]. For this analysis, BMI, WHR, and waist circumference were divided into two categories based on the WHO-standard for obesity in women using cutoff values of 30 kg/m2, 0.85, and 34.6 in., respectively [26]. We also conducted an analysis of the association between time-dependent diabetes and kidney cancer stratified by stage at diagnosis (in situ and local vs. regional and distant).
Finally, we analyzed the associations between duration of diabetes and kidney cancer using a nested-case control design within the IWHS cohort: 10 controls were matched to each case by age and vital status at the year of the cancer diagnosis. This resulted in 217 cases and 1977 controls for analysis. Total diabetes duration was calculated by subtracting each individual’s diabetes diagnosis year from their kidney cancer diagnosis year for cases or reference dates for controls.
3. Results
Among 36,975 women with a mean ± SD age of 61.7 ± 4.2 years, 2353 women (6.4%) reported having diabetes at baseline. An additional 3140 (8.5%) women reported a new diabetes diagnosis over the course of the follow-up period.
Women who reported diabetes at baseline and over the course of follow-up had a slightly higher BMI and WHR, were more likely to have hypertension and use diuretics, and had attained a lower level of education. The women with diabetes were also less likely to engage in high/moderate physical activity, and consume alcohol, calcium, and vitamin E compared to women without diabetes (Table 1).
Table 1.
Selected baseline characteristics according to self-reported diabetes status at baseline and during follow-up, IWHS (1986–2011).
| Baseline characteristicsa | Non-diabetes | Diabetes | |||
|---|---|---|---|---|---|
| At baseline | Over follow-up | P-valueb | P-valuec | ||
| N = 31,482 | N = 2353 | N = 3140 | |||
| Median person-years | 24.7 | 16.0 | 25.4 | < 0.01 | 0.01 |
| Lifestyle/Medical Factors | |||||
| Age, years (SD) | 61.7 (4.2) | 62.5 (4.2) | 61.5 (4.1) | < 0.01 | < 0.01 |
| Education more than high school (%) | 29.7 | 21.9 | 24.4 | < 0.01 | < 0.01 |
| Married (%) | 77.1 | 73.1 | 76.3 | < 0.01 | 0.31 |
| Current smoker (%) | 15.4 | 12.2 | 11.8 | < 0.01 | < 0.01 |
| Pack-years of smoking (SD) | 9.4 (17.6) | 10.0 (20.3) | 8.4 (16.8) | 0.14 | < 0.01 |
| Alcohol drinker (%) | 46.0 | 21.7 | 37.3 | < 0.01 | < 0.01 |
| BMI (kg/m2) (SD) | 26.4 (4.7) | 30.3 (6.4) | 30.3 (5.5) | < 0.01 | < 0.01 |
| WHR (SD) | 0.83 (0.08) | 0.90 (0.10) | 0.88 (0.08) | < 0.01 | < 0.01 |
| High/moderate physical activity (%) | 54.1 | 45.3 | 44.3 | < 0.01 | < 0.01 |
| Parity (# of births) (SD) | 3.0 (1.9) | 3.2 (2.1) | 3.3 (2.0) | < 0.01 | < 0.01 |
| Hormone therapy use (%) | 38.9 | 35.3 | 36.8 | < 0.01 | 0.02 |
| Diuretic use (%) | 33.0 | 58.2 | 49.1 | < 0.01 | < 0.01 |
| Hypertension (%) | 33.9 | 67.4 | 54.2 | < 0.01 | < 0.01 |
| Dietary Factors | |||||
| Total caloric intake (kcal/day) (SD) | 1784 (612) | 1737 (604) | 1836 (640) | < 0.01 | < 0.01 |
| Vitamin E intake (IU/day) (SD) | 68.2 (150.6) | 60.1 (146.4) | 59.0 (138.9) | 0.01 | < 0.01 |
| Fiber intake (g/day) (SD) | 5.3 (2.4) | 5.5 (2.4) | 5.4 (2.4) | < 0.01 | 0.71 |
| Calcium intake (mg/day) (SD) | 1094 (556) | 1066 (559) | 1071 (556) | 0.02 | 0.03 |
Continuous variables presented as mean (SD) and categorical variables presented as percentage.
P-value indicated the difference between individuals without diabetes and with diabetes at baseline. They were calculated using chi-square for categorical variables and t-test for continuous variables except median person-years. The P-value for median person-years was calculated by non-parametric test.
P-value indicated the difference between individuals without diabetes and with diabetes diagnosed over follow-up. They were calculated using chi-square for categorical variables and t-test for continuous variables except median person-years. The P-value for median person-years was calculated by non-parametric test.
Several characteristics were statistically significantly associated with kidney cancer risk (Table 2). Women who had BMI ≥ 30 kg/m2 had a 129% excess risk of kidney cancer [95% CI= (1.68, 3.12)] compared to those with BMI < 25 kg/m2 (p-trend < 0.01). Women with a WHR in the third and fourth quartiles versus those in the lowest quartile had a 93% and 136% excess risk of kidney cancer, respectively (p-trend < 0.01). Women who reported having hypertension versus those without hypertension had an 80% greater risk of kidney cancer [95% CI= (1.39, 2.32)]. Women who reported current use of diuretics versus non-users of diuretics had a 53% greater risk of kidney cancer (p-trend = 0.01). Finally, women consuming 10 or more grams of alcohol per day versus non-drinkers had a 60% less risk of kidney cancer [HR (95% CI) = 0.40 (0.22, 0.73)].
Table 2.
Age-adjusted Cox proportional hazards regression for the associations between baseline characteristics and kidney cancer risk, IWHS (1986–2011).
| Cases (N) | Person-years | Hazard ratio (95% CI) | P-value or P-trenda | |
|---|---|---|---|---|
| Lifestyle/Medical Factors | ||||
| Marital status | ||||
| Currently married | 185 | 581,800 | 1.00 (ref) | |
| Other | 60 | 161,919 | 1.14 (0.85, 1.53) | 0.39 |
| Education | ||||
| High school graduate | 128 | 392,321 | 1.00 (ref) | |
| Less than high school | 48 | 139,958 | 1.04 (0.75, 1.45) | |
| Some college | 69 | 215,238 | 0.98 (0.73, 1.31) | 0.75 |
| Smoking status Never smoker | 172 | 505,543 | 1.00 (ref) | |
| Former smoker | 39 | 137,494 | 0.86 (0.61, 1.22) | |
| Current smoker | 29 | 94,999 | 0.98 (0.66, 1.45) | 0.66 |
| Pack-years of smoking | 245 | 749,274 | 1.00 (0.99, 1.01) | 0.83 |
| Alcohol (g/day) | ||||
| 0 | 152 | 420,009 | 1.00 (ref) | |
| 0−10 | 81 | 244,698 | 0.92 (0.70, 1.21) | |
| ≥ 10 | 12 | 84,568 | 0.40 (0.22, 0.73) | < 0.01 |
| BMI (kg/m2) | ||||
| < 25 | 71 | 296,116 | 1.00 (ref) | |
| 25−29.9 | 81 | 281,853 | 1.19 (0.87, 1.64) | |
| ≥ 30 | 93 | 171,306 | 2.29 (1.68, 3.12) | < 0.01 |
| WHR | ||||
| Quartile 1 (0.34−0.77) | 39 | 195,407 | 1.00 (ref) | |
| Quartile 2 (0.78−0.83) | 52 | 190,170 | 1.36 (0.90, 2.06) | |
| Quartile 3 (0.84−0.89) | 72 | 186,270 | 1.93 (1.30, 2.85) | |
| Quartile 4 (0.90−2.84) | 82 | 174,678 | 2.36 (1.61, 3.47) | < 0.01 |
| Physical activity | ||||
| Low | 125 | 341,376 | 1.00 (ref) | |
| Moderate | 63 | 205,583 | 0.82 (0.61, 1.11) | |
| High | 51 | 187,951 | 0.73 (0.52, 1.00) | 0.04 |
| Parity (# of births) | 245 | 749,274 | 1.03 (0.96, 1.10) | 0.41 |
| Hormone therapy use | ||||
| Never | 134 | 459,639 | 1.00 (ref) | |
| Current / Former | 111 | 287,183 | 1.32 (1.03, 1.70) | 0.03 |
| Diuretic use | ||||
| No | 135 | 475,956 | 1.00 (ref) | |
| Yes, but not currently | 46 | 123,749 | 1.34 (0.96, 1.87) | |
| Yes, currently | 54 | 126,257 | 1.53 (1.11, 2.09) | 0.01 |
| Hypertension | ||||
| No | 118 | 470,843 | 1.00 (ref) | |
| Yes | 117 | 262,503 | 1.80 (1.39, 2.32) | < 0.01 |
| Dietary Factors | ||||
| Fiber intake | ||||
| 1st Quartile | 71 | 184,604 | 1.00 (ref) | |
| 2nd Quartile | 57 | 184,841 | 0.80 (0.56, 1.13) | |
| 3rd Quartile | 55 | 187,342 | 0.75 (0.53, 1.07) | |
| 4th Quartile | 62 | 192,487 | 0.82 (0.58, 1.15) | 0.23 |
| Calcium intake | ||||
| 1st Quartile (72.2−661.4) | 75 | 183,547 | 1.00 (ref) | |
| 2nd Quartile (661−1018) | 53 | 187,706 | 0.68 (0.48, 0.97) | |
| 3rd Quartile (1018−1395) | 63 | 190,048 | 0.81 (0.58, 1.13) | |
| 4th Quartile (1395−5076) | 54 | 187,974 | 0.70 (0.49, 0.99) | 0.09 |
| Total caloric intake | 245 | 749,274 | 1.00 (1.00, 1.00) | 0.98 |
P-value for continuous and binary variables and P-trend for categorical variables with more than two categories.
The associations of baseline and time-dependent diabetes with incident kidney cancer are shown in Table 3. Time-dependent diabetes was associated with an increased risk of kidney cancer in Models 1 and 2. However, these associations were attenuated and no longer statistically significant after adjustment for additional confounders [HR (95% CI) = 1.35 (0.94, 1.94)]. Results were not statistically significant for baseline diabetes in any of the models. Furthermore, we assessed whether BMI, WHR, waist circumference, or hypertension modified the relation between diabetes and kidney cancer but did not detect any statistically significant interactions (all P for interaction > 0.10) (Table 4). However, for the lower category of each risk factor, the risk of kidney cancer was increased by 47–118% for those with diabetes versus those without, with the statistically significant associations observed among those with lower BMI or lower waist circumference (Table 4). In the analysis stratified by cancer stage, diabetes was associated with increased risk of kidney cancer in an age-adjusted model for patients with regional and distant disease only, however, the association lost statistical significance after adjusting for additional confounders (Supplemental Table 1). The c-index for the base model was 0.59. After adding time-dependent diabetes and diabetes duration to the base model, the c-index improved to 0.62. Likewise, after adding weight measures (BMI and WHR) to the base model, the c-index improved to 0.63. The c-index for the fully adjusted model (base model + weight measures + time-dependent diabetes + diabetes duration) was 0.65. The AIC for the base model was 4329.818. After adding time-dependent diabetes to the base model, the AIC value decreased to 4324.593. After adding weight measures to the base model, the AIC value was 4303.866. The AIC for the fully adjusted model (base model + time-dependent diabetes + weight measures) was 4303.470.
Table 3.
Age and multivariable adjusted Cox proportional hazards regression for the association between diabetes and kidney cancer risk, IWHS (1986–2011).
| Variable | Cases (N) | Person-years | Model 1a | Model 2b | Model 3c |
|---|---|---|---|---|---|
| HR (95% CI) | HR (95% CI) | HR (95% CI) | |||
| Baseline Diabetes | |||||
| No | 227 | 712,121 | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) |
| Yes | 18 | 37,153 | 1.60 (0.99, 2.59) | 1.33 (0.82, 2.16) | 1.30 (0.79, 2.14) |
| Time-dependent Diabetes | |||||
| No | 202 | 672,466 | 1.00 (ref) | 1.00 (ref) | 1.00 (ref) |
| Yes | 43 | 76,808 | 1.76 (1.26, 2.45) | 1.43 (1.01, 2.01) | 1.35 (0.94, 1.94) |
Model 1 is adjusted for age (continuous).
Model 2 is adjusted for age (continuous) and BMI (< 25, 25–29.9, ≥30 kg/m2).
Model 3 is adjusted for age (continuous), BMI (< 25, 25–29.9, ≥30 kg/m2), WHR (continuous), pack-years of smoking (continuous), hypertension (yes/no), alcohol intake (yes/no), diuretic use (yes/no), high/moderate physical activity (yes/no), total caloric intake (continuous).
Table 4.
The interaction between diabetesa and weight characteristics and hypertension in relation to kidney cancer risk, IWHS (1986–2011).
| Variable | Cases (N) | Person-years | Hazard ratiob (95% CI) | Hazard ratioc(95% CI) |
|---|---|---|---|---|
| BMI < 30 kg/m2 | ||||
| No diabetes | 133 | 537,527 | 1.00 (ref) | 1.00 (ref) |
| Diabetes | 19 | 40,441 | 1.76 (1.08–2.85) | 1.82 (1.10, 3.00) |
| BMI ≥ 30 kg/m2 | ||||
| No diabetes | 69 | 134,939 | 1.00 (ref) | 1.00 (ref) |
| Diabetes | 24 | 36,367 | 1.22 (0.76, 1.94) | 1.24 (0.76, 2.02) |
| P for interaction | 0.30 | 0.34 | ||
| Waist Circumference < 34.6 in.d | ||||
| No diabetes | 82 | 391,922 | 1.00 (ref) | 1.00 (ref) |
| Diabetes | 9 | 19,713 | 1.94 (0.97, 3.88) | 2.18 (1.08, 4.38) |
| Waist Circumference ≥ 34.6 in. | ||||
| No diabetes | 111 | 256,151 | 1.00 (ref) | 1.00 (ref) |
| Diabetes | 33 | 53,804 | 1.36 (0.92, 2.01) | 1.32 (0.88, 2.00) |
| P for interaction | 0.31 | 0.25 | ||
| WHR ≤ 0.84d | ||||
| No diabetes | 101 | 421,970 | 1.00 (ref) | 1.00 (ref) |
| Diabetes | 10 | 25,674 | 1.43 (0.75, 2.75) | 1.60 (0.83, 3.12) |
| WHR > 0.84 | ||||
| No diabetes | 101 | 248,089 | 1.00 (ref) | 1.00 (ref) |
| Diabetes | 33 | 50,790 | 1.56 (1.05, 2.32) | 1.48 (0.98, 2.25) |
| P for interaction | 0.96 | 0.83 | ||
| No Hypertension | ||||
| No diabetes | 106 | 440,975 | 1.00 (ref) | 1.00 (ref) |
| Diabetes | 12 | 29,868 | 1.55 (0.85, 2.82) | 1.47 (0.78, 2.77) |
| Hypertension | ||||
| No diabetes | 86 | 217,418 | 1.00 (ref) | 1.00 (ref) |
| Diabetes | 31 | 45,085 | 1.67 (1.10, 2.52) | 1.76 (1.14, 2.71) |
| P for interaction | 0.83 | 0.64 |
Time-dependent diabetes.
Adjusted for age (continuous).
Adjusted for age (continuous), pack-years of smoking (continuous), hypertension (yes/no), alcohol intake (yes/no), diuretic use (yes/no), high/moderate physical activity (yes/no), and total caloric intake (continuous).
30 kg/m2, 0.84, and 34.6 in. was chosen as the cutoff for BMI, WHR, and waist circumference, respectively based on WHO standard for “obese” women.
In addition, we analyzed the association between duration of diabetes and kidney cancer risk using a matched case-control (1:10) study design (Table 5). Compared to people without diabetes, the odds ratios (95% CIs) for participants who had diabetes for less than 5 years and more than 5 years were 1.68 (0.88, 3.20) and 0.99 (0.59, 1.61) (p-trend = 0.8).
Table 5.
Odds ratio of kidney cancer in relation to diabetes duration in a matched case-control (1:10) study nested within IWHS, (1986–2011).
| Diabetes duration | Cases (N) | Controls (N) | Odds ratioa(95% CI) | P-trend |
|---|---|---|---|---|
| No diabetes | 179 | 1793 | 1.00 (Ref) | |
| < 5 years | 14 | 47 | 1.68 (0.88, 3.20) | |
| ≥ 5 years | 24 | 137 | 0.98 (0.60, 1.62) | 0.80 |
Adjusted for age (continuous), BMI (continuous), WHR (continuous), pack-years of smoking (continuous), hypertension (yes/no), alcohol intake (yes/no), diuretic use (yes/no), high/moderate physical activity (yes/no), total caloric intake (continuous).
4. Discussion
In this large prospective study of post-menopausal women, we found no significant associations of time-dependent diabetes with incident kidney cancer after multivariable adjustment for age, BMI, WHR, pack-years of smoking, hypertension, alcohol intake, diuretic use, high/moderate physical activity, and total caloric intake. However, among women with a BMI < 30 kg/m2 or a waist circumference < 34.6 in., time-dependent diabetes was associated with statistically significantly increased risk of kidney cancer even after adjusting for multiple confounders.
The association between diabetes and kidney cancer risk has been examined in many studies. Two meta-analyses summarized this association, but both of them included studies that varied in regard to the variables included in their models: some adjusted just for age while others adjusted for smoking status, BMI, diet factors, and medical history. The meta-analysis of nine cohort studies by Larsson et al. (2011) indicated a statistically significantly increased risk of kidney cancer associated with diabetes: RRs (95% CIs) were 1.70 (1.47, 1.97) for women and 1.26 (1.06, 1.49) for men. When restricting to the three studies which had adjusted for BMI or obesity, they found a null association [RR (95% CI) = 1.12 (0.99, 1.27)] [27]. The latest meta-analysis by Bao et al. (2013), which summarized findings from 11 cohort and 7 case-control studies, also reported a positive association between diabetes and an increased risk of kidney cancer that remained after adjustment for BMI. Similar to the meta-analysis by Larsson et al. (2011), the association was stronger in females [RR (95% CI) = 1.47 (1.18, 1.83)] versus males [RR (95% CI) = 1.28 (1.10, 1.48)] [14]. In parallel to our study, the Vitamin and Lifestyle study found a significant association in unadjusted but not in multivariable-adjusted model [HR (95% CI) = 1.39 (0.92, 2.09)] [28], while the population-based case-control study in Taiwan found no association in either unadjusted or multivariable-adjusted model [29]. In contrast, a recent prospective study including participants from the Nurses’ Health Study (NHS) and Health Professionals Follow-Up Study (HPFS), found a 53% increased risk of renal cell carcinoma among women with diabetes versus those without diabetes, after multivariable adjustment [19]. These inconsistencies may be explained by different study populations. Importantly, the parameter estimates were very similar in our and other studies and the meta-analyses.
Several mechanisms may explain the association between diabetes and kidney cancer risk. First, patients with type 2 diabetes (the majority of diabetes cases in our study) have high serum levels of insulin that promote the secretion and production of insulin-like growth factor 1 (IGF-1). IGF-1, which is important in the regulation of cell proliferation and differentiation, can promote the formation and growth of tumors [18,30]. Second, type 2 diabetes is associated with high blood glucose levels that may facilitate increased uncontrolled cell growth and division [30]. This mechanism is supported by findings from the clinical study of 310 kidney cancer patients that reported a more aggressive kidney cancer in diabetic versus non-diabetic cancer patients [31]. Additionally, the finding of an increased risk of kidney cancer associated with diabetes among normal weight women but not among obese women might reflect differences in the metabolic health among reference groups, i.e., non-diabetic women with normal and increased BMI. Obese women even without diabetes may have abnormal metabolic profiles that may result in their increased risk of cancer and lack of association between diabetes and kidney cancer risk in this group [32–36].
We found an inverse association between alcohol intake (dichotomized at 10 g/day) and the risk of kidney cancer. One potential explanation is that the diuretic effect of alcohol may decrease the concentration of carcinogens, so that the carcinogens stay in kidneys for a shorter time [37]. Moreover, moderate alcohol consumption (less than 30 g/day) was shown to be associated with reduced insulin concentration [38]. Therefore, alcohol intake may protect against kidney cancer through several potential mechanisms such as decreasing the carcinogen concentration or reducing insulin concentration [37].
The c-indexes for the models showed that the fully adjusted model provided the best prediction. The improvement of fit after adding time-dependent diabetes and diabetes duration is comparable to the improvement of fit that happened after adding weight measures to the model, which are known as a strong risk factor for kidney cancer. Likewise, the AIC values indicated the fully adjusted model has the best fit.
The lack of significance in our study may be attributed to the simultaneous adjustment for BMI, WHR, and hypertension. When adjusted only for (age and BMI) or (age and WHR), the association in our study was significant, which mirrored the results similar the meta-analysis by Bao et al. (2013) [14]. Also, the lack of an association in our study could be attributed to older age [median age at kidney cancer diagnosis = 75.5; range 56–91 years] of our study population. For instance, Japanese retrospective study reported that diabetes may be an independent predictor for recurrence of kidney cancer among patients younger than 65 years old, but not among older patients [39]. They suggested that younger people are in the earlier phase of type 2 diabetes and have an increased bioavailability of IGF-1, promoting the tumor growth, whereas older people who live with diabetes for a long time develop hypoinsulinemia that decreases the IGF-1 bioavailability. It is possible that diabetes which is not characterized by high IGF-1 levels is not associated with kidney cancer similar to what we observed in our cohort of older women [39]. Another potential explanation is that older people who, on average, live with diagnosed diabetes for a longer time have their diabetes under control. This explanation is in line with the stronger associations observed among those with shorter diabetes duration, i.e. those who are more likely to have uncontrolled diabetes, in our and several other prospective cohort studies [17–19]. However, contrary to our results, the NHS and NHS study combined with HPFF found an association even after simultaneous adjustment for BMI and hypertension [18,19]. Thus, the lack of a statistically significant association between diabetes and kidney cancer risk in our cohort may reflect a true null association after adjustment for all the confounders, may be explained by an older age of the IWHS participants including those with the kidney cancer, or may be driven by a limited number of kidney cancer cases in our study.
Another limitation in our study is that exposures were assessed using questionnaires which could lead to non-differential misclassification given that diabetes was ascertained before cancer diagnosis [24]. However, a validation study in the IWHS showed that measures of BMI and WHR were accurate and reliable [40], which minimizes the potential for bias. An additional limitation is that only 63.6% of diabetes cases were confirmed by a physician as shown by a small validation study of 44 self-reported diabetes cases at baseline. Hence, misclassification of diabetes is possible and could result in bias, most likely towards the null given that diabetes was ascertained before cancer diagnosis [24]. Furthermore, the IWHS has very limited data about hereditary disease; for instance, it did not collect information on Von Hippel-Lindau syndrome, which increases the risk of developing clear cell kidney cancer [41,42]. Finally, our study population was comprised of primarily non-Hispanic white post-menopausal women (99.2%). As a result, these findings may not be generalizable to males or other racial-ethnic groups. However, our study has several important strengths: long-term follow-up of up to 25 years with low loss to follow-up (1% annually), reliable ascertainment of cancer cases, detailed information about potential confounders, as well as information about diabetes duration.
In conclusion, an association between diabetes and kidney cancer was not statistically significant among the whole cohort. However, including diabetes in addition to weight measures (BMI and WHR) to the model slightly improved model fit and should be considered when predicting the risk of kidney cancer. In addition, a positive statistically significant association was observed among non-obese women (BMI < 30 kg/m2 or waist circumference < 34.6 in.), but the number of cases with diabetes in each category was limited. Although these findings should be validated in larger or pooled prospective studies, patients with new onset of diabetes may require more thorough surveillance for cancer including kidney cancer.
Supplementary Material
Acknowledgments
Funding
This research was supported by the National Cancer Institute at the National Institutes of Health under Grant R01 CA39742.
Footnotes
Ethical approval
The IWHS has been approved by the University of Minnesota’s Institutional Review Board.
Research data (data sharing and collaboration)
There are no linked research data sets for this paper. The data are confidential, and the authors do not have permission to share data.
Provenance and peer review
This article was not commissioned and was externally peer reviewed.
Declaration of Competing Interest
The authors report no declarations of interest.
Appendix A. Supplementary data
Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.maturitas.2020.07.010.
References
- [1].Ng CS, Wood CG, Silverman PM, Tannir NM, Tamboli P, Sandler CM, Renal cell carcinoma: diagnosis, staging, and surveillance, AJR Am. J. Roentgenol 191 (4) (2008) 1220–1232. [DOI] [PubMed] [Google Scholar]
- [2].Siegel RL, Miller KD, Jemal A, Cancer statistics, 2020, CA Cancer J. Clin 70 (1) (2020) 7–30. [DOI] [PubMed] [Google Scholar]
- [3].Renehan AG, Tyson M, Egger M, Heller RF, Zwahlen M, Body-mass index and incidence of cancer: a systematic review and meta-analysis of prospective observational studies, Lancet 371 (9612) (2008) 569–578. [DOI] [PubMed] [Google Scholar]
- [4].Adams KF, Leitzmann MF, Albanes D, Kipnis V, Moore SC, Schatzkin A, Chow WH, Body size and renal cell cancer incidence in a large US cohort study, Am. J. Epidemiol 168 (3) (2008) 268–277. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Pischon T, Lahmann PH, Boeing H, Tjonneland A, Halkjaer J, Overvad K, Klipstein-Grobusch K, Linseisen J, Becker N, Trichopoulou A, Benetou V, Trichopoulos D, Sieri S, Palli D, Tumino R, Vineis P, Panico S, Monninkhof E, Peeters PH, Bueno-de-Mesquita HB, Buchner FL, Ljungberg B, Hallmans G, Berglund G, Gonzalez CA, Dorronsoro M, Gurrea AB, Navarro C, Martinez C, Quiros JR, Roddam A, Allen N, Bingham S, Khaw KT, Kaaks R, Norat T, Slimani N, Riboli E, Body size and risk of renal cell carcinoma in the European Prospective Investigation into Cancer and Nutrition (EPIC), Int. J. Cancer 118 (3) (2006) 728–738. [DOI] [PubMed] [Google Scholar]
- [6].Nicodemus KK, Sweeney C, Folsom AR, Evaluation of dietary, medical and lifestyle risk factors for incident kidney cancer in postmenopausal women, Int. J. Cancer 108 (1) (2004) 115–121. [DOI] [PubMed] [Google Scholar]
- [7].Hu J, La Vecchia C, DesMeules M, Negri E, Mery L, Nutrient and fiber intake and risk of renal cell carcinoma, Nutr. Cancer 60 (6) (2008) 720–728. [DOI] [PubMed] [Google Scholar]
- [8].van Dijk BA, Schouten LJ, Oosterwijk E, Hulsbergen-van de Kaa CA, Kiemeney LA, Goldbohm RA, Schalken JA, van den Brandt PA, Carotenoid and vitamin intake, von Hippel-Lindau gene mutations and sporadic renal cell carcinoma, Cancer Causes Control 19 (2) (2008) 125–134. [DOI] [PubMed] [Google Scholar]
- [9].Lee JE, Giovannucci E, Smith-Warner SA, Spiegelman D, Willett WC, Curhan GC, Intakes of fruits, vegetables, vitamins A, C, and E, and carotenoids and risk of renal cell cancer, Cancer Epidemiol. Biomarkers Prev 15 (12) (2006) 2445–2452. [DOI] [PubMed] [Google Scholar]
- [10].Bertoia M, Albanes D, Mayne ST, Mannisto S, Virtamo J, Wright ME, No association between fruit, vegetables, antioxidant nutrients and risk of renal cell carcinoma, Int. J. Cancer 126 (6) (2010) 1504–1512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Huang TB, Ding PP, Chen JF, Yan Y, Zhang L, Liu H, Liu PC, Che JP, Zheng JH, Yao XD, Dietary fiber intake and risk of renal cell carcinoma: evidence from a meta-analysis, Med. Oncol 31 (8) (2014) 125. [DOI] [PubMed] [Google Scholar]
- [12].Hu J, Mao Y, White K, Diet and vitamin or mineral supplements and risk of renal cell carcinoma in Canada, Cancer Causes Control 14 (8) (2003) 705–714. [DOI] [PubMed] [Google Scholar]
- [13].Peterlik M, Grant WB, Cross HS, Calcium, vitamin D and cancer, Anticancer Res. 29 (9) (2009) 3687–3698. [PubMed] [Google Scholar]
- [14].Bao C, Yang X, Xu W, Luo H, Xu Z, Su C, Qi X, Diabetes mellitus and incidence and mortality of kidney cancer: a meta-analysis, J. Diabetes Complic 27 (4) (2013) 357–364. [DOI] [PubMed] [Google Scholar]
- [15].Inoue M, Iwasaki M, Otani T, Sasazuki S, Noda M, Tsugane S, Diabetes mellitus and the risk of cancer: results from a large-scale population-based cohort study in Japan, Arch. Intern. Med 166 (17) (2006) 1871–1877. [DOI] [PubMed] [Google Scholar]
- [16].Setiawan VW, Stram DO, Nomura AM, Kolonel LN, Henderson BE, Risk factors for renal cell cancer: the multiethnic cohort, Am. J. Epidemiol 166 (8) (2007) 932–940. [DOI] [PubMed] [Google Scholar]
- [17].Atchison EA, Gridley G, Carreon JD, Leitzmann MF, McGlynn KA, Risk of cancer in a large cohort of U.S. Veterans with diabetes, Int. J. Cancer 128 (3) (2011) 635–643. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Joh HK, Willett WC, Cho E, Type 2 diabetes and the risk of renal cell cancer in women, Diabetes Care 34 (7) (2011) 1552–1556. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Graff RE, Sanchez A, Tobias DK, Rodriguez D, Barrisford GW, Blute ML, Li Y, Sun Q, Preston MA, Wilson KM, Cho E, Type 2 diabetes in relation to the risk of renal cell carcinoma among men and women in two large prospective cohort studies, Diabetes Care 41 (7) (2018) 1432–1437. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Kushi LH, Kaye SA, Folsom AR, Soler JT, Prineas RJ, Accuracy and reliability of self-measurement of body girths, Am. J. Epidemiol 128 (4) (1988) 740–748. [DOI] [PubMed] [Google Scholar]
- [21].McCarl M, Harnack L, Limburg PJ, Anderson KE, Folsom AR, Incidence of colorectal cancer in relation to glycemic index and load in a cohort of women, Cancer Epidemiol. Biomarkers Prev 15 (5) (2006) 892–896. [DOI] [PubMed] [Google Scholar]
- [22].Kleinbaum DG, Klein M, Survival Analysis: A Self-learning Text, 2 ed., Springer-Verlag, New York, 2005. [Google Scholar]
- [23].Goran MI, Ball GD, Cruz ML, Obesity and risk of type 2 diabetes and cardiovascular disease in children and adolescents, J. Clin. Endocrinol. Metab 88 (4) (2003) 1417–1427. [DOI] [PubMed] [Google Scholar]
- [24].Kaye SA, Folsom AR, Sprafka JM, Prineas RJ, Wallace RB, Increased incidence of diabetes mellitus in relation to abdominal adiposity in older women, J. Clin. Epidemiol 44 (3) (1991) 329–334. [DOI] [PubMed] [Google Scholar]
- [25].Epstein M, Sowers JR, Diabetes mellitus and hypertension, Hypertension 19 (5) (1992) 403–418. [DOI] [PubMed] [Google Scholar]
- [26].Circumference Waist and Ratio Waist-Hip : Report of a WHO Expert Consultation, World Health Organization, Geneva, 2008, pp. 8–11 December 2008. https://apps.who.int/iris/handle/10665/44583. [Google Scholar]
- [27].Larsson SC, Wolk A, Diabetes mellitus and incidence of kidney cancer: a meta-analysis of cohort studies, Diabetologia 54 (5) (2011) 1013–1018. [DOI] [PubMed] [Google Scholar]
- [28].Macleod LC, Hotaling JM, Wright JL, Davenport MT, Gore JL, Harper J, White E, Risk factors for renal cell carcinoma in the VITAL study, J. Urol 190 (5) (2013) 1657–1661. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Lai SW, Liao KF, Lai HC, Tsai PY, Sung FC, Chen PC, Kidney cancer and diabetes mellitus: a population-based case-control study in Taiwan, Ann. Acad. Med. Singap 42 (3) (2013) 120–124. [PubMed] [Google Scholar]
- [30].Zelenko Z, Gallagher EJ, Diabetes and cancer, Endocrinol. Metab. Clin. North Am 43 (1) (2014) 167–185. [DOI] [PubMed] [Google Scholar]
- [31].Otunctemur A, Ozbek E, Sahin S, Dursun M, Besiroglu H, Koklu I, Erkoc M, Danis E, Bozkurt M, Gurbuz A, Diabetes mellitus as a risk factor for high grade renal cell carcinoma, Asian Pac. J. Cancer Prev 15 (9) (2014) 3993–3996. [DOI] [PubMed] [Google Scholar]
- [32].Bo S, Musso G, Gambino R, Villois P, Gentile L, Durazzo M, Cavallo-Perin P, Cassader M, Prognostic implications for insulin-sensitive and insulin-resistant normal-weight and obese individuals from a population-based cohort, Am. J. Clin. Nutr 96 (5) (2012) 962–969. [DOI] [PubMed] [Google Scholar]
- [33].Korduner J, Bachus E, Jujic A, Magnusson M, Nilsson PM, Metabolically healthy obesity (MHO) in the Malmö diet cancer study - Epidemiology and prospective risks, Obes. Res. Clin. Pract 13 (6) (2019) 548–554. [DOI] [PubMed] [Google Scholar]
- [34].Murphy N, Cross AJ, Abubakar M, Jenab M, Aleksandrova K, Boutron-Ruault MC, Dossus L, Racine A, Kühn T, Katzke VA, Tjønneland A, Petersen KE, Overvad K, Quirós JR, Jakszyn P, Molina-Montes E, Dorronsoro M, Huerta JM, Barricarte A, Khaw KT, Wareham N, Travis RC, Trichopoulou A, Lagiou P, Trichopoulos D, Masala G, Krogh V, Tumino R, Vineis P, Panico S, Bueno-de-Mesquita HB, Siersema PD, Peeters PH, Ohlsson B, Ericson U, Palmqvist R, Nyström H, Weiderpass E, Skeie G, Freisling H, Kong SY, Tsilidis K, Muller DC, Riboli E, Gunter MJ, A Nested Case-Control Study of Metabolically Defined Body Size Phenotypes and Risk of Colorectal Cancer in the European Prospective Investigation into Cancer and Nutrition (EPIC), PLoS Med 13 (4) (2016) e1001988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [35].Kabat GC, Kim MY, Stefanick M, Ho GYF, Lane DS, Odegaard AO, Simon MS, Bea JW, Luo J, Wassertheil-Smoller S, Rohan TE, Metabolic obesity phenotypes and risk of colorectal cancer in postmenopausal women, Int J Cancer 143 (3) (2018) 543–551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [36].Park YM, White AJ, Nichols HB, O’Brien KM, Weinberg CR, Sandler DP, The association between metabolic health, obesity phenotype and the risk of breast cancer, Int J Cancer 140 (12) (2017) 2657–2666. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [37].Cheng G, Xie L, Alcohol intake and risk of renal cell carcinoma: a meta-analysis of published case-control studies, Arch Med Sci 7 (4) (2011) 648–657. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [38].Davies MJ, Baer DJ, Judd JT, Brown ED, Campbell WS, Taylor PR, Effects of moderate alcohol intake on fasting insulin and glucose concentrations and insulin sensitivity in postmenopausal women: a randomized controlled trial, Jama 287 (19) (2002) 2559–2562. [DOI] [PubMed] [Google Scholar]
- [39].Fukushima H, Masuda H, Yokoyama M, Tatokoro M, Yoshida S, Ishioka J, Matsuoka Y, Numao N, Koga F, Saito K, Fujii Y, Kihara K, Diabetes mellitus with obesity is a predictor of recurrence in patients with non-metastatic renal cell carcinoma, Jpn. J. Clin. Oncol 43 (7) (2013) 740–746. [DOI] [PubMed] [Google Scholar]
- [40].Lindblad P, Chow WH, Chan J, Bergström A, Wolk A, Gridley G, McLaughlin JK, Nyren O, Adami HO, The role of diabetes mellitus in the aetiology of renal cell cancer, Diabetologia 42 (1) (1999) 107–112. [DOI] [PubMed] [Google Scholar]
- [41].Kaelin WG Jr., The von Hippel-Lindau gene, kidney cancer, and oxygen sensing, J. Am. Soc. Nephrol 14 (11) (2003) 2703–2711. [DOI] [PubMed] [Google Scholar]
- [42].Chou A, Toon C, Pickett J, Gill AJ, Von Hippel-Lindau syndrome, Front. Horm. Res 41 (2013) 30–49. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.