Dietary fats and their sources in association with the risk of bladder cancer: A pooled analysis of 11 prospective cohort studies

Abstract The effects of fat intake from different dietary sources on bladder cancer (BC) risk remains unidentified. Therefore, the present study aimed to investigate the association between fat intakes and BC risk by merging world data on this topic. Data from 11 cohort studies in the BLadder cancer Epidemiology and Nutritional Determinants (BLEND) study, provided sufficient information on fat intake for a total of 2731 BC cases and 544 452 noncases, which yielded 5 400 168 person‐years of follow‐up. Hazard ratios (HRs), with corresponding 95% confidence intervals (CIs), were estimated using Cox‐regression models stratified on cohort. Analyses were adjusted for total energy intake in kilocalories, gender, smoking status (model‐1) and additionally for sugar and sugar products, beers, wine, dressing and plant‐based and fruits intakes (model‐2). Among women, an inverse association was observed between mono‐unsaturated fatty acids (MUFAs) and BC risk (HR comparing the highest with the lowest tertile: 0.73, 95% CI: 0.58‐0.93, P‐trend = .01). Overall, this preventative effect of MUFAs on BC risk was only observed for the nonmuscle invasive bladder cancer (NMIBC) subtype (HR: 0.69, 95% CI: 0.53‐0.91, P‐trend = .004). Among men, a higher intake of total cholesterol was associated with an increased BC risk (HR: 1.37, 95% CI: 1.16‐1.61, P‐trend = .01). No other significant associations were observed. This large prospective study adds new insights into the role of fat and oils in BC carcinogenesis, showing an inverse association between consumption of MUFAs and the development of BC among women and a direct association between higher intakes of dietary cholesterol and BC risk among men.


| INTRODUCTION
According to the GLOBOCAN cancer statistics in 2020, bladder cancer is the 10th most commonly diagnosed cancer worldwide, with approximately 573 000 new cases and 213 000 deaths. 1 Approximately 75% of BC cases are nonmuscle invasive bladder cancer (NMIBC) characterized by frequent recurrences, which requires intensive treatments and follow-up measures, posing a large burden on the national health care budgets and patient's quality of life. 2 Several epidemiological studies have identified factors that potentially influence BC risk. These factors include gender, smoking, age and occupation. 2,3 In addition, evidence suggests that factors related to lifestyle, physical activity and diet, might also affect the risk of BC. 4,5 Previous research on diet and BC reported that higher intakes of fluid, fruit, vegetables and yoghurt are associated with a reduced risk of BC. 6 In addition, several dietary patterns have been associated with BC risk, 7,8 including a Western diet, which was shown to be associated with a higher BC risk, 9,10 and the Mediterranean diet, which was shown to be inversely associated with BC risk. 10,11 One of the major differences between the Western and the Mediterranean diet is the source of dietary fat. 12 Accordingly, while the Mediterranean dietary fat intake mainly derives from plants such as olives (high in monounsaturated fats), the dietary fat intake from the Western diet mainly derives from animal products (high in saturated fats). 13 This important difference may suggest that the sources of dietary fat might have different effects on BC risk. For example, in vitro studies on the effect of PUFAs on cancer may from anticarcinogenic to carcinogenic. PUFAs mainly found in vegetable oils (arachidonic acid) and meat (linoleic acid) are thought to induce tumor growth, by causing loss of cell viability and apoptosis, 14,15 while PUFAs are mainly found in ruminant meat (conjugated linoleic acid) and fat cold-water fish (docosahexaenoic acids) are thought to reduce tumor growth, by reducing cell proliferation. 16 Epidemiological evidence on the relation between dietary fat and BC and the various effects of different dietary fat sources, however, is scarce and inconclusive. While a Spanish case-control study found that the observed increased BC risk with high intake of monounsaturated fatty acids (MUFAs) disappeared after adjustment for saturated fat, 17 a Japanese case-control study reported an inverse association between both saturated and monounsaturated fat intakes and BC risk. 18 In addition, an observational study from Serbia highlighted the importance of the fat sources when establishing the effect of dietary fat intake on BC. 19 The authors reported an inverse association of sunflower oil and BC risk, while a positive association was observed for animal fat intake.
Due to this current lack of knowledge and contradictory evidence, the present study aims to investigate the association between dietary fat intake from major sources and BC risk by pooling data from 11 prospective cohort studies.

| Study sample
The study was conducted within the Bladder Cancer Epidemiology and Nutritional Determinants (BLEND) consortium. 20

| Data collection and coding
Details of the BLEND consortium protocol and methodology have been provided elsewhere. 20 All included studies used a selfadministered or interview administered food frequency questionnaire (FFQ) that was validated on either food groups, [23][24][25][26][27] and/or energy intake. 24,27,28 The collected dietary fat intake was harmonized and categorized by using the hierarchal Eurocode 2 food coding system developed by the European Union. 29 National specific standard portions sizes for each food item were used to calculate intake in gr/day. As a result of data availability, groups of fat and oils intakes were calculated in grams per day per 1000 kcal (g/1000 kcal/day, nutrient density method) to account for total energy intake and reduce extraneous variation in dietary intakes. 30,31 All fat and oils intakes were energy-adjusted using the nutrient density method (in g/1000 kcal/day) and were categorized into tertiles for individual fat types. 31 Dietary fats were classified as total lipids, total fatty acids, saturated fatty acids (SFAs), MUFAs, polyunsaturated fatty acids (PUFAs) and cholesterol. Also, dietary fat sources included: total fats and oils, plant-based fats and oils, animal fat, cream, butter, margarine, corn oil, soya bean oil, rapeseed oil, grape seed oil, peanut oil, sunflower oil and olive oil in g/1000 kcal/day. In addition, to the information on dietary intake, the BLEND dataset also includes data on study characteristics (eg, design, method of dietary assessment, recall period of dietary intake), geographical region, demographic information (age, gender and ethnicity) and smoking (current/former/never) and its quantity (packs/year), which were measured at the baseline.

| Statistical analysis
Baseline characteristics of the study participants, types of fat and oils and their dietary sources and other potential confounders were compared between case and noncase groups using analysis of variance or independent samples t-test for continuous variables or χ 2 or ANCOVA for categorical variables.
To assess the influence of the different sources of dietary fat and BC risk, Cox proportional hazard regression was used to obtain hazard ratios (HRs) and corresponding 95% confidence intervals (CIs). Based on the adjusted model 2, the P for heterogeneity was calculated using the Wald test. The proportional hazards assumptions were examined graphically 33 and no violation was observed.
Dietary fat intake was divided into three groups based on a tertile ordered distribution: low intake (tertile 1), medium intake (tertile 2) and high intake (tertile 3). The intake of some plant-based fat sources was not variable enough to be categorized into tertiles (ie, corn oil, soya bean oil, rapeseed oil, grapeseed oil, peanut oil and sunflower oil). For these sources we used the median intake as a cut-off to categorize the participants into low and high intake groups.
In the Cox regression model age was used as a time scale, thereby correcting for age in the analysis. Also, the effect of each study was analyzed as a random effect. The Cox regression models were fitted as crude, and adjusted models (adjusted for total energy intake in kilo- T A B L E 2 Hazard ratio (HR) and 95% confidence interval (CI) of the association of fat and oils types, and risk of BC based on tertile of fat and oils T A B L E 3 Hazard ratio (HR) and 95% confidence interval (CI) of the association of fat and oils intake, and risk of BC based on tertile of fat and oils 3 | RESULTS

| Baseline characteristics
Baseline characteristics of the study population are presented in Table 1. were more likely to be male (73% vs 28%). Cases were mainly current (39%) or former smokers (42%), while noncases were more likely to be never smokers (50% Additional baseline characteristics are provided in Table S1.

| Fat types and BC risk
The estimated HRs for the association between fat and oil intakes with BC are presented in Table 2. Overall, we found that higher con-

| Fat sources and BC risk
High consumption of animal fats showed to be associated with an showed to be associated with BC risk (Tables 3 and 4).

| Fat types and BC risk by gender and BC subtypes stratification
Significant heterogeneity between men and women was observed in the associations of MUFAs and total cholesterol intake with BC (P-het = .001 and <.001, respectively T A B L E 4 Hazard ratio (HR) and 95% confidence interval (CI) of the association of different vegetable oils intake according to median of intakes, and risk of BC information on BC subtypes), resulting in low power, thereby hampered the statistical power to find a small effect size.
The positive associations found between MUFA intake and BC risk are in agreement with a recent meta-analysis of observational studies and a Japanese case-control study, also suggesting an inverse association between high intake of MUFAs and BC risk. 18,38 In contrast, two previously conducted cohort studies on MUFAs intake and BC risk reported a null association. 37,39 Moreover, a Spanish multicenter case-control study found a slightly increased BC risk for high MUFA intake. 17 Interestingly, however, this initially found positive association disappeared after adjustment for saturated fat intake. A possible explanation for these controversial findings might be the source of the MUFAs. Monounsaturated fat can be obtained from either olive oil 34 or from animal sources, for example, beef, 12 which showed to have an opposite effect on BC risk. 40 In our study we observed an inverse association between plantbased fats and oils intakes and BC risk. This is in line with findings of the New Hampshire case-control study, also suggesting a decreased BC risk with high vegetable oil intake. 37 In addition, Brinkman et al, reported a clear reduced BC risk for high intakes of α-linolenic acid and vegetable fat. Furthermore, the same study showed a reduced BC risk was observed for polyunsaturated fat and linoleic acid. 37 The protective effects of plant-based oils, could be explained by its provision of various amounts of MUFAs, PUFAs and energy, which are potential antioxidants and chemopreventive factors that might affect the initiation, promotion and progression of cancers through several potential biologic mechanisms, including reduced cellular oxidative stress and probably decreased DNA damage. 41 In the last two decades, there have been puzzling results regarding the possible role of dietary olive oil in cancer prevention and treatment. 42 Oleic acid, which is a MUFA that is highly available in olive oil, canola oil, sunflower oil, soybeans oil, rapeseed oil and peanuts oil has been traditionally linked to a protective effect on cancer. 34 It is, therefore, surprising that the present study shows no effect of olive oil (MUFA: 73% vs PUFAs: 11%) and nor rapeseed oils (MUFA: 62% vs PUFAs: 32%) intakes on BC risk. This null-effect, however, has been observed in a previous study in which oils rich in MUFAs, derived from the seeds of soybean or grapeseed oil, did not exert health benefits and may not be associated with BC risk. 43 This may be extrapolated to olive oils. Our results further indicate that the protective effect of MUFA on BC risk is explained by dietary intake of multiple sources.
Interestingly, stratification for gender showed that a high intake of MUFAs may significantly decrease the risk of BC for women but not for men. Wakai et al also reported gender discrepancy in the association of MUFAs intake and BC risk. 18 This discrepancy might be related to overall gender differences in reporting diet 44  However, as mentioned before, data on BC subtype was mainly lacking, resulting in low power for detailed analysis. In our study it was shown that the total lipid intake was mainly derived from MUFA.
Since MUFAs are suggested to have a protective role against BC, the association found between total lipid intakes and NMIBC might be related to higher intakes of MUFAs.
ω-3 PUFAs have been reported as one potential modifiable protective factor against cancers. 47 Although not fully understood, it is suggested that n-3 PUFAs may possibly inhibit carcinogenesis through its anti-inflammatory activity. 39 Epidemiological studies on the intake of PUFAs and BC risk, however, showed inconsistent results. While some studies showed a null association between PUFA intake and BC risk, 17,18 others reported an inverse association. 35 In the present study a null association was observed. For fat and oil sources, which contain more PUFAs than MUFAs, a similar noneffect was shown for soybean Limited evidence and contradictory findings are available on the association between a high trans fatty acids (TFAs) intake and BC risk. 37,48,49 While several studies reported a direct association between higher TFAs intakes and BC risk, 48,49 others reported no significant association. 37,50 The present study also showed no evidence for an association between TFAs and BC risk, or for high intake of SFAs and PUFAs.
In recent years, cholesterol has received increasing attention due its role in carcinogenesis. 51 Clinical and experimental evidence suggest that an increased cholesterol level in blood is associated with a higher cancer risk and that cholesterol-lowering drugs (eg, statins) exhibit beneficial effects on bladder cancer development. 52  leading to resistance to apoptotic signals. 52 In the present study we found that cholesterol was associated with an increased BC risk among men but not among women. The null-association observed among women is in line with results from a Belgian case-control study and the New Hampshire case-control, showing an overall nullassociation between cholesterol intake and BC risk. 37,39 Since, evidence on the gender specific relations between cholesterol and BC risk are scarce, it remains unclear why in the present study a discrepancy between men and women was observed. However, the involvement of certain steroids, such as estrogen, in reducing the adverse effects of cholesterol, might explain the observed difference. Estrogen is proposed to protect against chronic diseases (ie, breast cancer and cardiovascular diseases or atherosclerosis) via its role in reverse cholesterol transport. 54 Furthermore, increased use of statins among men compared to women need to be taken into account in future research on the gender specific relation between cholesterol and BC.
It is suggested that animal fat increases oxidative stress and levels of reactive oxygen species (ROS) that interfere with cellular processes.
Healthy cells are attacked by free radicals, which cause peroxidation and eventually DNA damage. Thereby, ROS can lead to tumor initiation and progression of cancer cells. 55 The present study strengthens this hypothesis by showing an increased BC risk associated with an animal fat intake, which is in line with a previously conducted casecontrol study. 19 However, Brinkman et al showed a null association between intakes of animal fat and BC risk. 37 Again, this observed difference between the different studies might be due to the different type, composition and cooking method of the consumed animal fats included in the analysis.
No association was observed for higher intakes of total fats and oils and BC risk. Contrary to our finding, a meta-analysis revealed that the total dietary fat intake increases the BC risk. 38 However, this could only be observed among the European populations, while no association was reported for the North American populations. 38  is not expected to be significantly different between cases and noncases, and therefore the impact of information bias is expected to be minimal; (f) although we found similar results after adjusting for potential dietary risk factors, it remains possible that the observed associations were confounded by other dietary constituents or additives associated with fat intake; (g) the present study sample consists mostly of Caucasians, and this may limit the generalizability of our results to other racial/ethnic populations or geographic regions; (h) although status as well as duration and intensity of smoking were taken into account in our analysis, the adjustment for smoking might still be imperfect due to differences in smoking practices (eg, depth of inhalation or amount of inhalation), differences in types of smoke exposure or lack of information on passive smoking 58 ; (i) some tumor subtype information was missing, which hampered the statistical power required for stratified subgroup analyses.

| CONCLUSIONS
In