Body mass index and waist circumference in relation to risk of recurrence and progression after non‐muscle invasive bladder cancer

Abstract Background Obesity may be associated with increased risk of recurrence and progression in patients with non‐muscle invasive bladder cancer (NMIBC), but evidence is limited and inconsistent. We examined the associations of body mass index (BMI), waist circumference, and waist‐to‐hip ratio (WHR) with risk of recurrence and progression among patients with NMIBC. Methods This prospective study included 1029 patients diagnosed with primary NMIBC between 2014 and 2017. Patients reported weight 2 years before diagnosis at baseline, and weight, waist and hip circumference at 3 months postdiagnosis. Associations were quantified using Cox proportional hazard analyses, adjusted for clinical and lifestyle characteristics. Results More than half of patients were overweight (49%) or obese (19%) after diagnosis. During a median follow‐up time of 3.6 years, 371 patients developed ≥1 recurrence and 53 experienced progression. No associations with recurrence were observed for BMI (HRper 5 kg/m2 0.94; 95% CI 0.82, 1.07), waist circumference (HRper 10 cm 0.95; 95% CI 0.86, 1.05), or WHR (HRper 0.1 unit 0.90; 95% CI 0.76, 1.06). In contrast, higher BMI was associated with a 40% increased risk of progression, with only the 2‐year prediagnosis association reaching statistical significance (HRper 5 kg/m2 1.42; 95% CI 1.09, 1.84). No associations for pre‐to‐postdiagnosis weight change were found. Conclusion General and abdominal obesity were not associated with recurrence risk among patients with NMIBC, but might be associated with increased risk of progression. Studies with sufficient sample size to stratify by tumor stage and treatment are needed to better understand whether and how obesity could influence prognosis.


| INTRODUCTION
With 570,000 new cases each year, bladder cancer is the 10th most commonly diagnosed cancer worldwide and the 6th most common type of cancer in men. 1 Nearly 75% of diagnosed bladder cancer cases are non-muscle invasive. 2These patients have a good survival, but are at high risk of tumor recurrence. 3This necessitates a burdensome treatment and follow-up program which causes non-muscle invasive bladder cancer (NMIBC) to be one of the most expensive cancers in terms of lifetime treatment costs. 4For bladder cancer, acknowledged risk factors for worse oncologic outcomes are primary tumor characteristics, including higher tumor stage, higher grade, multiplicity, larger size, and concomitant carcinoma in situ. 5Lifestyle factors such as smoking and obesity may also be relevant for the prognosis of NMIBC and may contribute to more personalized follow-up schemes or interventions. 6here is limited and inconsistent evidence regarding the association between obesity and risk of NMIBC recurrence and progression.][9][10][11] In contrast, two studies reported no associations. 14,15A meta-analysis of five cohort studies concluded that obesity was associated with a 2-fold increased risk of recurrence and an 88% increased risk of progression compared to healthy weight patients with NMIBC. 16However, all studies only assessed BMI around time of diagnosis and most studies did not adjust for smoking status or other lifestyle factors, which might lead to biased outcomes.
To date, studies have not focused on other measures of obesity than BMI and have not included changes in weight over time.Using data of a prospective cohort study, we examined the associations between several measures of obesity (pre-and postdiagnosis BMI, postdiagnosis waist circumference and waist-to-hip ratio [WHR]) with the risk of recurrence and progression in patients with NMIBC, while adjusting for smoking and other lifestyle factors.Furthermore, we examined whether pre-to-post diagnosis weight change was associated with NMIBC recurrence and progression.

| Study design and population
We used data from the UroLife study, a prospective multicenter cohort study among patients diagnosed with primary NMIBC. 17Patients were recruited in 22 hospitals in the Netherlands between May 2014 and April 2017.Eligible patients were identified through the Netherlands Cancer Registry hosted by the Netherlands Comprehensive Cancer Organization using notification lists of the Pathological Anatomical National Automated Archive (PALGA Foundation).Patients were eligible if they were between 18 and 80 years old, Dutch speaking, diagnosed with a histologically confirmed primary stage Ta, T1, or Tis urothelial carcinoma of the tumor, and underwent a transurethral resection of the bladder tumor (TURBT).Exclusion criteria are shown in Figure 1.Approximately 4 weeks after diagnosis, patients were invited to participate in the UroLife study.Patients who agreed to participate provided written informed consent.Ethical approval was provided by the Committee for Human Research region Arnhem-Nijmegen (CMO 2013-494).

| Assessment of anthropometrics
Information on anthropometrics was collected at 2 time points via self-reported web-based or paper-and-pencil based questionnaires.First, at approximately 6 weeks after diagnosis (baseline), information on current height (and body weight) and body weight 2 years before diagnosis was collected.Second, at 3 months after diagnosis information on current body weight, waist circumference, and hip circumference was collected.Waist (midway between the lowest rib and the iliac crest) and hip (widest part) circumference were measured in centimeters with a tape sent to participants.Anthropometric measures were used as continuous and predefined categorical variables, that is, obesity (BMI ≥30.0), overweight (BMI ≥25-< 30), and healthy weight (BMI ≥18.5-< 25 kg/m 2 ).Patients with BMI <18.5 kg/m 2 (n = 10) were excluded from the analyses (Figure 1), as this underweight group was too small for performing statistical analyses.If BMI at 3 months after diagnosis was missing (n = 119), we used BMI assessed at baseline (correlation between both measures of BMI was 0.99).Pre-to-postdiagnosis weight change was calculated as the change in weight between 2 years prediagnosis and 3 months postdiagnosis and expressed as percentage weight change.Weight change was defined as weight loss ≥5%, weight stable −5% to 5%, and weight gain ≥5%, as ≥5% weight loss has been proven to have significant health benefits and is generally accepted as criterion for clinically meaningful weight change. 18Waist circumference was categorized into three groups based on sex-specific cutoff points commonly used for cancer prevention recommendations, 19 while WHR was categorized into three groups based on sex-specific tertiles.Waist circumference and WHR are measures of abdominal obesity, while BMI is a measure of general obesity.

| Outcome assessment
Information on recurrences and progression was collected from medical records by trained data managers from the Netherlands Cancer Registry in February-March 2021.A recurrence was defined as the first new bladder tumor after being tumor-free.Being tumor-free was defined by the date of radical TURBT (either primary TURBT or re-TURBT) in case of Ta and T1 tumors and by the date of the first tumor-negative cystoscopy after primary TURBT in case of (concomitant) carcinoma in situ (CIS).Progression was defined as the first occurrence of stage or grade progression. 20Patients were censored at (a) last contact between patient and urologist, (b) radical cystectomy in absence of recurrence/ progression, or (c) diagnosis of another type of cancer with metastasis, whichever came first.

| Covariate assessment
At baseline, patients reported sociodemographic information and whether they were ever diagnosed with diabetes mellitus.Smoking status was reported at baseline (reflecting pre-diagnosis smoking) and 3 months postdiagnosis.2][23][24][25][26][27] The reference period was a normal week in the months before diagnosis (at baseline) or the previous 3 months (during follow-up).Total minutes per week of moderate-to-vigorous physical activity were calculated based on leisure-time (cycling, gardening, odd jobs, and sports with a metabolic equivalent value ≥3) and commuting (walking and cycling) activities.9][30] The reference period for the FFQ was the year before diagnosis (at baseline) or the previous month (during follow-up).In the baseline FFQ, fruit and vegetable intake were queried separately for summer and winter.To limit seasonal variability between baseline and follow-up, we calculated baseline fruit and vegetable intake based on reported intake in the season that matched the season of follow-up assessment.To assess fruit and vegetable intake, alcoholic drinks, and total energy intake, we combined frequencies of intake with standard portion sizes and household measures. 28 Clinical data, such as tumor stage, differentiation grade, presence of (concomitant) CIS, tumor focality, and treatment, was collected from medical records by trained data managers of the Netherlands Cancer Registry.Participants were divided into low-, intermediate-and high-risk groups according to the European Association of Urology (EAU) guidelines 2019 based on stage, grade, CIS, and focality, 5 without considering tumor size (not available) and the recurrent nature of the tumor (only primary tumors included).

| Statistical analyses
First recurrence was the primary endpoint; progression was a secondary endpoint.Cox proportional hazard regression models were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations of anthropometry measures with recurrence and progression.Follow-up time began at baseline or follow-up questionnaire completion (progression) or at the first day of being tumor-free in case a person was not tumor-free at questionnaire completion (recurrence).In addition, we analyzed multiple recurrences with an extended Cox model for recurrent event data (gap time-unrestricted (GT-UR) model with a common baseline hazard and random effect). 32,33As the recurrence-specific baseline hazards were similar for all recurrence numbers, we selected a model with a common baseline hazard; analysis time was reset at each recurrence.The GT-UR model has a slightly different HR interpretation compared with the well-known Cox model; HRs are the increase or decrease of recurrence risk since the last event (either primary tumor or previous recurrence).The full models used to evaluate BMI 2 years prediagnosis were adjusted for age, sex, education level (low, intermediate, high), diabetes at diagnosis, EAU risk group (low, intermediate, high), initial treatment (only TURBT, only TURBT + single instillation, chemotherapy, BCG), baseline smoking status (never, former, current), moderate-to-vigorous physical activity, and fruit and vegetable, alcohol and energy intake.In the full models of postdiagnosis anthropometry (BMI, waist circumference, or WHR), we included the same covariates as in the analyses of prediagnosis BMI, except that smoking status, dietary intake, and physical activity were obtained from the postdiagnosis questionnaire.Weight change models were adjusted for the same covariates as the postdiagnosis models, with the addition of 2 years prediagnosis BMI.Besides diabetes, we did not consider any other components of the metabolic syndrome (like hypertension or hyperlipidemia) as these are not generally considered as risk factors for cancer recurrence and progression.To test for linear trends, the median score of each category was assigned to all participants within that category and entered as a continuous exposure in the Cox models.The proportional hazards assumption was tested using Schoenfeld's global test, and no statistically significant violations were detected.Multiple imputation (MICE) with 20 iterations, including Nelson-Aalen estimators and survival outcomes, was used to deal with missing self-reported data (exposures and covariates) in the Cox models. 34ubgroup analyses were performed to explore whether the associations between BMI at 3 months postdiagnosis and recurrence were modified by age at diagnosis, sex, smoking status, diabetes, EAU risk group, tumor stage, tumor grade, and initial treatment.

| RESULTS
Of the 2133 invited patients, 2025 patients were eligible and 1076 (53%) agreed to participate (Figure 1).A total of 1029 participants were eligible for the analyses of the current study.
Baseline characteristics for the total study population and according to BMI 3 months postdiagnosis are listed in Table 1.The median age at NMIBC diagnosis was 67 years and 80% was male.Most participants had stage Ta disease (76%).The median BMI 3 months postdiagnosis was 26.6 kg/ m 2 , with 49% being overweight and 19% being obese.Tumor characteristics did not differ by BMI classes.Patients with obesity were more likely to have diabetes while they were less likely to have a high educational level and to have never smoked, and they spent less time in moderate-to-vigorous physically activity compared with patients with a healthy weight.Participants included in the current analysis did not differ from eligible and invited non-participants with respect to age, sex, tumor stage, and tumor grade (data not shown).Pearson correlation coefficients of BMI 3 months postdiagnosis with waist circumference, WHR, and BMI 2year prediagnosis were 0.79, 0.38, and 0.89, respectively.

| General obesity and recurrence
BMI was not associated with risk of recurrence (Table 2).The adjusted HRs (model 3) for overweight and obesity at 3 months postdiagnosis, with healthy weight as reference, were 0.98 (95% CI: 0.77, 1.24) and 0.89 (95% CI: 0.65, 1.21), respectively.The analysis for continuous BMI at 3 months postdiagnosis yielded a HR of 0.94 (95% CI: 0.82, 1.07) per  2, model 1).Furthermore, results did not meaningfully differ when analyzing multiple recurrences instead of first recurrences (Table S1).Subgroup analyses showed no statistically significant effect modification of the associations between BMI 3 months postdiagnosis and recurrence by age, sex, smoking status, level of education, diabetes at diagnosis, EAU risk group, or treatment (Figure 2).However, higher BMI might be associated with an increased recurrence risk among the subgroup of patients with stage T1 NMIBC.

| Abdominal obesity and recurrence
Waist circumference and WHR at 3 months postdiagnosis were not associated with risk of recurrence (Table 2).The analysis for continuous waist circumference yielded a HR of 0.95 (95% CI: 0.86, 1.05) per 10 cm increase and for WHR a HR of 0.90 (95% CI: 0.76, 1.06) per 0.1-unit increase.Results did not differ meaningfully when we analyzed multiple recurrences instead of first recurrences (Table S1).
T A B L E 2 Hazard ratios (HR) for the association of obesity with risk of first bladder cancer recurrence.

| Obesity and
Higher BMI was associated with a 40% increased risk of progression, although only the 2 years prediagnosis association reached statistical significance (HR per 5 kg/m2 1.42; 95% CI 1.09, 1.84) (Table 3).The HR for waist circumference was 1.12 (95% CI: 0.87, 1.44) per 10 cm increase and for WHR 1.29 (95% CI: 0.84, 1.97) per 0.1-unit increase.T A B L E 3 Hazard ratios (HR) for the association of obesity with risk of bladder cancer progression.

| Pre-to-postdiagnosis weight change
Weight change from 2 years prediagnosis to 3 months postdiagnosis was neither associated with risk of recurrence (HR per 5% 1.02; 95% CI: 0.93, 1.13) nor progression per 5% 0.88; 95% CI 0.69, 1.13) (Table 4).[Correction added on November 16, 2023 after first online publication.The values in the previous sentence has been updated in this version.]However, it should be noted that only 28% of participants experienced a weight change ≥5%.

| DISCUSSION
In this prospective study among 1029 patients diagnosed with primary NMIBC, BMI, waist circumference, and WHR were not associated with risk of recurrence.In contrast, higher BMI tended to be associated with an increased risk of progression.Weight change from 2-years prediagnosis to 3 months postdiagnosis was neither associated with recurrence nor progression.

| In relation to current literature
][9][10][11][12][13] Similar to our results, two studies found no associations. 14,151][12] These studies slightly differed by the included study populations.][12] Our subgroup analyses among patients with stage T1 indicated that higher BMI was associated with a statistically nonsignificant increased recurrence risk, while this association was not observed among patients with stage Ta disease.No association for patients treated with BCG was found.However, power for these subgroup analyses was limited.From the previous studies reporting a statistically significant increased risk of recurrence with obesity, only one was adjusted for smoking status. 11Two other studies taking smoking into account reported either a statistically nonsignificant increased risk (HR 1.22; 95% CI 0.80, 1.87) 12 or no association (HR 1.02; 95% CI 0.76, 1.38) 14 with obesity.Our risk estimates did neither meaningfully change when smoking status was excluded from our model, nor by exclusion of other lifestyle factors (diet and physical activity).Variation in upper BMI range is not expected to explain differences in observed associations between studies.Studies showing that obesity was associated with increased risk of recurrence not only included study populations from the United States, 7,8 but also from Europe, 9 Turkey, 11 and China 10 where BMIs >35 are less common than in the United States.Furthermore, similar, but slightly lower, risk estimates were reported for overweight patients. 9,10,12,13Studies with sufficient sample size to stratify by tumor stage and treatment are needed to explain heterogeneity between study results.In line with our results, a meta-analysis of three cohort studies showed that obesity, compared to healthy weight, was associated with an increased of progression in patients with NMIBC. 16In contrast, two other studies reported either no association 14 or a protective association. 15e had limited power to study risk of NMIBC progression, as we observed relatively few events (n = 53).Therefore, anthropometrics were only analyzed as continuous variables (and not as categorical variables).Within our study, only the association between BMI 2-years prediagnosis and recurrence reached statistical significance.We cannot rule out that this positive association is due to chance.The recruitment of additional stage T1 patients and extension of follow-up time for our cohort will increase the power of future progression analyses.
This was the first study that examined the association between weight change and NMIBC prognosis.We found that weight change from 2 years prediagnosis to 3 months postdiagnosis was neither associated with risk of recurrence nor with progression.However, it should be noted that only 28% of participants experienced a weight change ≥5%.For cancer patients, it is important to know whether changing one's lifestyle after diagnosis can lower the risk of recurrence and progression.Additional studies are needed to further examine whether obesity after NMIBC diagnosis and changes therein can impact prognosis.

| Weaknesses and strengths of the study
Our study had several limitations.First, anthropometrics were self-reported at each time point.Generally, errors are present in self-reported anthropometric data; some people underreport, while others overreport and the magnitude of measurement error in waist circumference is unclear. 35owever, since anthropometric data were reported before the occurrence of an event, we do not expect these errors to be different between patients with or without an event.Second, although body weight was assessed at several timepoints, obesity at these timepoints might not reflect obesity later during the NMIBC trajectory.However, our group previously concluded that BMI did not significantly change between 3 and 15 months after NMIBC diagnosis. 36Third, we did not perform competing risk analyses, although death could prevent a recurrence diagnosis.As only 70 people died without recurrence within 5 years after diagnosis, it seems unlikely that our observed associations would be affected by the competing risk of death.Fourth, although the study was among the largest compared to other cohort studies on the same topic, power was still limited for subgroup analyses and for the outcome progression.Finally, as with all observational studies, we cannot eliminate the possibility of residual confounding even though we were able to include more possible confounders as most other studies on this topic.
Strengths of our study compared with previous studies are the inclusion of multiple measures of obesity and adjustment of our analyses for smoking status and other lifestyle factors.In contrast to previous studies, our study assessed the effects of both general and abdominal obesity.We observed similar associations with NMIBC outcomes for both general and abdominal obesity.Abdominal obesity, measured by waist circumference or WHR, is associated with adverse metabolic profiles such as insulin resistance and systemic inflammation, which are both linked to tumor growth. 37Furthermore, relative visceral fat area and degree of systemic inflammation may impact response to BCG treatment. 38However, mechanisms underlying the association of abdominal obesity with cancer outcomes may differ by cancer type, and the exact mechanisms for NMIBC recurrence and progression remain to be elucidated.In contrast to most previous studies, we adjusted our analysis for smoking status, dietary intake, and physical activity.However, adding these lifestyle factors to our models did not meaningfully change our risk estimates.This analysis was based on data from a multicenter population-based prospective cohort study.Although we do not know whether patients with NMIBC included in the current analysis differed from invited eligible patients with NMIBC regarding body weight, they were comparable with respect to age, sex, and tumor characteristics.

| CONCLUSION
In conclusion, general and abdominal obesity were not associated with an increased risk of recurrence among patients with NMIBC but tended to increase risk of progression.As current evidence is inconsistent, further studies with obesity measures, preferably also including CT-based visceral and subcutaneous fat, and with sufficient sample size to stratify by tumor stage and type of treatment are needed to understand whether and how obesity after diagnosis could influence risk of recurrence and progression.
T A B L E 1 5 Data of 116 participants were missing/unknown.6 Data of 82 participants were missing/unknown.7 Data of 123 participants were missing/unknown.8 Data of 124 participants were missing/unknown.9 Data of 126 participants were missing/unknown. 10Data of 130 participants were missing/unknown.
Multivariable HRs and 95% CIs of first recurrence per 5 kg/m 2 increase in BMI at 3 months after diagnosis, for the total study population and stratified by age at diagnosis, sex, smoking status, diabetes, EAU risk group, tumor stage, tumor grade, and initial treatment.Models were adjusted for age at diagnosis, sex, education, postdiagnosis smoking status, diabetes at diagnosis, postdiagnosis energy intake, EAU risk group (based on stage, grade, CIS, multifocality), and initial treatment (only TURBT, TURBT + single instillation, chemotherapy, BCG), except in models stratified by these variables.
Note: Model 1: adjusted for age at diagnosis and sex and using multiple imputation with 20 iterations for missing covariates and/or BMI 2y prediagnosis, waist circumference, waist-to-hip ratio.Note: Model 2: Model 1 + further adjustment for education, smoking status, diabetes at diagnosis, EAU risk category (based on stage, grade, CIS, multifocality), and initial treatment (only TURBT, single instillation, chemotherapy, BCG).
Model 1:adjusted for age at diagnosis and sex and using multiple imputation with 20 iterations for missing covariates and/or weight change.Model 2: Model 1 + further adjustment for education, smoking status, diabetes at diagnosis, EAU risk category (based on stage, grade, CIS, multifocality), and initial treatment (only TURBT, single instillation, chemotherapy, BCG).