Comparative efficacy of vitamin D status in reducing the risk of bladder cancer: A systematic review and network meta-analysis

Accepted Manuscript Comparative Efficacy of Vitamin D Status in Reducing the Risk of Bladder Cancer: A Systematic Review and Network Meta-Analysis Yue Zhao, MD, PhD, Changhao Chen, MD, PhD, Wenwei Pan, MD, PhD, Ming Gao, MD, He Wang, MD, PhD, Ren Mao, MD, PhD, Tianxin Lin, MD, PhD, Jian Huang, MD, PhD PII:

S0899-9007(15)00475-X

DOI:

10.1016/j.nut.2015.10.023

Reference:

NUT 9661

To appear in:

Nutrition

Received Date: 3 July 2015 Revised Date:

19 September 2015

Accepted Date: 31 October 2015

Please cite this article as: Zhao Y, Chen C, Pan W, Gao M, Wang H, Mao R, Lin T, Huang J, Comparative Efficacy of Vitamin D Status in Reducing the Risk of Bladder Cancer: A Systematic Review and Network Meta-Analysis, Nutrition (2016), doi: 10.1016/j.nut.2015.10.023. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

1

Comparative Efficacy of Vitamin D Status in Reducing the Risk of Bladder ACCEPTED MANUSCRIPT

2

Cancer: A Systematic Review and Network Meta-Analysis

3

Yue Zhao, MD, PhD1#; Changhao Chen, MD, PhD2#; Wenwei Pan, MD, PhD2#; Ming Gao, MD3; Wang He,

5

MD, PhD2; Ren Mao, MD, PhD1; Tianxin Lin, MD, PhD2*; Jian Huang, MD, PhD2*.

RI PT

4

6 7

1

8

China.

9

2

Department of Urology, Sun Yat-Sen Memorial Hospital, Guangzhou, China.

10

3

Department of Acupuncture, the People Hospital of Honghuagang District, Zunyi, China.

M AN U

SC

Department of Gastroenterology, the First Affiliated Hospital of Sun Yat-Sen University, Guangzhou,

11

#Yue Zhao, Changhao Chen and Wenwei Pan contributed equally to this study.

13

*Corresponding authors

14

*Corresponding authors, to whom request for reprints should be addressed:

15

Tianxin Lin MD, PhD

16

Department of Urology, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, 107 Yan-Jiang Xi Road,

17

Guangzhou, 510120, China.

18

Tel. +86 20 81332603; Fax: +86 20 81332853.

19

E-mail address: [email protected]

20

Jian Huang, MD, PhD

21

Department of Urology, Sun Yat-Sen Memorial Hospital, Sun Yat-Sen University, 107 Yan-Jiang Xi Road,

22

Guangzhou, 510120, China.

23

Tel.: +86 20 81332603; Fax: +86 20 81332853.

24

E-mail address: [email protected]

AC C

EP

TE D

12

1

ACCEPTED MANUSCRIPT

ABSTRACT

26

OBJECTIVE: The optimal concentration of individual vitamin D intake for preventing bladder cancer

27

has not been defined. To evaluate the comparative efficacy of different serum 25-hydroxyvitamin D

28

concentrations in preventing bladder cancer, we conducted a systematic search of the literature published

29

up to April 2015.

30

METHODS: We applied a pair-wise meta-analysis to estimate direct evidence from intervention-control

31

studies and a network meta-analysis within a Bayesian framework to combine direct and indirect evidence.

32

Moreover, a dose-response curve was utilized to predict the optimal median serum 25-hydroxyvitamin D

33

concentration based on the odds ratio (OR) for each quintile concentration.

34

RESULTS: Seven studies of a total of 90757 participants, including 2509 bladder cancer patients, were

35

included. Two prospective cohort studies with 57591 participants and 494 bladder cancer patients, and five

36

case-control studies with 33166 participants and 2264 bladder cancer patients. From the network

37

meta-analysis, we observed that sufficient serum 25-hydroxyvitamin D concentrations (> 75 nmol/L) were

38

superior to all other 25-hydroxyvitamin D concentrations in decreasing the risk of bladder cancer: OR=

39

0.68 and 95% credible interval (CrI) 0.52-0.87 compared with severely deficient concentrations (< 25

40

nmol/L); OR= 0.65 and 95% CrI 0.49-0.86 compared with moderately deficient concentrations (25-37.5

41

nmol/L); OR= 0.61 and 95% CrI 0.47-0.80 compared with slightly deficient concentrations (37.5-50

42

nmol/L); and OR= 0.65 and 95% CrI 0.48-0.85 compared with insufficient concentrations (50-75 nmol/L).

43

In addition, we noted a roughly inverse correlation between bladder cancer risk and 25-hydroxyvitamin D

44

concentrations (R2 = 0.98, p = 0.007).

45

CONCLUSIONS: Ensuring sufficient serum 25-hydroxyvitamin D concentrations might play an

46

important role in decreasing the risk of bladder cancer. The serum 25-hydroxyvitamin D concentration ≥

47

74 nmol/L was associated with a 60% lower risk of bladder cancer incidence.

AC C

EP

TE D

M AN U

SC

RI PT

25

2

ACCEPTED MANUSCRIPT

48 49

KEYWORDS: Serum 25-hydroxyvitamin D; Vitamin D; Bladder cancer; Network meta-analysis;

50

Systematic review.

51

ABBREVIATIONS: CrI, credible interval; ORs, odds ratios; HR, hazard ratio; CIs, confidence intervals;

53

BMI, body mass index; CV, coefficient of variation; QC, quality control.

54

INTRODUCTION

SC

55

RI PT

52

Bladder cancer is a major problem in developed countries, particularly for males, among whom the

M AN U

56

incidence is three- to five-fold greater than that in females[1, 2]. In the United States, bladder cancer is the

58

fourth most common cancer diagnosed in males and the eighth leading cause of cancer-related death.

59

Despite reduced exposure to established risk factors, such as smoking and occupational chemical

60

carcinogens, the incidence of bladder cancer remains high[3]. The term vitamin D refers to a group of fat

61

soluble secosteroids that play crucial roles in bone metabolism and the immune system; vitamin D is also

62

known to exert anti-cancer effects[4, 5]. Vitamin D can be hydroxylated in the liver into its circulating

63

form, 25-hydroxyvitamin D[6]. Serum 25-hydroxyvitamin D deficiency may be associated with an

64

increased risk for prostate cancer[7], lung cancer[8], breast cancer[9], colorectal cancer[10], and

65

non-Hodgkin’s lymphoma[11]. Several studies have examined the association between vitamin D status

66

and bladder cancer by measuring the serum 25-hydroxyvitamin D concentrations. These studies

67

demonstrated that serum 25-hydroxyvitamin D concentrations inversely correlated with the incidence of

68

bladder cancer[6, 12, 13]. In a recent pair-wise meta-analysis, Liao and colleagues observed that higher

69

serum 25-hydroxyvitamin D concentrations were associated with a reduced risk of subsequent bladder

70

cancer. However, pair-wise meta-analyses have provided only partial information in this context because

AC C

EP

TE D

57

3

71

they have only compared pairs of serum 25-hydroxyvitamin D concentrations, and consequently, the data ACCEPTED MANUSCRIPT

72

do not optimally inform decision-making.

Network meta-analyses have been recently developed to assess the comparative effectiveness of

74

several interventions and to synthesize evidence across a network of clinical studies [14, 15]. This method

75

combines the simultaneous analysis of direct evidence (from clinical studies that directly compare

76

treatments of interest) and indirect evidence (from clinical studies that compare treatments of interest with

77

a common comparator) to calculate a mixed effect size as the weighted average of the direct and indirect

78

evidence[16]. This method allows us to estimate the comparative treatment effects of two agents that have

79

not been directly compared with each other but have been compared with a common comparator[14].

80

Bayesian network meta-analysis combines all the evidence of the relative treatment effects to enable a

81

unified, coherent analysis of all clinical studies [17, 18].

M AN U

SC

RI PT

73

Thus, we constructed a comprehensive network that involved all candidate serum 25-hydroxyvitamin

TE D

82

D concentrations and conducted a systematic review and a network meta-analysis using an analytical

84

approach to evaluate the relative efficacy of vitamin D status in reducing bladder cancer risk.

85

METHODS

86

Literature Search and Eligibility Criteria

AC C

EP

83

We searched PubMed, EMBASE, MEDLINE, the Cochrane Library, and the Web of Science for

87 88

studies on the impact of vitamin D status on bladder cancer risk that were published from database

89

inception to April 18, 2015. We used MeSH combined free terms that were correctly adjusted for the

90

different databases in all the search strategies. The search strategy for PubMed was [Title/Abstract]:

91

[(25-hydroxyvitamin D/25-hydroxy-vitamin D/serum 25-hydroxyvitamin D/serum 25-hydroxyvitamin 4

D/vitamin D) and (bladder cancer/urinary bladder cancer/bladder cancer risk)]. The Related Articles ACCEPTED MANUSCRIPT

93

function was used to broaden the search. The computer search was supplemented with manual searches of

94

the reference lists of all the retrieved studies, review articles, and conference abstracts. No language

95

restrictions were applied. We also contacted the corresponding authors to acquire additional information if

96

the research results were unclear or more information was needed.

RI PT

92

The studies included in this network meta-analysis met the following inclusion criteria: (1) they

97

focused on measures of association by quantile of serum 25-hydroxyvitamin D in the prospective cohort

99

studies, nested case-control studies, or case-control studies; (2) they focused on the association between

100

different serum 25-hydroxyvitamin D concentrations and bladder cancer risk; (3) the outcome of interest

101

was bladder cancer, and pooled odds ratios (ORs) or hazard ratios (HR) were estimated with 95%

102

confidence intervals (CIs) of bladder cancer by quantile of serum 25-hydroxyvitamin D in both cohort

103

studies and case-control studies; and (4) the study assessed bladder cancer outcome or bladder

104

cancer-related progression or mortality. Letters, reports, laboratory studies, animal and nonhuman studies,

105

non-experimental studies, review articles, editorials, and articles that did not report the outcomes of

106

interest were excluded. For multiple reports from the same study, only the article with the largest dataset

107

was included in the network meta-analysis. Any disagreements were resolved via discussions to ultimately

108

reach a consensus.

109

Data Extraction and Quality Assessment

AC C

EP

TE D

M AN U

SC

98

The agreement between the two reviewers for the selection and validity assessment of the studies was

110 111

scored with the kappa coefficient (a measure of agreement). Two authors searched the publications

112

independently according to the PRISMA guidelines and collected information from these publications,

113

including the first author, study type, country, number of participants, follow up, median age, gender, body 5

mass index (BMI), tumor type, smoking status, and dietary vitamin D intake. Adjusted ORs and 95% CIs ACCEPTED MANUSCRIPT

115

for bladder cancer according to serum 25-hydroxyvitamin D concentrations were calculated using logistic

116

regression analysis in included studies. Adjustment was made for a number of possible confounding

117

variables including age, sex, region, cigarette smoking status, dairy consumption and season of blood

118

draw. The extracted data were crosschecked, and unresolved discrepancies were referred to an

119

adjudicating senior author; if necessary, problems were discussed in a panel meeting. To inform the

120

appropriateness of including studies in this network meta-analysis and to facilitate the assessment of the

121

strength of the evidence, we determined the risk of bias in each included study using the

122

Newcastle-Ottawa scale, which includes patient selection, the comparability of the study groups, and the

123

assessment of outcome. Each study received a score of 0-9 (indicated by stars; Tables S1 and S2). The

124

stars were used to quantitatively compare study quality. Studies that received a grade of six or more stars

125

were considered to be of high quality. In addition, the methodology, coefficient of variation (CV) and

126

quality control (QC) scheme for serum 25-hydroxyvitamin D concentrations were extracted from each

127

included study.

TE D

M AN U

SC

RI PT

114

Statistical Analysis

129

We conducted direct and network meta-analyses that compared each serum 25-hydroxyvitamin D

130

concentration to best summarize the totality of the available evidence. Groups were stratified by serum

131

25-hydroxyvitamin D concentrations: sufficient: > 75 nmol/L; insufficient: 50-75 nmol/L; slightly

132

deficient: 37.5-50 nmol/L; moderately deficient: 25-37.5 nmol/L; and severely deficient: < 25 nmol/L.

AC C

EP

128

Methods for Direct Treatment Comparisons

133

6

Traditional pair-wise meta-analyses were conducted for the high and low serum 25-hydroxyvitamin D ACCEPTED MANUSCRIPT

134

categories. To maintain uniformity, the ORs with 95% CIs that compared the highest serum

136

25-hydroxyvitamin D category with the lowest were used in the meta-analysis. Either fixed or random

137

effects models were used, and the degree of heterogeneity among studies was determined using Cochran's

138

Q statistic, with significance set at p < 0.10[19]. Statistical heterogeneity was assessed using the I2

139

statistic[20], which reports the total variation across studies that is not due to chance. An I2 statistic less

140

than 25% indicates a little inconsistency, a value between 25% and 50% suggests moderate heterogeneity,

141

and a value greater than 50% indicates considerable inconsistency[21]. The Egger regression test was

142

utilized to estimate the funnel plot asymmetry and evaluate the publication bias[22]. A two-sided p value

143

of less than 0.05 was considered significant. The direct meta-analysis was performed using Stata 12.0

144

(StataCorp, College Station, TX, USA).

M AN U

SC

RI PT

135

Methods for Indirect and Mixed Comparisons

146

Next, we fitted a network meta-analysis model to each outcome by combining direct evidence for

TE D

145

each comparison with indirect evidence for all pair-wise comparisons. The comparative efficacies of any

148

two treatments were determined as a function of each treatment relative to the reference treatment (i.e.,

149

severely deficient serum 25-hydroxyvitamin D concentrations). We evaluated the inconsistency by

150

comparing the estimated magnitude and direction of the point estimates from the direct and indirect

151

comparisons. Treatment effects were estimated by posterior means with corresponding 95% credible

152

intervals (CrIs), which can be interpreted similarly to conventional 95% CIs. We applied both fixed and

153

random effects models. The difference between the two models is that the latter considers between-study

154

variance, thereby producing wider CrIs, and is preferred in the presence of heterogeneity. The network

155

meta-analysis was conducted using the Bayesian Markov chain Monte Carlo method and was fitted using

AC C

EP

147

7

ADDIS software (Version 1.16.5; Drug Information Systems). For each outcome, we estimated the ACCEPTED MANUSCRIPT

157

probability that each intervention is the best (superior to all other interventions), second best, third best,

158

etc., based on the rank order of the treatments at each concentration of the Markov chain. The estimated

159

effect of p value and 95% CrI were generated for all possible serum 25-hydroxyvitamin D concentrations

160

pairwise comparisons despite not being evaluated directly in ahead-to-head fashion in included studies.

161

The pooled results were considered significant when p < 0.05 or if the 95% CrI excluded the value of 1.

RI PT

156

Methods for Dose–Response Gradient

163

Most studies identified the median value for serum 25-hydroxyvitamin D concentrations. If this value

SC

162

was not identified, midpoint values were calculated by computing the arithmetic mean of the upper and

165

lower bounds of the quantiles. Data reported in ng/mL were converted to nmol/L using the conversion 1

166

ng/mL = 2.5 nmol/L[23]. The vitamin D dose–response curves were standardized based on the

167

measurements provided by Carpenter et al.[24] and Steenland et al.[25]. The dose-response curve was

168

plotted using the ORs for each quintile of the pooled data. A least-squares trend line was constructed to

169

examine the dose-response relationship[24, 25]; p values for the trend were calculated using the

170

Mantel-Haenszel χ2 test. The serum 25-hydroxyvitamin D concentrations were determined by making a

171

vertical line from the point on the dose-response curve corresponding to OR = 0.60 to the point of

172

intersection with the horizontal axis.

173

RESULTS

174

Study Characteristics and Quality of Included Studies

Seven studies met the inclusion criteria and were included in this network meta-analysis[26-32]; these

175 176

AC C

EP

TE D

M AN U

164

studies provided data on a total of 90757 participates, including 2760 cases of bladder cancer. Two 8

prospective cohort studies with 57591ACCEPTED participants, including 494 bladder cancer patients, and five MANUSCRIPT

178

case-control studies with 33166 participants, including 2264 bladder cancer patients. A flow diagram

179

depicting the search and selection process is shown in Fig. 1. The search criteria yielded 87 entries; 42 of

180

these were excluded because they were duplicates. Thirty-five of the remaining publications were

181

excluded after screening the titles and abstracts because the topics of these studies were irrelevant or they

182

were non-comparative studies. The full text of the remaining 10 publications was screened, and three

183

additional articles were excluded. Two prospective cohort studies[26, 30], two nested case-control

184

studies[31, 32], and three case-control studies[27-29] were included. Table 1 summarizes the

185

characteristics of the studies included in the network meta-analysis. The studies were published from 2006

186

to 2013, and they varied in sample size (range: 561-47,800). The median age of included studies

187

population at baseline ranged from 57.7 to 67.6 years. The minimum and maximum ages were 20 and 81,

188

respectively. The range of estimated median BMI was narrow (range: 24.8 to 27.2 kg/m2), with 22 and 33

189

as the estimated minimum and maximum BMI values, respectively. The incidence of bladder cancer in

190

prospective studies were 0.6 per 1000 person-years (382/47800/14 years) with Giovannucci et al. study

191

and 0.4 per 1000 person-years (112/9791/28 years) with Afzal et al. study. A total of 2264 participants

192

were assigned to the cases group and 2258 participants to the controls group in case-control studies. The

193

follow-up period ranged from 12 to 28 years, and three of the studies lacked follow-up information. Three

194

studies were from the USA, and four were from European countries. All the studies provided adjusted ORs

195

for different serum 25-hydroxyvitamin D concentrations, but the adjusted confounding factors differed

196

between the studies (Table 1).

AC C

EP

TE D

M AN U

SC

RI PT

177

Fig. 2 shows the network of comparisons for each stratum of serum 25-hydroxyvitamin D

197 198

concentrations. The quality of the included studies was generally satisfactory, and only one study scored

199

lower than six stars. Besides, the quality of serum 25-hydroxyvitamin D measurement based on the 9

methodology, CVs and QC schemes was generally acceptable, although the two included study did not ACCEPTED MANUSCRIPT

201

provide the CVs[28, 29] (Supplementary Tables S3). Methods used to handle missing data and

202

intention-to-treat analyses were not sufficiently reported in most of the studies. Overall, the studies

203

appeared to be at low to moderate risk of bias. The study-level quality assessments are summarized in

204

Supplementary Tables S1 and S2.

205

Direct Meta-Analysis of Serum 25-Hydroxyvitamin D Concentrations and Bladder Cancer

RI PT

200

The results of the meta-analysis based on direct comparisons are presented in Fig. 4. Two prospective

SC

206

cohort studies and five case-control studies were included[26-32]. The pooled OR was 0.76 (95% CI,

208

0.67-0.87; p < 0.001), suggesting that a high serum 25-hydroxyvitamin D concentration significantly

209

decreased the risk of bladder cancer. Heterogeneity among studies was not detected (I2 = 1%) (Fig. 4).

210

Subgroup analyses were performed to determine the effect of matching variables, especially age and BMI.

211

The results indicated that adjustment for age and BMI factors did not alter the previous findings and there

212

was no statistical interaction between serum 25-Hydroxyvitamin D concentration and those factors

213

(Fig.S2). The funnel plot is shown to visually assess the publication bias (Fig. 6 and S1). The funnel plot

214

was symmetrical. All the studies were inside the 95% CIs with an even distribution around the vertical

215

axis, indicating no obvious publication bias according to Egger’s test (p = 0.651; Fig. 6).

216

Network Meta-Analysis of Serum 25-Hydroxyvitamin D Concentrations and Bladder Cancer

AC C

EP

TE D

M AN U

207

We established a network to compare the ability of sufficient, insufficient, slightly deficient,

217 218

moderately deficient, and severely deficient concentrations of serum 25-hydroxyvitamin D to reduce the

219

risk of bladder cancer. The two prospective cohort studies were excluded due to the absence of an

220

adequate control. So the remaining five case-control studies fulfilled our inclusion criteria in the network

10

meta-analysis[27-29, 31, 32]. The following OR and CrI values were obtained compared with severely ACCEPTED MANUSCRIPT

222

deficient 25-hydroxyvitamin D concentrations: moderately deficient, OR 0.90 and 95% CrI 0.73-1.12;

223

slightly deficient, OR 0.94 and 95% CrI 0.73-1.25; insufficient, OR 1.08 and 95% CrI 0.81-1.37; and

224

sufficient, OR 0.68 and 95% CrI 0.52-0.87. Only sufficient concentrations of serum 25-hydroxyvitamin D

225

significantly reduced the risk of bladder cancer (Fig. 3A).

RI PT

221

With respect to the comparative effectiveness of different serum 25-hydroxyvitamin D concentrations

226

based on the network meta-analysis, sufficient serum 25-hydroxyvitamin D concentrations were superior

228

to all other 25-hydroxyvitamin D concentrations at decreasing the risk of bladder cancer. Sufficient

229

concentrations of 25-hydroxyvitamin D significantly reduced the risk of bladder cancer compared with all

230

other concentrations: moderately deficient, OR 0.65 and 95% CrI 0.49-0.86; slightly deficient, OR 0.61

231

and 95% CrI 0.47-0.80; and insufficient, OR 0.65 and 95% CrI 0.48-0.85 (Fig. 3A). We confirmed that the

232

direct and indirect comparisons were coherent for all the endpoints; the node-splitting analysis did not

233

indicate significant inconsistencies (all p > 0.05). Furthermore, Fig. 3B graphically presents the rank order,

234

which reflects the relationship with decreased bladder cancer risk. Rank one is the best, and rank five is the

235

worst. Of all the examined 25-hydroxyvitamin D concentrations, sufficient concentrations reduced bladder

236

cancer risk the most, with a rank one probability of 99%, whereas the probability of the other

237

concentrations was close to 0%. Taken together, we demonstrated that sufficient concentrations of serum

238

25-hydroxyvitamin D are the efficacious at reducing bladder cancer risk.

239

Assessment of Serum 25-Hydroxyvitamin D Concentrations

AC C

EP

TE D

M AN U

SC

227

A linear dose-response curve was obtained for the pooled ORs for different concentrations of serum

240 241

25-hydroxyvitamin D (Fig. 5). We noted a roughly inverse correlation between bladder cancer risk and

242

25-hydroxyvitamin D concentrations (R2 = 0.98, p = 0.007). The results suggested that a serum 11

243

25-hydroxyvitamin D concentration ≥ACCEPTED 74 nmol/L wasMANUSCRIPT associated with a 60% lower risk of bladder cancer

244

incidence (Fig. 5).

245

DISCUSSION In this systematic review and network meta-analysis, we combined direct and indirect comparisons

247

from seven studies that involved 90757 participants, including 2509 bladder cancer patients, to evaluate

248

the comparative efficacy of different serum 25-hydroxyvitamin D concentrations in preventing bladder

249

cancer. It included five case-control studies with 33166 participants and 2264 bladder cancer patients, and

250

two prospective cohort studies with 57591 participants and 494 bladder cancer patients. We used ADDIS

251

to determine the concentrations that the effectively prevented bladder cancer. ADDIS is a powerful tool

252

that facilitated indirect pair-wise comparisons of interventions from various clinical studies and provided a

253

ranking of different serum 25-hydroxyvitamin D concentrations for a particular outcome. The principle

254

finding was that sufficient concentrations of serum 25-hydroxyvitamin D (> 75 nmol/L) were associated

255

with a significant reduction in bladder cancer risk compared with severely deficient concentrations (< 25

256

nmol/L), moderately deficient concentrations (25-37.5 nmol/L), slightly deficient concentrations (37.5-50

257

nmol/L), and insufficient concentrations (50-75 nmol/L). In addition, we demonstrated that severely

258

deficient serum 25-hydroxyvitamin D concentrations were a risk factor for bladder cancer (Fig. 3).

259

Moreover, bladder cancer risk inversely correlated with 25-hydroxyvitamin D concentration. A serum

260

25-hydroxyvitamin D concentration ≥74 nmol/L was associated with a 60% lower incidence risk of

261

bladder cancer(Fig. 5).

AC C

EP

TE D

M AN U

SC

RI PT

246

Our Bayesian network meta-analysis findings are consistent with those from previous studies, which

262 263

suggested that high vitamin D status protects patients from developing bladder cancer. A case-control

264

study by Mittal et al.[12] demonstrated that vitamin D and its analogues are potential anti-proliferative 12

agents that reduced the high mitotic rate of bladder cancer cells compared with normal cells. Animal and ACCEPTED MANUSCRIPT

266

in vitro studies[12, 33] have shown that vitamin D can suppress tumor progression by reducing cell

267

proliferation and invasiveness and stimulating apoptosis. Thierry-Palmer et al.[34, 35] showed that lower

268

25-hydroxyvitamin D concentrations might be caused by the loss of vitamin D metabolites via the urine,

269

which might be associated with an increased risk of bladder cancer. In addition, Chen et al[36].

270

demonstrated that vitamin D were inversely associated with bladder cancer risk in the meta-analysis. Liao

271

et al.[37] reported that high serum 25-hydroxyvitamin D concentrations was significantly associated with a

272

decreased risk of bladder cancer compared with low serum 25-hydroxyvitamin D concentrations in the

273

traditional meta-analysis.

M AN U

SC

RI PT

265

Most of the currently available data on the impact of high vitamin D concentrations on bladder cancer

274

risk involve comparisons to lower vitamin D concentrations. The present network meta-analysis is unique

276

in that we performed comparisons across five categories of serum 25-hydroxyvitamin D concentration.

277

When these different serum 25-hydroxyvitamin D concentrations were ranked, the sufficient

278

concentrations were always ranked first. Moderately deficient, slightly deficient, and insufficient

279

concentrations of serum 25-hydroxyvitamin D did not significantly reduce bladder cancer risk compared

280

with severely deficient concentrations. The results indicated that sufficient concentrations of serum

281

25-hydroxyvitamin D are associated with reducing the risk of bladder cancer compared with other

282

concentrations.

AC C

EP

TE D

275

Furthermore, this network meta-analysis revealed that dose adequacy is a critical issue. Although

283 284

optimal serum concentrations of 25-hydroxyvitamin D have not been established, a sufficient vitamin D

285

status has often been defined as a 25-hydroxyvitamin D concentration ≥ 30 ng/mL (75 nmol)[38-41]. This

286

definition is based on the finding that serum parathyroid hormone concentrations inversely correlate with 13

serum 25-hydroxyvitamin D concentrations and remain stable until serum 25-hydroxyvitamin D ACCEPTED MANUSCRIPT

288

concentrations fall below 30-40 ng/mL[38, 42]. In our study, we utilized a dose-response gradient to

289

evaluate the optimal serum concentrations of 25-hydroxyvitamin D for preventing bladder cancer. We

290

observed that the serum 25-hydroxyvitamin D concentration inversely correlated with the risk of bladder

291

cancer, and a serum 25-hydroxyvitamin D concentrations ≥ 74 nmol/L reduced the risk of bladder cancer

292

incidence by 60%.

RI PT

287

This review is characterized by numerous strengths. We conducted an extensive literature search to

294

maintain the quality of the studies and to include the most recent data. The strict inclusion and exclusion

295

criteria of this network meta-analysis were set to determine the quality of the included studies, which

296

enabled us to include the most valid and appropriate studies. In addition, the Bayesian model-based

297

network meta-analysis facilitated indirect comparisons between multiple concentrations, especially when

298

few studies that directly compared different concentrations of serum 25-hydroxyvitamin D were available.

299

Currently, network meta-analysis is the most appropriate method for multiple comparisons. Our study is

300

the first to rank the effects of different serum 25-hydroxyvitamin D concentrations on bladder cancer risk.

TE D

M AN U

SC

293

This study is also subject to several limitations. First, a network meta-analysis was applied to

EP

301

combine direct and indirect evidence, and the indirect effect could have influenced the overall effect when

303

the direct comparison was limited. Second, only seven studies were eligible for the network meta-analysis,

304

which prevented us from performing subgroup or sensitivity analyses. Third, although a network

305

meta-analysis of RCTs only would be ideal, the lack of available RCTs in this field prevented us from

306

reaching any definitive conclusions. Fourth, we extracted all the information from published data rather

307

than from individual patient data, which might have resulted in publication and reporting biases. Fifth, as

308

the included individuals were limited to narrow age and BMI ranges, the results might not be validated in

AC C

302

14

wider population group with extreme nutritional conditions. Finally, to minimize the risk of bias, we ACCEPTED MANUSCRIPT

310

restricted the inclusion criteria and comprehensively assessed the data. Excluding the studies by

311

Giovannucci et al. [30] and Afzal et al.[26] from the network meta-analysis may have negatively

312

influenced the validity of our findings. However, the pair-wise meta-analysis indicated that our results

313

remained valid when the aforementioned studies were included.

314

CONCLUSION

RI PT

309

In conclusion, we utilized a Bayesian network meta-analysis that compared five different

SC

315

concentrations of serum 25-hydroxyvitamin D to show that maintaining sufficient serum

317

25-hydroxyvitamin D concentration is associating with reducing the risk of bladder cancer. The

318

well-designed, large randomized control trials are needed to confirm these results.

319

ACKNOWLEDGMENTS

TE D

M AN U

316

The authors thank C. S. Sun and David L. Hill, Department of Psychology, University of Virginia,

320

Charlottesville, VA, for manuscript modification and research comments as well as Prof. J. X. Zhang,

322

Department of Medical Statistics and Epidemiology, School of Public Health, Sun Yat-Sen University,

323

Guangzhou, China, for statistical advice and research comments. None of these persons received

324

compensation for the work performed.

AC C

EP

321

The authors’ responsibilities were as follows: Jian Huang had full access to all the data in the study

325 326

and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept

327

and design: JH, CHC; Acquisition of data: YZ; Analysis and interpretation of data: WWP, CHC, MG;

328

Drafting of the manuscript: YZ, JH; Critical revision of the manuscript for important intellectual content:

329

CHC; Statistical analysis: JH; Obtaining funding: None; Administrative, technical, or material support: 15

330

MG; Supervision: JH; Read and approved the final manuscript: All authors. No funding entity had any role ACCEPTED MANUSCRIPT

331

in the design, implementation, analysis, or interpretation of the data. None of the authors reported a

332

conflict of interest related to the study.

RI PT

333

References

SC

[1] Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA: a cancer journal for clinicians. 2005;55:74-108. [2] Parkin DM. The global burden of urinary bladder cancer. Scandinavian journal of urology and nephrology Supplementum. 2008:12-20.

M AN U

[3] Holick CN, De Vivo I, Feskanich D, Giovannucci E, Stampfer M, Michaud DS. Intake of fruits and vegetables, carotenoids, folate, and vitamins A, C, E and risk of bladder cancer among women (United States). Cancer causes & control : CCC. 2005;16:1135-45. [4] Albert PJ, Proal AD, Marshall TG. Vitamin D: the alternative hypothesis. Autoimmunity reviews. 2009;8:639-44.

TE D

[5] Plum LA, DeLuca HF. Vitamin D, disease and therapeutic opportunities. Nature reviews Drug discovery. 2010;9:941-55.

EP

[6] Gallicchio L, Helzlsouer KJ, Chow WH, Freedman DM, Hankinson SE, Hartge P, et al. Circulating 25-hydroxyvitamin D and the risk of rarer cancers: Design and methods of the Cohort Consortium Vitamin D Pooling Project of Rarer Cancers. American journal of epidemiology. 2010;172:10-20.

AC C

[7] Donkena KV, Young CY. Vitamin d, sunlight and prostate cancer risk. Advances in preventive medicine. 2011;2011:281863. [8] Zhou W, Heist RS, Liu G, Asomaning K, Neuberg DS, Hollis BW, et al. Circulating 25-hydroxyvitamin D levels predict survival in early-stage non-small-cell lung cancer patients. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2007;25:479-85. [9] Goodwin PJ, Ennis M, Pritchard KI, Koo J, Hood N. Prognostic effects of 25-hydroxyvitamin D levels in early breast cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2009;27:3757-63. [10] Ng K, Meyerhardt JA, Wu K, Feskanich D, Hollis BW, Giovannucci EL, et al. Circulating 25-hydroxyvitamin d levels and survival in patients with colorectal cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2008;26:2984-91. 16

[11] Drake MT, Maurer MJ, Link BK, Habermann TM, Ansell SM, Micallef IN, et al. Vitamin D ACCEPTED MANUSCRIPT insufficiency and prognosis in non-Hodgkin's lymphoma. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2010;28:4191-8. [12] Mittal RD, Manchanda PK, Bhat S, Bid HK. Association of vitamin-D receptor (Fok-I) gene polymorphism with bladder cancer in an Indian population. BJU international. 2007;99:933-7. [13] Giovannucci E. The epidemiology of vitamin D and cancer incidence and mortality: a review (United States). Cancer causes & control : CCC. 2005;16:83-95.

RI PT

[14] Mills EJ, Ioannidis JP, Thorlund K, Schunemann HJ, Puhan MA, Guyatt GH. How to use an article reporting a multiple treatment comparison meta-analysis. Jama. 2012;308:1246-53.

SC

[15] Cipriani A, Higgins JP, Geddes JR, Salanti G. Conceptual and technical challenges in network meta-analysis. Annals of internal medicine. 2013;159:130-7. [16] Lu G, Ades AE. Combination of direct and indirect evidence in mixed treatment comparisons. Statistics in medicine. 2004;23:3105-24.

M AN U

[17] Caldwell DM, Welton NJ, Ades AE. Mixed treatment comparison analysis provides internally coherent treatment effect estimates based on overviews of reviews and can reveal inconsistency. Journal of clinical epidemiology. 2010;63:875-82. [18] Caldwell DM, Ades AE, Higgins JP. Simultaneous comparison of multiple treatments: combining direct and indirect evidence. BMJ. 2005;331:897-900.

TE D

[19] Jackson D, White IR, Riley RD. Quantifying the impact of between-study heterogeneity in multivariate meta-analyses. Statistics in medicine. 2012;31:3805-20. [20] Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Statistics in medicine. 2002;21:1539-58.

EP

[21] Ioannidis JP, Patsopoulos NA, Evangelou E. Uncertainty in heterogeneity estimates in meta-analyses. BMJ. 2007;335:914-6.

AC C

[22] Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997;315:629-34. [23] Gorham ED, Garland CF, Garland FC, Grant WB, Mohr SB, Lipkin M, et al. Optimal vitamin D status for colorectal cancer prevention: a quantitative meta analysis. American journal of preventive medicine. 2007;32:210-6. [24] Carpenter JR. A method for presenting and comparing dose-response curves. Journal of pharmacological methods. 1986;15:283-303. [25] Steenland K, Deddens JA. A practical guide to dose-response analyses and risk assessment in occupational epidemiology. Epidemiology. 2004;15:63-70. 17

[26] Afzal S, Bojesen SE, Nordestgaard BG. Low plasma 25-hydroxyvitamin D and risk of tobacco-related ACCEPTED MANUSCRIPT cancer. Clinical chemistry. 2013;59:771-80. [27] Amaral AF, Mendez-Pertuz M, Munoz A, Silverman DT, Allory Y, Kogevinas M, et al. Plasma 25-hydroxyvitamin D(3) and bladder cancer risk according to tumor stage and FGFR3 status: a mechanism-based epidemiological study. Journal of the National Cancer Institute. 2012;104:1897-904.

RI PT

[28] Brinkman MT, Buntinx F, Kellen E, Dagnelie PC, Van Dongen MC, Muls E, et al. Dietary intake of micronutrients and the risk of developing bladder cancer: results from the Belgian case-control study on bladder cancer risk. Cancer causes & control : CCC. 2011;22:469-78. [29] Brinkman MT, Karagas MR, Zens MS, Schned A, Reulen RC, Zeegers MP. Minerals and vitamins and the risk of bladder cancer: results from the New Hampshire Study. Cancer causes & control : CCC. 2010;21:609-19.

SC

[30] Giovannucci E, Liu Y, Rimm EB, Hollis BW, Fuchs CS, Stampfer MJ, et al. Prospective study of predictors of vitamin D status and cancer incidence and mortality in men. Journal of the National Cancer Institute. 2006;98:451-9.

M AN U

[31] Mondul AM, Weinstein SJ, Horst RL, Purdue M, Albanes D. Serum vitamin D and risk of bladder cancer in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening trial. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology. 2012;21:1222-5.

TE D

[32] Mondul AM, Weinstein SJ, Mannisto S, Snyder K, Horst RL, Virtamo J, et al. Serum vitamin D and risk of bladder cancer. Cancer research. 2010;70:9218-23. [33] Adorini L, Daniel KC, Penna G. Vitamin D receptor agonists, cancer and the immune system: an intricate relationship. Current topics in medicinal chemistry. 2006;6:1297-301.

EP

[34] Thierry-Palmer M, Carlyle KS, Williams MD, Tewolde T, Caines-McKenzie S, Bayorh MA, et al. Plasma 25-hydroxyvitamin D concentrations are inversely associated with blood pressure of Dahl salt-sensitive rats. The Journal of steroid biochemistry and molecular biology. 1998;66:255-61.

AC C

[35] Thierry-Palmer M, Doherty A, Bayorh MA, Griffin K. Dahl salt-sensitive rats excrete 25-hydroxyvitamin D into urine. The Journal of nutrition. 2003;133:187-90. [36] Chen F, Li Q, Yu Y, Yang W, Shi F, Qu Y. Association of vitamin C, vitamin D, vitamin E and risk of bladder cancer: a dose-response meta-analysis. Scientific reports. 2015;5:9599. [37] Liao Y, Huang JL, Qiu MX, Ma ZW. Impact of serum vitamin D level on risk of bladder cancer: a systemic review and meta-analysis. Tumour biology : the journal of the International Society for Oncodevelopmental Biology and Medicine. 2015;36:1567-72. [38] Looker AC, Gunter EW. Hypovitaminosis D in medical inpatients. The New England journal of medicine. 1998;339:344-5; author reply 5-6. [39] Malabanan A, Veronikis IE, Holick MF. Redefining vitamin D insufficiency. Lancet. 1998;351:805-6. 18

[40] Bischoff-Ferrari HA, Giovannucci E, Willett WC, Dietrich T, Dawson-Hughes B. Estimation of ACCEPTED MANUSCRIPT optimal serum concentrations of 25-hydroxyvitamin D for multiple health outcomes. The American journal of clinical nutrition. 2006;84:18-28. [41] Holick MF. High prevalence of vitamin D inadequacy and implications for health. Mayo Clinic proceedings. 2006;81:353-73.

SC

RI PT

[42] Chapuy MC, Preziosi P, Maamer M, Arnaud S, Galan P, Hercberg S, et al. Prevalence of vitamin D insufficiency in an adult normal population. Osteoporosis international : a journal established as result of cooperation between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA. 1997;7:439-43.

M AN U

FIGURE LEGENDS

Figure 1. Flow diagram of the identified, included, and excluded studies for the network meta-analysis.

Figure 2. Network established for each serum 25-hydroxyvitamin D comparison. Solid lines between drugs

TE D

represent the existence of direct comparisons. Each group is defined by a different serum 25-hydroxyvitamin D concentration: sufficient: > 75 nmol/L; insufficient: 50-75 nmol/L; slightly deficient: 37.5-50 nmol/L;

EP

moderately deficient: 25-37.5 nmol/L; and severely deficient: < 25 nmol/L.

Figure 3. The Bayesian network meta-analysis of bladder cancer risk in patients with different serum

AC C

25-hydroxyvitamin D concentrations based on combined direct and indirect evidence. (A) Pooled odds ratios (ORs) at each serum 25-hydroxyvitamin D concentration for bladder cancer risk. Upper triangles denote pooled ORs. The column treatment is compared with the row treatment. Numbers in parentheses indicate the corresponding 95% credible interval (CrI). Bold font numbers represent ORs with a Bayesian p < 0.05. Lower triangles denote no significant Bayesian inconsistency from the node-splitting analysis model (p > 0.05). (B) Rank probability for serum 25-hydroxyvitamin D concentration and the lowest risk (dark blue); the second (red), third (green), and fourth (purple) lowest risk; and the highest risk (light blue) for 19

bladder cancer. In this study, rank 1 isACCEPTED the best, and rank 5 is the worst. Each group is defined by different MANUSCRIPT serum 25-hydroxyvitamin D concentrations: sufficient: > 75 nmol/L; insufficient: 50-75 nmol/L; slightly deficient: 37.5-50 nmol/L; moderately deficient: 25-37.5 nmol/L; and severely deficient: < 25 nmol/L.

Figure 4. Forest plot of the highest versus the lowest serum 25-hydroxyvitamin D concentrations and

RI PT

bladder cancer risk.

Figure 5. Dose–response curve for bladder cancer according to serum 25-hydroxyvitamin D concentration

M AN U

D based on the combined data from the seven studies.

SC

(all seven studies were combined). The five points are the odds ratios for each quintile of 25-hydroxyvitamin

AC C

EP

TE D

Figure 6. Funnel plot of the publication bias and the p value from Egger's regression text.

20

ACCEPTED MANUSCRIPT

Table1 Basic characteristic of included studies Study

Study Design

Countr y

Participant s

Time of Grou follow-u p p (years)

N

Median age

Male/Femal e

BMI (kg Tumor type m 2)

Smoking status (never/occasional/former/curren t)

Years smokin g

Cigarettes/da y

Dietary vitamin D [µg/d(SD) ]

25(OH)D Quantile cutpoints (nmol/L)

Odds ratio P for by quantile trend (95% CI)

Low-grade :20

MT et al.2010[1]

Case-contro l

62 (9.2)

Cancer USA

561

MIBC

NA

(≥T2)

Control

239

60.7 (10.6)

138/

NA

101

Bladde r Mondul

Nested

AM et

Case-contro

59 (55-63)

29133

l 250

59 (56-63)

Bladde r

al.2011[3]

Case-contro l

Cancer Belgium

586

Bladde r

et al.2012[4]

l

Spain

2153

NA

27

377

112 5

64.2 (9.6)

68 (22–81)

226/

NA

Control

102 8

66 (20–81)

NA

(<25)

Low-grade

499

NMIBC

(25-26.99)

(TaG1/G2):579

986/

172

High-grade

139

(27-29.99)

NMIBC

123

(TaG3/T1) :20

60

5

32/166

156/221

NA

(>30)

NA

(Non-current/current)

NA

151

AC C

Cancer

Case-contro

171/

NA

NA

(23.5-28.4)

NA Control

Amaral AF

67.6 (9.9)

25.8

TE D

MT et

198

NA

56/0/162/104

95/0/103/41

NA

(23.3-27.8)

12 Control

Brinkman

NA

:93

NA

25.5

EP

al.2010[2]

250

Cancer Finland

High-grade :23

NA

85

M AN U

Brinkman

322

RI PT

r

1(ref)，

6

237/

SC

Bladde

(Non-current/current)

157/45/439/484

32.2 (14)

26.7 (14.8)

NA

10.7–24.3， 24.3-40.75 ≥40.75 27.4 (15)

40 (35-44)

38 (32-43)

NA

22.3 (15.6)

20 (15-25)

(27-29.99)

NA

20 (15-25)

1,

25-＜37.5，

1.01,

4.8

37.5-＜50，

0.97,

(3.5-6.9)

≥50

0.58

0.02

ug/d

NA

NA

NA

NA

NA

NA

(0.60–1.76)，

＜25， ug/d

7.24 (17.92)

8.54

1.0 (ref) 0.5-5

0.76

5-9.25

(0.47–1.24)

≥9.25

1.03

NA

0.93

(0.62–1.72)

(18.22)

NA

:246

290/82/383/273

0.22

(0.31–1.06)

＜25

1

25-＜34.5

0.82

34.5-＜50

0.78

50-＜75

0.77

≥75

0.52

＜25

1

25-＜37.5

0.81

37.5-＜50

1.28

(25-26.99)

119

1.03

(3.0-6.5)

415 169

(0.44–1.38)，

4.6

(<25)

909/

0.78

0.58

NA

MIBC

(≥T2)

0.95–10.7，

0.00 4

NA

136 (>30) 65 Mondul

Nested

AM et

Case-contro

al.2012[5]

l

Bladde USA

733

13

r Cancer

369

64 (60–68)

298/ 71

27.2 (24.4–29.5 )

NA

95/0/217/57

NA

NA

4.8 (3.2–7.1)

0.50

ACCEPTED MANUSCRIPT Control

i E et al.2006[6]

Prospective cohort

Afzal et

Prospective

al.2013[7]

cohort

64 (61–67)

294/ 70

(24.7–30.1

NA

160/0/169/35

NA

NA

)

4.9

USA

47,800

14

r

50-＜75

1.66

≥75

0.81

(3.4–7.6) NA

Bladde 382

NA

NA

NA

NA

NA

NA

NA

0.89(0.59-1.34

9.25 ug/d

)

Cancer

Denmark

9791

28

r Cancer

0.58

<12.5

Bladde 112

57.7(47-65

50/

)

62

NA

24.8

NA

AC C

EP

TE D

M AN U

SC

BMI: body mass index; NA: not applicate; MIBC = muscle-invasive bladder cancer; NMIBC = non-muscle-invasive bladder cancer

RI PT

Giovannucc

364

27.2

NA

NA

NA

12.5–<25

0.78(0.65-0.94

0.00

25–<50

)

8

>50

ACCEPTED MANUSCRIPT References [1] Brinkman MT, Karagas MR, Zens MS, Schned A, Reulen RC, Zeegers MP. Minerals and vitamins and the risk of bladder cancer: results from the New Hampshire Study. Cancer causes & control : CCC. 2010;21:609-19. [2] Mondul AM, Weinstein SJ, Mannisto S, Snyder K, Horst RL, Virtamo J, et al. Serum vitamin D and risk of bladder cancer. Cancer research. 2010;70:9218-23. [3] Brinkman MT, Buntinx F, Kellen E, Dagnelie PC, Van Dongen MC, Muls E, et al. Dietary intake of micronutrients and the risk of developing bladder cancer: results from the Belgian case-control study on bladder cancer risk. Cancer causes & control : CCC. 2011;22:469-78. [4] Amaral AF, Mendez-Pertuz M, Munoz A, Silverman DT, Allory Y, Kogevinas M, et al. Plasma 25-hydroxyvitamin D(3) and bladder cancer risk according to tumor stage and FGFR3 status: a mechanism-based epidemiological study. Journal of the National Cancer Institute. 2012;104:1897-904. [5] Mondul AM, Weinstein SJ, Horst RL, Purdue M, Albanes D. Serum vitamin D and risk of bladder cancer in the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening trial. Cancer epidemiology, biomarkers & prevention : a publication of the American Association for

RI PT

Cancer Research, cosponsored by the American Society of Preventive Oncology. 2012;21:1222-5. [6] Giovannucci E, Liu Y, Rimm EB, Hollis BW, Fuchs CS, Stampfer MJ, et al. Prospective study of predictors of vitamin D status and cancer incidence and mortality in men. Journal of the National Cancer Institute. 2006;98:451-9.

AC C

EP

TE D

M AN U

SC

[7] Afzal S, Bojesen SE, Nordestgaard BG. Low plasma 25-hydroxyvitamin D and risk of tobacco-related cancer. Clinical chemistry. 2013;59:771-80.

ACCEPTED MANUSCRIPT PubMed, EMBASE, MEDLINE, Cochrane, Web of Science

Studies identified through initial searches of

RI PT

electronic databases: n=87

Duplications reports: n = 42

SC

Titles and abstracts screened: n = 45

M AN U

Excluded studies: n = 35 -Editorials or letters: n= 6

-Reviews or meeting abstracts: n =11

Full-text articles screened n = 10

-Non-comparative studies: n= 18

TE D

Excluded studies: n= 3

-Irrelevant topics: n = 3

Included studies

EP

n=7

AC C

Figure 1. Flow diagram of the identified, included, and excluded studies for the network meta-analysis.

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 2. Network established for each serum 25-hydroxyvitamin D comparison. Solid lines between drugs represent the existence of

TE D

direct comparisons. Each group is defined by a different serum 25-hydroxyvitamin D concentrations: sufficient: > 75 nmol/L; insufficient:

AC C

EP

50-75 nmol/L; slightly deficient: 37.5-50 nmol/L; moderately deficient: 25-37.5 nmol/L; and severely deficient: < 25 nmol/L.

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 3. The Bayesian network meta-analysis of bladder cancer risk in patients with different serum 25-hydroxyvitamin D

EP

Concentrations based on combined direct and indirect evidence. (A) Pooled odds ratios (ORs) at each serum 25-hydroxyvitamin D concentration for bladder cancer risk. Upper triangles denote pooled ORs. The column treatment is compared with the row treatment. Numbers

AC C

in parentheses indicate the corresponding 95% credible interval (CrI). Bold font numbers represent ORs with a Bayesian p < 0.05. Lower triangles denote no significant Bayesian inconsistency from the node-splitting analysis model (p > 0.05). (B) Rank probability for serum 25-hydroxyvitamin D concentration and the lowest risk (dark blue); the second (red), third (green), and fourth (purple) lowest risk; and the highest risk (light blue) for bladder cancer. In this study, rank 1 is the best, and rank 5 is the worst. Each group is defined by different serum 25-hydroxyvitamin D levels: sufficient: > 75 nmol/L; insufficient: 50-75 nmol/L; slightly deficient: 37.5-50 nmol/L; moderately deficient: 25-37.5 nmol/L; and severely deficient: < 25 nmol/L.

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Figure 4. Forest plot of the highest versus the lowest serum 25-hydroxyvitamin D concentrations and bladder cancer risk.

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

Figure 5. Dose–response curve for bladder cancer according to serum 25-hydroxyvitamin D concentrations (all seven studies were

AC C

EP

TE D

combined). The five points are the odds ratios for each quintile of 25-hydroxyvitamin D based on the combined data from the seven studies.

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

Figure 6. Funnel plot of the publication bias and the p value from Egger's regression text.

ACCEPTED MANUSCRIPT

Highlights Based on the network meta-analysis, vitamin D was inversely associated with bladder cancer risk.




The sufficient serum 25(OH)D might play an important role in decreasing the bladder cancer risk.




The serum 25(OH)D concentration ≥ 74 nmol/L was associated with a 60% lower risk of bladder cancer.

AC C

EP

TE D

M AN U

SC

RI PT