Abstract
Background
Ursodeoxycholic acid (UDCA) was one of the earliest agents investigated as a drug for colorectal cancer prevention. However, UDCA failed to demonstrate efficacy to prevent the development of colorectal adenomas in a large phase III randomized controlled trial. We re-evaluated the effect of UDCA in men and women separately, based on sex-specific differences in bile acid metabolism and suspected variation in etiologic factors contributing to colorectal cancer risk.
Method
We conducted a secondary analysis of the efficacy of UDCA to prevent colorectal adenoma in men (n = 804) and women (n = 388).
Results
We found no reduction in risk of any metachronous adenoma with UDCA treatment in men or women. However, UDCA treatment significantly lowered the odds of advanced lesions (OR = 0.62, 95% CI = 0.43 to 0.89) in men, but not women. We also observed significantly higher odds of advanced lesions with UDCA treatment in women who were younger (age < 65 y; OR = 3.24, 95% CI = 1.10 to 9.56), obese (body mass index ≥ 30 kg/m2; OR = 5.45, 95% CI = 1.42 to 20.9), or in the highest tertile of total dietary fat (≥ 56.2 g/day; OR = 3.48, 95% CI = 1.35 to 8.95). The interactive effect of fat intake accounted for the modulating effects of age and BMI in women.
Conclusion
Our findings support the use of UDCA for preventing advanced colorectal adenomas in men. The increased odds of adenoma among women with high fat intake suggest a previously unrecognized harm that warrants further study, especially given some patients’ chronic exposure to UDCA for the treatment of primary biliary cirrhosis and the rising investigational use of this drug for several other conditions.
INTRODUCTION
Bile acids (BAs) are a group of acidic steroids produced in the liver, stored in the gall bladder, and secreted into the intestine to aid in the digestion of dietary lipids, where 90–95% are reabsorbed via enterohepatic recirculation (1). Secondary BAs have long been implicated in the etiology of colorectal cancer (CRC), particularly those arising in the proximal colon (2), for which women are at higher risk than men (3, 4). Ecological (5-8) and case-control studies (9-11), including a meta-analysis (12), support a positive association between fecal BA levels and colorectal neoplasia. Secondary BAs, including the hydrophobic deoxycholic acid (DCA), exhibit a spectrum of chemical and biologic properties including pro-tumorigenic effects, for which the exact pathway is unknown (13). One proposed mechanism is BA-induced DNA damage, which results in mutations and tumor initiation; alternatively, BA effects on cell signaling, including activation of β-catenin, and selection pressure for apoptosis resistance have been shown to promote development in a cancer-prone epithelium (14, 15).
One of the early, high-interest pharmacologic agents for chemoprevention of CRC was ursodeoxycholic acid (UDCA), a tertiary BA found naturally in low concentrations in humans (16). Early work suggested that long-term use of UDCA in patients with ulcerative colitis and primary sclerosing cholangitis reduced the risk of colonic neoplasia (17, 18). UDCA has been shown to prevent colorectal carcinogenesis in male Fisher rats receiving azoxymethane (19) and to counter the carcinogenic effects of secondary BAs, particularly DCA (4). UDCA also affects fecal BA levels, resulting proportionally lower concentrations of DCA relative to UDCA in some (20). In addition, a number of anti-tumor effects have been attributed to UDCA, including inhibitory activity for many of the same regulatory pathways activated by DCA, such as the mitogen-activated protein kinase pathway (21, 22). Furthermore, UDCA exhibits suppressive activity on arachidonic acid metabolism, including inhibition of cyclooxygenase-2 and inducible nitric oxide synthase (23, 24). These mechanistic actions made UDCA an attractive chemopreventive agent, particularly given its low toxicity (25).
In 2005, we reported the results of a large, phase III, double-blinded, placebo-controlled chemoprevention trial, which tested the efficacy of UDCA in persons with a history of colorectal adenoma (26). We did not detect a significant difference for the incidence of any metachronous adenoma between UDCA (daily 8–10 mg/kg of body weight) and placebo arms over three years. However, we observed a significant 39% reduction in colorectal adenoma with high-grade dysplasia. This result is consistent with previous studies that found decreased risk of CRC and high-grade dysplasia with UDCA use in patients with ulcerative colitis (18).
CRCs evolve through multiple (and possibly distinct) precursor pathways that manifest as differences in microsatellite stability, the degree of chromosomal instability, the extent of methylation in the CpG island of specific genes, and the presence and type of mutations in BRAF, KRAS, beta catenin, APC, and p53 (27). Differences in incidence rates, anatomic location (e.g., proximal versus distal), and risk factor profiles (e.g., bile acid exposure) for CRC subtypes in men and women suggest important sex-specific susceptibilities for the development of distinct morphologic and molecular subtypes (28). Thus, we re-analyzed data from the phase III UDCA chemoprevention trial in order to assess potential differences in efficacy by sex.
METHODS
Study population
A phase III clinical trial of UDCA to prevent metachronous adenomas was conducted at the Arizona Cancer Center, the details of which have been reported previously (26). Briefly, this double-blinded, placebo-controlled, randomized trial tested the efficacy of UDCA on metachronous adenoma among patients who had at least one colorectal adenoma removed during a pre-study colonoscopy. A total of 1,285 participants were randomized to daily UDCA or placebo, of which 1,192 (804 men and 388 women) had at least one follow-up colonoscopy.
Metachronous adenoma
As reported previously (26), participants were classified as having a metachronous adenoma if one or more colorectal adenomas or adenocarcinomas were discovered at least six months after the qualifying colonoscopy. Advanced adenomas were defined as those with adenocarcinoma, high-grade dysplasia, villous/tubulovillous histology, or a diameter ≥ 1 cm. Participants were classified as having a proximal adenoma if any of their metachronous adenomas occurred in the proximal colon. Multiplicity was defined as three or more metachronous adenomas.
Dietary measures
Study participants self-administered the Arizona Food Frequency Questionnaire at baseline (29). The questionnaire asked respondents to report how often they usually consumed each listed food item over the previous 12-month period. Subsequent nutrient calculations used age- and sex-specific portion size estimates.
Statistical methods
The association between UDCA treatment and any metachronous adenoma was assessed using logistic regression modeling, stratified by sex. For specific adenoma types, participants with a particular kind of lesion were compared to the combined group of those without any metachronous adenoma plus those with an adenoma that lacked the characteristic in question (e.g., participants with advanced lesion versus those without any metachronous adenoma plus those with non-advanced adenomas). Potential treatment-by-X [e.g., sex, age, body mass index (BMI), and dietary fat] interactions were assessed with likelihood-ratio tests, using both continuous and categorical versions of these variables in alternating models.
To investigate the potential relationship between multiple modulating factors, sex-stratified multivariate models testing the effect of treatment were generated for each of the adenoma-related endpoints. These models were adjusted for age, BMI, and dietary fat, allowing for the potential of two-way interactions between treatment and these three covariates. Manual backwards elimination variable selection started with all two-way interaction terms in the models and removed the interaction terms and covariates with Wald statistics P > 0.05 in a hierarchical manner. All statistical analyses were performed using Stata 10.1 (StataCorp, College Station, Texas), all reported P values are two-sided, and no adjustments were made for multiple comparisons.
RESULTS
Baseline characteristics
Two-thirds of the trial’s participants were male (n = 804, 67%; Table 1). Compared with women, male participants consumed more dietary calcium (1066 vs. 889 mg/day), had a higher BMI (28.5 vs. 27.5 kg/m2), and had a higher waist-to-hip ratio (0.96 vs. 0.81). Women were more likely than men to report a family history of CRC (34 vs. 24%), but they were less likely to report use of aspirin (19 vs. 32%) or BA sequestrants (29 vs. 37%). The proportion of white subjects (95%) did not differ by sex, and neither did the mean age (66 y) nor mean level of serum triglycerides (157 mg/dL). Half (51%) of the women reported using hormone replacement therapy. There were no substantial differences in the levels of any of these variables between the two treatment arms for either sex.
Table 1.
Characteristics | All participants (n = 1192) |
Men (n = 804) |
Women (n = 388) |
|||
---|---|---|---|---|---|---|
Placebo (n = 579) |
UDCA (n = 613) |
Placebo (n = 381) |
UDCA (n = 423) |
Placebo (n = 198) |
UDCA (n = 190) |
|
|
|
|
||||
Male sex, n (%) | 381 (47) | 423 (53) | --- | --- | --- | --- |
Age (y), mean ± SD | 66.5 ± 8.3 | 66.0 ± 8.6 | 66.6 ± 8.1 | 66.0 ± 8.6 | 66.3 ± 8.7 | 65.9 ± 8.7 |
White ethnicity/race*, n (%) | 535 (94) | 573 (95) | 351 (94) | 393 (95) | 184 (93) | 180 (96) |
Total dietary fat (g/day), mean ± SD | 60.7 ± 29.8 | 63.6 ± 33.5 | 68.3 ± 30.6 | 70.5 ± 34.7 | 46.1 ± 21.6 | 48.2 ± 24.4 |
Dietary calcium (mg/day), mean ± SD | 1003 ± 511 | 1014 ± 479 | 1046 ± 502 | 1084 ± 483 | 919 ± 517 | 858 ± 432 |
BMI (kg/m2)*, mean ± SD | 28.2 ± 4.8 | 28.1 ± 4.9 | 28.6 ± 4.3 | 28.4 ± 4.2 | 27.4 ± 5.5 | 27.5 ± 6.3 |
Waist-to-hip ratio*, mean ± SD | 0.91 ± 0.10 | 0.91 ± 0.09 | 0.96 ± 0.05 | 0.96 ± 0.06 | 0.81 ± 0.09 | 0.82 ± 0.07 |
Triglycerides (mg/dL), mean ± SD | 155 ± 93 | 160 ± 99 | 153 ± 90 | 163 ± 108 | 160 ± 98 | 153 ± 75 |
Family history of colorectal cancer†, n (%) | 171 (30) | 156 (25) | 102 (27) | 93 (22) | 69 (35) | 63 (33) |
Bile acid sequestrants§, n (%) | 204 (35) | 203 (33) | 145 (38) | 151 (36) | 59 (30) | 52 (27) |
Hormone replacement therapy§, n (%) | --- | --- | --- | --- | 101 (51) | 96 (51) |
Abbreviations: UDCA, ursodeoxycholic acid; SD, standard deviation; BMI, body mass index
Participants with missing data: ethnicity/race, 20; BMI, 26; waist-to-hip ratio, 105
Self-reported history of colorectal cancer in a parent, sibling, or child
Self-reported regular aspirin use within one month before randomization
Self-reported ever using bile acid sequestrants or hormone replacement therapy at any time during the trial
Effect modification by sex
Consistent with the original findings (26), we found no significant reduction in any metachronous adenoma among men nor women taking UDCA (Table 2). However, we detected effect modification by sex when examining specific types of adenomas. Men in the treatment group experienced a significant 57% reduction in the odds of high-grade dysplastic adenomas (OR = 0.43, 95% CI = 0.24 to 0.76), but there was no effect on such lesions among women (Pinteraction = 0.033). We also found a significant 39% reduction in the odds of large adenomas for men in the treatment group (OR = 0.61, 95% CI = 0.38 to 0.96), but there was no effect on large lesions among women (Pinteraction = 0.013). Furthermore, we detected a significant 38% reduction in the odds of advanced adenomas for men in the treatment group (OR = 0.62, 95% CI = 0.43 to 0.89), but there was no effect on advanced lesions among women (Pinteraction = 0.006). There were no significant treatment effects for adenomas defined by villous/tubulovillous histology, proximal location, or multiplicity (three or more adenomas) for men or women.
Table 2.
Men (n = 804) |
Women (n = 388) |
P interaction * | |
---|---|---|---|
|
|
|
|
Any adenoma | |||
UDCA, n (%) | 182/423 (43) | 69/190 (36) | |
Placebo, n (%) | 179/381 (47) | 75/198 (38) | |
OR (95% CI) | 0.85 (0.65 – 1.13) | 0.94 (0.62 – 1.41) | 0.714 |
High-grade dysplasia† | |||
UDCA, n (%) | 19/415 (5) | 14/188 (7) | |
Placebo, n (%) | 38/377 (10) | 12/197 (6) | |
OR (95% CI) | 0.43 (0.24 – 0.76) | 1.24 (0.56 – 2.76) | 0.033 |
Large (≥ 1 cm) | |||
UDCA, n (%) | 34/423 (8) | 20/190 (11) | |
Placebo, n (%) | 48/381 (13) | 12/197 (6) | |
OR (95% CI) | 0.61 (0.38 – 0.96) | 1.82 (0.87 – 3.84) | 0.013 |
Villous/tubulovillous histology | |||
UDCA, n (%) | 32/423 (8) | 18/190 (9) | |
Placebo, n (%) | 34/381 (9) | 11/198 (6) | |
OR (95% CI) | 0.84 (0.50 – 1.38) | 1.78 (0.82 – 3.87) | 0.107 |
Advanced lesion‡ | |||
UDCA, n (%) | 62/418 (15) | 37/189 (20) | |
Placebo, n (%) | 83//379 (22) | 27/197 (14) | |
OR (95% CI) | 0.62 (0.43 – 0.89) | 1.53 (0.89 – 2.64) | 0.006 |
Proximal location§ | |||
UDCA, n (%) | 133/420 (32) | 50/190 (26) | |
Placebo, n (%) | 132/380 (35) | 54/198 (27) | |
OR (95% CI) | 0.87 (0.65 – 1.17) | 0.95 (0.61 – 1.49) | 0.744 |
Multiple adenomas∥ | |||
UDCA, n (%) | 41/423 (10) | 7/190 (4) | |
Placebo, n (%) | 40/381 (11) | 14/198 (7) | |
OR (95% CI) | 0.91 (0.58 – 1.45) | 0.50 (0.20 – 1.27) | 0.250 |
Abbreviations: UDCA, ursodeoxycholic acid; OR, odds ratio; CI, confidence interval
P for treatment-by-sex interaction calculated using likelihood ratio tests
Missing dysplasia data for 12 men and 3 women
Advanced: adenocarcinoma, high-grade dysplasia, villous or tubulovillous histology, or large (≥ 1 cm); missing advanced lesion data for 7 men and 2 women
Proximal: any proximal adenoma; missing proximal location data for 4 men and 0 women
Multiple: three or more adenomas
Modulating effects of age, BMI, and dietary fat
Given the benefit of UDCA in men and not women for colorectal adenoma prevention, we hypothesized that factors that influence BA metabolism and secretion (30) may explain our gender-specific results. Therefore, we further investigated male-female differences by examining whether age, BMI, or total dietary fat modified the effect of UDCA treatment on metachronous adenoma. For men, there were no significant treatment-by-age interactions for any type of metachronous adenoma (Table 3). For women, however, we detected significantly increased odds of advanced lesions with UDCA treatment among women < 65 years old (OR = 3.24, 95% CI = 1.10 to 9.56; Pinteraction = 0.015). Similarly, there were no significant treatment-by-BMI interactions for any type of metachronous adenoma for men (Table 4). In contrast, we detected significantly increased odds of advanced lesions with UDCA treatment among obese women (BMI ≥ 30 kg/m2; OR = 5.45, 95% CI = 1.42 to 20.9, Pinteraction = 0.013).
Table 3.
Age (tertiles), y | Men |
Women |
||||
---|---|---|---|---|---|---|
40 – 64 (n = 276) |
65 – 71 (n = 291) |
72 – 81 (n = 237) |
40 – 64 (n = 142) |
65 – 71 (n = 126) |
72 – 81 (n = 120) |
|
|
|
|||||
Any adenoma | ||||||
UDCA, n (%) | 58/147 (39) | 65/158 (41) | 59/118 (50) | 23/71 (32) | 25/60 (42) | 21/59 (36) |
Placebo, n (%) | 57/129 (44) | 59/133 (44) | 63/119 (53) | 25/71 (35) | 20/66 (30) | 30/61 (49) |
OR (95% CI) | 0.82 (0.51 – 1.33) | 0.88 (0.55 – 1.40) Pinteraction* = 0.318 (0.973) |
0.89 (0.53 – 1.48) | 0.88 (0.44 – 1.77) | 1.64 (0.79 – 3.42) Pinteraction = 0.362 (0.131) |
0.57 (0.27 – 1.19) |
Advanced lesion† | ||||||
UDCA, n (%) | 20/146 (14) | 25/155 (16) | 17/117 (15) | 14/71 (20) | 11/59 (19) | 12/59 (20) |
Placebo, n (%) | 26/129 (20) | 28/133 (21) | 29/117 (5) | 5/71 (7) | 9/66 (14) | 13/60 (22) |
OR (95% CI) | 0.63 (0.33 – 1.19) | 0.72 (0.40 – 1.31) Pinteraction = 0.466 (0.762) |
0.52 (0.27 – 1.00) | 3.24 (1.10 – 9.56) | 1.45 (0.56 – 3.79) Pinteraction = 0.015 (0.194) |
0.92 (0.38 – 2.23) |
Proximal location‡ | ||||||
UDCA, n (%) | 34/145 (23) | 54/158 (34) | 45/117 (38) | 15/71 (21) | 19/60 (32) | 16/59 (27) |
Placebo, n (%) | 43/128 (34) | 43/133 (32) | 46/119 (39) | 16/71 (23) | 15/66 (23) | 23/61 (38) |
OR (95% CI) | 0.61 (0.36 – 1.03) | 1.09 (0.67 – 1.77) Pinteraction = 0.773 (0.243) |
0.99 (0.59 – 1.68) | 0.92 (0.42 – 2.04) | 1.58 (0.71 – 3.48) Pinteraction = 0.438 (0.245) |
0.61 (0.28 – 1.33) |
Multiple adenomas§ | ||||||
UDCA, n (%) | 8/147 (5) | 17/158 (11) | 16/118 (14) | 3/71 (4) | 1/60 (2) | 3/59 (5) |
Placebo, n (%) | 10/129 (8) | 16/133 (12) | 14/119 (12) | 3/71 (4) | 3/66 (5) | 8/61 (13) |
OR (95% CI) | 0.68 (0.26 – 1.79) | 0.88 (0.43 – 1.82) Pinteraction = 0.888 (0.679) |
1.18 (0.55 – 2.53) | 1.00 (0.19 – 5.13) | 0.36 (0.04 – 3.52) Pinteraction = 0.264 (0.602) |
0.35 (0.09 – 1.41) |
Abbreviations: UDCA, ursodeoxycholic acid; OR, odds ratio; CI, confidence interval
P for treatment-by-age interaction calculated using likelihood ratio tests with continuous (and categorical) age
Advanced: adenocarcinoma, high-grade dysplasia, villous or tubulovillous histology, or large (≤ 1 cm); missing advanced lesion data for 7 men and 2 women
Proximal: any proximal adenoma; missing proximal location data for 4 men and 0 women
Multiple: three or more adenomas
Table 4.
BMI*, kg/m2 | Men |
Women |
||||
---|---|---|---|---|---|---|
18 – <25 (n = 160) |
25 – <30 (n = 373) |
≥ 30 (n = 256) |
18 – <25 (n = 144) |
25 – <30 (n = 129) |
≥ 30 (n = 101) |
|
|
|
|||||
Any adenoma | ||||||
UDCA, n (%) | 28/82 (34) | 87/194 (45) | 64/137 (47) | 23/66 (35) | 25/68 (37) | 19/46 (41) |
Placebo, n (%) | 33/78 (42) | 87/179 (49) | 56/119 (47) | 29/78 (37) | 25/61 (41) | 20/55 (36) |
OR (95% CI) | 0.71 (0.37 – 1.34) | 0.86 (0.57 – 1.29) Pinteraction† = 0.773 (0.721) |
0.99 (0.60 – 1.61) | 0.90 (0.46 – 1.79) | 0.84 (0.41 – 1.70) Pinteraction = 0.261 (0.763) |
1.23 (0.55 – 2.75) |
Advanced lesion‡ | ||||||
UDCA, n (%) | 7/81 (9) | 34/193 (18) | 21/135 (16) | 10/66 (15) | 15/67 (22) | 11/46 (24) |
Placebo, n (%) | 9/78 (12) | 41/177 (23) | 31/119 (26) | 17/78 (22) | 7/60 (12) | 3/55 (5) |
OR (95% CI) | 0.73 (0.26 – 2.05) | 0.71 (0.43 – 1.18) Pinteraction = 0.408 (0.733) |
0.52 (0.28 – 0.97) | 0.64 (0.27 – 1.52) | 2.18 (0.82 – 5.79) Pinteraction = 0.013 (0.014) |
5.45 (1.42 – 20.9) |
Proximal location§ | ||||||
UDCA, n (%) | 17/81 (21) | 69/192 (36) | 44/137 (32) | 16/66 (24) | 18/68 (26) | 15/46 (33) |
Placebo, n (%) | 16/78 (21) | 65/178 (37) | 48/119 (40) | 18/78 (23) | 20/61 (33) | 16/55 (29) |
OR (95% CI) | 1.03 (0.48 – 2.22) | 0.98 (0.64 – 1.49) Pinteraction = 0.270 (0.562) |
0.70 (0.42 – 1.17) | 1.07 (0.49 – 2.31) | 0.74 (0.35 – 1.58) Pinteraction = 0.285 (0.684) |
1.18 (0.51 – 2.75) |
Multiple adenomas∥ | ||||||
UDCA, n (%) | 7/82 (9) | 19/194 (10) | 14/137 (10) | 1/66 (2) | 4/68 (6) | 2/46 (4) |
Placebo, n (%) | 6/78 (8) | 18/179 (10) | 15/119 (13) | 5/78 (6) | 6/61 (10) | 3/55 (5) |
OR (95% CI) | 1.12 (0.36 – 3.49) | 0.97 (0.49 – 1.92) Pinteraction = 0.749 (0.865) |
0.79 (0.36 – 1.71) | 0.22 (0.03 – 1.97) | 0.57 (0.15 – 2.13) Pinteraction = 0.509 (0.644) |
0.79 (0.13 – 4.93) |
Abbreviations: UDCA, ursodeoxycholic acid; BMI, body mass index; OR, odds ratio; CI, confidence interval
Missing data on BMI for 15 men and 11 women; there are 0 men and 3 women (2 placebo, 1 UDCA) with BMI < 18.5
P for treatment-by-BMI interaction calculated using likelihood ratio tests with continuous (and categorical) BMI
Advanced: adenocarcinoma, high-grade dysplasia, villous or tubulovillous histology, or large (≥ 1 cm); missing advanced lesion data for 6 men and 2 women
Proximal: any proximal adenoma; missing proximal location data for 4 men and 0 women
Multiple: three or more adenomas
We also found significant treatment-by-fat interactions in women for any metachronous adenoma (Pinteraction = 0.001), advanced lesions (Pinteraction = 0.028), proximal location (Pinteraction = 0.006), and multiplicity (Pinteraction = 0.030), but not for men (Table 5). The adverse effect of UDCA in women was restricted to those in the highest tertile of total dietary fat (≥ 52.6 g/day). Women in the lowest tertile (< 35.5 g/day) showed borderline significant protection against any adenoma with UDCA treatment (OR = 0.51, 95% CI = 0.25 to 1.03), compared with significantly increased odds in the highest tertile (OR = 2.50, 95% CI = 1.21 to 5.18). Furthermore, we found significantly increased odds of advanced lesions for women in the highest tertile of total dietary fat (OR = 3.48, 95% CI = 1.35 to 8.95).
Table 5.
Dietary fat (tertiles), g/day |
Men |
Women |
||||
---|---|---|---|---|---|---|
< 51.1 (n = 268) |
51.1 – < 76.5 (n = 268) |
≥ 76.5 (n = 268) |
< 35.5 (n = 130) |
35.5 – < 52.6 (n = 129) |
≥ 52.6 (n = 129) |
|
|
|
|||||
Any adenoma | ||||||
UDCA, n (%) | 58/142 (41) | 65/139 (47) | 59/142 (42) | 21/60 (35) | 15/64 (23) | 33/66 (50) |
Placebo, n (%) | 60/126 (48) | 55/129 (51) | 53/126 (42) | 36/70 (51) | 21/65 (32) | 18/63 (29) |
OR (95% CI) | 0.76 (0.47 – 1.23) | 0.84 (0.52 – 1.35) Pinteraction* = 0.130 (0.765) |
0.98 (0.60 – 1.59) | 0.51 (0.25 – 1.03) | 0.64 (0.29 – 1.40) Pinteraction = 0.001 (0.004) |
2.50 (1.21 – 5.18) |
Advanced lesion† | ||||||
UDCA, n (%) | 16/141 (11) | 25/137 (18) | 21/140 (15) | 11/60 (18) | 6/63 (10) | 20/66 (30) |
Placebo, n (%) | 24/126 (19) | 33/127 (26) | 26/126 (21) | 11/69 (16) | 9/65 (14) | 7/63 (11) |
OR (95% CI) | 0.54 (0.27 – 1.08) | 0.64 (0.35 – 1.14) Pinteraction = 0.473 (0.893) |
0.68 (0.36 – 1.28) | 1.18 (0.47 –2.96) | 0.65 (0.22 – 1.96) Pinteraction = 0.028 (0.055) |
3.48 (1.35 –8.95) |
Proximal location‡ | ||||||
UDCA, n (%) | 43/141 (31) | 48/138 (35) | 42/141 (30) | 17/60 (28) | 11/64 (17) | 22/66 (33) |
Placebo, n (%) | 42/126 (33) | 50/128 (39) | 40/126 (32) | 24/70 (34) | 18/65 (28) | 12/63 (19) |
OR (95% CI) | 0.88 (0.53 – 1.47) | 0.83 (0.51 – 1.37) Pinteraction = 0.470 (0.969) |
0.91 (0.54 – 1.54) | 0.76 (0.36 – 1.60) | 0.54 (0.23 – 1.26) Pinteraction = 0.006 (0.050) |
2.13 (0.94 –4.78) |
Multiple adenomas§ | ||||||
UDCA, n (%) | 7/142 (5) | 17/139 (12) | 17/142 (12) | 2/60 (3) | 0/64 (0) | 5/66 (8) |
Placebo, n (%) | 10/126 (8) | 15/129 (12) | 15/126 (12) | 7/70 (10) | 5/65 (8) | 2/63 (3) |
OR (95% CI) | 0.60 (0.22 – 1.63) | 1.06 (0.51 – 2.22) Pinteraction = 0.581 (0.636) |
1.01 (0.48 – 2.11) | 0.31 (0.06 – 1.55) | n/a Pinteraction = 0.030 (0.062) |
2.50 (0.47 – 13.4) |
Abbreviations: UDCA, ursodeoxycholic acid; OR, odds ratio; CI, confidence interval
P for treatment-by-age interaction calculated using likelihood ratio tests with continuous (and categorical) age
Advanced: adenocarcinoma, high-grade dysplasia, villous or tubulovillous histology, or large (≥ 1 cm); missing advanced lesion data for 7 men and 2 women
Proximal: any proximal adenoma; missing proximal location data for 4 men and 0 women
Multiple: three or more adenomas
Dietary fat explains the modulating effects of age and BMI
When examined separately, age, BMI, and total dietary fat each modified the effect of UDCA treatment in women. Age and BMI were both significantly correlated with total dietary fat (ρ = −0.18 and 0.20, respectively; both P < 0.001), but not with each other (P = 0.594), in women. Thus, we sought to determine whether or not the modulating effects of age and BMI were explained by fat intake, using sex-stratified multivariate models. For any adenoma, proximal location, and multiplicity, only the treatment-by-fat interaction remained significant (P < 0.05), whereas both the treatment-by-age and treatment-by-BMI interactions were not (P > 0.05). Age continued to be significantly associated with all of these outcomes, and BMI remained significantly associated with only proximal location. Thus, for these outcomes, the modulating effects of age and BMI were explained by the interaction of treatment with fat intake. For men, in contrast, all three interaction terms were removed from the multivariate models for all adenoma-related outcomes (P > 0.05).
Our results were not substantially different when we used saturated fat intake in these multivariate models instead of total dietary fat; the one exception was for the proximal location outcome in women, in which the treatment-by-fat interaction was no longer significant. We did not model both fat variables simultaneously due to their extremely high correlation (ρ = 0.96, P < 0.001). Furthermore, we explored other treatment-by-X interactions for family history of CRC, aspirin use, BA sequestrants, and hormone replacement therapy; we found no significant interactions with any of these variables (data not shown).
DISCUSSION
In a secondary analysis of a randomized UDCA trial, we found that sex significantly modified the effect of UDCA treatment on certain subtypes of metachronous colorectal adenomas. UDCA significantly reduced the odds of large, high-grade dysplastic, and advanced adenomas for men, but not women. We also observed significantly increased odds of advanced lesions among women in the treatment group who were young (age < 65 y), obese (BMI ≥ 30 kg/m2), or in the highest tertile of total dietary fat (≥ 52.5 g/day). Multivariate modeling showed that the modulating effects of age and BMI in women could be explained by total dietary fat, as these variables were strongly correlated. To our knowledge, no previous studies have evaluated sex-specific benefit or risk of UDCA in any setting, including use in patients with ulcerative colitis who were receiving UDCA for the treatment of primary sclerosing cholangitis (17, 31, 32). These studies, however, are limited by small sample sizes.
Recent advances in the molecular classification of CRCs have shown that colorectal tumors comprise a heterogeneous group of diseases. These tumors are thought to arise through molecularly distinct precursors that exhibit anatomic preference in the colon (i.e., proximal vs. distal) and for which men and women experience differential risk (27). Thus, UDCA may act in only a subgroup of adenoma formers, which led us to re-evaluate UDCA benefit in our trial, using existing information on factors that may influence subtype-specific risk (e.g., sex, age, anatomic location, development of multiple lesions, and family history of CRC). Our finding that the odds of advanced adenomas were reduced by 38% in men receiving UDCA supports our a priori hypothesis that UDCA acts as a chemopreventive agent for CRC in humans. We found no evidence of a differential effect by anatomic location or the development of multiple adenomas.
UDCA was originally introduced as a drug for gallstone dissolution, acting to reduce stone formation by lowering the cholesterol saturation index of bile, protecting against cholesterol nucleation, and possibly by increasing gallbladder volume and prolonging emptying (33). Female sex hormones and obesity are two of the best recognized risk factors for gallstone formation (34). Our observation of an adverse effect of UDCA in the same subgroup as those at highest risk for gallstones leads us to hypothesize that UDCA may increase exposure of the intestinal tract to lithogenic and potentially pro-tumorigenic and/or proinflammatory BAs, which may derive from small, asymptomatic gallstones. Alternatively, in persons at high risk for gallstones, UDCA might have a differential adverse effect on the composition of secondary and tertiary BAs in the colon. Examples of gender differences influencing composition, conjugation, and transport/uptake of BAs have been described in animal models (35, 36) and humans (37, 38), though sex-specific analyses are limited.
We attempted to address the potential contribution of existing gall bladder disease as a risk factor in our study by assessing participant use of BA sequestrants, but we found no supportive evidence of an interaction with UDCA (data not shown). Data on cholecystectomy were not available, so we were unable to assess the potential role of gall bladder removal. Of the other putative modifiers of BA exposure for which we had data (e.g., dietary calcium, serum cholesterol, and serum triglycerides), only dietary fat modified the effect of UDCA among women; furthermore, it appeared to account for the observed treatment interactions with age and BMI. Thus, the observed modulating effects of younger age and BMI in women are likely related to their association with higher intakes of dietary fat, which have been suggested to be associated with more lithogenic bile in animal models (39). Results from studies in humans are equivocal, where recognized gender differences and broad inter-individual variability limit interpretation of association studies and small feeding trials (40, 41).
The exact mechanism of action of UDCA for the prevention of high-grade dysplasia observed in our study (26) and others (17, 18) is unknown. Biologic differences between pro-tumorigenic and anti-tumorigenic BAs are subtle (42). Extensive overlap in the chemical and signaling properties of pro-tumorigenic hydrophobic (e.g., DCA) and anti-tumorigenic hydrophilic (e.g., UDCA) BAs and effects on nuclear receptors, such as the nuclear BA receptor farnesoid X receptor (15), challenge our understanding of the opposing effects of these molecules. UDCA has potent cytoprotective effects in hepatocytes and cholangiocytes, as well as colon cancer cell lines, which is counterintuitive for an inherent anti-cancer activity (15, 22). Anti-apoptotic effects that have been attributed to UDCA include inhibition of p53 transcription, resulting in suppression of Bax translocation from the cytosol to the mitochondrial membrane, cytochrome c release, caspase activation, mitochondrial membrane perturbation, and pore formation (14, 15, 43). Under certain circumstances, UDCA may offer a survival advantage to premalignant lesions, as observed here, thereby countering any anti-tumor activity. This pro-tumorigenic effect might occur via effects of sex on the metabolism and/or transport of UDCA and its conjugated forms; alternatively, the interaction of cofactors (e.g., steroid hormones or dietary fat) on signaling pathways could modulate the effect of UDCA. Additional studies are needed to separate the anti-tumor action of UDCA from possible pro-tumorigenic activities and to understand their relationship to sex and potential signaling cofactors, like dietary fat.
Regardless of the biological mechanism, our finding of a potential adverse effect of UDCA on colorectal adenoma in women is important because UDCA is already approved for a number of uses. Most commonly, UDCA is prescribed for the treatment of primary bile duct disease, particularly primary biliary cirrhosis, which is substantially more common in women than men (44). Beneficial effects of UDCA for the treatment of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis have also been reported (45). In addition, the potent anti-apoptotic activities of taurine-conjugated UDCA (tauroursodeoxycholic acid) have prompted its study for the prevention of apoptotic-mediated disease, including amyotrophic lateral sclerosis (http://clinicaltrials.gov/ct2/show/NCT00877604) and Alzheimer’s disease (46), the latter of which represents a sizable patient population that might be exposed. Furthermore, tauroursodeoxycholic acid has been demonstrated to alleviate endoplasmic reticulum stress and restore glucose homeostasis in a mouse model of type 2 diabetes (47). The fact that patients are already using UDCA, with increasing clinical uses on the horizon, highlights the necessity of understanding whether or not this agent exhibits pro-tumorigenic effects.
To our knowledge, no previous studies have reported any serious adverse effects of UDCA in the clinical setting; however, the evaluable data come from studies with limited use for cancer (or adenoma) endpoints, and the sample sizes are generally too small for stratification by age or sex. Our findings are the first to demonstrate a potential adverse effect of UDCA for adenoma development in the colorectum. Though the trial was not powered for sex-stratification, we were able to detect significantly increased odds of advanced lesions in young and obese women and in women consuming fat in the upper tertile of intake. Due to our small sample size, however, we were not able to stratify by age, BMI, and dietary fat simultaneously. Additionally, we chose not to adjust for multiple comparisons, as we envisioned these analyses as hypothesis-generating and not hypothesis-testing. We were also unable to investigate the potential role of gall bladder removal, as this information was not available for our study. Furthermore, the three-year trial duration may be too short to capture the important effects of UDCA on various types of colorectal adenomas, let alone cancer.
Even though additional studies are certainly needed, the significantly higher odds of adenoma in women with the highest dietary fat intake with UDCA exposure suggest the potential for a pro-neoplasia effect of UDCA in women, though UDCA treatment may protect low fat-intake women against metachronous adenomas. Furthermore, UDCA seems protective in men for advanced adenomas. Although these results support a chemopreventive effect of UDCA for colorectal adenoma, they highlight a need for additional efforts to separate the anti-tumor properties of UDCA from potential pro-neoplasia effects. Further investigation is needed to confirm our results prior to recommending any change in the clinical use of UDCA. In the interim, female patients taking UDCA for primary biliary cirrhosis—women suffer this condition far more frequently than do men and generally take UDCA for the rest of their lives—and participants in clinical trials of UDCA for other conditions should follow current CRC screening recommendations in order to detect advanced adenomas and possible cancers early and thus mitigate any potential harm from UDCA.
Acknowledgments
FUNDING
This work was supported by the National Institutes of Health and the National Cancer Institute [P01-CA-41108]. ET Jacobs is supported by a Career Development Award from the National Cancer Institute [1K07CA10629-01A1].
GRANT SUPPORT
Colon Cancer Prevention Program Project Grant [NCI/NIH PO1 CA4108]
REFERENCES
- 1.Martinez-Augustin O, Sanchez de Medina F. Intestinal bile acid physiology and pathophysiology. World J Gastroenterol. 2008;14:5630–40. doi: 10.3748/wjg.14.5630. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.McMichael AJ, Potter JD. Host factors in carcinogenesis: certain bile-acid metabolic profiles that selectively increase the risk of proximal colon cancer. J Natl Cancer Inst. 1985;75:185–91. [PubMed] [Google Scholar]
- 3.Distler P, Holt PR. Are right- and left-sided colon neoplasms distinct tumors? Dig Dis. 1997;15:302–11. doi: 10.1159/000171605. [DOI] [PubMed] [Google Scholar]
- 4.Bernstein H, Bernstein C, Payne CM, Dvorakova K, Garewal H. Bile acids as carcinogens in human gastrointestinal cancers. Mutat Res. 2005;589:47–65. doi: 10.1016/j.mrrev.2004.08.001. [DOI] [PubMed] [Google Scholar]
- 5.Hill MJ, Aries VC. Faecal steroid composition and its relationship to cancer of the large bowel. J Pathol. 1971;104:129–39. doi: 10.1002/path.1711040207. [DOI] [PubMed] [Google Scholar]
- 6.Jensen OM, MacLennan R, Wahrendorf J. Diet, bowel function, fecal characteristics, and large bowel cancer in Denmark and Finland. Nutr Cancer. 1982;4:5–19. doi: 10.1080/01635588209513733. [DOI] [PubMed] [Google Scholar]
- 7.McKeigue PM, Adelstein AM, Marmot MG, et al. Diet and fecal steroid profile in a South Asian population with a low colon-cancer rate. Am J Clin Nutr. 1989;50:151–4. doi: 10.1093/ajcn/50.1.151. [DOI] [PubMed] [Google Scholar]
- 8.Reddy BS, Hedges A, Laakso K, Wynder EL. Fecal constituents of a high-risk North American and a low-risk Finnish population for the development of large bowel cancer. Cancer Lett. 1978;4:217–22. doi: 10.1016/s0304-3835(78)94622-0. [DOI] [PubMed] [Google Scholar]
- 9.Hill MJ, Drasar BS, Williams RE, et al. Faecal bile-acids and clostridia in patients with cancer of the large bowel. Lancet. 1975;1:535–9. doi: 10.1016/s0140-6736(75)91556-1. [DOI] [PubMed] [Google Scholar]
- 10.Imray CH, Radley S, Davis A, et al. Faecal unconjugated bile acids in patients with colorectal cancer or polyps. Gut. 1992;33:1239–45. doi: 10.1136/gut.33.9.1239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Nordling MM, Glinghammar B, Karlsson PC, de Kok TM, Rafter JJ. Effects on cell proliferation, activator protein-1 and genotoxicity by fecal water from patients with colorectal adenomas. Scand J Gastroenterol. 2003;38:549–55. doi: 10.1080/00365520310002913. [DOI] [PubMed] [Google Scholar]
- 12.Tong JL, Ran ZH, Shen J, Fan GQ, Xiao SD. Association between fecal bile acids and colorectal cancer: a meta-analysis of observational studies. Yonsei Med J. 2008;49:792–803. doi: 10.3349/ymj.2008.49.5.792. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Perez MJ, Briz O. Bile-acid-induced cell injury and protection. World J Gastroenterol. 2009;15:1677–89. doi: 10.3748/wjg.15.1677. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Amaral JD, Castro RE, Sola S, Steer CJ, Rodrigues CM. p53 is a key molecular target of ursodeoxycholic acid in regulating apoptosis. J Biol Chem. 2007;282:34250–9. doi: 10.1074/jbc.M704075200. [DOI] [PubMed] [Google Scholar]
- 15.Amaral JD, Viana RJ, Ramalho RM, Steer CJ, Rodrigues CM. Bile acids: regulation of apoptosis by ursodeoxycholic acid. J Lipid Res. 2009 doi: 10.1194/jlr.R900011-JLR200. R900011-JLR200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Paumgartner G, Beuers U. Ursodeoxycholic acid in cholestatic liver disease: mechanisms of action and therapeutic use revisited. Hepatology. 2002;36:525–31. doi: 10.1053/jhep.2002.36088. [DOI] [PubMed] [Google Scholar]
- 17.Tung BY, Emond MJ, Haggitt RC, et al. Ursodiol use is associated with lower prevalence of colonic neoplasia in patients with ulcerative colitis and primary sclerosing cholangitis. Ann Intern Med. 2001;134:89–95. doi: 10.7326/0003-4819-134-2-200101160-00008. [DOI] [PubMed] [Google Scholar]
- 18.Pardi DS, Loftus EV, Jr., Kremers WK, Keach J, Lindor KD. Ursodeoxycholic acid as a chemopreventive agent in patients with ulcerative colitis and primary sclerosing cholangitis. Gastroenterology. 2003;124:889–93. doi: 10.1053/gast.2003.50156. [DOI] [PubMed] [Google Scholar]
- 19.Earnest DL, Holubec H, Wali RK, et al. Chemoprevention of azoxymethane-induced colonic carcinogenesis by supplemental dietary ursodeoxycholic acid. Cancer Res. 1994;54:5071–4. [PubMed] [Google Scholar]
- 20.Hess LM, Krutzsch MF, Guillen J, et al. Results of a Phase I Multiple-Dose Clinical Study of Ursodeoxycholic Acid. Cancer Epidemiol Biomarkers Prev. 2004;13:861–7. [PubMed] [Google Scholar]
- 21.Qiao D, Stratagouleas ED, Martinez JD. Activation and role of mitogen-activated protein kinases in deoxycholic acid-induced apoptosis. Carcinogenesis. 2001;22:35–41. doi: 10.1093/carcin/22.1.35. [DOI] [PubMed] [Google Scholar]
- 22.Im E, Martinez JD. Ursodeoxycholic acid (UDCA) can inhibit deoxycholic acid (DCA)-induced apoptosis via modulation of EGFR/Raf-1/ERK signaling in human colon cancer cells. J Nutr. 2004;134:483–6. doi: 10.1093/jn/134.2.483. [DOI] [PubMed] [Google Scholar]
- 23.Wali RK, Khare S, Tretiakova M, et al. Ursodeoxycholic acid and F(6)-D(3) inhibit aberrant crypt proliferation in the rat azoxymethane model of colon cancer: roles of cyclin D1 and E-cadherin. Cancer Epidemiol Biomarkers Prev. 2002;11:1653–62. [PubMed] [Google Scholar]
- 24.Ikegami T, Matsuzaki Y, Shoda J, Kano M, Hirabayashi N, Tanaka N. The chemopreventive role of ursodeoxycholic acid in azoxymethane-treated rats: suppressive effects on enhanced group II phospholipase A2 expression in colonic tissue. Cancer Lett. 1998;134:129–39. doi: 10.1016/s0304-3835(98)00248-1. [DOI] [PubMed] [Google Scholar]
- 25.Combes B, Carithers RL, Jr., Maddrey WC, et al. A randomized, double-blind, placebo-controlled trial of ursodeoxycholic acid in primary biliary cirrhosis. Hepatology. 1995;22:759–66. [PubMed] [Google Scholar]
- 26.Alberts DS, Martinez ME, Hess LM, et al. Phase III trial of ursodeoxycholic acid to prevent colorectal adenoma recurrence. J Natl Cancer Inst. 2005;97:846–53. doi: 10.1093/jnci/dji144. [DOI] [PubMed] [Google Scholar]
- 27.Jass JR. Molecular heterogeneity of colorectal cancer: Implications for cancer control. Surg Oncol. 2007;16:S7–9. doi: 10.1016/j.suronc.2007.10.039. [DOI] [PubMed] [Google Scholar]
- 28.Jacobs ET, Thompson PA, Martinez ME. Diet, gender, and colorectal neoplasia. J Clin Gastroenterol. 2007;41:731–46. doi: 10.1097/MCG.0b013e3180338e56. [DOI] [PubMed] [Google Scholar]
- 29.Martinez ME, Marshall JR, Graver E, et al. Reliability and validity of a self-administered food frequency questionnaire in a chemoprevention trial of adenoma recurrence. Cancer Epidemiol Biomarkers Prev. 1999;8:941–6. [PubMed] [Google Scholar]
- 30.Lambou-Gianoukos S, Heller SJ. Lithogenesis and Bile Metabolism. Surgical Clinics of North America. 2008;88:1175–94. doi: 10.1016/j.suc.2008.07.009. [DOI] [PubMed] [Google Scholar]
- 31.Pardi DS, Loftus EV, Kremers WK, Keach J, Lindor KD. Ursodeoxycholic acid as a chemopreventive agent in patients with ulcerative colitis and primary sclerosing cholangitis. Gastroenterology. 2003;124:889–93. doi: 10.1053/gast.2003.50156. [DOI] [PubMed] [Google Scholar]
- 32.J MW, L AR, B AL. The impact of ursodeoxycholic acid on cancer, dysplasia and mortality in ulcerative colitis patients with primary sclerosing cholangitis. Alimentary Pharmacology & Therapeutics. 2005;22:783–8. doi: 10.1111/j.1365-2036.2005.02650.x. [DOI] [PubMed] [Google Scholar]
- 33.Venneman NG, van Erpecum KJ. Gallstone disease: Primary and secondary prevention. Best Pract Res Clin Gastroenterol. 2006;20:1063–73. doi: 10.1016/j.bpg.2006.03.008. [DOI] [PubMed] [Google Scholar]
- 34.Shaffer EA. Gallstone disease: Epidemiology of gallbladder stone disease. Best Pract Res Clin Gastroenterol. 2006;20:981–96. doi: 10.1016/j.bpg.2006.05.004. [DOI] [PubMed] [Google Scholar]
- 35.Simon FR, Fortune J, Iwahashi M, Bowman S, Wolkoff A, Sutherland E. Characterization of the mechanisms involved in the gender differences in hepatic taurocholate uptake. Am J Physiol. 1999;276:G556–65. doi: 10.1152/ajpgi.1999.276.2.G556. [DOI] [PubMed] [Google Scholar]
- 36.Wang DQ, Lammert F, Paigen B, Carey MC. Phenotypic characterization of lith genes that determine susceptibility to cholesterol cholelithiasis in inbred mice. Pathophysiology Of biliary lipid secretion. J Lipid Res. 1999;40:2066–79. [PubMed] [Google Scholar]
- 37.Bennion LJ, Drobny E, Knowler WC, et al. Sex differences in the size of bile acid pools. Metabolism. 1978;27:961–9. doi: 10.1016/0026-0495(78)90140-3. [DOI] [PubMed] [Google Scholar]
- 38.Lampe JW, Fredstrom SB, Slavin JL, Potter JD. Sex differences in colonic function: a randomised trial. Gut. 1993;34:531–6. doi: 10.1136/gut.34.4.531. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Robins SJ, Fasulo J. Mechanism of lithogenic bile production: studies in the hamster fed an essential fatty acid-deficient diet. Gastroenterology. 1973;65:104–14. [PubMed] [Google Scholar]
- 40.Yago MD, González V, Serrano P, et al. Effect of the type of dietary fat on biliary lipid composition and bile lithogenicity in humans with cholesterol gallstone disease. Nutrition. 2005;21:339–47. doi: 10.1016/j.nut.2004.06.028. [DOI] [PubMed] [Google Scholar]
- 41.Cuevas A, Miquel JF, Reyes MS, Zanlungo S, Nervi F. Diet as a Risk Factor for Cholesterol Gallstone Disease. J Am Coll Nutr. 2004;23:187–96. doi: 10.1080/07315724.2004.10719360. [DOI] [PubMed] [Google Scholar]
- 42.Powell AA, Akare S, Qi W, et al. Resistance to ursodeoxycholic acid-induced growth arrest can also result in resistance to deoxycholic acid-induced apoptosis and increased tumorgenicity. BMC Cancer. 2006;6:219. doi: 10.1186/1471-2407-6-219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Botla R, Spivey JR, Aguilar H, Bronk SF, Gores GJ. Ursodeoxycholate (UDCA) inhibits the mitochondrial membrane permeability transition induced by glycochenodeoxycholate: a mechanism of UDCA cytoprotection. J Pharmacol Exp Ther. 1995;272:930–8. [PubMed] [Google Scholar]
- 44.Hohenester S, Oude-Elferink R, Beuers U. Primary biliary cirrhosis. Seminars in Immunopathology. doi: 10.1007/s00281-009-0164-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Younossi ZM. Review article: current management of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis. Aliment Pharmacol Ther. 2008;28:2–12. doi: 10.1111/j.1365-2036.2008.03710.x. [DOI] [PubMed] [Google Scholar]
- 46.Ramalho RM, Viana RJ, Low WC, Steer CJ, Rodrigues CM. Bile acids and apoptosis modulation: an emerging role in experimental Alzheimer’s disease. Trends Mol Med. 2008;14:54–62. doi: 10.1016/j.molmed.2007.12.001. [DOI] [PubMed] [Google Scholar]
- 47.Ozcan U, Yilmaz E, Ozcan L, et al. Chemical chaperones reduce ER stress and restore glucose homeostasis in a mouse model of type 2 diabetes. Science. 2006;313:1137–40. doi: 10.1126/science.1128294. [DOI] [PMC free article] [PubMed] [Google Scholar]