Author Affiliations: Department of General Surgery, Xinhua Hospital, Shanghai Jiaotong University, School of Medicine, Shanghai 200092, China (Zhou D, Wang JD, Yang Y and Quan ZW); Department of Biliary Tract Surgery II, Eastern Hepatobiliary Surgery Hospital, Second Military Medical University, Shanghai 200438, China (Yu WL and Zhang YJ)

Corresponding Author: Zhi-Wei Quan, MD, FACS, Department of General Surgery, Xinhua Hospital, Shanghai Jiaotong University, School of Medicine, Shanghai 200092, China (Tel: +86-21-25077905; Fax: +86-21-65795173; Email: zhiwquan@163.com)

 

© 2016, Hepatobiliary Pancreat Dis Int. All rights reserved.

doi: 10.1016/S1499-3872(16)60073-5

Published online February 24, 2016.

 

 

Acknowledgments: The authors thank Dr. Yao Zhu from Fudan University Shanghai Cancer Center and Prof. Chun-Bo Li from Shanghai Mental Health Center for providing statistical assistance in the development and validation of our nomogram.

Contributors: ZD, WJD, ZYJ and QZW conceived the study. All authors collected and analyzed the data. ZD, ZYJ and QZW wrote the first draft. ZD and WJD contributed equally to this article. All authors contributed to the design and interpretation of the study and to further drafts. QZW is the guarantor.

Funding: This study was supported by grants from the National Natural Science Foundation of China (81401932, 81272747 and 81372642).

Ethical approval: This study was approved by the Committee for Ethics of Xinhua Hospital Affiliated Shanghai Jiaotong University, School of Medicine and Eastern Hepatobiliary Surgery Hospital Affiliated Second Military Medical University.

Competing interest: No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.

 

 

BACKGROUND: Early diagnosis of gallbladder cancer (GBC) can remarkably improve the prognosis of patients. This study aimed to develop a nomogram for individualized diagnosis of stage I-II GBC in chronic cholecystitis patients with gallbladder wall thickening.

 

METHODS: The nomogram was developed using logistic regression analyses based on a retrospective cohort consisting of 89 consecutive patients with stage I-II GBC and 1240 patients with gallbladder wall thickening treated at one biliary surgery center in Shanghai between January 2009 and December 2011. The accuracy of the nomogram was validated by discrimination, calibration and a prospective cohort treated at another center between January 2012 and December 2014 (n=928).

 

RESULTS: Factors included in the nomogram were advanced age, hazardous alcohol consumption, long-standing diagnosed gallstones, atrophic gallbladder, gallbladder wall calcification, intraluminal polypoid lesion, higher wall thickness ratio and mucosal line disruption. The nomogram had concordance indices of 0.889 and 0.856 for the two cohorts, respectively. Internal and external calibration curves fitted well. The area under the receiver-operating characteristic curves of the nomogram was higher than that of multidetector row computed tomography in diagnosis of stage I-II GBC (P<0.001).

 

CONCLUSION: The proposed nomogram improves individualized diagnosis of stage I-II GBC in chronic cholecystitis patients with gallbladder wall thickening, especially for those the imaging features alone do not allow to confirm the diagnosis.

 

(Hepatobiliary Pancreat Dis Int 2016;15:180-188)

KEY WORDS: chronic cholecystitis; gallbladder wall thickening; gallbladder cancer; individualized diagnosis; nomogram

Gallbladder cancer (GBC) is an uncommon malignancy with an incidence of 3/100 000 each year. The 5-year survival rate of GBC is 3%-5%, which is the lowest among gastrointestinal cancers.^{[1, 2]} To date, curative (R0) resection has only been possible in less than 10% of patients with GBC.^[3] Although this disease is generally considered rare, its incidence rate is more than doubled in China in the past three decades.^[4]

 

The carcinogenesis of GBC follows a progressive sequence of chronic inflammation, gallbladder wall thickening (GBWT), metaplasia, dysplasia, carcinoma in situ and invasive cancer.^[5] This progression occurs insidiously, resulting in the late diagnosis of this disease. Therefore, successful individualized early diagnosis is essential to improve the prognosis. The identification of factors related to carcinogenesis may be helpful for the early diagnosis. Approximately 85% of GBC patients are rooted by gallstone disease which is widely considered the most significant risk factor for this malignancy.^[6] Age, gender, ethnicity, chronic infection, obesity and primary sclerosing cholangitis are also correlated with GBC.^[7] Gallbladder carcinogenesis might be influenced by many factors and their complex interactions, which cannot be accurately assessed purely through laboratory testing, complicating diagnosis of GBC in clinical practice. Furthermore, factors related to GBC may vary geographically. For example, anomalous junction of the pancreaticobiliary duct is found in 10%-38% of Japanese GBC patients but is very rare in other countries.^[8]

 

Currently, detection of GBC is highly dependent on computed tomography (CT). However, the accuracy of preoperative diagnosis of GBC on conventional horizontal CT only reaches 52.6%-71.7%.^{[9, 10]} Recently, multidetector row computed tomography (MDCT) combined with longitudinal multiplanar reformatted imaging (LMPRI) has been reported to be highly accurate in diagnosis of abdominal mass due to its ability of reconstructing multidirectional planes to target organs without losing their resolution.^[11] Thus, the use of LMPRI might improve the effectiveness in differentiating stage I (T1N0M0) and II (T2N0M0) GBC from benign GBWT.

 

Considering the above mentioned issue, integrated statistical assessment of patients’ detailed medical histories, potential risk factors, and high quality imaging findings may improve the diagnosis of GBC. The nomogram is a statistical model for predicting the disease occurrence and prognosis that has a strong ability to obviate the effect of spectrum bias when potential predictors are stratified.^[12] Nomogram can provide superior individualized risk assessments that facilitate disease diagnosis and treatment-related decisions.

 

In China, approximately 10% of the general population suffer from chronic cholecystitis.^[13] However, most patients with long-standing diagnosed stones, and even those accompanied with GBWT are unwilling to undergo cholecystectomy, thus leading to the risk of GBC. In this study, we attempted to develop a nomogram to improve the accuracy of diagnosis of stage I-II GBC.

 

 

During the first stage of our research, we developed a nomogram based on a retrospective case-control study. Between January 2009 and December 2011, 1550 consecutive patients with primary diagnosis of gallstones with GBWT (>3 mm) underwent surgical treatment at the Department of Biliary Tract Surgery II, Eastern Hepatobiliary Surgery Hospital affiliated Second Military Medical University. Patients who underwent radical cholecystectomy with a definitive pathological diagnosis of stage I-II GBC were recruited as participants in the case group and the remaining patients with simple GBWT served as controls. During the second stage of the study, the nomogram was validated using a prospective cohort including 928 chronic cholecystitis patients diagnosed primarily with gallstones and GBWT treated at Department of General Surgery, Xinhua Hospital, Shanghai Jiaotong University School of Medicine between January 2012 to December 2014.

 

Pathological staging of GBC was based on the AJCC Manual 7th Edition.^[14] All participants were observed once every three months during the first two years after surgery and every six months thereafter. The exclusion criteria were as follows: (1) stages III and IV GBC; (2) insufficient clinicopathological data; (3) liver or other types of biliary tumors metastasizing or invading to the gallbladder; and (4) diagnosis or history of other concurrent malignant tumors.

 

This study was conducted in accordance with the guidelines described in the statement on strengthening the reporting of observational studies in epidemiology (STROBE). Participants were interviewed by two well-trained interviewers independently through the administration of a structured questionnaire. In order to ensure the completeness and accuracy of the medical history collection, a detailed medical record review was also performed based on the Medical Institutions Information Inquiry System (the internet systems engineering developed by the government which realize the sharing of information on patients’ medical history, follow-up data, laboratory, imaging and pathological findings among different hospitals) to obtain data on the following candidate GBC-related factors: age, gender, and body mass index (BMI), heavy smoking, hazardous alcohol consumption, diabetes, menopause, the course of diagnosed gallstones, atrophic gallbladder, number of gallstones, gallstones diameters, imaging findings (malformation of the biliary system, gallbladder wall calcification, intraluminal polypoid lesion, higher wall thickness ratio and mucosal line disruption), levels of tumor markers in plasma (CA19-9, CA72-4, CA24-2 and CEA) and pathological diagnosis. The concordance between data obtained by the two interviewers through the administration of the structured questionnaire was 100%.

 

Pathological diagnosis was conducted and validated independently by two experienced pathologists. Stage I (T1N0M0) GBC was defined as tumors confined to the lamina propria or muscular layer of the gallbladder wall without lymph nodes or distant metastases. Stage II (T2N0M0) GBC was defined as tumors grown into perimuscular connective tissue without lymph node or distant metastases. BMI was defined as the ratio of human body weight (kg) to squared height (m²). Patients who smoked 20 or more cigarettes a day were considered as heavy smokers.^[15] Hazardous alcohol consumption was diagnosed according to the International Classification of Diseases (ICD-10) developed by the World Health Organization (WHO) and a modified Alcohol Use Disorders Identification Test (AUDIT) in China. The AUDIT score of 7 or more was considered as an indicator of hazardous and harmful alcohol use.^[16-18] The length of the course of diagnosed gallstones was calculated when gallstones were first detected by imaging examinations until the time of cholecystectomy. Physical factors in the nomogram were collected based on the results of preoperative imaging examinations, and history was determined by standardized preoperative interviews for all patients in both nomogram development cohort and validation cohort. The included patients in the two cohorts received ultrasound examination. Furthermore, Mirizzi’s syndrome, abnormal opening of the cyst duct, the left or right hepatic bile ducts and anomalous pancreaticobiliary ductal junction were detected by MRI preoperatively and finally confirmed during the surgery. Atrophy of the gallbladder was diagnosed on the basis of both ultrasonic and pathological diagnostic criteria.^[19] The number of gallstones (muddy, single or multiple) and stone diameters were confirmed by radiologists as well as pathologists. The diameter of the largest stone was recorded when a patient had multiple gallstones. MDCT, always combined with LMPRI, was used when gallbladder wall calcification, intraluminal polypoid lesion, gallbladder wall thickness ratio and mucosal line disruption were difficult to detect only using ultrasound. The thickest part of the gallbladder wall >3 mm was considered as wall thickening. Gallbladder wall thickness ratio was defined as the ratio of the thickest to the thinnest part of the gallbladder wall. The above protocol was designed and implemented in the period of the study including the nomogram development stage as well as the validation stage.

 

The primary endpoint of this study was confirmation of the stage I-II GBC-related factors for diagnosis and construction of the nomogram for this disease.

 

Binary variables were compared using the Chi-square test, Mantel-Haenszel test or Fisher’s exact test, and continuous variables were compared using Student’s t test. The Mann-Whitney U test was used for variables with an abnormal distribution. Univariate analysis and logistic regression with interactive analysis were performed to determine which indicators could be included in the nomogram model. Multivariate analyses were used to minimize confounding bias. Crude or adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. The logistic regression model with interaction analysis was developed using the backward model selection procedure. The results of the logistic regression model were used to construct the nomogram.

 

The established nomogram was internally and externally validated by discrimination and calibration, respectively. The discrimination of the model was measured by Harrell’s concordance index (C-index),^[20] which is positively correlated with prediction accuracy. Calibration charts were developed using the bootstrapping method to assess the capability of the calibration.^{[21, 22]}

 

SPSS 17.0.1 software for Windows (SPSS Inc., Chicago, IL, USA) was used for univariate and logistic regression analyses and receiver-operating characteristic curve (ROC) generation. The development and validation of the nomogram model were conducted using open-source statistical software R version 3.1.1 (R Foundation for Statistical Computing, Vienna, Austria) and additional packages (Frank Harrell, Vanderbilt University, Nashville, TN, USA). A P value <0.05 for a two-tailed test was considered statistically significant.

 

 

The clinicopathological features of the GBC patients are summarized in Table 1. Of the 1550 initially enrolled patients in the nomogram development cohort, 175 in the case group with stage III and IV GBC were excluded, in addition to 36 in the control group who had other concurrent malignant tumors. We also excluded 10 patients for insufficient clinicopathological data. Thus, 89 patients with stage I-II GBC and 1240 controls with chronic cholecystitis were finally included in the present study. Of these patients, 37 and 52 were pathologically diagnosed with stage I and II GBC, respectively. In the prospective training cohort, there were no patients lost to follow-up. Among 928 patients with primary diagnosis of GBWT, postoperative pathology showed 14 patients with stage I and 21 patients with stage II GBC, respectively.

 

Pathological examination revealed that all of the GBC patients also had gallstones. There were no differences between the case group and control group when come to the proportion of patients who presented with upper right abdominal pain (87.64% vs 91.77%, P=0.170 for nomogram development cohort and 85.71% vs 91.27%, P=0.232 for training cohort, respectively) or abdominal distension (15.73% vs 9.84%, P=0.100 for nomogram development cohort and 17.14% vs 11.76%, P=0.296 for training cohort, respectively). There was no asymptomatic case in our study. More than 70% of the patients were first diagnosed with gallstones just after appearance of typical abdominal pain or an initial attack of acute cholecystitis and the remaining patients were diagnosed at first with gallstones within three or six months after onset of their initial symptoms.

 

In either retrospective or prospective period of the study, no notable differences were seen in the proportion of patients between the two groups who had MDCT combined with LMPRI [34.83% (31/89) vs 36.69% (455/1240), P=0.725] for nomogram development cohort and [31.43% (11/35) vs 28.56% (255/893), P=0.712] for training cohort.

 

In addition, the baseline characteristics of either GBC patients or GBWT controls between the two cohorts were also comparable (data not shown).

 

The results of univariate and logistic analysis are listed in Tables 1 and 2, respectively. The following 8 variables were finally identified as significant clinical indicators for individualized diagnosis of GBC: advanced age (P=0.000), hazardous alcohol consumption (P=0.039), course of diagnosed gallstones (P=0.006), atrophic gallbladder (P=0.000), imaging findings characteristic of gallbladder wall calcification (P=0.000), intraluminal polypoid lesion (P=0.000), higher wall thickness ratio (P=0.000), and mucosal line disruption (P=0.000). There were no interactive effects detected among the above indicators (P>0.05).

 

Based on the logistic regression model, a nomogram with point scales of the above 8 clinical indicators was constructed (Fig. 1A). The patient risk score for each indicator in the nomogram was summed and plotted on the “total points” axis. An individual’s estimated risk probability of developing GBC could be measured by drawing a vertical line from the plotted “total points” axis down to the “predicted risk” axis. For example, a 75-year-old (83 points) patient with a 15-year long history of diagnosed gallstones (19 points), whose imaging findings suggest an intraluminal polypoid lesion (38 points) with the wall thickness ratio of 6 (33 points), has a total score of 173 points that converts to a probability of developing stage I-II GBC of more than 80% (Fig. 1B).

 

Internal validation using the bootstrap method with 200 repetitions yielded a high bias-corrected C-index of 0.889. The nomogram was also externally validated with an independent prospective cohort based on 928 patients treated at another center from January 2012 to December 2014, and a satisfactory bias-corrected C-index of 0.856 was obtained, indicating good discrimination. In the calibration graphs of internal and external validation (Fig. 2), the bias-corrected probability curves were close to the ideal curves, which also demonstrated that the scoring system of our nomogram effectively discriminated the probability of stage I-II GBC occurrence.

 

ROC analyses were performed to compare the discriminatory power of the nomogram for diagnosis of stage I (Fig. 3A) and II (Fig. 3B) GBC with that of imaging examinations, respectively, and significantly higher area under the curve (AUC) values were obtained for the nomogram than for the MDCT imaging methods (P<0.001).

 

 

Preoperative identification of stage I-II GBC in gallstones patients with GBWT is still challenging. The present study attempted to achieve individualized diagnosis of incidental stage I-II GBC by generating a nomogram and comprehensively assessing patients’ related risk factors and imaging findings.

 

The incidence of GBC tends to increase with age. In 2008, the Memorial Sloan-Kettering Cancer Centre reported a median age of 67 years in a review of 435 patients with GBC.^[23] In our study, the median age of the GBC patients was 63 years, which was significantly older than that in the controls (55 years, P=0.000).

 

Among the various risk factors for GBC, gender is predominant because women have a greater risk of GBC than men.^[24] Our results showed female-to-male ratios of 2.56 and 1.90 for the GBC patients and the control group, respectively. Although the two ratios were generally comparable (P=0.247), the proportion of female GBC patients rose more sharply with increasing age, particularly after menopause (P=0.019). Multivariate logistic regression analysis might eliminate “menopause” from the statistical model because it had a co-linear relationship with the parameter “age”. The true role of estrogen in the development of GBC remains unclear, but biliary sludge, which promotes cholesterol crystal formation, develops in up to 30% of women during pregnancy.^{[25, 26]}

 

In China, long-standing gallstones are common. However, the exact natural course of gallstones is difficult to determine in clinical practice. Therefore, it is important to ascertain when the first diagnosis was made and when first symptoms occurred. Using the same sample, we established the initial nomogram separately in our study according to the above two factors for estimating the course of gallstones and no significant differences could be detected between GBC patients and controls. This might be due to the patients who received their first gallstones diagnosis just after an initial attack of acute cholecystitis or within 3-6 months after the onset of their first symptoms but most patients did not accept cholecystectomy until several years after their first diagnosis. Thus, a relatively short-time interval between the onset of symptoms and diagnosis had a limited impact on the scoring system of the nomogram. It has been proven that repeated physical irritation to the gallbladder mucosa caused by long-term gallstones leads to a state of continuous inflammatory trauma. Moreover, long-standing gallstones can aggravate the above mentioned pathological changes by producing carcinogens, such as secondary bile acids.^[27] As expected, our data revealed that patients with a history of diagnosed gallstones of >10 years and >20 years had overall 2.344-fold (P=0.001) and 5.250-fold (P=0.001) higher risks of GBC, respectively, when compared with those with a history of <10 years (Table 1). Notably, the follow-up results showed that 8.57% (15/175) of the patients with stage III-IV cancer diagnosed at initial visit were accompanied with asymptomatic stones. This unfortunate reality suggests that the proposed nomogram is not an ideal tool for predicting GBC in population without typical symptoms or evidence of gallstones and that regular health check-up for early screening and prompt treatment of gallstones might be the most effective way to prevent GBC.

 

Evidences from Western countries support an association between large gallstones (e.g., >3 cm) and a high risk of GBC, but whether large gallstones can be considered as an indication for cholecystectomy requires further confirmation.^[28] In the present study, neither the size (P=0.698) nor the number of gallstones (P=0.717) was a significant risk predictor of GBC.

 

In the present study, we utilized MDCT combined with LMPRI for approximately 30% of the patients who were difficult to be differentiated from stage I-II GBC ones. LMPRI was reported to have higher diagnostic accuracies for various digestive tract cancers because it can depict the gradual transition from normal to abnormal tissue as estimated on the longest plane.^[29]

 

Notably, there were approximately 40%-50% of the GBC or GBWT patients in whom differential diagnosis for symptomatic gastric ulcers was needed in our study. Gastroscopy was performed in patients who had continuous symptoms of middle upper abdomen pain, nausea, vomiting, short-term weight loss, malnutrition and any other existing symptoms which were not completely compatible with the diagnosis of cholecystitis. The number of patients who had gastroscopy before surgery was comparable between the GBC and control groups in both two cohorts [44.94% (40/89) vs 42.10% (522/1240), P=0.599 for nomogram development cohort and 51.43% (18/35) vs 49.05% (438/893), P=0.782 for training cohort, respectively]. Among these patients, those who were accompanied with symptomatic gastric ulcer had an increased risk of developing GBC [17.50% (7/40) vs 4.02% (21/522), P=0.002 for nomogram development cohort and 22.22% (4/18) vs 5.94% (26/438), P=0.024 for training cohort, respectively]. These data were somewhat similar to the findings of a population-based study conducted in 2011; however, in that study,^[30] hazardous alcohol consumption and gastric ulcer had synergistic effects on the development of GBC, whereas our interactive analysis indicated that alcohol was an independent risk factor for this disease (P=0.039). Alcohol promotes the movement of bile salts into the stomach which stimulates the production of gastric acid and colonization of H. pylori in the stomach and these are the three main causes of ulceration.^{[31, 32]} H. pylori has also been linked to gallstones and GBC in many countries such as South America and east, south and west Asia.^[33] The mechanism underlying this phenomenon might be related to the induction of duodenal stimulation, pyloric dysfunction, gastrointestinal hormone secretion disorder and increased abdominal pressure.^{[34, 35]} Although gastric ulcer could not be included as a related factor or a variable in our nomogram because gastroscopy was not performed in all the patients, the above initial evidences indicate that further research is required to clarify whether pathological conditions of the gastroduodenal tract and biliary system should be considered as indivisible entities when evaluating the risk of GBC.

 

Our study has several limitations. First, our data for constructing the nomogram were collected by a retrospective study. The statistical power of data obtained from the medical records is still inevitably limited by the design of this study. Second, most of the included patients and controls were from Shanghai, China; therefore, our findings may not be popularized to other geographic locations. Third, the present nomogram is not applicable to GBC patients with no typical symptoms or evidence of gallstones. Finally, our research cannot clarify the relationship between the exposure duration of some risk factors (e.g. alcohol consumption) and the start of the malignant transformation process of gallbladder mucosa.

 

In conclusion, the proposed nomogram is helpful to assess gallstones patients with GBWT for individualized diagnosis of stage I-II GBC. MDCT combined with LMPRI could help to differentiate stage I-II GBC from benign GBWT. 

 

 

1 Lazcano-Ponce EC, Miquel JF, Muñoz N, Herrero R, Ferrecio C, Wistuba II, et al. Epidemiology and molecular pathology of gallbladder cancer. CA Cancer J Clin 2001;51:349-364. PMID: 11760569

2 Kapoor VK, McMichael AJ. Gallbladder cancer: an ‘Indian’ disease. Natl Med J India 2003;16:209-213. PMID: 14606770

3 Sheth S, Bedford A, Chopra S. Primary gallbladder cancer: recognition of risk factors and the role of prophylactic cholecystectomy. Am J Gastroenterol 2000;95:1402-1410. PMID: 10894571

4 Hsing AW, Gao YT, Devesa SS, Jin F, Fraumeni JF Jr. Rising incidence of biliary tract cancers in Shanghai, China. Int J Cancer 1998;75:368-370. PMID: 9455795

5 Misra S, Chaturvedi A, Misra NC, Sharma ID. Carcinoma of the gallbladder. Lancet Oncol 2003;4:167-176. PMID: 12623362

6 Singh V, Trikha B, Nain C, Singh K, Bose S. Epidemiology of gallstone disease in Chandigarh: a community-based study. J Gastroenterol Hepatol 2001;16:560-563. PMID: 11350554

7 Zatonski WA, Lowenfels AB, Boyle P, Maisonneuve P, Bueno de Mesquita HB, Ghadirian P, et al. Epidemiologic aspects of gallbladder cancer: a case-control study of the SEARCH Program of the International Agency for Research on Cancer. J Natl Cancer Inst 1997;89:1132-1138. PMID: 9262251

8 Misra SP, Gulati P, Thorat VK, Vij JC, Anand BS. Pancreaticobiliary ductal union in biliary diseases. An endoscopic retrograde cholangiopancreatographic study. Gastroenterology 1989;96:907-912. PMID: 2914651

9 Kalra N, Suri S, Gupta R, Natarajan SK, Khandelwal N, Wig JD, et al. MDCT in the staging of gallbladder carcinoma. AJR Am J Roentgenol 2006;186:758-762. PMID: 16498103

10 Kim SJ, Lee JM, Lee JY, Choi JY, Kim SH, Han JK, et al. Accuracy of preoperative T-staging of gallbladder carcinoma using MDCT. AJR Am J Roentgenol 2008;190:74-80. PMID: 18094296

11 Iwama A, Yamazaki S, Mitsuka Y, Yoshida N, Moriguchi M, Higaki T, et al. A longitudinal computed tomography imaging in the diagnosis of gallbladder cancer. Gastroenterol Res Pract 2015;2015:254156. PMID: 26064088

12 Gold JS, Gönen M, Gutiérrez A, Broto JM, García-del-Muro X, Smyrk TC, et al. Development and validation of a prognostic nomogram for recurrence-free survival after complete surgical resection of localised primary gastrointestinal stromal tumour: a retrospective analysis. Lancet Oncol 2009;10:1045-1052. PMID: 19793678

13 Zhu L, Aili A, Zhang C, Saiding A, Abudureyimu K. Prevalence of and risk factors for gallstones in Uighur and Han Chinese. World J Gastroenterol 2014;20:14942-14949. PMID: 25356055

14 Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A, eds. AJCC Cancer Staging Manual, 7th edition. New York, NY: Springer;2010:237-246.

15 Qian J, Cai M, Gao J, Tang S, Xu L, Critchley JA. Trends in smoking and quitting in China from 1993 to 2003: National Health Service Survey data. Bull World Health Organ 2010;88:769-776. PMID: 20931062

16 Saunders JB, Aasland OG, Babor TF, de la Fuente JR, Grant M. Development of the Alcohol Use Disorders Identification Test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol consumption--II. Addiction 1993;88:791-804. PMID: 8329970

17 Li B, Sheng YC, Zhang BQ, Zheng XH, Wang XG. The test of the Alcohol Use Disorders Identification Test (AUDIT) in China. Chinese Mental Health Journal 2003;17:1-3.

18 Li Q, Babor TF, Hao W, Chen X. The Chinese translations of Alcohol Use Disorders Identification Test (AUDIT) in China: a systematic review. Alcohol Alcohol 2011;46:416-423. PMID: 21467046

19 Kanoh K, Shimura T, Tsutsumi S, Suzuki H, Kashiwabara K, Nakajima T, et al. Significance of contracted cholecystitis lesions as high risk for gallbladder carcinogenesis. Cancer Lett 2001;169:7-14. PMID: 11410319

20 Harrell FE Jr, Califf RM, Pryor DB, Lee KL, Rosati RA. Evaluating the yield of medical tests. JAMA 1982;247:2543-2546. PMID: 7069920

21 Harrell FE Jr, Lee KL, Mark DB. Multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat Med 1996;15:361-387. PMID: 8668867

22 Sauerbrei W, Schumacher M. A bootstrap resampling procedure for model building: application to the Cox regression model. Stat Med 1992;11:2093-2109. PMID: 1293671

23 Duffy A, Capanu M, Abou-Alfa GK, Huitzil D, Jarnagin W, Fong Y, et al. Gallbladder cancer (GBC): 10-year experience at Memorial Sloan-Kettering Cancer Centre (MSKCC). J Surg Oncol 2008;98:485-489. PMID: 18802958

24 Gupta P, Agarwal A, Gupta V, Singh PK, Pantola C, Amit S. Expression and clinicopathological significance of estrogen and progesterone receptors in gallbladder cancer. Gastrointest Cancer Res 2012;5:41-47. PMID: 22690257

25 Randi G, Franceschi S, La Vecchia C. Gallbladder cancer worldwide: geographical distribution and risk factors. Int J Cancer 2006;118:1591-1602. PMID: 16397865

26 Maringhini A, Ciambra M, Baccelliere P, Raimondo M, Orlando A, Tinè F, et al. Biliary sludge and gallstones in pregnancy: incidence, risk factors, and natural history. Ann Intern Med 1993;119:116-1120. PMID: 8512160

27 Shaffer EA. Gallbladder cancer: the basics. Gastroenterol Hepatol (N Y) 2008;4:737-741. PMID: 21960896

28 Hundal R, Shaffer EA. Gallbladder cancer: epidemiology and outcome. Clin Epidemiol 2014;6:99-109. PMID: 24634588

29 Kumano S, Okada M, Shimono T, Kuwabara M, Yagyu Y, Imaoka I, et al. T-staging of gastric cancer of air-filling multidetector-row CT: comparison with hydro-multidetector-row CT. Eur J Radiol 2012;81:2953-2960. PMID: 22304982

30 Andreotti G, Liu E, Gao YT, Safaeian M, Rashid A, Shen MC, et al. Medical history and the risk of biliary tract cancers in Shanghai, China: implications for a role of inflammation. Cancer Causes Control 2011;22:1289-1296. PMID: 21744094

31 Franke A, Teyssen S, Singer MV. Alcohol-related diseases of the esophagus and stomach. Dig Dis 2005;23:204-213. PMID: 16508284

32 Zhang L, Eslick GD, Xia HH, Wu C, Phung N, Talley NJ. Relationship between alcohol consumption and active Helicobacter pylori infection. Alcohol Alcohol 2010;45:89-94. PMID: 19808941

33 Zhou D, Wang JD, Weng MZ, Zhang Y, Wang XF, Gong W, et al. Infections of Helicobacter spp. in the biliary system are associated with biliary tract cancer: a meta-analysis. Eur J Gastroenterol Hepatol 2013;25:447-454. PMID: 23470268

34 Solaini L, Sharma A, Watt J, Iosifidou S, Chin Aleong JA, Kocher HM. Predictive factors for incidental gallbladder dysplasia and carcinoma. J Surg Res 2014;189:17-21. PMID: 24589178

35 Madura JA. Primary bile reflux gastritis: diagnosis and surgical treatment. Am J Surg 2003;186:269-273. PMID: 12946831