The stage of liver fibrosis at the time of primary

Transcription

1 LIVER DISEASE Smoking and Increased Severity of Hepatic Fibrosis in Primary Biliary Cirrhosis: A Cross Validated Retrospective Assessment Claudia O. Zein, 1,2 Kristi Beatty, 1 Anthony B. Post, 1 Laurie Logan, 1 Sara Debanne, 3 and Arthur J. McCullough 4 An epidemiological association between cigarette smoking and primary biliary cirrhosis (PBC) has been demonstrated. Our aim was to determine the relationship between smoking and severity of liver fibrosis at presentation in patients with PBC. All patients with PBC seen at the three major teaching hospitals of Case Western Reserve University between October 1998 and December 2005 were identified. Data obtained at the time of the first evaluation leading to the PBC diagnosis on 97 patients were collected. The cumulative number of cigarette packs smoked per year (pack-years) was calculated. Advanced histological disease was defined as Ludwig stages 3 or 4. Analyses were performed to determine associations between advanced histological disease, smoking and other variables related to liver fibrosis. Smoking history was more common (P.0008) in patients with advanced histological disease at presentation compared to those with early disease. Among smokers, mean lifetime tobacco consumption was higher (P.04) in cases with advanced histological disease at presentation (30 pack-years) compared to cases with early disease (17 pack-years). Logistic regression demonstrated a significant association between a lifetime tobacco consumption of >10 pack-years and advanced histological disease at presentation (OR 13.3). The association remained significant after adjusting for age, gender, and alcohol intake. The validity of these results was corroborated by cross-validation in an independent confirmatory set of 172 patients with PBC. In conclusion, smoking may accelerate the progression of PBC. This could be induced by exposure to chemicals in cigarette smoke. (HEPATOLOGY 2006;44: ) See Editorial on Page 1394 Abbreviations: PBC, primary biliary cirrhosis; BMI, body mass index. From the 1 Division of Gastroenterology and Hepatology, University Hospitals of Cleveland; 2 Division of Gastroenterology and Hepatology, Louis Stokes Cleveland Veterans Affairs Medical Center; 3 Department of Biostatistics and Epidemiology, Case Western Reserve University; and 4 Division of Gastroenterology and Hepatology, Cleveland Clinic, Cleveland, OH. Received June 5, 2006; accepted September 1, This work was presented, in part, at the Digestive Disease Week 2006 (DDW 2006), May 20-25, Los Angeles, California. Address reprint requests to: Claudia O. Zein, M.D., M.Sc., Division of Gastroenterology and Hepatology, Louis Stokes Veterans Affairs Medical Center, East Boulevard 111 E (W), Cleveland, OH Claudia.Zein@va.gov; fax: Copyright 2006 by the American Association for the Study of Liver Diseases. Published online in Wiley InterScience ( DOI /hep Potential conflict of interest: Nothing to report. The stage of liver fibrosis at the time of primary biliary cirrhosis (PBC) diagnosis varies between patients. Advanced age, male gender, and history of significant alcohol intake have been associated with more severe progression of fibrosis in chronic liver diseases in general, including PBC. 1-4 In patients with PBC, therapy with ursodeoxycholic acid has been associated with slower progression of fibrosis. 5 However, other host or environmental factors specifically associated with the progression of fibrosis in patients with PBC have not been defined. Although the exact etiology of PBC is still unknown, it is believed that environmental factors may induce this disease in genetically susceptible individuals An epidemiological association between cigarette smoking and PBC has been shown in population based case control studies. 6-9 One study that described this epidemiological association noted that smoking history was, in general, more frequent among PBC patients compared to controls. In that study, the odds ratio for PBC was even higher when the smoking history was longer than 20 years. 6 A subsequent survey study in the United States found elevated odds ratios regarding smoking history for PBC cases compared to siblings and to friends without PBC. 7 Recently, two questionnaire-based case-control studies aimed at defining the epidemiological risk factors associated with PBC, confirmed that a history of smoking was significantly associated with an increased risk of 1564

2 HEPATOLOGY, Vol. 44, No. 6, 2006 ZEIN ET AL PBC. 8,9 However, the association between smoking and PBC has not been further characterized. The aims of our study were to: (1) assess the relationship between smoking history and the severity of histological liver disease at presentation in a cohort of patients with PBC, and (2) cross-validate the results in an independent patient cohort. Patients and Methods Patient Population. A total of 269 patients with PBC seen at the four major academic medical institutions in the city of Cleveland were included in the study. Ninety-seven patients with the diagnosis of PBC who were seen at the three major teaching hospitals of Case Western Reserve University in Cleveland, OH (University Hospitals of Cleveland, the Louis Stokes Cleveland Veterans Affairs Medical Center, and MetroHealth Medical Center) between January 1, 1998 and October 31, 2005 were identified as the original cohort. Data on an independent cohort of 172 patients seen at the Cleveland Clinic between January 1, 1998 and March 30, 2006 were used to validate the results. Data Collection. Electronic and paper medical records of all patients who had received ICD-9 code were obtained and reviewed. Cases with incomplete, unavailable, or unclear data were excluded. Cases with a confirmed diagnosis of PBC were identified and clinical data was extracted with special attention to smoking history. The following information corresponding to the time of the evaluation which prompted the PBC diagnosis was collected for each patient: gender, race, age at diagnosis, age at presumed onset of symptoms, biochemical test results at presentation including aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, bilirubin, albumin, prothrombin time, platelet count, antimitochondrial antibody, creatinine level, clinical parameters, and liver biopsy results. Smoking history and lifetime quantitative smoking consumption were obtained by meticulous chart review specifically seeking out self-reported smoking history data, and/or information obtained at the time of completion by the patient of past history questionnaires, as well as information obtained by direct interview of the patient at the time of liver clinic evaluation, pretransplant evaluation, cardiopulmonary evaluation, or other evaluations where a detailed smoking history is compulsory. Smoking data abstracted included history of current, former or never smoking, pack-years, number of years smoked, and date of quitting smoking for former smokers. History of alcohol use was also obtained by chart review including selfreported information in past history questionnaires, as well as information obtained by the providers at the time of liver clinic, pre-transplant, or other evaluations where alcohol use history was relevant. Alcohol use data abstracted included history of current, former, or never using alcohol, average number of drinks per day, type of alcoholic beverage, and date of quitting if applicable. The average amount of alcohol consumed per week was calculated for each patient based on the obtained information. Data entry was completed into a spread sheet database program. The study was approved by the institutional review boards of all the institutions involved. Definition of PBC. The diagnosis of PBC was made in the absence of other known etiology of liver disease and based on the following criteria: (1) detectable antimitochondrial antibody; (2) cholestatic biochemical profile for at least 6 months; and (3) compatible liver histopathology. Definition of Smoking History. A positive smoking history was defined as current or past smoking of 100 cigarettes (i.e., 5 packs) 11 at anytime during the patient s lifetime up to the time of PBC diagnosis. The amount of cigarettes smoked up to the time of PBC diagnosis was quantified in pack-years (number of packs smoked per day multiplied by the number of years smoked). The pack-years were entered in the database as a continuous variable for each patient. A subsequent categorization was made of patients between those who had a history of smoking 10 pack-years and those who did not. For the purposes of this study, a significant history of smoking was defined as a history of consumption of 10 pack-years. Definition of Significant Alcohol Intake. History of alcohol use was considered significant if a history of alcohol intake of 140g/week (females) or 210 g/week (males) was present during 6 months at any time during the patient s lifetime prior to the PBC diagnosis. Definition of Advanced Histological Disease. Liver fibrosis was staged based on the criteria of Ludwig et al. 12 Advanced stage of fibrosis was defined as Ludwig stages 3 or 4. Liver biopsy adequacy for assessment of fibrosis stage was judged by a length of at least 12 mm based on the histopathology report specimen description. In addition, the presence of an adequate number of portal tracts for evaluation, as judged by the interpreting pathologist, was also considered. In general, the presence of at least 8 complete portal tracts was considered sufficient for assessment of fibrosis stage in the standard practice of pathologists at the institutions involved. Since inadequate samples can underestimate stage of disease, in cases where the liver biopsy specimen measured less than 12 mm or where the specimen was considered inadequate for accurate assessment of fibrosis by the interpreting pathologist, the stage of fibrosis was considered not available.

3 1566 ZEIN ET AL. HEPATOLOGY, December 2006 Statistical Analysis: Original Cohort. The constructed spreadsheet database was transferred to a statistical analysis software program for analysis. Descriptive statistical analysis of demographic, laboratory, and other pertinent variables was performed. Univariate and multivariate analyses were performed to determine associations between variables including smoking history and advanced histological disease. Student t test, or non-parametric tests where applicable, were used to compare continuous variables and chi-squared test was used to compare categorical Logistic regression analysis was done to investigate the role of a history of smoking and a significant history of smoking (history of consumption of 10 pack years) in predicting advanced stage of liver fibrosis at presentation while controlling for factors potentially associated with advanced liver fibrosis including age, gender, and history of significant alcohol use. The results are presented in tables with categorical variables presented as number and percentage and continuous variables presented as mean and standard error of the mean. Cross-Validation Data. With the purpose of crossvalidating our results in an independent set of patients, we identified all patients who were listed under ICD-9 code at the Cleveland Clinic between January 1, 1998 and March 30, Paper and electronic medical records were reviewed and a diagnosis of PBC was confirmed in 172 cases. Information was collected on each patient regarding all the variables of interest. For the cross-validation cohort, information regarding Medicaid coverage (versus private insurance) as a proxy for disadvantaged socioeconomic status was abstracted. In addition, for the cross-validation cohort, information regarding weight and height at the time of PBC diagnosis was collected in order to calculate the body mass index (BMI). Also, information was collected regarding the coexistence of a diagnosis of Diabetes Mellitus Type 2. Therapy with anti-diabetic agents (oral hypoglycemiants or insulin) was considered valid evidence of an established diagnosis of diabetes. Also, based on current American Diabetes Association criteria, patients with fasting plasma glucose 126 mg/dl on at least two occasions were considered as fulfilling diagnostic criteria for diabetes. Statistical Analysis: Cross-Validation Cohort. Descriptive statistical analysis of demographic, laboratory, and other pertinent variables was performed. The characteristics of the cross-validation cohort were compared to those of the original cohort. Student t test or non-parametric tests, and chi-squared test were used where indicated to compare variables and assess potential associations between advanced liver fibrosis at presentation and variables of interest including smoking history, history of consumption of 10 pack-years, Table 1. Characteristics at Presentation of 97 (Original Cohort) and 172 (Cross-Validation Cohort) Patients With PBC* age, gender, history of significant alcohol use, Medicaid coverage, BMI, and diabetes mellitus type 2. The role of smoking history in predicting advanced liver fibrosis was assessed in the cross-validation group by logistic regression analysis while controlling for other variables of interest. Results Original Cohort (n 97) Cross Validation Cohort (n 172) P Female gender 85/97 (88%) 157/172 (91%) NS Caucasian race 89/97 (92%) 157/164 (96%) NS Age at PBC diagnosis (years) 53 ( 1.2) 52 ( 0.9) NS History of smoking 47/92 (51%) 81/161 (50.3%) NS History of 10 pack-years 40/90 (44%) 67/156 (43%) NS Average pack-years (smokers) 30 ( 4.9) 28.2 ( 2.9) NS Significant alcohol use 9/96 (9.4%) 12/171 (7%) NS BMI 26.7 ( 0.5) BMI 30 31/136 (22.8%) Diabetes mellitus type 2 14/168 (8.3%) AST (IU/dL) 77 ( 8) 83 ( 5) NS Alkaline phosphatase (IU/dL) 374 ( 33.8) 439 ( 28) NS Bilirubin (mg/dl) 1.2 ( 0.2) 1.5 ( 0.2) NS Albumin (g/dl) 3.9 ( 0.1) 3.9 ( 0.04) NS Platelet count ( 103) 260 ( 9.4) 231 ( 7.6) 0.02 Protime (seconds) 11 ( 0.2) 11 ( 0.1) NS Mayo Risk Score 4.5 ( 0.2) 4.7 ( 0.2) NS Advanced fibrosis (stages 3-4) 40/81 (49%) 85/165 (51.5%) NS Demographic and Other Characteristics. Of 97 patients included in the original cohort, 85 (88%) were female. Mean age at diagnosis was years. Advanced histological disease at presentation (Ludwig stages 3-4) was documented in 49% (40/81) of patients. Stage of liver fibrosis was not available in 16 patients. Demographic, clinical and laboratory characteristics were similar between patients in whom liver fibrosis stage was available compared to those for whom fibrosis stage was not available, with the exception of age (mean years for those with available stage of fibrosis versus years for those without available stage of fibrosis; P.02). Forty-five patients (49%) were never smokers, whereas 47 (51%) had a history of smoking. Smoking history was not available in 5 cases. The demographic, biochemical, and histologic data of these 97 patients are summarized in Table 1. Predictors of Advanced Liver Fibrosis. Table 2 shows differences regarding variables of interest in the 97 patients according to early versus advanced stage of fibro-

4 HEPATOLOGY, Vol. 44, No. 6, 2006 ZEIN ET AL Table 2. Characteristics at Presentation of 97 Patients With PBC, According to Early (Stages 0-2) Versus Advanced (Stages 3-4) Stage of Fibrosis* Early Fibrosis (Stages 0-2) Advanced Fibrosis (Stages 3-4) Female gender 35/40 (88%) 36/41 (88%).97 Caucasian race 39/40 (98%) 37/41 (90%).16 Age at PBC diagnosis (years) 50 ( 1.8) 53 ( 1.8).3 History of smoking 13/39 (33%) 27/38 (71%).0008 History of 10 pack-years 7/37 (19%) 27/38 (71%).0001 Average pack-years (smokers) 17 ( 5.3) 30 ( 3.6).04 Significant alcohol use 4/39 (10%) 3/41 (7.3%).64 AST (IU/dL) 62 ( 12.8) 98.2 ( 12.8).05 Alkaline phosphatase (IU/dL) 351 ( 52) 399 ( 52).5 Bilirubin (mg/dl) 0.9 ( 0.2) 1.3 ( 0.2).14 Albumin (g/dl) 4.1 ( 0.1) 3.8 ( 0.1).003 Platelet count ( 103) 282 ( 15) 231 ( 14.5).02 Protime (seconds) 10.9 ( 0.23) 10.9 ( 0.2).8 Mayo Risk Score 3.9 ( 0.3) 4.8 ( 0.3).03 sis. Anticipated differences between these two subgroups of patients including lower albumin, lower platelet count, and higher Mayo Risk Score among patients with advanced fibrosis compared to those with early stages of fibrosis were noted. Smoking history was more common (P.0008) in patients with advanced stage of fibrosis at presentation (27/38, 71%) compared to those with early stage of fibrosis at presentation (13/39, 33%). Similarly, the frequency of a history of smoking 10 pack-years was significantly higher (P.0001) among patients with advanced stage of fibrosis at presentation (27/38, 71%) versus those with early stage of fibrosis at presentation (7/37, 19%). As shown in Fig. 1, among smokers, mean lifetime tobacco consumption was higher (P.04) in cases with advanced histological disease at presentation (average of pack-years) compared to cases with early disease (average of ). Smoking as a Risk Factor for Advanced Fibrosis. Similar findings to those described above were noted P when patients were classified based on smoking history. The frequency of advanced liver fibrosis at presentation was significantly higher (P.0008) in those with a history of smoking (27/40, 68%) compared to those who never smoked (11/37, 30%). Figure 2 shows the significantly larger (P.0001) proportion of patients with advanced liver fibrosis at presentation among those who had a history of smoking 10 pack-years in comparison to those who did not. The only significant difference between the characteristics of patients with smoking history and those who never smoked regarded the higher frequency of advanced liver fibrosis at presentation in the first group. Table 3 shows a comparison between the characteristics of patients with a history of smoking 10 pack-years compared to those who did not. The presence of advanced liver fibrosis (P.0001) at presentation and a history of significant alcohol use (P.04) were the only two variables significantly associated with a history of smoking 10 pack-years. Multivariate analysis demonstrated a predictive role of lifetime tobacco consumption in pack-years for the presence of advanced liver fibrosis at presentation. In Table 4, the logistic regression model demonstrating the role of a smoking history of 10 pack-years in predicting advanced liver fibrosis at presentation in patients with PBC (OR 13.3) is shown. This association remained significant after adjusting for age, gender, and history of significant alcohol intake. Cross-Validation of Results. Among the independent group of 172 patients with PBC, 157 (91%) were female and the mean age was years. Advanced liver fibrosis at presentation was present in 85 of 165 patients (51.5%). Stage of liver fibrosis was not available in 7 patients. Demographic, smoking history, alcohol intake history, biochemical, and histological characteristics of this independent group of patients are summarized in Table 1. Smoking history was not available in 9 patients. Fig. 1. The average pack-years among smokers according to advanced versus early liver fibrosis at presentation was significantly different in the original cohort ( vs , P.04) (A) and in the cross-validation cohort ( vs , P.006) (B).

5 1568 ZEIN ET AL. HEPATOLOGY, December 2006 Fig. 2. Original cohort: Different frequencies (P.0001) of advanced fibrosis in patients with PBC according to history of exposure of 10 pack-years of tobacco or not. In 5 of the 81 patients with a history of smoking, quantitative lifetime pack-years information was not available. The original and cross-validation groups were similar regarding all variables of interest except for platelet count (Table 1). Table 5 shows the differences between patients with early versus advanced stage of fibrosis at presentation in the cross-validation cohort. Similar to the original cohort, anticipated differences were seen regarding liver tests in patients with advanced stage of fibrosis including lower albumin and platelet count, as well as higher bilirubin and Mayo Risk Score. A higher (P.04) frequency of Medicaid coverage (as a proxy for low socioeconomic status) among patients with advanced fibrosis at presentation was noted. A trend towards significance by univariate analysis was noted in the frequency of diabetes mellitus type 2 in patients with advanced fibrosis compared to those with early fibrosis. Significantly higher frequencies of smoking history (P.0001) and of smoking 10 pack-years (P.0001) were seen among patients with advanced versus early liver fibrosis in the cross-validation cohort. Figure 1B shows that, similar to what was seen in the original cohort, smokers in the cross-validation cohort had higher (P.006) average lifetime tobacco consumption in cases Table 3. Characteristics of 97 Patients With PBC, According to Smoking History of >10 Pack-Years* <10 Pack- Years >10 Pack- Years Female gender 45/50 (90%) 33/40 (83%).3 Caucasian race 47/50 (94%) 35/40 (88%).28 Age at PBC diagnosis (years) 54.2 ( 1.8) 54.2 ( 1.9).4 Significant alcohol use 2/50 (4%) 7/40 (17.5%).03 AST (IU/dL) 78 ( 11.7) 81 ( 13).9 Alkaline phosphatase (IU/dL) 343 ( 48) 431 ( 54).2 Bilirubin (mg/dl) 1.3 ( 0.3) 1.1 ( 0.3).7 Albumin (g/dl) 3.9 ( 0.06) 3.8 ( 0.07).3 Platelet count ( 103) 257 ( 14) 263 ( 15).8 Protime (seconds) 10.9 ( 0.2) 11.1 ( 0.3).6 Mayo Risk Score 4.3 ( 0.3) 4.9 ( 0.3).1 Advanced fibrosis (stages 3-4) 11/41 (27%) 27/34 (79%).0001 P with advanced histological disease at presentation compared to those with early histological disease at presentation ( vs pack years). Analogous to what was observed in the original cohort, the frequency of advanced fibrosis at presentation was significantly higher (P.0001) among patients with a history of smoking (51/77, 66%) compared to never smokers (19/78, 24%). As shown in Fig. 3, the frequency of advanced liver fibrosis at presentation was significantly higher (P.0001) among patients who had a history of smoking 10 pack-years (48/65, 74%) compared to those who did not (20/85, 24%). Table 6 shows the characteristics of the patients in the cross-validation cohort classified based on a history of smoking 10 pack years. Besides advanced liver fibrosis, male gender, a history of significant alcohol use, and Medicaid coverage (as a proxy for low socioeconomic status) were the only three other variables associated with a smoking history of 10-pack years in the cross-validation cohort (Table 6). However, multivariate analysis showed no significant role for these three variables in predicting advanced liver fibrosis at presentation. In addition, although there was no significant association between diabetes mellitus type 2 and advanced liver fibrosis by univariate analysis, given the trend for a higher frequency of diabetes among patients with advanced fibrosis, an additional multivariate logistic regression analysis was performed including diabetes (logistic regression model not shown). The results of that analysis demonstrated no significant role for diabetes mellitus type 2 in predicting advanced liver fibrosis. Table 4. Logistic Regression Model for Advanced Histological Disease at Presentation in 97 Patients With PBC* SE P OR (95% CI) 10 pack-year smoking history ( ) Age at PBC diagnosis ( ) Male gender ( ) Significant alcohol history ( ) *Whole Model Test

6 HEPATOLOGY, Vol. 44, No. 6, 2006 ZEIN ET AL Table 5. Cross-Validation: Characteristics at Presentation of 172 Patients With PBC, According to Early (Stages 0-2) Versus Advanced (Stages 3-4) Stage of Fibrosis* Table 7 shows the multivariate logistic regression model for the cross-validation cohort showing that a history of smoking 10 pack-years predicts the presence of advanced liver fibrosis at presentation (OR of 8.5) while adjusting for age, gender, Medicaid coverage, and history of significant alcohol intake. Discussion Early Fibrosis (Stages 0-2) Advanced Fibrosis (Stages 3-4) Female gender 80/85 (94%) 70/80 (88%).14 Caucasian race 75/78 (96%) 75/79 (95%).71 Age at PBC diagnosis (years) 50.7 ( 1.1) 52.6 ( 1.1).24 History of smoking 26/85 (31%) 51/70 (73%).0001 History of 10 pack-years 17/82 (21%) 48/68 (71%).0001 Average pack-years (smokers) 19.8 ( 3.9) 33.3 ( 2.7).006 Significant alcohol use 5/85 (6%) 7/80 (9%).48 Medicaid coverage 2/84 (2.4%) 8/79 (10%).04 BMI 27.3 ( 0.7) 26.3 ( 0.7).3 BMI 30 19/69 (27.5%) 11/64 (17.2%).15 Diabetes mellitus type 2 4/84 (5%) 10/78 (13%).07 AST (IU/dL) 74 ( 6.5) 93 ( 6.4).04 Alkaline phosphatase (IU/dL) 499 ( 40) 373 ( 41).03 Bilirubin (mg/dl) 0.8 ( 0.3) 2.0 ( 0.3).02 Albumin (g/dl) 4.0 ( 0.1) 3.8 ( 0.1).0004 Platelet count ( 103) 263 ( 11) 206 ( 10).0002 Protime (seconds) 11.0 ( 0.14) 11.1 ( 0.13).5 Mayo Risk Score 4.0 ( 0.2) 5.3 ( 0.17).0001 P Table 6. Cross-Validation: Characteristics of 172 Patients With PBC According to Smoking History of >10 Pack-Years* <10 Pack- Years >10 Pack- Years Female gender 87/89 (98%) 56/67 (84%).002 Caucasian race 79/84 (94%) 64/66 (97%).39 Age at PBC diagnosis (years) 50.7 ( 1.1) 52.8 ( 1.3).22 Significant alcohol use 2/89 (2.3%) 10/67 (15%).003 Medicaid coverage 2/88 (2.3%) 8/66 (12%).01 BMI 27.1 ( 0.6) 26.3 ( 0.8).44 BMI 30 20/74 (27%) 8/49 (16%).17 Diabetes mtllitus Type 2 5/88 (6%) 9/66 (13.6%).09 AST (IU/dL) 87.7 ( 6.4) 74.4 ( 7).9 Alkaline phosphatase (IU/dL) ( 41) ( 45).2 Bilirubin (mg/dl) 1.6 ( 0.3) 1.3 ( 0.3).5 Albumin (g/dl) 4.0 ( 0.06) 3.9 ( 0.06).22 Platelet count ( 103) 235 ( 11.3) 235 ( 11.7).98 Protime (seconds) 11.2 ( 0.14) 11.1 ( 0.14).56 Mayo Risk Score 4.5 ( 0.2) 4.8 ( 0.2).31 Advanced fibrosis (stages 3-4) 20/85 (24%) 48/65 (74%).0001 The results of this study demonstrate a significant association between smoking history and severity of liver fibrosis at presentation in patients with PBC. These findings were confirmed by cross-validation in an independent confirmatory patient cohort. The epidemiological link between smoking and PBC has been suggested by questionnaire-based studies that reported a higher frequency of smoking among patients with PBC compared to subjects without PBC. 8,9 However, this epidemiological association had not been further characterized. This study demonstrated an association between a history of smoking and the severity of liver fibrosis at presentation in PBC. Interestingly, an association between smoking history and severe liver fibrosis in patients with chronic hepatitis C has been suggested by recent studies Those findings as well as the findings of this study support the hypothesis that accelerated progression of liver fibrosis might be induced by smoking. In the studies of patients with hepatitis C, the predictive role of smoking history for advanced liver fibrosis was similar or less significant than that of other known predictors of advanced liver fibrosis in patients with hepatitis C, such as male gender, viral load, and age. 15 However, in both of our populations of patients with PBC, besides the well-recognized clinical P Fig. 3. Cross-validation cohort: Different frequencies (P.0001) of advanced fibrosis according to smoking history of 10 pack-years.

7 1570 ZEIN ET AL. HEPATOLOGY, December 2006 Table 7. Logistic Regression Model for Advanced Histological Disease at Presentation in 172 Patients With PBC*: Cross-Validation Cohort SE P OR (95% CI) 10 pack-year smoking History ( ) Age at PBC diagnosis ( ) Male gender ( ) Medicaid coverage ( ) Significant alcohol history ( ) *Whole Model Test and laboratory markers of advanced liver disease, the only variable significantly associated with advanced liver fibrosis at presentation was a history of smoking. Furthermore, a history of smoking 10 pack-years was the only significant predictor of advanced liver fibrosis at presentation in patients with PBC even when age, gender, and history of significant alcohol use were included in the model. Although defining the potential mechanisms involved in this observation is beyond the scope of this study, various possibilities are plausible. It is known that smoking is associated with an altered balance of the T-helper-1/Thelper-2 response and altered cytokine levels. 16,17 Levels of interleukin-13, which has been associated with progression of liver fibrosis, are known to be elevated in smokers compared to non-smokers. 17 In their study of patients with hepatitis C, Dev et al. recently suggested hypoxia in smokers as a potential mechanism for progression of enhanced liver fibrosis. 15 They found that elevated levels of proangiogenic factors, which could lead to enhanced liver fibrosis, were higher in hepatitis C patients who smoked and were associated with more severe liver fibrosis. Another factor that could potentially aggravate liver fibrosis in smokers is insulin resistance. It is known that insulin resistance is associated with a higher likelihood of progression to severe hepatic fibrosis in certain types of chronic liver disease. 21 In addition, it has been shown that smoking can lead to the development of insulin resistance. 22 In this study, although there was a trend towards an increased frequency of diabetes among patients with advanced liver fibrosis, the association between smoking and severe liver fibrosis was independent of other variables including coexisting diabetes mellitus type 2. The definitive mechanisms behind the apparent accelerated progression of liver fibrosis in patients with PBC who smoked are still unclear and should be further studied. Our study is potentially affected by all the limitations inherent to a retrospective study design. However, the importance of retrospective studies for the characterization of relatively rare liver diseases, such as PBC, has been recently underscored by the National Institutes of Health in their Action Plan for Liver Disease Research. 23 Obtaining accurate smoking history data, especially quantitative information, is difficult in retrospective studies. In an attempt to be as accurate as possible in this regard, our study design was meticulous regarding the definition of smoking history and the methodology for data abstraction. Detailed self-reported information and/or through smoking history available in the medical chart were sought out. If information was not available or ambivalent regarding either variable, the data were considered not available. Another limitation of the study was the inability to obtain a history of exposure to secondary smoke or passive smoking. Although one questionnaire-based study did not find a higher exposure to secondary smoke among patients with PBC compared to controls, 8 we believe that the association between significant exposure to passive smoke and advanced liver fibrosis at presentation in PBC should be prospectively assessed. We attempted to control for variables potentially associated with more severe liver fibrosis at presentation including age, gender, and a history of significant alcohol use. Disadvantaged socioeconomic status can be associated with both smoking and a potentially delayed access to healthcare. In order to try to control for this potential confounder in the cross-validation cohort, we used Medicaid coverage (versus private insurance) as a proxy marker for low socioeconomic status. Although an association was noted between low socioeconomic status and both advanced liver fibrosis at presentation and smoking history of 10 pack-years by univariate analysis, no significant association between low socioeconomic status and advanced fibrosis at presentation in PBC was demonstrated by multivariate analysis. In summary, our study demonstrates that a history of smoking is associated with a higher risk of advanced liver fibrosis in patients with PBC. A history of smoking 10 pack-years predicts a higher risk of advanced liver fibrosis at presentation in PBC while controlling for other potential predictors. Prospective epidemiological studies to further confirm our findings as well as studies of the potential mechanisms behind the accelerated progression of liver fibrosis apparently induced by smoking in patients with PBC are needed. References 1. Newton JL, Jones DE, Metcalf JV, Park JB, Burt AD, Bassedine MF, et al. Presentation and mortality of primary biliary cirrhosis in older patients. Age Ageing 2000;29: Poynard T, Mathurin P, Lai CL, Guyader D, Poupon R, Tainturier MH, et al; PANFIBROSIS Group. A comparison of fibrosis progression in chronic liver diseases. J Hepatol 2003;38:

8 HEPATOLOGY, Vol. 44, No. 6, 2006 ZEIN ET AL Di Martino V, Lebray P, Myers RP, Pannier E, Paradise V, Charlotte F, et al. Progression of liver fibrosis in women infected with hepatitis C; long term benefit of estrogen exposure. HEPATOLOGY 2004;40: Ong JP, Elariny H, Collantes R, Younoszai A, Chandhoke V, Reines HD, et al. Predictors of nonalcoholic steatohepatitis and advanced fibrosis in morbidly obese patients. Obes Surg 2005;15: Corpechot C, Carrat F, Bonnand AM, Poupon RE, Poupon R. The effect of ursodeoxycholic acid therapy on liver fibrosis progression in primary biliary cirrhosis. HEPATOLOGY 2000;32: Howel D, Fischbacher CM, Bhopal RS, Gray J, Metcalf JV, James OF. An exploratory population based case control study of primary biliary cirrhosis. HEPATOLOGY 2000;31: Parikh-Patel A, Gold EB, Worman HJ, Krivy KE, Gershwin MD. Risk factors for primary biliary cirrhosis in a cohort of patients from the United States. HEPATOLOGY 2001;33: Gershwin ME, Selmi C, Worman HJ, Gold EB, Watnik M, Utts J, et al. Risk factors and comorbidities in primary biliary cirrhosis: A controlled interview-based study of 1032 patients and 1041 random-digit dialed matched controls from the United States. HEPATOLOGY 2005;42: James O, Ducker S, Prince M. Case control studies support the association of environmental and genetic risk factors with primary biliary cirrhosis [Abstract 33]. HEPATOLOGY 2005;42(Suppl):209A. 10. Ala A, Stanca CM, Bu-Ghanim M, Ahmado I, Branch AD, Schiano TD, et al. Increased prevalence of primary biliary cirrhosis near Superfund toxic waste sites. HEPATOLOGY 2006;43: CDC. Cigarette smoking among adults United States, MMWR Morb Mortal Wkly Rep 2005;54: Ludwig J, Dickson ER, McDonald GS. Staging of chronic nonsuppurative destructive cholangitis (syndrome of primary biliary cirrhosis). Virchows Arch A Pathol Anat Histol 1978;379: Pessione F, Ramond MJ, Njapoum C, Duchatelle V, Degott C, Erlinger S, et al. Cigarette smoking and hepatic lesions in patients with chronic heaptitis C. HEPATOLOGY 2001; 34: Hezode C, Lonjon I, Roudot-Thoraval F, Mavier JP, Pawlotsky JM, Zafrani ES, et al. Impact of Smoking of histological liver lesions in chronic hepatitis C. Gut 2003;52: Dev A, Patel K, Conrad A, Blatt A, McHutchison JG. Relationship between smoking and fibrosis in patients with chronic hepatitis C. Clin Gastroenterol Hepatol 2006;4: Cozen W, Diaz-Sanchez D, Gauderman W, Zadnick J, Cockburn MG, Gill PS, et al. Th1 and Th2 cytokines and IgE levels in identical twins with varying levels of cigarette consumption. J Clin Immunol 2004;24: Whetzel CA, Corwin EJ, Klein LC. Disruption in Th1/Th2 immune response in young adult smokers. Addict Behav In press. 18. Kaviratne M, Hesse M, Leusink M, Cheever AW, Davies SJ, McKerrow JH, et al. IL-13 activates a mechanism of tissue fibrosis that is completely TGF-beta independent. J Immunol 2004;173: Reiman RM, Thompson RW, Feng CG, Hari D, Knight R, Cheever AW, et al. Interleukin (IL-5) augments the progression of liver fibrosis by regulating IL-13 activity. Infect Immun 2006;74: Sugimoto R, Enjoji M, Nakamuta M, Ohta S, Kohjima M, Fukushima M, et al. Effect of IL-4 and IL-13 on collagen production in cultured L190 human hepatic stellate cells. Liver Int 2005;25: Angulo P, Keach JC, Batts KP, Lindor KD. Independent predictors of liver fibrosis in patients with nonalcoholic steatohepatitis. HEPATOLOGY 1999; 30: Houston TK, Person SD, Pletcher MJ, Liu K, Iribarren C, Kiefe CI. Active and passive smoking and development of glucose intolerance among young adults in a prospective cohort: CARDIA study. BMJ 2006;332: Action Plan for Liver Disease Research. US Department of Health and Human Services, National Institutes of Health. December NIH Publication No