fax: (358) 9 47176089
Liver Failure and Liver Disease
Article first published online: 23 MAR 2006
Copyright © 2006 American Association for the Study of Liver Diseases
Volume 43, Issue 4, pages 723–728, April 2006
How to Cite
Ropponen, A., Sund, R., Riikonen, S., Ylikorkala, O. and Aittomäki, K. (2006), Intrahepatic cholestasis of pregnancy as an indicator of liver and biliary diseases: A population-based study. Hepatology, 43: 723–728. doi: 10.1002/hep.21111
See Editorial on Page 723
Potential conflict of interest: Nothing to report.
- Issue published online: 23 MAR 2006
- Article first published online: 23 MAR 2006
- Manuscript Accepted: 28 DEC 2005
- Manuscript Received: 29 MAY 2005
- Finnish state. Grant Numbers: TYH2210, TYH5102
- Sigrid Juselius Foundation
Intrahepatic cholestasis of pregnancy (ICP) is a liver disorder, thought to be specific for pregnancy and to spontaneously resolve after delivery. Increased rates of gallstone formation and hepatitis C have previously been associated with ICP. However, there are no longitudinal studies to determine its significance as an indicator of subsequent liver or biliary diseases. In this retrospective cohort study with cases and controls we assessed the risk of liver and biliary diseases in 21,008 women, 10,504 with a history of ICP during the years 1972-2000 (cases) and 10,504 with a normal pregnancy (controls). Cases and controls were matched for age, time of delivery, and place of delivery. The diagnoses of liver and biliary disease were traced from the Finnish Hospital Discharge Register with an almost 100% coverage. Several liver and biliary diseases were found to have a significantly higher incidence in patients with ICP than in controls. The rate ratio for hepatitis C was 3.5 (CI 1.6-7.6; P < .001), for nonalcoholic liver cirrhosis 8.2 (CI 1.9-35.5; P < .05), for gallstones and cholecystitis 3.7 (CI 3.2-4.2; P < .001) and for nonalcoholic pancreatitis 3.2 (CI 1.7-5.7; P < .001). In conclusion, there is an association of ICP with several liver and biliary diseases. Some patients with ICP are at risk of the subsequent development of cirrhosis and other severe chronic diseases. Contrary to what has been previously thought, follow-up may need to be considered for these patients. (HEPATOLOGY 2006;43:723–728.)
Intrahepatic cholestasis of pregnancy (ICP) complicates approximately 1 of 100 pregnancies.1–3 It usually manifests in the third trimester of pregnancy as pruritis with elevation of the serum level of liver enzymes and bile acids. It resolves spontaneously in a couple of weeks after delivery and may recur during subsequent pregnancies. The cause of ICP is unknown but both hereditary and environmental factors are thought to play a role. Although most cases are sporadic, ICP is known to run in families.4–7 Mutations in the ABCB4 gene, encoding a member of the ATP binding cassette (ABC) family of membrane transporters,6, 8–10 and variants in the ATP8B1 gene11, 12 have been identified in a small number of patients with ICP. One of the possible underlying causes may be an insufficient liver capacity to metabolize the high amounts of placenta-derived sex steroids during pregnancy.13 The findings that the incidence is higher in multiple pregnancies2, 13 support this hypothesis. Sensitivity to estrogens can also manifest during the use of ethinylestradiol-containing contraceptive pills as a rise in levels of liver enzymes4, 14, 15 and impaired hepatic excretory function.16
ICP is associated with a predisposition for cholesterol gallstones.2, 13, 17 Interestingly, mutations in the ABCB4 gene, causing susceptibility to ICP, also cause susceptibility to cholesterol gallstones.18 Recently, the incidence of ICP has been found to be higher in women positive for hepatitis C.19, 20
Although the association of gallstones and hepatitis C with ICP has been previously noted, there are no previous longitudinal studies to assess the possible risk of liver and biliary diseases in patients with ICP. Therefore, we designed a population-based study to analyze 2the occurrence of these diseases in women with ICP and in controls.
Materials and Methods
The study population of women diagnosed with ICP during the years 1972-2000 was identified from the Finnish Hospital Discharge Register. Each record in the register includes data such as patient and hospital ID numbers, age, sex and diagnosis codes as well as dates of admission and discharge. Data on all inpatient hospital care in Finland are included in the register. Several studies have concluded that the register has reasonably good validity for epidemiological research.21–24 During the years 1972-2000, three versions of the International Classification of Diseases (ICD-8, -9, and -10) were used in Finland. The diagnosis related codes used in the identification of the study population were as follows: during ICD-8 (1972-1986) 63,900 (hepatosis gravidarum, n = 3,831)/63,901 (icterus gravis gravidarum, n = 132)/63,909 (necrosis acuta et subacuta hepatis gravidarum, n = 162); during ICD-9 (1987-1995) 6467A (hepatosis gravidarum, n = 4,226)/6,467X (hepatopathia alia gravidarum, n = 72); and during ICD-10 (1996-2000) O26.6 (intrahepatic cholestasis of gravidarum, n = 2,081). Altogether 10,504 women with a history of ICP were found. For each woman with ICP a matched control without ICP was also identified from the register. To eliminate the potential bias owing to differences in individual characteristics or in the practice of using the ICD codes, controls from the register were selected in relation to the first ICP-related delivery of each (randomly ordered) case using the site of delivery (the same hospital), age at delivery (in years), and time of delivery (date) as the matching criteria. If no exact match was available, the candidate with closest delivery date fulfilling the other criteria was selected. Data on all inpatient hospital care of the study population were extracted from the register for the years 1972-2000 using the unique personal identification numbers of patients as the linkage key. The final study data set contained 138,605 records for 21,008 women. The number of women with no subsequent hospitalizations after the index event was 2,329 among cases and 3,123 among controls.
The ICD codes used in the identification of disease groups of interest are given in Table 1. Due to the limits of the eighth revision of the ICD (ICD-8), it was possible to define all diseases of interest identically using only ICD-9 and ICD-10. Instead of comparing only individual diseases, as many of these are rare, groups of diseases were created for comparison. Different types of hepatitis were classified into two groups: hepatitis A and B, or hepatitis C with non-A–non-B hepatitis, because these were mostly cases of hepatitis C during the earlier years of ICD-9 when the virus could not be tested. A hepatitis not-specified group consisted of other viral and active or chronic autoimmune hepatitis. Alcohol-related cirrhosis was separated from other forms of cirrhosis, that include primary biliary and other nonalcoholic cirrhosis (micro- and macro-nodular, cryptogenic, portal, and of unknown origin) as well as liver fibrosis and sclerosis.
|Hepatitis A and B||0700A; 0701A; 0702A; 0703A; 5714C||B15.0; B15.9; B16.0; B16.1; B16.2; B16.9, B18.0, B18.1|
|Hepatitis non-A-non-B||0704A; 0705A||B17.0; B17.2|
|Hepatitis C||B17.1; B18.2|
|Hepatitis not specified—hepatitis, virus induced—other hepatitis||0706X; 0709X 5714A; 5714B; 5714X||B17.8; B18.8-9; B19.0; B19.9 K73.0-2; K73.8-9; K75.2-3; K75.8-9|
|Cirrhosis hepatitis—non-alcoholic—primary biliary cirrhosis—fibrosis non-alcoholic||5715A 5716A; 5716X 5719X||K74.6 K74.3-5 K72.1; K72.9; K74.0-2|
|Alcohol related liver disorders||5710A; 5711A; 5712A; 5713X||K70.0-4; K70.9|
|Miscellanous liver disorders Bilirubin metabolism disorders||5709A,B; 5718X; 5720A; 5721A; 5722A; 5723A; 5724A; 5728X; 5734A; 5738A,X; 5739X 2774A,X||K72.0; K75.0; K76.0-9 E80.4-7|
|Cholelithiasis and cholecystitis||5740A,B; 5741A,B; 5742A,B; 5743A,B; 5744A,B; 5745A,B; 5750A; 5751A||K80.0-2; K80.4-5; K80.8; K81.0-1; K81.8-9|
|Biliary system disorders||5753A; 5754A; 5755A; 5756A; 5758X; 5759X; 5761A,B,C,X; 5762A; 5763A; 5764A;5765A; 5768X; 5769X||K80.3; K82.0-4; K82.8-9; K83.0-5; K83.8-9|
|Disorders of the pancreas|
|Pancreatitis, non-alcoholic||5770A,B,C; 5771A,B||K85; K86.1|
|Pancreatitis, alcoholic||5770D,E,F; 5771C,D||K86.0|
|Other||5772A,B; 5778A,X; 5779X||K86.2-3; K86.8-9|
Each woman was followed from the beginning of the year 1987 onwards or from the first ICP-associated pregnancy (or corresponding pregnancy used in matching for controls), if such was not detected before 1987, until the last recorded patient encounter or first occurrence of the disease of interest. The median follow-up time was approximately 5.2 person-years per woman, resulting in a median follow-up time of 54,800 person-years for cases and controls per disease. Rates, rate ratios, and confidence intervals were calculated using standard techniques.25 Because the analyses incorporated significance tests for multiple but equally important endpoints, the P values were Bonferroni adjusted in order to detect any possible bias.26
Cumulative probabilities of certain disorders were also calculated using the product-limit estimator, and differences between cases and controls were tested using the log-rank-test. Only the cases (and corresponding controls) having their first ICP occurrence during 1987-2000 (n = 2 × 6,320) were included in these calculations.
This study was approved by the Ethics Committee of the Department of Obstetrics and Gynecology of the Helsinki University Central Hospital.
Statistically significant differences in the incidence of several liver and biliary diseases were found (Table 2). The rate ratios for hepatitis C with non-A–non-B hepatitis and for nonspecific hepatitis (including autoimmune hepatitis) were 3.5 (CI 1.6-7.6; P < .001) and 4.2 (CI 1.7-10.2; P < .001), respectively. There were no statistically significant differences in the occurrence of hepatitis A and B between patients with ICP and controls. The rate ratio for nonalcoholic cirrhosis with liver fibrosis was 8.2 (CI 1.9-35.5; P < .05) and for primary biliary cirrhosis (PBC) alone 5.8 (CI 0.7-48; not significant, data not shown). The risk for the group of miscellaneous liver disorders (ICD codes given in Table 1) was also increased with a rate ratio of 6.1 (CI 2.6-14.5; P < .001), whereas alcohol-related liver disorders were not (rate ratio 0.4, CI 0.2-1.3). There was no difference in the occurrence of diseases of bilirubin metabolism (rate ratio 2.4, CI 0.5-12.4).
|Disease||ICP (n = 10504)||Control (n = 10504)||rr*||CI**||P|
|Hepatitis A and B||4||3||1.3||0.3-5.7||NS|
|Hepatitis C and non-A-non-B||29||8||3.5||1.6-7.6||<.001|
|Alcohol related liver disorders||5||11||0.4||0.2-1.3||NS|
|Miscellanous liver disorders||38||6||6.1||2.6-14.5||<.001|
|Cholelithiasis and cholecystitis||965||260||3.7||3.2-4.2||<.001|
|Biliary system disorders||45||15||2.9||1.6-5.2||<.001|
Cholecystitis and cholelithiasis occurred more often in women with ICP. The rate ratio for the entire study groups was 3.7 (CI 3.2-4.2; P < .001). When analyzed separately for different age groups, the difference was bigger in younger women (Table 3). The rate ratio for other biliary system disorders was 2.9 (1.6-5.2; P < .001). The combined rate ratio for cholangitis and PBC, two chronic cholestatic liver diseases, was 5.5 (1.6-18.6; P < .05).
|Age Group (y)||N*||Number of Occurrences**ICP||Disease Control||Rate Ratio||95% CI||P|
|<20||149||3 (1.2)||1 (1.0)||2.6||0.3-25.1||NS|
|20-24||1135||41 (1.2)||4 (0.8)||10.2||3.6-28.4||<.001|
|25-29||2745||146 (2.1)||20 (3.5)||7.2||4.5-11.4||<.001|
|30-34||3051||235 (4.5)||43 (5.2)||5.4||3.9-7.5||<.001|
|35-39||2218||229 (6.9)||61 (8.4)||3.8||2.8-5.0||<.001|
|40-44||935||140 (10.9)||54 (12.4)||2.6||1.9-3.5||<.001|
|>44||271||171 (17.4)||77 (17.7)||2.2||1.7-2.9||<.001|
Nonalcoholic pancreatitis was more common in patients with ICP than in controls, rate ratio 3.2 (CI 1.7-5.7; P < .001) whereas the rate ratio for alcohol-induced pancreatitis was 0.4 (CI 0.2-1.1; not significant).
Cumulative probabilities of disease occurrence from the first diagnosis of ICP compared with controls were calculated for hepatitis C (with non-A–non-B hepatitis), nonspecific hepatitis, and cholelithiasis and cholecystitis (Fig. 1). The cumulative probabilities of disease occurrence were systematically and significantly higher in women with a history of ICP in all these diseases.
This data set represents the largest collection of women with a history of intrahepatic cholestasis of pregnancy in the medical literature. The study is based on the analysis of the occurrence of liver, biliary, and pancreatic diseases in total of 21,008 women, with equal numbers of cases and controls using the Finnish Hospital Discharge Register as the source.
The main finding of the study was the increased occurrence of several liver, biliary, and pancreatic diseases in women with ICP compared with controls. Hepatitis C infection was over threefold more common in patients with ICP than in controls, whereas there was no difference for hepatitis A and B. This confirms the previous finding that hepatitis C infection is associated with ICP.19, 20 It has been suggested that hepatitis C virus would downregulate the expression of the ABC transporter MRP2 in the liver, thus inducing a failure in the transport of various toxic substances.27 This failure along with high circulating estrogen levels during pregnancy may increase the risk of ICP.
Nonalcoholic cirrhosis (rate ratio 8.2) occurred more often in patients with ICP. For PBC alone, when analyzed separately, the rate ratio was 5.8 (CI 0.7-48), but did not reach statistical significance because of the small number of occurrences. The risk of PBC is nearly 100-fold higher in the first-degree relatives of patients than it is in the general population,28 suggesting inherited susceptibility. The main hypothesis for pathogenesis is autoimmune mechanism because mitochondrial antibodies are detected in about 90% of PBC patients.28 It has also been suggested that viral infections could trigger PBC.29, 30 During pregnancy, it is possible that PBC is misdiagnosed as ICP.31 In a Puerto Rican family with four sisters having prolonged ICP, cirrhosis or periportal fibrosis was present in all four, suggesting that rather than ICP these patients may have had an underlying undiagnosed liver disease.32, 33 To study this in our series, we analyzed the time difference between the diagnosis of ICP and cirrhosis. This analysis could only be done for the subpopulation with first ICP pregnancy during 1987-2000. In all these cases (n = 6) cirrhosis presented within 5 years after the pregnancy with ICP. It is therefore possible that these patients had an underlying liver disease first presenting during pregnancy.
The incidence of cholelithiasis and cholecystitis was almost fourfold higher in women with ICP than in controls. Advanced age, female gender, obesity, and pregnancy are risk factors for gallstones.34–36 Ultrasound examination of the upper abdomen is routinely performed in many women with ICP whereas asymptomatic gallstones are likely to go undiagnosed in an otherwise normal pregnancy. Therefore, asymptomatic gallstones are more likely to be found in patients with ICP than controls. This correlation, however, may explain the difference only partly, as it also remained significant in women aged 45 years or older (Table 3). Interestingly, a genetic factor connecting cholelithiasis with ICP are mutations in the ABCB4 gene, which have been found in patients with gallstones and in patients with ICP.6, 18
Cholelithiasis is the main reason for nonalcoholic pancreatitis. The higher rate of nonalcoholic pancreatitis in women with ICP (rate ratio 3.2, CI 1.7-5.7, P < .001) may be explained by the higher frequency of gallstones. However, the ATP8B1 gene is strongly expressed in extrahepatic tissues such as the pancreas, small intestine, and kidney11 and the presence of diarrhea and pancreatitis in FIC1 transporter dysfunction suggests a role in extrahepatic tissues.37 Sequence variations in ATP8B1 gene have been identified in patients with ICP in Finland and in the United Kingdom.11, 12
It is known that the use of secondary register data in the analyses imposes several challenges and potential sources of bias to the study.38–42 The validity of the Finnish Hospital Discharge Register has been evaluated in several studies.21–24 Virtually all inpatient hospital care periods in Finland are recorded in the register because the reporting is compulsory. The accuracy of most variables utilized in this study (personal and hospital ID numbers, age, sex, and dates of admission and discharge) is at least 95%. The recorded main diagnosis is typically consistent (about 95%) with the medical record and for the subsidiary diagnoses the consistency is also good.
In this study, the delivery of first pregnancy with ICP diagnosis was chosen to represent the index event. The controls were selected by matching for age as well as time and place of delivery. Matching for age and delivery was done to adjust for individual level factors, and for time and place to adjust for the regional factors as well as minimize the systematic bias in clinical and coding practices.
The diseases of interest could be defined identically only for ICD-9 and ICD-10 classifications. The recorded register diagnoses are known to correspond to information in medical records, but incomplete recording of subsidiary diagnoses may distort the results. Therefore, sensitivity analyses were performed using only the main diagnoses. The only change in the results was for the nonspecific hepatitis which barely turned nonsignificant. Most subsidiary diagnoses of nonspecific hepatitis were recorded during the ICP pregnancies. There is no reason to believe that any bias attributable to the identification of diseases from the register would be systematically different among the women with ICP than among controls, because the data were produced in equal manner for both groups. However, any disease occurrence, which did not lead to inpatient care, cannot be found through this register. Therefore, the rates of occurrence for any diseases determined in this study are likely to be the minimum rates. If disease occurrence in outpatient care differs between women with ICP and controls, rate ratios could be affected.
Because age and pregnancy are potential risk factors for many diseases of interest, the unbalanced distribution of these factors between women with ICP and controls in relation to events of disease occurrence may affect the results. To detect this bias, the follow-up time was first stratified according to the age group and pregnancy status and then rate ratios were calculated using the Poisson regression analysis adjusting simultaneously for age group and pregnancy. The results remained practically identical, suggesting that such bias was not present.
The analyses were further complemented by examining the disease occurrences from the beginning of 1987 until the index event. It turned out that for hepatitis C and non-A–non-B and biliary system disorders several women with ICP who had the disorder recorded after the index event also had the condition before the index event. There were no women with ICP who had a diagnosis of nonalcoholic cirrhosis or miscellanous liver disorder before the index event. However, register data did not include information concerning outpatient visits, and therefore it is possible that some patients have had treatment because of liver disorder before the first diagnosis of ICP. Finally, the time axis was reversed and rate ratios for disease of interest occurring between the beginning of 1987 and index event were calculated. It turned out that women with ICP had significantly higher risk than controls for cholelithiasis and cholecystitis as well as for other biliary system disorders even before the first occurrence of ICP.
In summary, after all conservative analyses were performed in this study, we concluded there is an association between ICP and several liver and biliary diseases. The etiology of ICP remains unknown and the cause of its association with other diseases is also presently unknown. It is possible that some diseases increase the risk of ICP or that there is a common cause for ICP and some other liver and biliary diseases. Furthermore, in some cases an underlying liver disorder could be diagnosed as ICP during pregnancy. Nevertheless, these results suggest that in a subgroup of patients ICP may not be solely a self-limiting, spontaneously resolving condition, but also an indicator of subsequent, often more serious diseases. To resolve the nature of the association we have identified, further studies are clearly indicated.
- 21Finnish health and social welfare registers in epidemiological research. Norsk Epidemiologi 2004; 14: 113–120., .
- 25Essential Medical Statistics. 2nd edition. Malden: Blackwell Science, 2003., .
- 36Pregnancy and the biliary tract. Rev Can J Gastroenterol 2000; 14: 5559., .
- 42Utilisation of administrative registers using scientific knowledge discovery. Intelligent Data Analysis 2003; 7: 501–519..