Vascular involvement of the liver in Turner's syndrome



Unexplained liver test abnormalities are frequent in patients with Turner's syndrome. This cohort study was performed to clarify the histopathologic features, causes, and long-term outcome of liver involvement in these patients. Thirty patients with persistently abnormal liver test results were followed-up for 8.8 ± 5.2 years. Liver specimens were available in 27 patients. Marked architectural changes were present in 10 patients, including nodular regenerative hyperplasia in six, multiple focal nodular hyperplasia in two, and cirrhosis in two patients. These changes frequently were associated with obliterative portal venopathy lesions and with aortic malformations. There was mild to moderate portal fibrosis in 15 of the 17 other patients, inflammatory infiltrates in nine patients, and nonalcoholic fatty liver disease in 11 patients. Bile duct alterations resembling small duct sclerosing cholangitis were observed in 21 patients (with or without architectural changes). There was no viral, alcoholic, autoimmune, or drug-induced liver damage. Portal hypertension was observed in four patients with marked architectural changes, including three in whom refractory ascites or recurrent variceal bleeding developed, one of whom underwent transplantation. None of the patients without marked architectural changes experienced progressive or decompensated liver disease. There was no evidence of liver toxicity from estrogen replacement therapy. In conclusion, the main causes of liver involvement in Turner's syndrome are vascular disorders, probably of a congenital origin, and nonalcoholic fatty liver disease. In patients with vascular disorders, severe liver disease requiring liver transplantation may develop. Estrogen therapy does not appear to be pathogenically implicated. (HEPATOLOGY 2004;39:239–247.)

Turner's syndrome (TS) is one of the most frequent sex chromosome abnormalities, affecting approximately 1 in 2,000 females.1 Typical TS is characterized by the association of short stature, ovarian dysgenesia, dysmorphic syndrome, and various abnormalities of the X chromosome. Other somatic abnormalities, such as cardiovascular or renal malformations, are common.1

Liver involvement appears to be frequent in TS patients. The prevalence of liver test abnormalities ranges from 20% to 80% depending on the patient's age, with the highest values in the oldest patients.2–5 The cause of liver test abnormalities in TS patients has been debated. Excess weight and estrogen replacement therapy both have been suggested, but there is no definite evidence for either factor.3, 6 Cirrhosis of unexplained origin and nodular regenerative hyperplasia (NRH) have been both described in TS patients.7–11 However, there are no cohort studies that evaluate histopathologic features.

Because of the lack of information on histopathologic features, the causes, or prognosis of liver involvement in TS patients, we performed this cohort study of 30 patients followed-up for an average of approximately 9 years. In most cases, one or several liver biopsies were obtained and revealed major changes in one third of the patients.


TS, Turner's syndrome; NRH, nodular regenerative hyperplasia; GGT, gamma-glutamyl transferase; HIV, human immunodeficiency virus; FNH, focal nodular hyperplasia.

Patients and Methods

Patient Population.

Thirty TS patients with liver test abnormalities, who were referred to one of the six participating liver units between 1995 and 2002, were enrolled. Diagnosis of TS was based on clinical data and chromosome X abnormalities. Associated malformations and diseases, clinical features of liver disease, and type and duration of treatments were recorded.

The following laboratory tests were performed in all patients during the first liver biopsy: blood cell counts, serum levels of aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, gamma-glutamyl transferase (GGT), total bilirubin, prothrombin time, factor V, cholesterol, triglycerides, glucose, ceruloplasmin, α-1-antitrypsin, ferritin, thyroid-stimulating hormone; immunologic tests for hepatitis B surface antigen, antibodies to hepatitis B surface antigen, antibodies to hepatitis B core antigen, antibodies to hepatitis C virus, antibodies to human immunodeficiency virus, anti-thyroglobulin, anti-thyroperoxidase, anti-nuclear, anti-smooth muscle, anti-mitochondrial, and anti-liver kidney microsomal antibodies.

An abdominal ultrasound examination was performed in all patients, computed tomography scan in seven patients, and coeliomesenteric arteriography in two patients. Upper gastrointestinal endoscopy was available in 13 patients and colonoscopy in six patients. Endoscopic retrograde cholangiopancreatography was performed in 10 patients and endoscopic ultrasonography in two patients with cholestatic features.

Patients were followed-up for a median of 8.8 years (range, 7 months-19 years). Clinical examination and liver tests were performed twice yearly.

Histopathologic Analysis.

Liver biopsy was performed in patients with liver test abnormalities that had persisted for at least 1 year or who had signs of portal hypertension. Twenty-seven patients underwent at least one liver biopsy. Nine patients had two biopsies, and three biopsies were available for two patients. Two patients had a surgical liver biopsy. In one patient, who underwent orthotopic liver transplantation, the entire explanted liver was examined. Specimens were fixed in formalin or Bouin's solution, paraffin embedded, and stained with hematoxylin-eosin-safran, Masson's trichrome, or picrosirius red and Perls. All specimens were reviewed by the same pathologist (CD), who was unaware of the clinical data. Fibrosis was graded according to the Metavir score12 and was considered to be moderate for a score of F1-F2 and marked for a score of F3-F4. Nodular regenerative hyperplasia was defined as the presence of multiple small parenchymal nodules without annular fibrosis and with conserved portal tracts. Focal nodular hyperplasia (FNH) was defined by the presence of a large nodule with or without intranodular fibrosis, alternating with fibrotic areas containing numerous abnormal arteries and ductules. Parenchymal plates within the nodule are composed by normal hepatocytes and have a width of one to two hepatocytes. The presence of multiple nodules was defined as multiple FNH. For the purpose of this study, NRH, FNH, and cirrhosis collectively were referred to as marked architectural changes, and patients were divided into two groups based on the presence or absence of these changes (group A and group B, respectively). Changes in the intrahepatic portal veins, including thrombosis, intimal thickening, or complete obstruction and replacement by a fibrous scar containing numerous vessels were considered to be features of obliterative portal venopathy.13 Ductopenia was defined as the absence of bile ductules in more than 50% of the portal spaces. Steatosis was graded as absent, mild to moderate (involving less than 30% of the hepatocytes), or severe (30% of the hepatocytes or more).


In most cases, the diagnosis of TS (at the age of 13.7 ± 8.0 years) preceded that of liver involvement (at the age of 32.4 ± 14.5 years). In five patients, who were short with no other clinical abnormalities, abnormal liver tests led to the diagnosis of TS. Liver involvement was found only during adulthood in one patient. In most cases, liver involvement was asymptomatic and was discovered during systematic blood testing. Hepatomegaly and ascites were present in two patients. A detailed chromosome X analysis was available in 18 patients. Complete loss of one X chromosome (45,X0) was present in 38% of the cases, mosaïcism (45X0, 46XX) in 44%, and other X-chromosome abnormalities in 18%. Associated malformations and diseases are presented in Tables 1 and 2. Six patients had aortic abnormalities: three patients had aortic coarctation, two had aortic stenosis, and two had aortic bicuspidia (one patient had both aortic coarctation and bicuspidia).

Table 1. Clinical and Biological Data in Turner's Syndrome Patients With Liver Involvement
 All Patients (N = 30)*Group A: Patients With Liver Architectural Changes (n = 10; Nos. 1–10)Group B: Patients Without Liver Architectural Changes (n = 17; Nos. 11–27)
  • *

    Liver histologic data are not available for three patients; significant difference between group A and B.

  • P < .02.

  • P < .01.

Age (yrs)43.0 ± 15.341.3 ± 17.145.7 ± 14.0
Body-mass index (kg/m2)23.9 ± 4.722.0 ± 2.424.3 ± 4.1
Associated diseases   
 Arterial Hypertension (no.)715
 Diabetes (no.)403
 Dyslipidemia (no.)715
 Hashimoto thyroiditis (no.)312
Associated malformations   
 Aortic malformations [no. (%) (Coarctation, bicuspidia, stenosis)]6 (20)5 (50)1 (6)
 Renal malformations [no. (Horseshoe kidney, hypoplasia)]413
 Other malformations [no. (Bonnerie-Ulrich lymphoedema, Coats retinitis, pyloric stenosis)]523
Portal hypertension [no. (%)]4 (13)4 (40)0 (0)
 Ascites (no.)220
 Oesophageal varices (no.)330
AST − x ULN2.1 ± 1.52.7 ± 2.41.8 ± 0.8
ALT − x ULN2.5 ± 2.03.1 ± 2.52.5 ± 1.7
Alkaline phosphatase − x ULN1.9 ± 1.22.3 ± 1.31.8 ± 1.1
γ-glutamyltransferase − x ULN5.2 ± 3.76.8 ± 5.24.6 ± 2.7
Patients with elevation of AST or ALT and with normal alkaline phosphatase [no. (%)]10 (33)26
Patients with elevation of alkaline phosphatase and γ-glutamyltransferase and with normal AST and ALT [no. (%)]4 (13)13
Patients with elevation of AST or ALT, and γ-glutamyltransferase or alkaline phosphatase [no. (%)]16 (53)78
Table 2. Liver Histologic Findings in Turner's Syndrome Patients With Hepatic Involvement: Association With Aortic Malformations
Patients With Liver Architectural Changes (patient no.)Nodular LesionsVascular LesionsPortal FibrosisInflammatory LesionsSteatosisPeriductal FibrosisAortic Malformations
1Nodular regenerative hyperplasiaAbsentModerateModerateModeratePresentAbsent
2Nodular regenerative hyperplasiaObliterative portal venopathyAbsentAbsentAbsentPresentAbsent
3Nodular regenerative hyperplasiaAbsentModerateAbsentAbsentPresentAbsent
4Nodular regenerative hyperplasiaAbsentModerateModerateModeratePresentAbsent
5Nodular regenerative hyperplasiaAbsentAbsentAbsentAbsentAbsentStenosis
6Nodular regenerative hyperplasiaObliterative portal venopathyAbsentAbsentAbsentPresentCoarctation Bicuspidia
7*§¥Multiple focal nodular hyperplasiaObliterative portal venopathyAbsentAbsentAbsentPresentCoarctation
8§Multiple focal nodular hyperplasiaObliterative portal venopathy Sinusoidal dilatationAbsentAbsentAbsentPresentBicuspidia
9*CirrhosisSinusoidal dilatationExtensiveModerateSeverePresentCoarctation
10CirrhosisObliterative portal venopathyExtensiveModerateAbsentAbsentAbsent
Patients Without Liver Architectural Changes (patient no.)       
  • *

    Surgical liver biopsy.

  • Explanted liver.

  • Associated ductopenia.

  • §

    Hepatic nodules visible at US examination and CT scan.

  • Endoscopic esophageal varices.

  • ¥

    Intrahepatic portosystemic shunts.

11No lesionAbsentModerateAbsentAbsentPresentAbsent
12No lesionAbsentModerateModerateSeverePresentAbsent
13No lesionSinusoidal dilatationModerateModerateModeratePresentAbsent
14No lesionAbsentAbsentAbsentModeratePresentAbsent
15No lesionAbsentModerateModerateAbsentAbsentAbsent
16No lesionAbsentModerateModerateModeratePresentAbsent
17No lesionObliterative portal venopathyExtensiveModerateModerateAbsentAbsent
18No lesionAbsentModerateModerateSevereAbsentStenosis
19No lesionAbsentModerateAbsentModeratePresentAbsent
20No lesionObliterative portal venopathyModerateAbsentAbsentPresentAbsent
21No lesionAbsentAbsentAbsentAbsentPresentAbsent
22No lesionAbsentModerateModerateSeverePresentAbsent
23No lesionAbsentModerateAbsentAbsentPresentAbsent
24No lesionObliterative portal venopathyModerateModerateModeratePresentAbsent
25No lesionAbsentModerateModerateModerateAbsentAbsent
26No lesionAbsentModerateAbsentModeratePresentAbsent
27No lesionAbsentModerateAbsentAbsentPresentAbsent

No evidence of alcoholism was found in any of the patients. Twenty-one patients received sex hormone replacement therapy. Five patients were treated with levothyroxine, five received growth hormone, and one received a combination of growth hormone and oxandrolone.

Serum bilirubin levels were normal in all patients. A moderate elevation in serum levels of alkaline phosphatase and GGT, and a mild to moderate increase in aspartate aminotransferase and/or alanine aminotransferase were found in 16 patients. Four patients had increased serum levels of alkaline phosphatase and GGT with normal aminotransferases. Ten patients had increased serum aminotransferases and/or GGT with normal serum alkaline phosphatase levels (Table 1). No sign of hepatic failure was found in any of the patients. Hepatitis B surface antigen, antibodies to hepatitis B core antigen, antibodies to hepatitis C virus, and antibodies to human immunodeficiency virus antibodies were negative in all patients. Serum levels of ceruloplasmin, α-1-antitrypsin, and ferritin were normal in all patients. Isolated antinuclear or smooth-muscle antibodies were found in two patients. Antithyroid antibodies were present in three other patients.

The biliary tract seemed normal by abdominal ultrasound examination except in patients 17 and 20 (listed in Table 2), who had gallbladder stones, and patient 23, who had intrahepatic stones. Parenchymal liver abnormalities included a hyperechoic appearance in two patients, multiple angiomas in one patient, nodular lesions in two patients, and features of portal hypertension associated with a dysmorphic liver and an enlarged spleen in four patients. Coeliomesenteric angiography showed a complete intrahepatic portacaval fistula in patient 7.

Moderate to large esophageal varices were found in three patients. Colonoscopy showed nonulcerative colitis in one patient and colonic diverticula in another. Endoscopic retrograde cholangiopancreatography and biliary endoscopic ultrasonography showed a normal biliary tree in the 10 patients who were evaluated.

Histopathologic data are shown in Table 2. The changes in the 10 patients with marked liver architectural changes (group A) included NRH in six patients, multiple FNH in two patients, and cirrhosis in two patients (illustrated in Fig. 1). One patient with multiple FNH (patient 7) had multiple liver biopsies during laparotomy, allowing the analysis of large samples of several nodules in segments II, III, and IV. The explanted liver of patient 8 showed a combination of multiple FNH, obliterative portal venopathy, and sinusoidal dilatation. In four of the nine other group A patients, changes corresponding to obliterative portal venopathy were found at biopsy.

Figure 1.

Liver architectural changes and vascular lesions in patients with Turner's syndrome. (A) Multiple focal nodular hyperplasia; (inset) fibrotic area containing abnormal vessels; (B) nodular regenerative hyperplasia; (C) numerous vessels within fibrosis.

There was mild to moderate portal fibrosis in 15 of the 17 group B patients (without marked architectural changes), microvesicular or macrovesicular steatosis in 11 (mild or moderate in eight patients and severe in three patients), and inflammatory infiltrates in nine patients. Histologic features were consistent with nonalcoholic fatty liver disease in 11 of these 17 patients, including isolated steatosis in one and steatosis with inflammation and/or fibrosis in the others (illustrated in Fig. 2A,B). Four of the 17 Group B patients showed sinusoidal dilatation and changes corresponding to obliterative portal venopathy. One of these patients (Patient 17) had both non-alcoholic fatty liver disease and obliterative portal venopathy.

Figure 2.

Liver steatosis and biliary lesions in patients with Turner's syndrome. (A) Microvesicular steatosis; (B) macrovesicular steatosis and portal inflammatory infiltrate (arrow); (C) periductal fibrosis (arrow) and sinusoidal dilatation.

Periductal fibrosis (Fig. 2C) was observed in 21 patients, including eight in group A and 13 in group B. In surgical biopsy specimens and in the explanted liver, pericanalar fibrosis involving most interlobular bile ducts and some septal ducts was observed. The same pattern of pericanalar fibrosis involving small ducts was found in both needle and surgical biopsies. In 20 of 21 patients, biochemical or clinical features of cholestasis matched the presence of histologic bile duct abnormalities. No iron deposits were observed in any sample.

Patients in group A showed a significantly higher frequency of aortic bicuspidia, coarctation, and stenosis (5/10 vs. 1/17; P < .02, Fisher test) as well as of portal hypertension (4/10 vs. 0/17; P < .01). There was no significant difference in the degree of liver test abnormalities between the two groups.

Major complications occurred in 3 of the 10 group A patients. Patient 10 died of uncontrolled refractory ascites with pleural effusion and cardiac failure. Patient 8 experienced uncontrolled variceal bleeding and intractable cholestasis, requiring orthotopic liver transplantation 6 years after the diagnosis of liver involvement. Patient 6 underwent surgical portocaval shunting for recurrent variceal bleeding. Patient 7, who had multiple FNH, showed marked enlargement of one of the nodules (from 7 to 13 cm). Although patient 9 remained clinically stable, her fibrosis score increased in a second biopsy performed 2 years later. Two patients in group A were treated with ursodeoxycholic acid (8–12 mg/kg daily) without any effect on liver test results.

There were no liver-related complications in any of the 17 group B patients. Hormone replacement therapy was discontinued for 4 to 6 months in seven of these patients without any change in liver test results. Within a few weeks, serum aminotransferases and alkaline phosphatase levels returned to normal in each of the seven patients of this group who received ursodeoxycholic acid, although GGT remained slightly increased. A second liver biopsy was performed in two patients treated with ursodeoxycholic acid for 8 and 7 years, respectively, showing no change in patient 12 and revealing NRH in patient 3. A second histologic evaluation also was available in four patients who were not treated with ursodeoxycholic acid. No change was observed in patients 23 and 26, whereas steatosis decreased in patient 19, who had a body-mass index reduction from 30 to 28 kg/m2. An increased amount of fibrous tissue was observed in patient 20.


Although liver test abnormalities are believed to be common in TS patients, only 26 case studies including histopathologic results have been reported in the literature (Table 3). Hepatic changes included minimal abnormalities,14 steatosis,6 steatohepatitis,4 biliary involvement,6, 15–17 cirrhosis,7–9 and NRH.10, 11, 18, 19 It has been suggested that the actual incidence of severe hepatic involvement is high.14 Indeed, a fivefold increased risk of “cirrhosis” was recently reported in TS patients, compared with control patients.20 In this study, we report the results of a comprehensive study in a cohort of 30 patients referred to liver units for abnormal liver test results.

Table 3. Liver Histological Findings in Turner's Syndrome Patients: Case Reports
Reference No.Nodular LesionsVascular LesionsFibrosisInflammatory InfiltratesSteatosisBiliary InvolvementMinimal Lesions
  1. Abbreviation: NA, not available.

10Nodular regenerative hyperplasiaNAAbsentAbsentAbsentCholestasis 
11Nodular regenerative hyperplasiaNAAbsentAbsentAbsentAbsent 
19Nodular regenerative hyperplasiaNAExtensiveAbsentAbsentAbsent 
18Focal nodular hyperplasiaHepatoportal sclerosisAbsentAbsentAbsentAbsent 
16Uncharacterized nodulesNAExtensiveAbsentAbsentDuctopenia 
4No lesionNAModeratePresentPresentAbsent 
4No lesionNAModeratePresentPresentAbsent 
4No lesionNAModeratePresentPresentAbsent 
6No lesionNAAbsentPresentAbsentPeriductal fibrosis 
6No lesionNAAbsentAbsentModeratePeriductal fibrosis 
6No lesionNAModerateAbsentAbsentCholestasis 
14No lesionNAAbsentAbsentAbsentAbsentPresent
14No lesionNAAbsentAbsentAbsentAbsentPresent
14No lesionNAAbsentAbsentAbsentAbsentPresent
14No lesionNAAbsentAbsentAbsentAbsentPresent
15No lesionNAModerateAbsentAbsentDuctopenia 
15No lesionNAAbsentAbsentAbsentCholestasis 
17No lesionSinusoidal dilatationAbsentAbsentAbsentCholestasis 

The main finding was the unexpected occurrence of marked liver architectural changes in 30% of patients. These alterations included cirrhosis, NRH, and multiple FNH. In 4 of these 10 patients, portal hypertension developed, leading to life-threatening complications in three patients. Ultrasound revealed hepatic nodules or features of portal hypertension in half of these patients. However, in the five other patients, laboratory and imaging features did not differentiate patients with and without abnormal liver architecture, and thus the patients with a potentially poor outcome. Therefore, liver biopsy should be recommended for diagnostic and prognostic purposes in TS patients with persistently abnormal liver tests. Several findings suggest that a primary vascular involvement was the cause of the marked architectural changes. Nodular regenerative hyperplasia, which was documented in 6 of these 10 patients, currently is thought to be an adaptation to microcirculatory disturbances that result in heterogeneous distribution of intrahepatic blood flow.21–26 According to this view, NRH combines hepatocyte atrophy in areas of decreased perfusion, with hepatocyte hyperplasia in areas of normal perfusion. Focal nodular hyperplasia, the histologic lesions found in two other group A patients, also is regarded as a focal hyperplastic hepatocellular response resulting from focal deprivation of portal perfusion and enhanced arterial perfusion in the corresponding area.27

Confirming the potential link between vascular involvement and architectural changes is the case of patient 7, who had intrahepatic shunt and multiple FNH. The association of nodules and large portosystemic shunts has been documented under a variety of names, including congenital absence of portal vein and patent ductus venosus (in the case of large shunts), nodular hyperplasia, or partial nodular transformation.28–30

The hypothesis of a TS-associated vascular liver disorder is further supported by the findings in one explanted liver showing areas of multiple FNH coexisting with areas of sinusoidal dilatation and obliterative portal venopathy. In two patients with cirrhosis, the lack of evidence for a known cause of chronic liver disease suggests that a vascular disorder may also be involved. Moreover, vascular abnormalities (including aortic coarctation, aortic bicuspidia, cerebral vessel aneurysm, and gastrointestinal telangiectasia) are common in TS31–33 and were found more frequently in patients with marked architectural changes (5 of the 10 patients). Therefore, there is circumstantial but compelling evidence to suggest that some hepatic changes in TS patients are part of a general disorder involving vessels of different sizes, types, and locations. Obviously, a congenital origin would be a likely hypothesis to explain this vascular disorder.

Noninflammatory, concentric fibrosis of small intrahepatic bile ducts resembling primary sclerosing cholangitis frequently was found in patients with and without marked architectural changes. Turner's syndrome patients have a higher than expected incidence of inflammatory bowel disease,34, 35 a condition that is frequently associated with primary sclerosing cholangitis. In the present study, however, none of the patients with ductal fibrosis had inflammatory bowel disease. In addition, extrahepatic bile duct involvement, which is common in primary sclerosing cholangitis, was not found in this cohort. Ductal fibrosis frequently occurs in patients with hepatoportal sclerosis and related conditions36 and in patients with damaged peribiliary arterioles.36–38 Thus, concentric biliary fibrosis may be related to an altered blood supply. If this hypothesis is correct, biliary lesions found in patients without marked architectural alterations could correspond to one end of a spectrum of vascular-related anomalies, with marked architectural changes on the other end. Indeed, biliary abnormalities were reported in 8 of the 21 TS patients of the literature. Whether minor lesions remain stable or evolve to more severe forms requires longer follow-up.

In the 17 patients without marked architectural changes, histopathologic findings included steatosis, steatohepatitis, and steatofibrosis. Insulin resistance syndrome has recently been recognized as a common cause of nonalcoholic fatty liver disease.39, 40 Body mass index values more than 25 kg/m2 frequently are reported in TS patients20 and were found in 9 of these 17 patients. Six of them underwent liver biopsy, and nonalcoholic fatty liver disease was found in five. Thus, nonalcoholic fatty liver disease seems to be the second most common cause of liver involvement in TS patients.

Because TS can be associated with autoimmune disorders, we looked for the presence of circulating autoantibodies. Although five patients had significant titers of autoantibodies, none of them had hypergammaglobulinemia or histologic liver features consistent with autoimmune liver disease.

Estrogen-induced hepatotoxicity has been considered to be the main cause of liver test abnormalities in TS patients receiving hormone replacement therapy.41, 42 However, the causative role of estrogens is not well established. Indeed, alterations in liver test results have been reported regardless of whether TS patients were treated with estrogens, and these alterations were not improved by cessation of replacement therapy.2, 11 Interestingly, nine patients in the present cohort never received replacement therapy, including six patients with marked architectural changes (patients 1, 5, 6, 8, 9, and 10). Therefore, definitive discontinuation of replacement therapy is not needed unless interruption followed by a rechallenge shows improvement in liver test anomalies followed by recurrence.

In this cohort, ursodeoxycholic acid frequently was prescribed because of the cholestatic profile of liver test alterations and of intrahepatic bile duct changes. The findings in this study, although limited, suggest that ursodeoxycholic acid therapy has little effect on liver histologic features, despite occasional improvement in hepatic biochemistry.

In conclusion, hepatic histologic changes in adults with TS are related to one or both of two distinct but not exclusive entities: (i) marked architectural changes, probably related to congenitally abnormal vessels and associated with a risk of severe liver-related complications; and (ii) steatosis, steatofibrosis, and steatohepatitis probably related to metabolic disorders. There is no evidence of liver toxicity from estrogen replacement therapy.


The authors thanks Drs. Michel Beaugrand (Bondy), Thierry Coste (Bobigny), Tony Andreani (Paris), Dominique Wendum (Paris), and Paulette Bioulac-Sage (Bordeaux) for their contribution to this study. We thank Christine Coulondou for her assistance in the preparation of the manuscript.