MATERIALS AND METHODS
This study included mothers of infants with cholestatic liver disease, biliary atresia, and other liver diseases not of infectious or iatrogenic origin. Volunteers for this study were recruited from mothers of children attending The Children's Hospital Pediatric Liver Center Clinic. All children in the hepatobiliary disease groups had diagnoses established by standard criteria, based on physical findings, biochemical analyses, serologies, abdominal ultrasonography, hepatobiliary scintigraphy, percutaneous or surgical liver biopsy, and exploratory laparotomy when indicated. Fifty mothers of infants and children with liver diseases were included in the study.
Mothers were divided into three groups based on their infant's hepatobiliary disease: idiopathic neonatal cholestasis (INC), biliary atresia (BA), and a liver disease control group (LDC). The LDC group included mothers of infants with various liver diseases, including idiopathic liver failure (2 cases), breast milk jaundice, Alagille's syndrome (3 cases), Byler's syndrome, cirrhosis, Neiman-Pick disease (4 cases), an uncertain metabolic disorder, and cystic fibrosis. Mothers of infants who were receiving total parenteral nutrition, who had received known hepatotoxic medications, or who had infectious hepatitis were excluded.
The normal control group (NC) consisted of mothers of healthy infants born within the last 5 years. These mothers were recruited from the Denver metropolitan area. Excluded from the control group were mothers who had a child with liver disease or serious illness during the first year of life. Twenty-two mothers of healthy infants and children were enrolled in the study. In total, 72 mothers participated.
Signed consent was obtained from all participants. This study was reviewed and approved by the Combined Multiple Institutional Review Board for the protection of human subjects of the University of Colorado Health Sciences Center.
Mothers completed a questionnaire about their health and the health of their child. Approximately 10 mL of blood was drawn from each mother for analysis. Because several of the infants included in the study were older than 6 months at the time of enrollment, transplacentally acquired maternal autoantibody would be expected to be undetectable, so blood samples were not taken from the children.
Maternal sera were examined for autoantibodies using the following general approaches: testing in a commercial laboratory using a standard panel of assays for autoantibodies associated with NLE, SLE, and related diseases; testing in a research laboratory for NLE-associated autoantibodies using ELISA; and testing in a research laboratory for organ-specific autoantibodies. The standard panel of assays in the commercial laboratory consisted of the following: antinuclear antibody (ANA) with titer and pattern; anti-dsDNA and anti-ssDNA; antimitochondrial antibodies; and precipitating antibodies to Ro, La, Sm, U1RNP, PmScl, ribosomal P, Mi2, and Jo1. An ANA was considered positive if antibodies were detected at a titer of ≥ 1:120.
Maternal sera were tested by ELISA for autoantibodies to 60 kD Ro, 52 kD Ro, and La (15–17). A positive ELISA reading indicated that the serum had an optical density (OD) > 2 SD above the mean of a control panel of normal sera tested concurrently and that the reaction was > 50% inhibitable by purified antigen.
The sera were tested by indirect immunofluorescence for organ-specific autoantibodies using as substrates cultured rat cholangiocytes and a human cholangiocarcinoma cell line (both provided by A. P. Feranchak, MD). Positive results were considered possibly organ specific if the serum reacted with the cholangiocyte or cholangiocarcinoma cells but not with the Hep-2 substrate used for the ANA testing, or if the serum reacted with the cholangiocyte or cholangiocarcinoma cells at a concentration at least fourfold more dilute than with the Hep-2 substrate. Results consistent with possible organ-specific autoantibodies were verified using immunoblotting.
Proportions were compared between groups using Fisher exact test or the uncorrected χ2 test if appropriate. Confidence intervals for the frequency differences were calculated using a large sample delta approximation based on the log((1 + x)/(1 − x)) transformation. For statistical analyses, the S-Plus 6.0 statistical software package (Insightful Corporation, Seattle, WA) was used.
A large, multicenter study of ANA positivity in the general population (18) tested sera at dilutions of 1:40, 1:80, 1:160, 1:320, etc., whereas the commercial laboratory used in this study tested sera at dilutions of 1:40, 1:120, 1:360, etc. To compare results, despite differences in dilutions tested, we used the following method: the difference in frequency of ANA between subjects in our study and the population was obtained by taking the difference of the frequency we observed in our study mothers at the titer of 1:120, and the mean of the frequencies observed in the study by Tan et al. (18) at dilutions of 1:80 and 1:160. The variance of this rate estimate was calculated using first principles of mathematical statistics, by which the formula p1(1 − p1)/n + [r1(1 − r1) + r2(1 − r2) + r1 + 2(r2 − r1r2)]/(4m) was derived; where p1 is the frequency of positives at 1:120 in the mothers of children with liver disease, r1 is the frequency of positives in controls at 1:80, r2 is the frequency of positives in controls at 1:160, n is the size of our sample, and m is the size of the sample from which the control frequency is estimated.
Sera were obtained from mothers of children who had the following diagnoses: INC (N = 11), BA (N = 25), LDC (N = 14), and NC without liver disease (N = 22). The average age of the child at the time of sampling of the maternal serum was: 8.6 months (range, 1–25 months) for INC, 26.7 months (range, 1–81 months) for BA, 36.1 months (range, 1–150 months) for LDC, and 30.1 months (range, 8–68 months) for NC. The average age of the mother at the time of sampling was: 21.6 years for INC, 30.1 years for BA, 30.3 years for LDC, and 32.4 years for NC.
The characteristic serologic findings of NLE are precipitating antibodies to Ro and/or La in the maternal serum. These findings were absent in all mothers in our study. Serum samples from mothers in all groups were also negative for precipitating antibodies to Sm, U1RNP, PmScl, ribosomal P, Mi2, or Jo1. Sera from three mothers (one BA, one LDC, and one NC) contained an unidentified precipitin. All maternal serum samples were negative for antimitochondrial antibodies. Maternal sera were negative for antibodies to dsDNA; however, one maternal sample (BA group) contained antibodies to ssDNA. No organ-specific antibodies were detected in any maternal sera by direct immunofluorescence using cultured rat cholangiocytes or a human cholangiocarcinoma cell line as substrates.
Although precipitating antibodies to Ro were not detected, several maternal sera contained lower titer, nonprecipitating antibodies to 60 kD Ro, 52 kD Ro, and La, detectable by ELISA. The distribution of low-titer maternal anti-60 kD Ro antibodies in the four study groups is shown in Figure 1. There were no significant differences in the frequency of positivity for anti-60 kD Ro between the NC group and any of the three liver disease groups (INC, BA, or LDC), either separately (P = 0.12, 0.21, 0.99) or when all liver disease groups were combined (P = 0.26). The difference in frequency of positivity between the combined three liver disease groups (INC, BA, and LDC) and the NC group was 15% (95% CI, −7% to 36%).
The distribution of maternal anti-52 kD Ro antibodies detected by ELISA is shown in Figure 2. The differences in the frequency of positivity for anti-52 kD Ro between the NC and INC groups (P = 0.016) and between the NC and the LDC groups (P = 0.0014) were significant. There was no significant difference between the BA group and the NC group (P = 0.17). The difference between the combined three liver disease groups (INC, BA, and LDC) and the NC group was significant (P = 0.014). There were no significant differences in the frequency of maternal anti-La antibodies among the groups.
Twelve of the 72 mothers had a positive ANA (defined as titer of ≥ 1:120). The finding of a positive ANA was not associated with any particular disease group (Fig. 3). We did discover, however, a notable difference in frequency of positive ANA in the mothers of infants with liver disease of any type (INC, BA, and LDC) compared with the mothers of infants without liver disease (NC). Twenty-two percent of mothers of infants with liver disease (11/50) had a positive ANA. Five percent of mothers of infants without liver disease (1/22) had a positive ANA.
The difference in ANA positive frequency between the mothers of children in the three liver disease groups combined and the NC group was 17% (95% CI, 1% to 31%; P = 0.09). Because of the concern that the small size of our control group may overestimate this difference found in our data, we compared the positive ANA frequency in the mothers of children in the three liver disease groups to the frequency previously observed in a large sample of healthy people (18). The calculated frequency (based on the data of Tan et al. (18)) for a positive ANA in a normal population at a dilution of 1:120 is approximately 9%. Thus, the difference in ANA positive frequency between mothers of children in the liver disease groups combined and the expected frequency of a positive ANA at 1:120 in the general population is approximately 13% (i.e., 22%–9%, 95% CI, 0.1% to 25.5%; P = 0.04).
Evaluation of the questionnaire responses revealed no symptoms or mild, nonspecific symptoms in 70 of 72 mothers, with no correlation to autoantibody results. Two mothers had a constellation of symptoms suggestive of connective tissue disease: one had a positive ANA, and the other had a negative ANA.
This was a preliminary study to determine whether NLE could be a cause of INC. Almost two decades ago, the same question was raised regarding idiopathic congenital heart block. In those studies, it was found that most cases of idiopathic congenital heart block were in fact cases of NLE (19,20). INC shares clinical features with the originally reported cases of NLE hepatobiliary disease, i.e., otherwise unexplained conjugated hyperbilirubinemia, a self-limited course, and a generally good prognosis. With the clinical similarity, it was reasonable to expect that some cases of INC would prove to be NLE. However, none of 11 serum samples from mothers of infants with INC in our series had the characteristic maternal serologic findings of NLE, providing a calculated 95% CI for the occurrence of NLE in our series of patients with INC of 0% to 24%. Stated another way, as few as none to as many as one-fourth of the INC cases may still prove to be NLE based on our data. Thus, it appears that NLE is not responsible for most cases of INC.
We were also interested in the possibility that maternal autoantibodies might be associated with BA. The finding of a T-cell inflammatory infiltrate in the extrahepatic bile duct of patients with BA, resembling that observed in primary sclerosing cholangitis, has led investigators to suggest that BA may be caused by an autoimmune phenomenon (21). Because the characteristic maternal serologic findings associated with NLE were absent in our study, it seems very unlikely that NLE is a frequent cause of BA. However, the increased frequency of low titers of autoantibodies in mothers of children with any neonatal liver disease leaves open the possibility that BA and other diseases may have a maternally derived autoimmune component.
Most testing for antibodies to Ro and La has traditionally been performed using an immunodiffusion (Ouchterlony) assay for precipitating antibodies. The formation of a precipitin line requires high titers of antibodies. While not a sensitive test, immunodiffusion is a test with well-defined clinical relevance (22). In this study, we performed not only the immunodiffusion assay but also ELISA assays for autoantibodies to 60 kD Ro, 52 kD Ro, and La. The ELISA assay is quite sensitive and is capable of detecting far lower titers of autoantibodies than can the immunodiffusion assay. However, the clinical relevance of low titers of anti-Ro and La autoantibodies is not yet well defined because these assays are relatively new, and few studies have been done specifically examining their clinical relevance. Although we realized that the interpretation of positive results in ELISA would be subject to the above limitation, we did not wish to overlook a possible clinical correlation with low titer ELISA autoantibody responses, so we determined to examine the sera comprehensively.
The major finding in the ELISA assays was a statistically significant difference between the detection of maternal autoantibodies to 52 kD Ro in the INC and LDC groups compared with the control group. This finding could indicate a specific association between a maternal immune response to Ro and the development of neonatal liver disease, perhaps stimulated by passively transferred maternal IgG. Alternatively, the anti 52 kD Ro antibodies may simply be an indication of an autoimmune diathesis in the mother that is more common in families of children with neonatal liver disease but is not involved in its pathogenesis.
It is intriguing that serologic evidence of autoimmunity, specifically positive ANAs and autoantibodies to 52 kD Ro, occurred frequently in mothers of infants in all of the liver disease groups in our study. This raises the possibility that autoimmunity in the mother may increase the risk for the development of liver disease in the fetus. Alternatively, liver disease in the fetus may induce autoimmunity in the mother. It is noteworthy that adults who have a variety of liver diseases may have a positive ANA (23–25). In some liver diseases, autoimmune processes are clearly involved and a positive ANA is not unexpected. For other liver diseases, such as alcoholic liver disease, the explanation for increased frequency of positive ANA is not clear. Our findings raise the possibility that hepatobiliary disease in the fetus may play a role in the induction of autoimmunity in the mother. However, it must be emphasized that the autoimmunity found in the maternal groups consisted of serologic autoimmunity, not definable autoimmune disease.
This study is subject to several limitations. This was a small, single-center trial studying a relatively rare condition. The study design used a cross-sectional evaluation of mothers of young children with known neonatal liver disease, thus not all maternal serum specimens were drawn during the neonatal period. Therefore it is not possible to determine when maternal ANA and anti 52 kD Ro antibodies became positive. However, for the purposes of examination of maternal antibodies associated with NLE, retrospective examination of maternal serum appears to be satisfactory. Although some autoantibody responses vary with time, autoantibody responses to Ro/La have been noted to be relatively stable in titer over time, independent of disease activity (14). Thus, it is unlikely that a woman would have significant titers of anti-Ro/La during pregnancy and not continue to have significant titers for several years after pregnancy.
In summary, in our study, no cases of INC were confirmed to be NLE. It is possible that rare cases of INC represent NLE, but this study indicates that the majority do not. Many mothers of infants with liver disease had serologic evidence of autoimmunity. We postulate that liver disease in the fetus stimulates autoantibody production in the mother, or that maternal autoimmunity predisposes to neonatal liver disease. Additional prospective studies are needed to address the prevalence and underlying cause of serologic autoimmunity in mothers of infants with liver disease.
The authors are deeply indebted to the nursing staff of The Children's Hospital Liver Center, in particular Nancy Butler-Simon, RN, MS, and Debra Smith, RN, MS. Angela Marchbank assisted with sera processing and the assays for liver-specific autoantibodies. Victoria McCubbin and Jeanette Osban refined the anti-52 kD Ro ELISA assays. Timothy E. Grayson assisted with the immunofluorescence studies.
1. Lee LA. Neonatal lupus erythematosus. J Invest Dermatol 1993; 100:9S–13S.
2. Buyon JP. Neonatal lupus. Curr Opin Rheumatol 1996; 8( 5):485–90.
3. Miranda-Carus ME, Boutjdir M, Tseng CE, et al. Induction of antibodies reactive with SSA/Ro-SSB/La and development of congenital heart block in a murine model. J Immunol 1998; 161( 11):5886–92.
4. Franco HL, Weston WL, Peebles C, et al. Autoantibodies directed against sicca syndrome antigens in the neonatal lupus syndrome. J Am Acad Dermatol 1981; 4( 1):67–72.
5. Miyagawa S, Kitamura W, Yoshioka J, et al. Placental transfer of anticytoplasmic antibodies in annular erythema of newborns. Arch Dermatol 1981; 117:569–72.
6. Laxer RM, Roberts EA, Gross KR, et al. Liver disease in neonatal lupus erythematosus. J Pediatr 1990; 116( 2):238–42.
7. Lee LA, Reichlin M, Ruyle SZ, et al. Neonatal lupus liver disease. Lupus 1993; 2:333–8.
8. Evans N, Gaskin K. Liver disease in association with neonatal lupus erythematosus. J Paediatr Child Health 1993; 29:478–80.
9. Rosh JR, Silverman ED, Groisman G, et al. Intrahepatic cholestasis in neonatal lupus erythematosus. J Pediatr Gastroenterol Nutr 1993; 17( 3):310–2.
10. Taylor-Albert E, Reichlin M, Toews WH, et al. Delayed dilated cardiomyopathy as a manifestation of neonatal lupus: case reports, autoantibody analysis, and management. Pediatrics 1997; 99:733–5.
11. Lee LA, Sokol RJ, Buyon JP. Hepatobiliary disease in neonatal lupus: prevalence and clinical characteristics in cases enrolled in a national registry. Pediatrics 2002; 109( 1):E11.
12. Schoenlebe J, Buyon JP, Zitelli BJ, et al. Neonatal hemochromatosis associated with maternal autoantibodies against Ro/SS-A and La/SS-B ribonucleoproteins. Am J Dis Child 1993; 147( 10):1072–5.
13. Selander B, Cedergren S, Domanski H. A case of severe neonatal lupus erythematosus without cardiac or cutaneous involvement. Acta Paediatr 1998; 87( 1):105–7.
14. Maddison PJ, Reichlin M. Deposition of antibodies to a soluble cytoplasmic antigen in the kidneys of patients with systemic lupus erythematosus. Arthritis Rheum 1979; 22:858–63.
15. Reichlin M, Harley JB. Detection by ELISA of antibodies to small RNA protein particles in systemic lupus erythematosus patients whose sera lack precipitins. Transactions of the Association of American Physicians 1986; 99:161–71.
16. Itoh Y, Reichlin M. Autoantibodies to the Ro/SSA antigen are conformation dependent. I: Anti-60 kD antibodies are mainly directed to the native protein; anti-52 kD antibodies are mainly directed to the denatured protein. Autoimmunity 1992; 14:57–65.
17. Lee LA, Frank MB, McCubbin VR, et al. The autoantibodies of neonatal lupus erythematosus. J Invest Dermatol 1994; 102:963–6.
18. Tan EM, Feltkamp TE, Smolen JS, et al. Range of antinuclear antibodies in `healthy' individuals [see comments]. Arthritis Rheum 1997; 40( 9):1601–11.
19. Scott JS, Maddison PJ, Taylor PV, et al. Connective-tissue disease, antibodies to ribonucleoprotein, and congenital heart block. N Engl J Med 1983; 309:209–12.
20. Reed BR, Lee LA, Harmon C, et al. Autoantibodies to SS-A/Ro in infants with congenital heart block. J Pediatr 1983; 103( 6):889–91.
21. Silveira TR, Salzano FM, Donaldson PT, et al. Association between HLA and extrahepatic biliary atresia. J Pediatr Gastroenterol Nutr 1993; 16( 2):111–3.
22. Maddison PJ. Autoantibodies in SLE. Disease associations. Adv Exp Med Biol 1999; 455:141–5
23. Bailey RJ, Krasner N, Eddleston AL, et al. Histocompatibility antigens, autoantibodies, and immunoglobulins in alcoholic liver disease. Br Med J 1976; 2( 6038):727–9.
24. Gluud C, Tage-Jensen U. Autoantibodies and immunoglobulins in alcoholic steatosis and cirrhosis. Acta Med Scand 1983; 214( 1):61–6.
25. Konikoff F, Isenberg DA, Barrison I, et al. Antinuclear autoantibodies in chronic liver diseases. Hepatogastroenterology 1989; 36( 5):341–5.