Maternal anti-Ro autoantibodies are associated with cardiac manifestations of neonatal lupus (cardiac NL), yet only 2% of women with this reactivity have an affected child. Identification of a more specific marker would channel intense monitoring to fetuses at greater risk. This study aimed to determine whether autoantibodies against Ro 52 amino acids 200–239 (p200) confer added risk over autoantibodies to full-length Ro 52, Ro 60, or La.
Anti-Ro–exposed pregnancies resulting in cardiac NL or no cardiac manifestations were identified from the Research Registry for Neonatal Lupus and the PR Interval and Dexamethasone Evaluation study. Umbilical cord (n = 123) and maternal (n = 115) samples were evaluated by enzyme-linked immunosorbent assay.
The frequencies of p200, Ro 52, Ro 60, and La autoantibodies were not significantly different between affected and unaffected children. However, neonatal anti–Ro 52 and Ro 60 titers were highest in cardiac NL and their unaffected siblings compared to unaffected neonates without a cardiac NL sibling. Although both maternal anti–Ro 52 and p200 autoantibodies were less than 50% specific for cardiac NL, anti-p200 was the least likely of the Ro autoantibodies to be false-positive in mothers who have never had an affected child. Titers of anti–Ro 52 and p200 did not differ during a cardiac NL or unaffected pregnancy from the same mother.
Maternal reactivity to p200 does not confer an added risk to fetal conduction defects over full-length Ro 52 or Ro 60 autoantibodies. Mothers who may never be at risk for having an affected child have lower anti–Ro 60 titers and may require less stringent echocardiographic monitoring compared to women with high-titer autoantibodies.
One of the strongest clinical associations with autoantibodies directed to components of the SSA/Ro–SSB/La RNP complex is the development of neonatal lupus (NL) in an offspring. Cardiac and cutaneous diseases are the seminal manifestations of NL. In contrast to the reversible short-lived cutaneous disease, the cardiac disease of NL (cardiac NL) is fatal in nearly one-fifth of cases and most surviving children require a pacemaker for life (1). Cardiac NL develops during 18–24 weeks of gestation and is typically characterized by fibrosis of the atrioventricular node, which can extend to the working myocardium and endocardium (2). The rapidity of clinically detectable injury is supported by the reports of normal sinus rhythm progressing to irreversible third-degree block within 1–2 weeks (3, 4). At present, serial echocardiographic evaluation of all fetuses exposed to anti-SSA/Ro autoantibodies is recommended to detect potentially reversible incomplete blocks (5). Identification of a specific biomarker of cardiac NL would channel intense monitoring to those fetuses at greatest risk of disease.
Autoantibodies to the 52-kd SSA/Ro (Ro 52), 60-kd SSA/Ro (Ro 60), and 48-kd La proteins were first associated with cardiac NL more than 2 decades ago (6). Two non–mutually exclusive hypotheses have been proposed to explain the pathogenic mechanism by which these autoantibodies to normally sequestered intracellular antigens initiate injury in the fetal heart. The first posits that the intracellular Ro/La antigens translocate to the surface of cardiomyocytes undergoing apoptosis during physiologic remodeling and are bound by autoantibodies. The formation of pathogenic autoantibody–apoptotic cell immune complexes promotes proinflammatory and profibrotic responses (7–9). The second hypothesis is based on molecular mimicry wherein autoantibodies cross-react with L-type calcium channels and cause dysregulation of calcium homeostasis (10–12). While several studies have attempted to identify specific epitopes within the Ro and La antigens that are associated with cardiac NL, most of these studies report epitopes common to the anti-Ro/La response regardless of the fetal outcome. Moreover, different autoantibody subsets are identified depending on the immunoassay employed. For example, the sensitivity of peptide or recombinant protein enzyme-linked immunosorbent assays (ELISAs) for anti–Ro 60 autoantibodies is low and may result in false-negatives (13–16).
There has been recent excitement in the autoantibody response against the p200 epitope, spanning Ro 52 amino acids 200–239, as a candidate biomarker conferring an increased maternal risk for the development of cardiac NL in an offspring (17, 18). While several groups have confirmed the high prevalence of the p200 response in women giving birth to a child with cardiac NL, there have been inconsistencies regarding its utility in high-risk assessment relative to the pregnancy exposure (19). Consensus has not been reached as to whether this autoantibody response is also similarly observed in anti-Ro–exposed healthy children when all other maternal autoantibody reactivities to components of the Ro/La complex are equivalent. Moreover, it has not been determined whether autoantibodies to the p200 region of Ro 52 confer any added risk over that observed to full-length Ro 52. A limitation of most previous studies is that the prevalence and titer of maternal autoantibodies have not been measured during the time of fetal exposure. To address the clinical utility of the p200 response as a diagnostic indicator of cardiac NL, this study evaluated umbilical cord blood and maternal serum during affected and unaffected pregnancies for reactivity to p200, full-length Ro 52, Ro 60, and La.
Significance & Innovations
This study is the first assessment of maternal anti–Ro 52 and p200 autoantibodies in umbilical cord blood from cardiac manifestations of neonatal lupus (cardiac NL) and unaffected siblings, thereby eliminating any doubt as to exposure to autoantibody.
For a mother with a child with cardiac NL, the frequencies and titers of anti–Ro 52 and p200 autoantibodies are not informative with regard to the risk of recurrence.
Although both anti–Ro 52 and p200 autoantibodies were of low specificity for cardiac NL, anti-p200 was the least likely of the Ro autoantibodies to be false-positive in mothers who have never had an affected child. However, the sensitivity of anti-p200 was lower than Ro 52 autoantibodies; therefore, testing solely for reactivity to p200 is insufficient.
Mothers with low-titer anti–Ro 60 may require less stringent echocardiographic monitoring compared to women with high-titer autoantibodies.
MATERIALS AND METHODS
The eligibility criterion for this study was enrollment in the Research Registry for Neonatal Lupus (20), which contains neonates with documented cardiac or cutaneous NL and their unaffected siblings, or enrollment in the PR Interval and Dexamethasone Evaluation study (3), which required mothers to be pregnant and positive for anti-Ro autoantibodies as determined by the Clinical Laboratory Improvement Amendments (CLIA)–approved commercial immunology laboratory of New York University (NYU)/Langone Medical Center. Umbilical cord plasma and the matched maternal serum (when available) were obtained with informed consent. Three groups of pregnancies were defined in this study: group 1 comprised pregnancies in which the offspring had documented cardiac NL (samples were available from 48 mothers [50 pregnancies] and 59 neonates); group 2 comprised unaffected pregnancies subsequent to a previous affected pregnancy (30 mothers [32 pregnancies] and 35 neonates); and group 3 comprised unaffected pregnancies where all previous pregnancies (at least 1) were unaffected (31 mothers [33 pregnancies] and 29 neonates). Maternal samples were taken between gestational week 20 and birth. Maternal diagnoses pertained to the last followup visit. Serum was also obtained from 28 healthy controls (anti-Ro/La negative). Studies were approved by the institutional review board.
Recombinant full-length Ro 52 and p200 synthetic peptide.
Recombinant human Ro 52 was generated using the expression plasmid pET28 system in Escherichia coli BL21 (DE3; Novagen) and affinity purified by nickel column chromatography (21). A synthetic peptide representing amino acids 200–239 of Ro 52 was purchased with biotin conjugated at the NH2-terminus (Mimotopes).
Ro 52 and p200 ELISAs.
Full-length Ro 52 (3 μg/ml) in phosphate buffered saline (PBS) was coated onto 96- well microtiter plates overnight at 4°C. After washing (PBS/0.1% Tween 20), nonspecific sites were blocked with 0.1% gelatin/PBS. Human sera were applied (1/500 in blocking buffer) for 2 hours at room temperature. Alkaline phosphatase–conjugated rabbit anti-human IgG (γ-chain specific; Sigma) was used (1:3,000) with phosphatase substrate tablets and optical density (OD) at 405 nm was determined after 15 minutes. The p200 ELISA was done as described (18), with minor modifications. Biotinylated p200 peptide (3 μg/ml in 0.03M Na2CO3, 0.07M NaHCO3, 0.1% NaN3) was applied to preblocked streptavidin microplates (R&D Systems) overnight. Plates were washed (0.15M NaCl, 0.006M NaH2PO4, 20% NaN3/0.05% Tween 20/2% bovine serum albumin), human sera were applied (1/500), and binding of human IgG was detected as described above. Reactivity was expressed as binding units based on a ratio with 1 high-titer patient selected as a standard (binding units = [(OD of sample)/OD standard] × 100). If the OD of the sample was greater than the standard, the sample would be further diluted and binding units multiplied by the dilution factor. Sera were considered positive if the binding units were 3 SDs above the mean binding units of 28 healthy controls.
Ro 60 and La ELISAs.
Due to the low sensitivity of recombinant Ro 60 ELISA for detecting anti–Ro 60 (15, 16), a commercial ELISA using Ro 60 purified from bovine spleen and thymus was employed. Anti-La was also detected by a commercial ELISA using La purified from calf and rabbit thymus according to the manufacturer's recommendations (Diamedix). Ro 60 and La ELISAs were done by the NYU/Langone Medical Center CLIA-approved immunology laboratory.
Calculations were performed using GraphPad Prism software, version 5.04. The prevalence of autoantibodies was compared by Fisher's exact test. Anti–Ro 52, p200, Ro 60, and La autoantibody titers were compared in samples that were positive for the respective autoantibodies by Mann-Whitney test or paired t-test, where appropriate. Linear regression was used to evaluate correlations. P values less than 0.05 were considered significant.
p200 as an immunodominant epitope in the Ro 52 autoantibody response regardless of fetal clinical outcome.
Umbilical cord blood from 123 autoantibody-exposed infants was studied: 59 with cardiac NL (group 1), 35 without cardiac NL but who had a sibling with cardiac NL (group 2), and 29 without cardiac NL and had only healthy siblings (group 3). Sex and race were not significantly different among the neonates studied. As expected, children with cardiac NL were of lower birth weight than their unaffected siblings (P = 0.03) and frequently born prematurely compared to unaffected siblings (P = 0.004) and unaffected nonsiblings (P = 0.005). All neonates were positive for anti–Ro 60 and 120 (98%) were positive for anti–Ro 52. Autoantibodies against p200 were found in 87% of the neonates with anti–Ro 52 reactivity. Anti-p200 autoantibodies were less frequent in unaffected nonsiblings (group 3) compared to affected neonates and siblings (groups 1 and 2); however, this trend did not reach significance (Table 1).
Table 1. Clinical and demographic characteristics of neonates*
Group 2: unaffected with at least 1 sibling with cardiac NL; cutaneous NL (n = 4).
Group 3: unaffected with unaffected siblings; cutaneous NL (n = 3) and hepatic/hematologic complications (n = 1).
Sex of child
Gestational age, mean ± SD weeks/total neonates
36.9 ± 0.3/55
37.9 ± 0.3/35
38.3 ± 0.4/15
Birth weight, mean ± SD gm/total neonates
2,746 ± 104/34
3,123 ± 147/18
3,078 ± 181/9
Maternal serum from 115 pregnancies was assessed: 50 cardiac NL pregnancies (group 1), 32 unaffected pregnancies subsequent to a cardiac NL pregnancy (group 2), and 33 unaffected pregnancies where all previous pregnancies were unaffected (group 3). The clinical diagnosis and race of the mothers were not significantly different across the groups. All mothers were positive for anti–Ro 60 and 111 (97%) were positive for anti–Ro 52 autoantibodies. Reactivity against the p200 epitope was more frequent in group 1 (88%) and group 2 (97%) compared to group 3 (67%). The presence of anti-p200 had 88% sensitivity and 33% specificity for cardiac NL (P = 0.026, group 1 versus group 3), while full-length Ro 52 autoantibody testing showed 100% sensitivity and 12% specificity for cardiac NL (P = 0.022, group 1 versus group 3). There was no significant difference between groups 1 and 2 for p200 or Ro 52 reactivity. Anti-La was equivalent across all of the groups (Table 2).
Table 2. Clinical and demographic characteristics of pregnant mothers*
Group 2: unaffected pregnancy, mother had a previous child with cardiac NL; healthy (n = 29) and cutaneous NL (n = 4).
Group 3: unaffected pregnancy, mother has only unaffected children; cutaneous NL (n = 3) and hepatic/hematologic complications (n = 1).
Association of high-titer anti–Ro 52 and Ro 60 autoantibodies with cardiac NL.
The titer of anti-p200, anti–Ro 52, anti–Ro 60, and anti-La was next evaluated in umbilical cord samples that were positive for the respective autoantibodies. Median ± SEM titers of anti-p200 were higher in neonates with cardiac NL (group 1; 40 ± 7) and unaffected siblings (group 2; 54 ± 9) compared to unaffected nonsiblings (group 3; 24 ± 17); however, this trend did not reach significance. Median ± SEM anti–Ro 52 titers were also higher in group 1 (85 ± 11) and group 2 (89 ± 17) compared to group 3 (54 ± 15; P = 0.02, group 1 versus 3 and P = 0.002, group 2 versus group 3). Higher titer was not restricted to Ro 52 autoantibodies because the median ± SEM anti–Ro 60 titer was also increased in group 1 (9,216 ± 1,796) and group 2 (11,904 ± 3,254) compared to group 3 (3,232 ± 1,425; P = 0.03, group 1 versus group 3 and P = 0.002, group 2 versus group 3). In contrast, median ± SEM titers of anti-La were not significantly different (2,240 ± 482 for group 1, 1,357 ± 312 for group 2, and 1,392 ± 1,648 for group 3) (Figure 1). Linear regression showed that anti-p200, Ro 52, Ro 60, and La autoantibody titers were independent of gestational age (P = 0.09, R2 = 0.07 for anti-p200; P = 0.57, R2 = 0.008 for anti–Ro 52; P = 0.10, R2 = 0.07 for anti–Ro 60; and P = 0.70, R2 = 0.004 for anti-La). Titers of anti–Ro 52 were positively correlated with p200 autoantibodies (P < 0.01, R2 = 0.2) in all of the neonates.
In serum from pregnant mothers, median ± SEM titers of anti-p200 were higher in mothers who were pregnant with a child with cardiac NL (group 1; 44 ± 4) or an unaffected sibling of a child with cardiac NL (group 2; 75 ± 5) compared to pregnant mothers who only had unaffected children (group 3; 30 ± 27); however, this trend did not reach significance. The median ± SEM maternal titers of anti–Ro 52 paralleled p200 autoantibody titers (87 ± 15 for group 1, 93 ± 14 for group 2, and 75 ± 20 for group 3; P = not significant for all comparisons). Median ± SEM levels of anti–Ro 60 were significantly higher in group 1 (14,336 ± 4,896) and group 2 (16,896 ± 12,125) compared to group 3 (2,624 ± 3,245; P = 0.0002, group 1 versus group 3 and P < 0.0001, group 2 versus group 3). Median ± SEM titers of La autoantibodies were not significantly different for group 1 (3,904 ± 1,472), group 2 (1,944 ± 963), or group 3 (752 ± 1,677) (Figure 2).
There was a positive correlation between maternal and neonatal autoantibody titer for p200 (P < 0.0001, R2 = 0.7), anti–Ro 52 (P = 0.0008, R2 = 0.4), and anti–Ro 60 (P < 0.0001, R2 = 0.8). The median ± SEM ratios of maternal to fetal autoantibody titers were as follows: anti-p200: 1.7 ± 0.4 for group 1, 1.4 ± 0.1 for group 2, 1.4 ± 0.3 for group 3; anti–Ro 52: 1.5 ± 0.2 for group 1, 1.4 ± 0.2 for group 2, 1.5 ± 0.3 for group 3; anti–Ro 60: 3.1 ± 1.2 for group 1, 2.0 ± 0.1 for group 2, 1.9 ± 0.4 for group 3; and anti-La: 2.6 ± 0.4 for group 1, 2.0 ± 0.2 for group 2, 1.6 ± 0.2 for group 3.
Equivalence of anti-p200 and Ro 52 autoantibody titers in cardiac NL and unaffected pregnancies from the same mother and in twins discordant for cardiac NL.
Serum samples from 11 mothers obtained during 2 pregnancies, cardiac NL and unaffected, were available for study. For the groups as a whole, there were no significant differences in mean ± SD anti-p200 titers during a pregnancy complicated by cardiac NL (61 ± 31) or resulting in a healthy child (66 ± 29; P = 0.2). Mean ± SD titers of Ro 52 autoantibodies were also equivalent (cardiac NL versus healthy 73 ± 23 versus 80 ± 27; P = 0.2) (Table 3).
Table 3. Anti–Ro 52 and p200 antibody titers in serum from mothers during pregnancy*
Pregnancy 1: cardiac NL
Pregnancy 2: unaffected
Cardiac NL = cardiac manifestations of neonatal lupus; DOB = date of birth.
Umbilical cord blood was available from 9 sibling pairs, including 4 twin sets, of whom 1 twin had cardiac NL and the other was unaffected. Mean ± SD titers of anti-p200 were equivalent among cardiac NL (47 ± 28) and unaffected siblings (50 ± 27; P = 0.4). Mean ± SD anti–Ro 52 titers were also equivalent in cardiac NL (84 ± 49) and unaffected siblings (97 ± 73; P = 0.2) (Table 4).
Table 4. Anti–Ro 52 and p200 antibody titers in umbilical cord blood from sibling (1–5) or twin pairs (6–9) discordant for cardiac NL*
Sibling 1: cardiac NL
Sibling 2: healthy
Cardiac NL = cardiac manifestations of neonatal lupus; DOB = date of birth.
The search for specific biomarkers that predict mothers at high risk of delivering a child with cardiac NL is challenging, given the rarity of the disease. Definitive conclusions regarding whether the anti-p200 response significantly increases the risk of cardiac NL over that associated with autoantibodies to full-length Ro 52 or other components of the Ro/La complex are limited by several factors. None of the previous studies restricted the evaluation of maternal sera to that obtained during pregnancy, assessed umbilical cord blood from both affected and unaffected siblings, or simultaneously evaluated maternal and neonatal sera for both anti-p200 and full-length Ro 52 responses. Accordingly, each of these limitations was addressed in the current study. Anti–Ro 52 and Ro 60 autoantibodies were prevalent in the mothers during their pregnancies complicated by cardiac NL and their subsequent pregnancies of healthy children, findings that were paralleled by equivalent frequencies in the neonates. The presence of both anti–Ro 52 and p200 autoantibodies was more common in mothers pregnant with a child with cardiac NL or who had previously had a child with cardiac NL compared to mothers who never had a child with cardiac NL. Although both anti–Ro 52 and p200 autoantibodies were of low specificity for cardiac NL, anti-p200 was the least likely of the Ro autoantibodies to be false-positive in mothers who have never had an affected child. However, because the sensitivity of anti-p200 was lower than Ro 52 autoantibodies, testing solely for reactivity to p200 is insufficient for pregnant mothers of unknown autoantibody status. Based on a 2% prevalence of cardiac NL in anti–Ro 60–positive mothers (22), the presence of anti–Ro 52 and anti-p200 minimally increases the risk of cardiac NL to 2.2% and 2.6%, respectively, in 1,000 pregnancies. Overall, the presence of p200 autoantibodies was common in each group, suggesting that p200 is a dominant epitope in the anti–Ro 52 response regardless of fetal outcome.
One of the difficulties in comparing the frequency and titer of autoantibody specificities between mothers of affected and unaffected children is that the recurrence rate of cardiac NL, although 10-fold higher than the overall risk of disease, is not 100%. If a change in autoantibody profile from one pregnancy to another is absent, as demonstrated for the majority of families studied herein, other factors must contribute to defining risk for a fetus. However, at least from the perspective of defining a high-risk autoantibody profile potentially independent of the fetal component, comparison to mothers who have never had a child with cardiac NL and have had at least 2 healthy pregnancies is likely to be informative. Although it is acknowledged that cardiac NL can occur even after 5 healthy children (23), the group 3 neonates and their mothers did reveal lower titers of autoantibodies against Ro 52 and Ro 60 compared to the cardiac NL neonates and their unaffected siblings. Most studies attempting to identify high-risk reactivities have compared affected pregnancies to those in which the fetus remains healthy and the mother has never had a child with cardiac NL. The findings reported herein are consistent with the literature. Specifically, in a smaller initial study by Salomonsson and colleagues, an increased titer of anti-p200, anti–Ro 60, and anti-La was reported in cases of heart block compared to mothers who gave birth to unaffected children (17). More recently, Jaeggi and colleagues reported that cardiac complications in fetuses were associated with higher titers of maternal anti-Ro autoantibodies (did not distinguish between anti–Ro 52 or anti–Ro 60) compared to mothers carrying an unaffected child, with the exception that mothers of unaffected children who had previously had a child with cardiac NL also had high-titer anti-Ro comparable to affected pregnancies (24). These observations indicate that measuring p200, Ro 52, or Ro 60 autoantibody titers in a mother who has previously had a child with cardiac NL will not add value in predicting the outcome of any subsequent pregnancies. However, mothers who may never be at risk for having an affected child have lower autoantibody frequencies and titers.
It has been previously reported that titers of autoantibodies against epitopes of Ro 52 were increased in mothers between 18 and 30 weeks of gestation (13) and that Ro 52 autoantibodies gradually decline from early to late pregnancy (25). Moreover, hemodilution during pregnancy secondary to the expected physiologic increase of plasma volume may result in changes in autoantibody concentrations. These findings were not observed in the current study because consecutive pregnancies were exposed to a similar frequency and titer of maternal autoantibodies regardless of the time of maternal serum sampling. Moreover, equivalent levels of anti-p200 and Ro 52 autoantibodies were observed in each twin pair. This finding was not consistent with an earlier study by Harley and colleagues in which the fraternal twin with cardiac NL had only marginally detectable anti-Ro autoantibodies compared to high-titer anti-Ro autoantibodies in the unaffected twin (26). However, our findings were in agreement with a subsequent study of twins discordant for cardiac NL in which equivalent levels of anti-Ro and La autoantibodies were reported (27).
Given that p200 autoantibodies are absent in 15% of the cardiac NL neonates reported in this study and 40% of anti–Ro 52–positive mothers who had a child with heart block in a recent population-based study (28), it is unlikely that anti-p200 autoantibodies are the sole cause of autoimmune heart block in fetuses. In a murine study, passive transfer of anti-Ro and anti-La antibodies from mothers of children with heart block resulted in bradycardia and PR interval prolongation in 70% and 90% of pups, respectively (29). In contrast, active immunization with the p200 peptide results in 20% of rat pups with first-degree block and none with complete block (30). These data suggest that autoantibodies to p200 are neither necessary nor sufficient to account for all cases of cardiac NL. A more recent study in which pregnant rats were passively immunized with monospecific autoantibodies against human p200 identified a 100% penetrance of first-degree block and 81% sinus bradycardia in pups (31). These observations raise the possibility that p200 autoantibodies may be an important initiator of injury, but that autoantibodies against Ro 60 (and perhaps La) RNA-binding antigens are required for full expression of atrioventricular nodal disease and advanced block. The clinical significance of first-degree block remains controversial, since it has not been rigorously established whether first-degree block progresses to second or third degree. In a recent prospective interinstitutional study of 150 fetuses with a persistently normal PR interval throughout the observation period of gestational weeks 19–24, a diagnosis of complete block was subsequently made in 1 fetus. Of 15 untreated fetuses either with PR interval prolongation or with type 1 second-degree block, progressive heart block developed in none of them. Three (20%) of these 15 fetuses had a neonatal diagnosis of first-degree block that spontaneously resolved or had not progressed on followup examinations (32).
Autoantibodies against both full-length Ro 52 and native Ro 60 have a higher sensitivity for cardiac NL compared to p200. However, the specificity of these responses is low, suggesting that other fetal or environmental factors are required for full disease expression. These factors are likely to be fetal genes that might amplify inflammatory and fibrosing responses (33–35) or protective factors that inhibit autoantibody-mediated tissue injury (36, 37).
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Reed had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Reed, Clancy, Buyon.
Acquisition of data. Reed, Lee, Saxena, Izmirly, Buyon.