Objective Cutaneous neonatal lupus (cNL) occurs in possibly 5%–16% of anti-Ro±anti-La antibody–exposed infants. Data suggest in utero exposure to hydroxychloroquine (HCQ) may prevent cardiac NL. The aim was to assess whether in utero exposure to HCQ decreases the risk of cNL and/or delays onset.
Methods A multicentre case–control study was performed with 122 cNL cases and 434 controls born to women with a rheumatological disease who had documentation of maternal anti-Ro±anti-La antibodies at pregnancy and confirmation of medication use and the child’s outcome. A secondary analysis was performed on 262 cNL cases, irrespective of maternal diagnosis, to determine if HCQ delayed time to cNL onset.
Results Twenty (16%) cNL cases were exposed to HCQ compared with 146 (34%) controls (OR 0.4 (95% CI 0.2 to 0.6); p<0.01). Exposure to HCQ was associated with a reduced risk of cNL; exposure to anti-La antibody and female gender were associated with an increased risk of cNL. Exposure to HCQ remained significantly associated with a reduced cNL risk in the analyses limited to mothers with systemic lupus erythematosus and those who developed rash ≤1 month. When analysing all 262 cNL cases, HCQ-exposed infants were older (6.0 (95% CI 5.7 to 6.3) weeks) at cNL onset versus HCQ-non-exposed infants (4.4 (95% CI 3.9 to 5.0) weeks), but the difference was not statistically significant (p=0.21).
Conclusion Exposure to HCQ was associated with a reduced risk of cNL. Among cNL cases, those exposed to HCQ tend to have later onset of rash. Both findings suggest a protective effect of HCQ on cNL.
- systemic lupus erythematosus
- Sjøgren's syndrome
Statistics from Altmetric.com
What is already known about this subject?
In utero exposure to hydroxychloroquine (HCQ) has been associated with a reduced risk of cardiac neonatal lupus. No studies have yet examined the impact of prenatal HCQ exposure on cutaneous neonatal lupus (cNL).
What does this study add?
This multicenter study showed that prenatal exposure to HCQ was associated with a reduced risk of cNL. Among cNL cases, those exposed to HCQ tend to have later onset of rash.
How might this impact on clinical practice or future developments?
Preventing cNL could reduce the occurrence of permanent scarring (telangiectasia, epidermal atrophy and cutaneous pigmentation changes).
Neonatal lupus (NL) is an autoimmune disease associated with the transplacental passage of maternal anti-Ro±anti-La antibodies. Cutaneous involvement is one of the most common non-cardiac manifestations of NL, affecting possibly 5%–16% of anti-Ro±anti-La antibody–exposed infants1 2 and the recurrence rate of cutaneous NL (cNL) approaches 23%.3 Biopsy specimens of affected areas usually show interface dermatitis.4 cNL is a transient condition that generally heals spontaneously as maternal autoantibodies are cleared from the child’s circulation,5 although permanent scarring can develop in cases where the lesions are extensive.6 In addition, if misdiagnosed, it may lead to unnecessary investigations and treatments.
Hydroxychloroquine (HCQ) is a drug frequently used in women with systemic autoimmune rheumatic diseases (SARDs).7 HCQ has been shown to prevent disease flares in pregnant women with systemic lupus erythematosus (SLE).8 Transplacental passage of HCQ has been demonstrated in pregnant women with detectable levels in cord blood and is considered safe during pregnancy.9 10 HCQ is thought to act via inhibition of toll-like receptors (TLRs), which have been implicated in the pathogenesis of cardiac NL and cutaneous lupus.11–13 Although HCQ has not been specifically studied for prevention of cNL, there are encouraging data on its association with a reduced risk of cardiac NL.14–17 Accordingly, the primary objective of this study was to assess if prenatal exposure to HCQ lowered the risk of cNL with the secondary objective evaluating for a delay in cNL rash onset in HCQ-exposed infants who develop cutaneous involvement.
All cases of cNL and controls found in three data sources were collected for this multicentre retrospective study: the SickKids Neonatal Lupus Erythematosus Database (SickKids NLE Database), the Research Registry for Neonatal Lupus (RRNL) and the French Registry of Neonatal Lupus (French RNL). The SickKids NLE Database was created in 1984 and contains prospectively collected data on anti-Ro±anti-La antibody–exposed infants and their mothers living in the Greater Toronto Area, irrespective of the infants’ NL status.17 The RRNL, established in 1994, and the French RNL, established in 2000, both enrol women that are anti-Ro±anti-La antibody positive, living in the USA and France, respectively, and have had at least one child with any manifestation of NL. In addition, these two registries collected data on whether siblings of the affected NL children had NL. These registries comprise data collected both prospectively and retrospectively.18 19
Inclusion and exclusion criteria
Inclusion criteria for the case–control analyses were (1) infant born to a woman positive for anti-Ro±anti-La antibodies, documented prior to or at the time of pregnancy, (2) known age of onset of cNL and (3) documentation of medication taken during pregnancy. For the primary analysis, a maternal diagnosis of a SARD in which HCQ could be used was required: SLE, Sjogren’s syndrome (SS), dermatomyositis, rheumatoid arthritis or juvenile idiopathic arthritis. Diagnosis had to be made prior to conception to allow time for maternal HCQ blood level to increase. For the secondary analysis related to onset of rash, all cases of cNL irrespective of the maternal diagnosis were included, which meant that mothers could be clinically asymptomatic. The diagnosis of cNL in both analyses had to be made by a rheumatologist, dermatologist±paediatrician. Infants with cardiac NL were excluded from all analyses because data from the three registries regarding the effects of HCQ on cardiac NL have already been reported.14 15 17
Study design, outcome measure and data collection
The primary analysis was a case–control study to determine whether exposure to HCQ in mothers with a known SARD reduced the risk of cNL. The primary outcome was the development of cNL. A secondary analysis used a cohort of all children whot developed cNL, irrespective of maternal diagnosis, to assess if in utero exposure to HCQ delayed the onset of cNL.
Information was extracted on infants (gender, year of birth, age of onset of cNL and siblings with cNL±cardiac NL) and their mothers (age at delivery, diagnosis, anti-Ro±anti-La antibody positivity, medication intake including HCQ, fluorinated and non-fluorinated steroids, azathioprine, intravenous immunoglobulins (IVIGs) and plasmapheresis). Children were considered exposed to HCQ if their mother had documented intake of ≥200 mg per day throughout the entire pregnancy. Maternal intake of fluorinated and non-fluorinated steroids, azathioprine and IVIG was defined as intake of any dose at least once during pregnancy. Anti-Ro and anti-La antibodies were measured locally using the following assays: RRNL, commercial Clinical Laboratory Improved Amendments (CLIA)-approved laboratories but in majority of cases confirmed in Buyon Clancy laboratory using recombinant Ro52 and La48 and native Ro60 in ELISA; SickKids NLE Database, commercially available ELISAs using affinity-purified Ro and La antigens; and French RNL, counterimmunoelectrophoresis and/or by using INNO-LIAT ANA Update (Innogenetics, provided by InGen, Rungis, France in most of the cases). Research ethics board approval was obtained at each participating institution.
For the primary analyses, characteristics of cases and controls were compared using χ2, Fisher’s exact and Mann-Whitney U tests as appropriate. To account for within-family data correlation, generalised estimating equations (GEEs) were used for data analyses.20 21 The following 10 covariates were tested in univariable GEE as potential factors associated with cNL: maternal age at delivery, maternal diagnosis of primary or secondary SS (higher levels of anti-Ro±anti-La antibodies±epitope specificity±genetic background in women with SS may put them at a higher risk of having an affected child),22–24 maternal anti-La antibody positive, maternal use of HCQ, maternal use of fluorinated steroids±IVIG±plasmapheresis (use to treat fetuses), maternal intake of non-fluorinated steroids±azathioprine (use to treat mothers), infant’s gender, infant born ≥year 2000 (a time when the safety of HCQ use during pregnancy had been recognised), sibling with cNL±cardiac NL (identify families that may be at increased risk of NL) and registry source (RRNL and French RNL grouped as have same enrolment criteria). Variables with a p value ≤0.05 were selected for inclusion in the multivariable GEE model. A p value <0.05 in the final multivariable model was considered statistically significant. Two a priori subgroup analyses were performed. The first included only infants born to SLE mothers, to account for possible confounding by indication, as women with SLE may be more likely to be prescribed HCQ than women with other types of SARD. The second, limiting cases to those who developed cNL within ≤4 weeks of life. The rationale was that neonates are not continued on HCQ after delivery, therefore HCQ levels are expected to decline.
The secondary analysis was limited to cNL cases, irrespective of maternal health. Kaplan-Meier survival analysis was performed to assess if HCQ delayed the onset of cNL. A p value <0.05 by the Gehan-Breslow-Wilcoxon test was considered statistically significant. Data analysis was performed using IBM SPSS Statistics V.21.0 (IBM) and R V.3.1.1 (R Foundation for Statistical Computing, Vienna, Austria).
Included and excluded patients
A total of 737 infants were screened (figure 1). Forty-one were excluded due to missing key variables. The overall study population consisted of 696 infants: 262 cNL cases and 434 controls. Among the 262 cNL cases, 122 infants were born to women with a SARD diagnosed before pregnancy and were therefore included into the primary analysis. The first subgroup analysis included only infants born to women with SLE (cNL cases=85; controls=300). The second subgroup analysis was restricted to infants who developed cNL within 4 weeks of life (cNL cases=48; controls=434). All 262 cNL cases were included into the secondary analysis.
Primary analysis: exposure to HCQ in mothers with a known SARD reduces the risk of cNL
The study population consisted of 556 infants: 122 cNL cases and 434 controls. There were 10 twin pairs: cNL status was concordant in 8/10 pairs (both twins unaffected). Thirty-eight per cent (N=209) of the study population were siblings: 89 sibling pairs, 7 groups of 3 siblings and 2 groups of 5 siblings. Thirty-seven per cent (N=208) of the study population had at least one sibling with cutaneous (N=85), cardiac (N=107) or both NL features (N=16).
Demographics of the primary analysis study population are shown in table 1. A diagnosis of SLE (48.4%) and SS (48.4%) was found with equal frequency in mothers of cNL cases as compared with mothers of controls for which SLE was the leading diagnosis (62.0%). Nearly all women were anti-Ro antibody positive (99%), but the proportion of women with positive anti-La antibody was significantly higher in mothers of cNL cases (73%) than mothers of controls (48%); p<0.01. In utero exposure to HCQ and to non-fluorinated steroids±azathioprine was higher in controls (34% HCQ exposed and 44% non-fluorinated steroids±azathioprine exposed) than cNL cases (16% HCQ exposed and 25% non-fluorinated steroids±azathioprine exposed); p<0.01. Characteristics of the 556 infants were similar. The RRNL and French RNL had more cNL cases (65%) versus the SickKids NLE Database (35%). The median age of onset of cNL was 6 (range 0–21) weeks old.
Main analysis: factors associated with development of cNL
Association between the 10 predefined covariates and cNL was first explored with univariable GEE analyses (table 2). Six variables were found to have a p value ≤0.05 and were selected for the multivariable GEE model. In the final model, the odds of in utero exposure to HCQ were significantly lower in cNL cases than in controls (odds ratio (OR) 0.4 (95% CI 0.2 to 0.7); p<0.01) (table 3). A similar association was found for exposure to non-fluorinated steroids±azathioprine (OR 0.5 (95% CI 0.3 to 0.8); p=0.01). Cases with cNL were more likely to be female (OR 1.7 (95% CI 1.1 to 2.6); p=0.02), exposed to maternal anti-La antibody (2.7 (95% CI 1.7 to 4.3); p<0.01) and come from the French or RRNL Registries (OR 2.0 (95% CI 1.1 to 3.6); p=0.02).
First subgroup analysis: Infants born to women with SLE
This analysis included the 385 infants (cNL cases=85, controls=300) born to women with SLE. Results of the univariable and multivariable GEEs were similar to those observed for the main analysis (tables 4 and 5) with significantly lower odds of HCQ exposure in cNL cases than in controls (OR 0.4 (95% CI 0.2 to 0.7); p<0.01). In the subgroup of SLE mothers, secondary SS was positively associated with cNL (OR 2.4 (95% CI 1.1 to 5.3); p=0.03), but female gender was no longer significant in the multivariable analysis.
Second subgroup analysis: infants who developed cNL within 4 weeks of life
A total of 482 infants were included (cNL cases=48, controls=434). In univariable analysis, cNL cases were more likely to be exposed to maternal anti-La antibody (OR 3.7 (95% CI 1.7 to 7.8); p<0.01) and come from the French or RRNL Registries (OR 2.1 (95% CI 1.1 to 4.0); p=0.02) but less likely to be exposed to HCQ (OR 0.2 (95% CI 0.1 to 0.6); p<0.01) (table 6). Maternal anti-La antibody positivity (OR 3.2 (95% CI 1.5 to 6.8); p<0.01) and exposure to HCQ (OR 0.2 (95% CI 0.1 to 0.5); p<0.01) remained significant in the multivariable GEE model (table 7).
Secondary analysis: in utero exposure to HCQ and timing of cNL
The study population consisted of 262 patients with cNL. Twenty-three (8.8%) infants were exposed to HCQ. Demographics of the study cohort are shown in table 8. Characteristics of infants exposed to HCQ differed from those non-exposed with respect to the following variables: maternal diagnoses of SLE or cutaneous lupus (52.2% vs 24.4%; p<0.01) or SS (43.5% vs 21.8%; p=0.04) were more frequent, but the proportion of asymptomatic/undifferentiated autoimmune syndrome was lower (4.3% vs 51.7%; p<0.01), and prenatal exposure to non-fluorinated steroids±azathioprine (43.5% vs 9.2%; p<0.01) and proportion of infants born ≥year 2000 (91.3% vs 53.1%; p<0.01) were higher.
Effect of in utero exposure to HCQ on the age of onset of cNL
Kaplan-Meier analysis showed no statistically significant difference in the median number of weeks to cNL onset: HCQ exposed 6.0 (95% CI 5.7 to 6.3) versus non-exposed 4.4 (95% CI 3.9 to 5.0) weeks; p=0.21 (figure 2). Although no statistically significant difference was seen, the survival curves showed a delayed onset of rash during the first 6 weeks in infants exposed to HCQ.
Cutaneous involvement is one of the most common manifestations of NL. In prospective studies, the reported incidence has ranged from 5% to 16%.1 2 Although cutaneous NL is transient and is considered a benign condition, if misdiagnosed, it may lead to unnecessary investigations and treatments. In addition, telangiectasia, epidermal atrophy and/or cutaneous pigmentation changes may appear in 10%–38% of affected children.5 25 26 There is currently no therapy known to prevent cNL. In this multicentre study of 556 infants born to women with SARDs and exposed to anti-Ro±anti-La antibodies, the odds of exposure to HCQ was lower in cNL cases than in controls, suggesting HCQ may protect against cNL.
HCQ may exert a protective effect through inhibition of TLR activation and signalling.27 In lupus-prone NZBxNZW F1 mice, TLR-7 and TLR-9 were shown to be involved in the initiation and maintenance of interface dermatitis, which is similar to histopathological changes found in cNL.4 13 Moreover, inhibiting TLR-7 and TLR-9 signalling improved dermatitis in these mice. In addition, HCQ is effective in treating subacute cutaneous lupus, which closely resembles cNL.7 Another encouraging link derives from data on the protective effect of HCQ on cardiac NL. Four reports have shown that prenatal HCQ exposure is associated with a lower risk of cardiac NL.14–17
Anti-La antibody positivity has been associated with higher incidence of cNL and maternal high titres of anti-La antibody were found to be more common in infants with non-cardiac NL, who predominantly had cutaneous involvement.1 28 Binding of human anti-La antibodies to apoptotic cells at the dermal–epidermal junction and around the basal keratinocytes of murine BALB/c fetuses has been demonstrated.29 In our study, maternal anti-La antibody positivity was significantly associated with cNL in all of our multivariable models, with ORs ranging from 2.7 to 3.4. Infant’s female gender was found to be associated with cNL in our main analysis. This finding may be secondary to the observed increased expression of Ro and La antigen on the membrane surface of keratinocytes exposed to oestrogen.30 31
The impact of maternal health status on the risk of developing cNL remains to be characterised. To our knowledge, there are no studies that have demonstrated that women with SS have a higher risk of having a child with cNL as compared with women with SLE. Our subgroup analysis performed in 385 infants born to women with SLE showed that secondary SS was associated with cNL. Maternal and/or fetal genetic factors could explain the mechanism by which maternal disease modulates the risk of skin involvement in their offspring.22–24 Both the main case–control analysis and the subgroup analysis restricted to mothers with SLE identified a lower exposure of non-fluorinated steroids±azathioprine in infants with cNL. A direct effect of these drugs in the infants is not anticipated since the placenta acts as a relative barrier to azathioprine’s main metabolites and inactivates non-fluorinated steroids.32 33 Indirect effects through modulation of maternal disease, lowering of autoantibody titres and/or residual confounding due to the association with maternal disease subtypes are other possible explanations for these findings. In addition, our study was not designed to assess the effect of non-fluorinated steroids±azathioprine on cNL; therefore, no conclusion on the effect of these two drugs on cNL should be made. Infants from the RRNL and French RNL were more likely to have cNL than those enrolled into the SickKids NLE Database. This is most likely a result of the different enrolment criteria between registries. Inclusion criteria for both the RRNL and French RNL require at least one infant with NL. In contrast, the SickKids NLE Database includes all children exposed in utero to maternal anti-Ro±anti-La antibodies, irrespective of the infant’s NL status. Consequently, the SickKids NLE Database had significantly more controls.
The secondary analysis of this study explored the effect of HCQ on the time of onset of cNL. The retrospective multicentre analysis of the 262 cNL infants irrespective of maternal diagnosis suggests that in utero exposure to HCQ did not significantly delay the onset of rash. Despite having a relatively large sample size, there were only 23 cNL infants exposed to HCQ. The lack of a statistically significant effect of HCQ may be explained by the latter limitation. Interestingly, the Kaplan-Meier curves showed a delayed onset of rash in HCQ-exposed infants during the first 6 weeks of life. This suggests that HCQ could have a protective effect early after birth during the time neonatal blood levels are still within a detectable range. Transplacental passage of HCQ has been demonstrated in neonates with measurable drug levels in cord blood.9 The pharmacokinetics and pharmacodynamics of HCQ in fetuses and neonates has not been studied. Therefore, the potential therapeutic dose range required to prevent cNL and time at which infants’ blood levels fall below this threshold remain unknown. HCQ has been shown to be transfered into breast milk at very low concentrations resulting in estimated HCQ ingested doses by breastfed infants of 0.1–0.2 mg/kg/day.9 34 Although we did not collect data on breast feeding, we believe it is unlikely that the amount of HCQ found into breast milk had an impact on timing of cNL.
Our study is the first to address the effect of HCQ exposure on cNL. Cutaneous involvement in NL may result in unnecessary medical consultations, investigations and treatments and lead to permanent scarring.5 25 26 Moreover, the risk of cardiac NL in a pregnancy following the birth of a child with cNL has been estimated to be 13%.3 Using an inexpensive and safe medication to prevent cNL and therefore minimise these issues is definitely appealing. In addition, we were able to assemble a relatively large sample size for a NL study as our data came from three of the largest NL database/registries. Despite inconclusive findings regarding HCQ's effect on timing of cNL, results of this work are valuable as the survival curves showed a delayed onset of rash during the first 6 weeks of life in infants exposed to HCQ in utero. All analyses performed suggest that prenatal exposure to HCQ may have an impact on the development of cNL and support the need for further studies addressing the underlying pathophysiology of cNL and potential preventive use of maternal HCQ for cNL.
Potential limitations need to be acknowledged. Medication adherence in mothers was not measured specifically during pregnancy, but women were queried about medication intake during routine prenatal rheumatology follow-up and on enrolment into their respective database. It is therefore possible that some children categorised as exposed to HCQ were not exposed throughout gestation due to maternal non-adherence to therapy.35 Because women with SLE are more likely to be prescribed HCQ than women with other SARDs, confounding by indication could have occurred in our study. To account for this potential issue, a subgroup analysis was performed and results were consistent with that of the primary analysis. For the secondary analysis, a larger sample size may have been required to show a significant difference. The age of onset of cNL was obtained from parents during postnatal clinic follow-up. Because early cNL may be very subtle, parents might not have noticed the cutaneous lesions when they first appeared and the age of onset reported could have been incorrect.
In this large multicentre study, exposure to HCQ in utero was associated with a reduced risk of cNL, and this association remained in subgroup analyses limited to mothers with SLE and in infants who developed cNL within 4 weeks of life. In addition, although not statically significant, cNL cases exposed to HCQ tend to have later onset of cNL perhaps due to protection conferred by HCQ when neonatal HCQ blood levels still remain within a detectable range.
The authors thank the following individuals for assitance with data collection and extraction: SickKids NLE Database, Michelle Wang and Shazia Ali; RRNL, Tishaun Middleton and French RNL, Ada Clarke. We wish to thank all the rheumatologists working in the Greater Toronto Area for referring patients to Dr. Laskin and all of the obstetricians at Mount Sinai Hospital High Risk Unit. Similarly, we want to acknowledge physicians who refered patients to the RRNL and French Registry of Neonatal Lupus. Lastly, the authors thank families who agreed to be part of all three registries.
Handling editor Josef S Smolen
Contributors Study conception and design: JB, NC-C, JPB, EDS, PMI. Acquisition of data: JB, NC-C, AB, CF-N, US, CAL, NM, KL, JPB, EDS, PMI. Analysis and interpretation of data: JB, NC-C, PB, JPB, EDS, PMI. All authors were involved in drafting the article and/or revising it critically for important intellectual content, and all authors approved the final version. JB and PMI had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Funding Funding for RRNL was provided by National Institute of Arthritis and Musculoskeletal and Skin Disease (NIAMS) contract N01-AR-4-2220-11-0-1 and grant 5R37AR042455, Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) grants R01HD079951-01A1 and R03 HD069986, and a Lupus Foundation of America LFA Lifeline Program grant to JPB.
Competing interests None declared.
Patient consent Not required.
Ethics approval The study was approved by the Hospital for Sick Children Ethics Boards (REB1000037488), RRNL IRB (protocol no. 7820), French RNL IRB Pitié-Salpetrière.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement There are no unpublished data provided in the manuscript.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.