The current study is embedded within the Pregnancy-induced Amelioration of Rheumatoid Arthritis (PARA) Study, which is a prospective cohort study on pregnancy and RA 12 . In this study, patients with RA are visited preferably before pregnancy, three times during pregnancy and three times postpartum. The disease activity of RA was scored using the internationally recognized Disease Activity Score 28 (DAS28) with three variables (swollen joint count, tender joint count and C-reactive protein (CRP) level), since this variant of the DAS28 is the most reliable during pregnancy 13 . Some patients were analyzed during more than one pregnancy. Controls were followed from the first trimester of pregnancy onwards. Data from a total of 216 Caucasian patients with RA (patients) and 31 healthy pregnant Caucasian volunteers without an adverse obstetric history (controls) were included in the study. Informed patient consent was obtained for the study. The study is in compliance with the Helsinki Declaration and was approved by the Ethics Review Board at the Erasmus MC University Medical Center, Rotterdam, the Netherlands.

Available data for patients differed for each research question. To investigate whether MBL genotypes are associated with changes in RA disease activity, data for a maximum of 181 patients were available. Patients who experienced a miscarriage were excluded. To investigate whether MBL genotypes are associated with changes in IgG galactosylation, data for a maximum of 145 patients were available. The association between MBL genotypes and pregnancy outcomes, including miscarriages, could be studied in 214 patients. For the analyses of the pregnancy outcome measures: birth weight and gestational age, non-Caucasians, twin pregnancies and pregnancies that resulted in the birth of a child with a malformation were excluded, resulting in 184 patients to be studied. Data were analyzed with or without the patients who participated twice or more.

In accordance with the European League Against Rheumatism (EULAR) criteria, remission of RA was defined as DAS28 <2.6 and intermediate and high disease activity as DAS28 >3.2. Improvement of disease activity during pregnancy was defined according to the EULAR criteria as 'good', 'moderate' (combined with 'responders') or 'nonresponders'. In line with the EULAR criteria, the response criteria can be applied only to those patients with an initial DAS28 >3.2 in the first trimester (n = 84). Deterioration of disease activity after delivery was defined according to so-called reversed EULAR criteria 12 . Since there is no baseline DAS28 requirement for these criteria, this classification was applied to all patients. An early flare was defined as the beginning of deterioration between 6 weeks and 3 months postpartum, and a late flare was defined as disease deterioration between 3 and 6 months postpartum.

Genotyping was performed using LightCycler Real-Time PCR (Roche Applied Science, Almere, The Netherlands) techniques as described previously 2 . Genotyping included the wild type (A allele) and the three SNPs of the first exon of the structural gene: codon 52 (D allele, rs5030737), codon 54 (B allele, rs1800450) and codon 57 (C allele, rs1800451) and two of the SNPs in the promoter region, codon -550 (H/L, rs11003125) and codon -221 (X/Y, rs7096206) of the MBL2 gene. On the basis of the haplotypes, individuals can be categorized into three groups that correlate best with MBL serum levels 10 11 : the high MBL production group, A; the intermediate MBL production group, B; and the low or deficient MBL production group, C. Using such an approach in patients with RA, a very good correlation between MBL genotype groups and MBL serum concentrations has been shown (Spearman's ρ = 0.82, P < 0.0001) 11 .

IgG was purified from the sera of patients and controls as described previously 4 14 . Next, IgG galactosylation was analysed by performing matrix-assisted laser desorption/ionization time of flight mass spectrometry to detect tryptic glycopeptides, and mass spectra of IgG1 and IgG2 were processed using FlexAnalysis software (Bruker Daltonics, Wormer, The Netherlands).

Preterm birth was defined as gestational age <37 weeks of gestation, and low birth weight was defined as a birth weight <2,500 g. As described before, the birth weights analyzed were corrected for gestational age and the sex of the child by using the birth weight standard deviation (SD) score. Birth weight SD scores as well as uncorrected birth weights are added to the analyses 15 . Hypertensive disorders were scored according to the criteria of the International Society for the Study of Hypertension and Pregnancy 7 . Miscarriage was scored in case of a spontaneous loss of a pregnancy before the 20th week.

Statistical analysis was performed using SPSS version 15.0 software (SPSS, Inc., Chicago, IL, USA) and SAS version 9.1 software (SAS Inc., Cary, NC, USA). A two-sided P value ≤0.05 was considered statistically significant. The disease activity (DAS28, patients) and galactosylation profile (patients and controls) were estimated using a linear mixed model. By using this model, we investigated possible associations between the MBL genotype groups and DAS28 or IgG galactosylation. χ² analysis was performed to compare MBL genotype groups of responders versus nonresponders, patients with a flare versus no flare postpartum, patients which had a miscarriage versus those who had not had a miscarriage, patients which had a preterm birth versus those who did not have a preterm birth and patients with a child with low birth weight versus those who did not have a child with low birth weight.

Logistic regression analysis was performed for dichotomous variables, and linear regression analysis was performed for linear data. On the basis of the literature, we considered the following variables a priori to be possible confounders (when applicable): gestational age, maternal smoking during pregnancy, maternal age at delivery, sex of the child, prednisone use during the first trimester, parity and disease activity in the first trimester (DAS28). First, simple regression analyses were performed to determine which confounders had to be included in the multiple regression analyses.

The characteristics of the patients and controls are given in Table 1.

MBL genotypes were determined in 216 patients and 31 controls. In two patients, the promoter SNPs could not be determined. Therefore, these patients could not be assigned to one of the MBL genotype groups and were excluded from analyses, resulting in a total of 214 patients and 31 controls to be analyzed.

No significant differences in DAS28 levels were observed between MBL genotype groups A, B and C at all time points during pregnancy and the postpartum period (P = 0.899) (Figure 1a). Also, no differences were observed between the MBL genotype groups when patients were categorized into responder and nonresponder groups during pregnancy, among patients who had an early or late postpartum flare and among patients who did not have a postpartum flare (responders vs. nonresponders: MBL genotype group A versus MBL groups B and C, odds ratio (OR) 0.91, 95% confidence interval (95% CI) 0.58 to 1.42; early flare vs. no flare: MBL genotype group A versus MBL genotypes B and C, OR 0.69, 95% CI 0.39 to 1.25; late flare vs. no flare: MBL genotype group A versus MBL genotype groups B and C, OR 1.00, 95% CI 0.51 to 1.98). Similar results were found when groups A, B and C were analyzed separately (data not shown) and when patients who participated twice or more were excluded.

No significant differences in IgG1 galactosylation levels were observed between MBL genotype groups A, B and C at all time points during the pregnancy and postpartum periods as shown in Figure 1b (P = 0.75, patients). Similar nonsignificant results were obtained for IgG2 as well as in controls (data not shown). Univariate and multivariate analyses revealed that MBL genotype groups do not affect IgG galactosylation levels, even when corrections for possible confounders such as medication use, disease activity and clinical characteristics are applied.

In RA, the gestational age or birth weight did not differ significantly among the MBL genotype groups (P = 0.78 and P = 0.95, respectively). Accordingly, there was a similar distribution of preterm birth and low birth weight infants among MBL genotype groups (P = 0.75 and P = 0.68, respectively). The distribution of miscarriages (23 of 201 patients) was also not significantly different among the MBL genotype groups (P = 0.81).

All logistic regression and linear regression analyses could not show an association between MBL genotype groups and the pregnancy outcome measures preterm birth, low birth weight, hypertensive disorders, miscarriage and gestational age, birth weight SD score or birth weight (Tables 2 and 3), even after correction for multiple possible confounders as described above. Subgroup analysis for nulliparous women as well as for patients who did not use prednisone in the first trimester of pregnancy did not reveal any effect of MBL on gestational age, birth weight or birth weight SD score (data not shown). Grouping of the intermediate and low MBL genotype groups B and C in all linear and logistic analyses did not reveal a different effect (data not shown).