The effect of pregnancy on lupus nephritis

Authors


Abstract

Objective

To evaluate the effect of pregnancy on lupus nephritis with respect to renal activity and renal deterioration.

Methods

Seventy-eight pregnancies occurred in 53 women with systemic lupus erythematosus (SLE) and renal disease. Seventy-eight nonpregnant SLE patients with evidence of renal disease were matched to the study population by age at the time of each pregnancy and by the presence of a renal manifestation at the beginning of the study. The nonpregnant controls were seen within 2 years of the assessment dates of the pregnant patients with whom they were matched. Renal activity was defined as the presence of active urine sediment or proteinuria, and changes in these parameters were monitored throughout the study period in both study populations. Renal deterioration was defined as an increase in the serum creatinine level that was >20% above the baseline value or an increase to >120 mmoles/liter.

Results

Renal disease activity patterns were available for 74 pairs of pregnancies and controls. Renal disease became active during the study period in 33 pregnancies (44.6%) and 31 controls (41.9%). Serial serum creatinine levels were available for 75 study pairs, among which 62 pregnancies (82.7%) and 57 controls (76.0%) showed no deterioration. Comparison of the treatments received by both the pregnant and the nonpregnant patients showed no significant difference in the amount of steroids taken. A significantly lower amount of immunosuppressive and antimalarial agents were taken during the pregnancies.

Conclusion

During pregnancy in patients with SLE and renal disease, changes in renal disease activity and deterioration in renal function are similar to those which occur in nonpregnant patients with lupus nephritis.

Systemic lupus erythematosus (SLE) is an autoimmune disease that most commonly strikes women in their childbearing years. With the advent of new immunosuppressive drugs with which to better manage the condition, improvements in the obstetric care of women with medical complications, and the many studies indicating the safety of pregnancy in SLE, the number of pregnancies in these patients has increased severalfold. While many investigators have examined the impact of lupus activity on pregnancy, data examining the impact of pregnancy on renal disease are scarce.

Nephritis is known to be one of the most serious complications of SLE and is a strong predictor of poor outcome (1). It is therefore important to inform women of the influence of lupus nephritis on pregnancy. In a previous study of women with SLE, we found that maternal renal disease was a strong predictor of adverse fetal outcome (2), but studies of the effect of pregnancy on the renal disease itself have yielded conflicting results. Studies evaluating pregnancies for which conception occurred during active renal disease have reported flare rates of 48–62% in the pregnancies studied, whereas studies evaluating pregnancies initiated during remission of nephritis indicate flare rates between 7.4% and 32%. Thus, the natural history of renal disease in pregnancy remains to be elucidated (3–6).

The purpose of the present study was to evaluate the effects of pregnancy on lupus nephritis with respect to renal activity and renal deterioration using a nested case–control design.

PATIENTS AND METHODS

Selection of patients.

The University of Toronto Lupus Clinic was established in 1970 and has since been enrolling patients with SLE as part of an ongoing prospective study. Between July 1970 and August 2001, 201 pregnancies occurred in 1,010 patients (886 women) during their enrollment in our clinic. Only pregnancies occurring after enrollment into the clinic were included in this study. All patients fulfilled the American College of Rheumatology (formerly, the American Rheumatism Association) 1982 criteria for a diagnosis of SLE (7). Of these 201 pregnancies, 78 occurred in 53 patients who had lupus-related renal disease. Renal disease was defined as the presence of proteinuria, hematuria, sterile pyuria, casts, nephrotic syndrome, a serum creatinine level >120 mmoles/liter, or the need for dialysis or a renal transplant. Attribution of the above findings to the lupus nephritis was made on the basis of other clinical, serologic, or biopsy evidence of active lupus.

Selection of controls.

As a comparison group, 78 nonpregnant women with SLE and renal disease who were receiving care at our clinic were selected. For each pregnancy that occurred in the pregnant patient population, 1 control subject was selected and matched for age (within 5 years) and for the presence of a renal manifestation at their first study visit. Controls were seen within 2 years of the assessment dates of the pregnant patients with whom they were matched, and they had not been pregnant within 12 months of their first assessment.

Assessments.

Pregnant patients were seen at least once during the prenatal period of each of her pregnancies, once during each pregnancy, and once during each 3-month postpartum period. At every visit, a standardized medical history and physical examination were completed, and routine blood studies and serologic assessments were performed. Data related to lupus disease activity, renal involvement, and to the pregnancy and its outcome were collected prospectively using a standardized protocol and were stored in an Oracle database. Similar assessments in the controls were recorded over a 12-month period. The first 3 months of study being equivalent to the prepartum period and the next 9 months being equivalent to the 9 months of pregnancy.

Renal disease activity was documented as the presence or absence of hematuria, sterile pyuria, casts, or proteinuria occurring in each trimester of the pregnancy, as well as in the prepregnancy and postpregnancy periods. Each pregnancy was then categorized according to the renal disease activity: remaining active, becoming active, or becoming inactive relative to the first assessment for that pregnancy. The investigator who assigned the categories (AT) was blinded to the study group. For example, in patients with hematuria during their prepregnancy or first study visit, but without hematuria throughout the pregnancy and at the postpregnancy visit (or study period), the renal activity was categorized as having become inactive if no other renal variables were present. Renal deterioration was defined as an increase in the serum creatinine level that was either >20% higher than the baseline value or >120 mmoles/liter if the initial value was in the normal range.

Disease activity scores were calculated at each assessment. Both the Systemic Lupus Erythematosus Disease Activity Index 2000 (SLEDAI-2K), a validated measure of disease activity in SLE (8), and the adjusted mean SLEDAI (AMS) (9), a measure that takes into account variable intervals between visits, were calculated for both the patients and the controls over the 12-month study period. Treatment modalities, including the use of steroids and their dosages at the start of the study, the mean dosages throughout the study, and the maximum dosage taken, were recorded for the prepregnancy and pregnancy periods. Steroid use in the controls was similarly recorded. The use of immunosuppressive and antimalarial agents by both populations throughout the study period was also recorded.

Statistical analysis.

Each pregnancy was treated as a separate observation for purposes of analysis. Paired t-tests were used for continuous data. Bowker's test of symmetry was used to analyze renal activity and McNemar's test was used for 2 × 2 categorical data. To address the issue of multiple pregnancies in the same patient, we also analyzed only first pregnancies and their controls. We also performed an analysis using generalized estimating equations. In addition, we did a trend analysis by decade for outcomes, disease activity, and treatment in each group and in the two groups combined.

RESULTS

There were 78 pregnancies occurring in the presence of renal disease in 53 SLE patients during the years 1970–2001. The mean ± SD age of the patients at the time of their pregnancies and of the nonpregnant controls at the start of the study period was 28.1 ± 4.6 years and 29.1 ± 4.9 years, respectively (P = 0.221). There were no statistically significant differences between the control and patient populations with respect to demographics (Table 1). Pregnant patients had a mean of 5.1 ± 2.1 visits and controls had a mean of 5.7 ± 1.3 visits during the course of the study. The time from the onset of renal disease to the beginning of the study in the pregnant patients and their controls was similar (mean ± SD 3.0 ± 3.2 years and 2.4 ± 3.3 years, respectively; P = 0.39).

Table 1. Demographic characteristics of the patients and controls*
 PatientsControlsP
  • *

    A total of 78 pregnancies occurred in 53 patients with lupus and renal disease. A nonpregnant control patient was matched with the patients for each of the 78 pregnancies by age and by the presence of a renal manifestation at their first study visit. SLEDAI-2K = Systemic Lupus Erythematosus Disease Activity Index 2000 update.

  • Anti-DNA antibody was determined by the Farr assay (normal <8 units/ml).

  • Proteinuria was defined as a urinary protein level >500 mg in 24 hours.

Age at start of study, mean ± SD years28.1 ± 4.629.1 ± 4.90.22
Disease duration at start of study, mean ± SD years6.8 ± 5.26.0 ± 4.90.32
Ethnicity, no. (%)   
 Caucasian41 (77.3)56 (71.8) 
 Black3 (5.6)6 (7.7) 
 Chinese2 (3.8)12 (15.4) 
 Other7 (13.2)4 (5.1) 
Age at diagnosis, mean ± SD years21.3 ± 4.223.1 ± 6.10.04
C3, mean ± SD gm/liter0.86 ± 0.250.86 ± 0.310.91
C4, mean ± SD gm/liter0.22 ± 0.080.25 ± 0.140.09
Anti-DNA antibody, mean ± SD units/ml22.9 ± 19.219.3 ± 22.80.16
Serum creatinine, mean ± SD IU/liter85.4 ± 41.886.3 ± 92.90.78
Proteinuria, no. positive/no. tested (%)16/75 (21.3)19/75 (25.3)0.49
SLEDAI-2K6.12 ± 4.577.19 ± 6.020.19
SLEDAI-2K, excluding renal manifestations3.55 ± 3.64.05 ± 3.800.40

The renal manifestations observed in the pregnant and control patients are shown in Table 2. Of the 53 pregnant patients, 19 underwent renal biopsy prior to the study. Biopsy samples were graded according to the World Health Organization (WHO) criteria, with class I in 2 patients, class II in 5, class III in 7, class IV in 4, and class V in 1. Of the 78 controls, 31 underwent renal biopsy prior to the study. Ten had WHO class II, 5 had class III, 9 had class IV, and 7 had class V. Although the outcome of the pregnancies was not an objective of the study, each outcome was documented. Forty-six pregnancies resulted in a live birth, 3 resulted in a stillbirth, 16 resulted in a spontaneous abortion (defined as spontaneous termination of the pregnancy at <20 weeks of gestation), and 13 resulted in a therapeutic abortion.

Table 2. Renal manifestations during the study
 Total no.No. with 1 featureNo. with 2 featuresNo. with 3 featuresNo. with 4 featuresNo. with 5 features
PatientsControlsPatientsControlsPatientsControlsPatientsControlsPatientsControlsPatientsControls
Nephrotic syndrome160010010302
Increased serum creatinine870231332001
Proteinuria43331471915654402
Hematuria454418211915444301
Pyuria253391489454302
Casts682112102302

Renal disease activity patterns were available on 74 pairs of controls and pregnant patients. Table 3 shows the renal disease activity in both control and patient (pregnancies) populations during the study period. As compared with the first assessment, renal disease remained active throughout the study period in 32 (43.2%) pregnancies and 30 (40.5%) controls. Of the 33 pregnancies categorized as becoming active, 13 became active in the postpregnancy period only, 17 were active only during pregnancy, and 3 were active in the pregnancy and postpregnancy period. There was no difference in the number of patients (pregnancies) and controls in whom renal disease became active or inactive. This result was not affected by removing proteinuria as a criterion for active renal disease.

Table 3. Patterns of renal activity in pregnant patients and controls*
 No. (%) of pregnanciesNo. (%) of controlsBowker's test of symmetry
  • *

    Information on both the patients and their matched controls was available for 74 of the 78 pregnancies and for 47 of the 49 full-term plus stillbirth pregnancies.

74 pairs of pregnancies and controls   
 Active throughout32 (43.2)30 (40.5)P = 0.45
 Became inactive9 (12.2)13 (17.6)
 Became active33 (44.6)31 (41.9)
47 pairs of full-term pregnancies and controls   
 Active throughout14 (29.8)14 (29.8)P = 0.77
 Became inactive6 (12.8)5 (10.6)
 Became active27 (57.5)28 (59.6)

Renal deterioration, defined as a >20% increase in the serum creatinine level compared with that at the first assessment of the study period or a serum creatinine level >120 mmoles/liter, was seen in 13 pregnancies (17.3%) and in 18 controls (24.0%) (Table 4). To ensure that both patients and controls had the same overall disease activity, their AMS were compared and showed approximately equal disease activity in the pregnant (6.08 ± 3.36) and nonpregnant (6.12 ± 3.59) groups (P = 0.942) during the study interval. Similar observations were made when only the 47 pairs of pregnant patients with live births or stillbirth alone and their controls were analyzed.

Table 4. Renal deterioration in pregnant patients and controls*
 No. (%) of pregnanciesNo. (%) of controlsMcNemar's test
  • *

    Information on both the patients and their matched controls was available for 74 of the 78 pregnancies and for 47 of the 49 full-term plus stillbirth pregnancies.

75 pairs of pregnancies and controls   
 Unchanged62 (85.7)57 (76.0)P = 0.34
 Worsened13 (17.3)18 (24.0)
47 pairs of full-term pregnancies and controls   
 Unchanged37 (78.7)36 (76.6)P = 0.81
 Worsened10 (21.3)11 (23.4)

The numbers of patients studied provided a power of 80% to detect an increase in the frequency of renal disease becoming active in the pregnancy group from 40% (seen in the controls) to 61%. Likewise, considering patients who had active renal disease or whose renal disease became active during the period of study (80% in the controls), there was 80% power to detect an increase to 94% in the pregnancy group. With regard to renal deterioration, which occurred in 24% of the controls, our study had 80% power to detect an event rate in the pregnancy group of 44%. Thus, there was enough power to detect clinically important differences between the pregnancy group and their controls in this study.

Fifty-four pregnancies (69.2%) were conceived while the patient was taking prednisone. The same number of controls were taking prednisone at the beginning of the study (Table 5). A significantly higher number of controls were taking immunosuppressive and antimalarial drugs at the beginning of the study. During the study, a similar number of pregnant patients and controls took steroids, while a significantly higher number of controls took immunosuppressive and antimalarial drugs (Table 5).

Table 5. Treatment patterns in patients and controls, in both the prepregnancy period and the entire study*
TreatmentNo. (%) of patientsNo. (%) of controls (first visit)McNemar's test
  • *

    Prepregnancy medication information was not available for 2 patients.

During the prepregnancy period   
 Steroids54 (71.1)54 (69.2)P = 0.85
 Immunosuppressives12 (15.8)26 (33.3)P = 0.01
 Antimalarials19 (25.0)34 (43.6)P = 0.004
During the entire study   
 Steroids60 (76.9)60 (76.9)P = 1.0
 Immunosuppressives13 (16.7)33 (42.3)P = 0.001
 Antimalarials25 (29.5)43 (55.1)P = 0.001

The average steroid dosage at the start of the study period in the patient population was 14.6 ± 13.6 mg (mean ± SD), with the steroid dosage over the entire study period being 13.5 ± 10.0 mg. There was no statistically significant difference between study groups in the maximum and mean dosages of steroids taken during the study or the dosage of steroids taken at the start of the study (Table 6). The only statistically significant difference between medication use in the prepregnancy period in the patient population and the first visit of the controls was the use of antimalarials. Forty-three percent of the control population were using these drugs on their first visit while 25% of pregnant patients were using this medication.

Table 6. Steroid dosages taken during the study period in the patients and controls
Steroid dosagePatients, mean ± SD mg/dayControls, mean ± SD mg/dayPaired t-test
Dosage at study start14.6 ± 13.617.8 ± 13.5P = 0.29
Maximum dosage22.0 ± 17.922.8 ± 16.9P = 0.73
Mean dosage13.5 ± 10.015.2 ± 13.7P = 0.29

There was no decade effect (e.g., patients seen during the 70s, 80s, 90s, and beyond) with regard to fetal outcomes, SLE disease activity, and therapy in the pregnancy group, the control group, or the combined group.

We addressed the issue of nonindependence of multiple pregnancies in the same patient in 2 ways. We analyzed only the first pregnancy in each patient and their respective control, and we obtained results similar to those reported above (P = 0.77 for renal disease activity and P = 0.66 for renal deterioration). In addition, we used generalized estimating equations to compare the pattern of renal activity and renal deterioration during pregnancies and in controls. The P values were not significant (P = 0.87 for activity and P = 0.29 for renal deterioration).

DISCUSSION

The impact of pregnancy on the course of lupus has been a subject of debate for decades. While there have been a large number of studies done, no consensus has emerged. In studies with carefully matched controls, as reported by Lockshin et al (6), Mintz et al (10), and our study (11), no excessive rate of lupus flares was noted during pregnancy. Studies in which the patients were used as their own controls, however, such as those by Petri et al (12) and Ruiz-Irastorza et al (13), have shown an excessive rate of renal flare. The discrepancy in these various results may be related to the definition of a true lupus flare in pregnancy. While some have used serologic features to define a flare, others have used clinical features. Unfortunately, the many complications that can occur during pregnancy in general, such as the hemolysis, elevated liver enzymes, and low platelets syndrome or severe eclampsia, may mimic lupus flares as well (14).

Women with renal disease from any cause have long been thought to be at significant risk of poor maternal and fetal outcomes. Many physiologic changes during pregnancy, including an increase in the glomerular filtration rate and renal plasma flow, theoretically render a woman with renal disease vulnerable to deterioration (15). In a study evaluating 82 pregnancies in the presence of primary renal disease with moderate or severe renal insufficiency, both an increase in serum creatinine concentrations and an increase in hypertension were seen throughout the pregnancies (16). The frequency of high-grade proteinuria also rose from a baseline of 23% to a level of 43%. Postpartum, 43% of the women continued to experience deterioration in renal function. Similar findings have been reported in previous studies (17–19).

Evaluation of the effects of pregnancy on lupus nephritis has been subject to limitations, making it quite difficult to reach a conclusion. While the fertility of patients with lupus is unaltered by the disease (20), the number of women with pregnancies in the presence of renal disease is small, thereby making it very difficult to gather enough patients to provide sufficient power for a study to be able to reach a definite conclusion. The use of cyclophosphamide, which leads to premature ovarian failure (21, 22), the increase in fetal loss, and the reluctance of clinicians to advise pregnancy may be a few of the factors that would account for the difficulties in studying the impact of pregnancy on lupus nephritis.

Both prospective and retrospective studies have generated varying results. One of the first large studies was performed by Hayslett and Lynn in 1980 (3). Their study of 65 pregnancies in 47 patients illustrated a parallel between worsening renal function during pregnancy and the degree of lupus activity. Thirty-two percent of the pregnancies conceived when lupus nephritis had been in remission for 6 months showed an exacerbation of nephritis, compared with 48% of pregnancies that were conceived during active disease (3). A retrospective study by Le Thi Huong et al (23) examined 32 pregnancies in histologically proven SLE nephritis in a single center in France. Those authors concluded that from the perspective of worsening renal function, pregnancy need not be discouraged in patients with mild renal impairment, but that the women were at higher risk for preeclampsia and premature delivery. Each pregnancy, however, was conceived when the patients serum creatinine concentration was normal (>100 μmoles/liter), which may underestimate both the renal and fetal complications when applied to the entire lupus population. Another retrospective study published more recently evaluated maternal outcome in 70 pregnancies in patients with lupus nephritis. In that study, Moroni et al (24) showed that the incidence of renal flares observed in the period from the diagnosis of lupus nephritis to the beginning of the pregnancy was similar to the incidence observed during pregnancy and 6 months after delivery.

The present nested case–control study matched all controls within 5 years of age and within 2 years of the first assessment date for the pregnancy in order to control for changing treatment modalities between 1970 and 2001. To the best of our knowledge, this is the first case–control study evaluating the effects of pregnancy on lupus nephritis. A total of 78 pregnancies were evaluated, representing one of the largest cohorts of lupus nephritis patients studied in a single center. All patients were followed up in the same clinic, making certain that followup was similar in both study populations. All changes during pregnancy were recorded over time, which allowed for the development of a new categorization system for monitoring renal activity. This clear documentation also allowed the investigators to monitor disease activity with the SLEDAI-2K over the study period as well as to document changes in treatments.

Since proteinuria may result from non-lupus causes during pregnancy, we compared the 2 populations without using proteinuria as a defining characteristic of renal activity. There were no statistically significant differences between the control and study population in the percentages of pregnancies during which renal disease became active, remained active, or became inactive over the study period. This observation was true for all pregnancies and for first pregnancies only. Although this study did not specifically consider whether the severity of renal disease influences the effects of pregnancy on renal disease, the patient and control groups were similar in terms of renal manifestations, with evidence of elevated serum creatinine levels, proteinuria, hematuria, as well as the total number of renal manifestations and the number with WHO classes III and IV on renal biopsy.

Forty-one percent of the pregnancies studied did not result in a live birth. This restricts our evaluation of these pregnancies to the short interval that the women were actually pregnant. Given that our study sought to determine the effect of pregnancy on renal disease rather than fetal outcome, the short followup period can still provide valid data. Moreover, exclusion of pregnancies that resulted in therapeutic or spontaneous abortions did not alter the results.

In summary, in our comparison of pregnant and nonpregnant patients with lupus nephritis, we found no significant difference in the exacerbation of renal activity nor did we find any difference in the number of cases of renal deterioration seen over an interval equivalent to a prenatal, pregnancy, and postpartum period. To ensure that both pregnant and nonpregnant lupus patients had the same disease severity, the AMS was measured over the 12-month study period and was found to be comparable. To account for the possibility that the pregnant women were treated more aggressively, we analyzed the use of medications as well as the steroid dosages in each population, and no statistically significant difference was seen. We can therefore conclude that renal disease in lupus patients need not be a contraindication to pregnancy.

Ancillary