SEARCH

SEARCH BY CITATION

Keywords:

  • Systemic lupus erythematosus;
  • Disease activity;
  • Pregnancy

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Objective

To validate the Lupus Activity Index in Pregnancy (LAI-P) scale as a diagnostic tool for lupus flares during pregnancy and the puerperium.

Methods

The LAI-P is a modified activity scale specific for pregnancy. Thirty-eight pregnant women with systemic lupus erythematosus (SLE) were prospectively followed in 3 clinics specific for lupus in pregnancy. On each visit, LAI-P was calculated. A modified physician global assessment (m-PGA) scale was used as gold standard (0 = no activity, 1 = mild-moderate activity, 2 = severe activity). A change ≥ 0.25 in LAI-P was predefined as a flare according to previous studies in nonpregnant patients. For the purposes of the study, each visit was considered as an independent case.

Results

During the study period, 158 visits took place for a total 621 patient-weeks. Sensitivity to change was high (standardized response mean for LAI-P = 1.6). We found a significant association between LAI-P and m-PGA (P < 0.002 in all regression models performed). Sensitivity, specificity, and positive and negative predictive values were 0.93, 0.98, 0.88, and 0.99. Positive and negative likelihood ratios were 49 and 0.07, respectively.

Conclusions

LAI-P has a high sensitivity to changes in lupus activity, a significant correlation with m-PGA, and high sensitivity, specificity, predictive values, and likelihood ratios for diagnosing SLE flares during pregnancy and the puerperium.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Over the last 20 years, pregnancy has become a frequent event in women with systemic lupus erythematosus (SLE). During this period, the effect of pregnancy and the puerperium on lupus activity has been matter of debate. Several controlled studies have tried to address this issue with conflicting results (1–6). The reasons for this heterogeneity have been discussed elsewhere (7, 8), one of the most important being the lack of a common working definition for a “lupus flare.” Some of the studies employed lupus activity scales already in use (4–6), such as the Lupus Activity Index (LAI) or the SLE Disease Activity Index (SLEDAI), despite the fact that these scales have been validated in nonpregnant lupus populations, which also included men (9). Because some physiologic changes in pregnancy may mimic lupus activity and also may limit the validity of some clinical and biochemical findings (7), it seemed appropriate to construct lupus activity scales specific for pregnant women.

A proposal for the definition of 3 activity scales specific for pregnancy (SLE Pregnancy Disease Activity Index [SLEPDAI], Lupus Activity Index in Pregnancy [LAI-P], and modified Systemic Lupus Activity Measure [m-SLAM]) was published in 1999 (10). They were adaptations of SLEDAI, LAI and SLAM, respectively. Although the SLEPDAI has already been used in 1 research article (11), the validity of these scales as sensitive and specific tools to measure lupus activity during pregnancy has not been established. We therefore tested the usefulness of LAI-P in a multicenter, prospective study of pregnant women with lupus attending 3 combined lupus-pregnancy clinics at St. Thomas' Hospital, London, UK, the Birmingham Women's Hospital, Birmingham, UK, and the Hospital for Special Surgery, New York, NY.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Patients

From October 1999 to March 2001, 44 unselected, consecutive pregnant women were recruited. All fulfilled at least 4 of the 1997 modified criteria of the American College of Rheumatology for the classification of SLE (12, 13). Women in whom the diagnosis of SLE had not been made before the index pregnancy were excluded from the study.

Six women had a miscarriage or failed to attend the clinic after the first visit, therefore 38 women (27 from London, 7 from Birmingham, and 4 from New York) were available for the study. At the time of inclusion, 26 women were in the first trimester of pregnancy, 11 in the second, and 1 in the third. Demographic data of the study population are shown in Table 1.

Table 1. Demographic data
Age, years, median (range)30.5 (19–41)
Disease duration, years, median (range)4.5 (1–14)
Race 
 White, n (%)28 (73)
 Black, n (%)5 (13)
 Hispanic, n (%)3 (8)
 Asian, n (%)1 (3)
 Indian, n (%)1 (3)

Activity scales

The LAI was originally defined and the limits for a lupus flare established in 1991 by Petri et al (9). Ruiz-Irastorza and Khamashta modified it with the name of LAI-P (10). Complete definitions of LAI-P have been published (10) and are shown in the Appendix (Copy available at the Arthritis Care & Research Web site at http://www.interscience.wiley.com/jpages/0004-3591:1/suppmat/index.html.) In brief, the original visual analog scale (VAS) was replaced by a graded scale for each item. Original vague items, such as “fatigue,” were withdrawn, and other more objective terms, such as “vasculitis,” “fever,” or “myositis,” were added. Items were divided into 4 groups (minor clinical features, major clinical features, changes in medication, and laboratory measures), the final score being the arithmetic mean of the values for each group. Of note, we decided to withdraw the physician assessment, which was the first item in the original LAI, to decrease the degree of subjectivity of LAI-P.

Manifestations related to antiphospholipid syndrome, such as cerebrovascular accidents or thrombocytopenia, were not scored in LAI-P in patients with antiphospholipid antibodies unless other signs of lupus activity were present.

A 0–2 VAS modified from Petri et al (9), the modified physician global assessment (m-PGA), was used as the gold standard to determine SLE disease activity (see below). The 3 possible values were 0 = no activity; 1 = mild-moderate activity; 2 = severe activity.

Study protocol

Baseline variables, including clinical manifestations of lupus and treatment at enrollment, were collected. Following the first visit, patients were seen according to the protocols of each center, which were not modified for the purposes of this study. At each visit, the m-PGA and LAI-P were completed. The LAI-P and m-PGA were recorded by independent investigators; those calculating m-PGA were involved in decisions regarding therapeutic changes. Laboratory determinations, including complete blood cell count, C3 and C4 levels, and anti-DNA antibodies, were performed at each center's laboratory on each visit. Urinalysis was also done on every visit, 24-hour protein excretion in urine being measured when the urine dipstick was positive for protein. Anti-DNA antibodies were detected by enzyme-linked immunoassay, Crithidia lucillae test, or Farr assay.

Definitions of lupus flare

The gold standard for SLE flare was an increase of m-PGA ≥ 1 from baseline values (i.e., from 0 to 1 or 2 or from 1 to 2). Based on previous works (9, 10), an increase of LAI-P ≥ 0.25 was defined as a flare.

Statistical analysis

This study aimed to validate LAI-P in 3 different ways: 1) showing its sensitivity to clinically relevant changes in lupus activity; 2) establishing the degree of association of LAI-P with m-PGA; and 3) calculating its accuracy as a tool to diagnose a lupus flare.

For all calculations, each visit was considered as an individual case, i.e., every patient could contribute more than 1 visit to the analysis. This approach has been used in recent articles addressing the utility of complement levels (14) and anti-DNA antibodies (15) as markers of SLE activity.

Sensitivity to changes in lupus activity was evaluated using the standardized response mean (SRM) (16). Baseline values of m-PGA and LAI-P were established for each patient, that is, the lowest scores found during all the followup. In every situation in which a flare occurred (i.e., a change in m-PGA of at least 1 point from baseline values), we calculated the difference between the values of LAI-P obtained at the same 2 visits. SRM was calculated as the ratio between the mean and the standard deviation of the differences. SRM values above 0.8 denote high sensitivity to change (16).

The association between m-PGA and LAI-P was established by performing a polytomous logistic regression using m-PGA as the dependent variable with 3 possible levels (0, 1, and 2). We analyzed m-PGA and LAI-P scores at visits 1, 2, 3, and 4, (n = 38, 38, 30, 19, respectively). The P value for the overall effect of LAI-P scores in the prediction of m-PGA values was obtained by the logistic likelihood ratio test with 2 degrees of freedom. The strength of association was measured by the statistic R2, which expresses the proportion of the variation explained by the regression model (17).

Finally, the utility of LAI-P as a diagnostic tool for a lupus flare was established by calculating its sensitivity, specificity, predictive values, and likelihood ratios (18). The positive likelihood ratio is defined as

  • equation image

and the negative likelihood ratio corresponds to

  • equation image

Likelihood ratios close to 1 imply low diagnostic power of a positive or negative test result. On the other hand, a positive likelihood ratio above 10 or a negative likelihood ratio below 0.1 greatly increases the posttest probability of having or not having, respectively, the studied condition—a lupus flare in this case (19).

Using the lowest score during the followup as the baseline value, the difference with the scores at each visit was calculated. This was confronted with the variation in m-PGA between the same 2 visits. Any difference of at least 0.25 points in LAI-P was considered a flare according to our working definition. A variation of at least 1 point in the m-PGA was considered the gold standard for SLE flare (see above). Thus, 4 possible situations could happen at each visit: 1) Change in m-PGA and flare according to LAI-P; true-positive result. 2) No change in m-PGA and no flare according to LAI-P; true-negative result. 3) No change in m-PGA and flare according to LAI-P; false-positive result. 4) Change in m-PGA and no flare according to LAI-P; false- negative result.

The sum of individual results was used to build a 2 × 2 table and to calculate the sensitivity, specificity, predictive values, and likelihood ratios of LAI-P.

Data filing and processing and statistical calculations were carried out using the package StatView, version 5.0.1 for Power Macintosh (SAS Institute Inc., Cary, NC). Sensitivity, specificity, predictive values, and likelihood ratios were calculated using the EBM calculator available at the web page of the Centre for Evidence Based Medicine (http://www.cebm.utoronto.ca/practise/ca/statscal.htm).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

Followup

The 38 women attended 158 visits during the study period (median 4 visits, range 2–8), for a total 621 patient-weeks.

Clinical status at the first visit

Thirty-three women were considered to be in remission at enrolment (m-PGA = 0). The remaining 5 patients presented with signs of lupus activity: 2 women with cutaneous vasculitis, 1 with skin rashes, 1 with arthritis and rash, and 1 with central nervous system involvement. Twenty patients (53%) were taking prednisolone at a mean dosage (SD) of 11 (5.3) mg/day. Twelve women (32%) were taking hydroxychloroquine and 11 (29%) azathioprine. Other medications included low-dose aspirin (25 women; 66%), heparin (8 patients; 21%) and nonsteroidal antiinflammatory drugs (1 patient; 2.6%).

Lupus flares

A flare of lupus, according to the gold standard (m-PGA increase ≥1), happened in 10 patients (26%), 3 of them having 2 flares. Flares consisted of cutaneous vasculitis (n = 4), arthritis (n = 3, 1 with skin rash), rash (n = 2), pleurisy (n = 2), nephritis (n = 1), and neuropsychiatric lupus (n = 1). Three flares happened in the first trimester, 6 in the second, and 4 in the third. As a whole, an m-PGA increase ≥1 over the baseline score happened in 15 visits because lupus activity lasted for 2 visits in 2 women.

Sensitivity to change

The SRM for LAI-P was 1.6. If we limit this analysis to 1 flare per patient (n = 10), the value of SRM was 1.7. Thus, LAI-P showed a high sensitivity to changes in lupus activity.

Association between LAI-P and m-PGA

Logistic regression found a significant association between LAI-P and m-PGA scores at visits 1, 2, 3, and 4 (P = 0.0002, 0.0003, 0.0014, and 0.0004, respectively). The respective R2 values obtained were 0.48, 0.66, 0.40, and 0.76.

Sensitivity, specificity, predictive values, and likelihood ratios

Fifteen visits were classified as SLE flares according to the variation in m-PGA. LAI-P correctly identified 14 of these flares (true-positive results) and failed to identify the remaining 1 flare (false-negative result). One hundred and five visits were considered no flares according to m-PGA. Of these, LAI-P correctly classified 103 visits (true-negative results) and identified as flares 2 visits (false-positive results). The resulting values for sensitivity, specificity, predictive values, and likelihood ratios of LAI-P are shown in Table 2.

Table 2. Sensitivity, specificity, predictive values, and likelihood ratios of LAI-P*
  • *

    LAI-P = Lupus Activity Index in Pregnancy; 95% CI = 95% confidence interval; PPV = positive predictive value; NPV = negative predictive value; LHR + = positive likelihood ratio; LHR − = negative likelihood ratio.

Sensitivity (95% CI)0.93 (0.89–0.98)
Specificity (95% CI)0.98 (0.96–1.00)
PPV (95% CI)0.88 (0.82–0.93)
NPV (95% CI)0.99 (0.97–1.00)
LHR +49
LHR −0.07

Description of false-negative and false-positive results

LAI-P failed to identify 1 flare in a woman who had subacute cutaneous lupus rash at 13 weeks that resolved with topical steroids. m-PGA increased 1 point from baseline and the LAI-P increase was 0.125.

LAI-P gave 2 false-positive results. The first 1 was an increase of 0.29 points in a woman having persistent hypocomplementemia associated with mild facial rash that did not require treatment. The second case was an increase in protein excretion more than double over baseline values in a woman with previous lupus nephritis and residual proteinuria that was not considered to be due to active lupus (m-PGA increase 0; LAI-P increase 0.42).

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

This study confirms that LAI-P is an accurate tool for measuring SLE activity and diagnosing lupus flares during pregnancy and the puerperium. This scale is sensitive to changes in disease activity (as measured by SRM) and has a strong association with m-PGA. Regarding the utility of LAI-P for diagnosing lupus flares, both sensitivity and specificity of LAI-P were well above 90%. This is reflected in the positive and negative likelihood ratios, which were higher than 10 and lower than 0.1, respectively.

One of the main difficulties in validating lupus activity scales is the lack of a good gold standard. Traditionally, the treating physician opinion, expressed in VAS and usually named PGA has been used to define the degree of lupus activity. Although this is rather subjective a gold standard—thus any result of validation studies using it must be taken with caution—it has been widely used (9, 20, 21). Indeed, no other available standards (patient global assessments, quality of life, etc.) have proved superior to PGA (16).

This limitation also applies to this work. In fact, because the opinion of the treating physician regarding activity usually has therapeutic consequences, there is a bias favoring scales that contain treatment items, such as the LAI-P (20). Specifically, minor increases in activity affecting only 1 organ that required changes in treatment were recognized as flares by LAI-P. False-positive results were also few, due in part to the modifications made in the original LAI to make unlikely diagnosing as lupus flare changes in anti-DNA and complement levels without any concurrent clinical manifestations. Interestingly, the utility of both immunologic tests as predictors of SLE flares has been put into question in 2 recent, well-designed studies (14, 15). Also, manifestations of APS, such as stroke and thrombocytopenia, cannot be labeled as lupus by LAI-P because its definitions specifically exclude these issues from scoring items (10). This is particularly relevant in pregnant patients, because combined lupus-pregnancy clinics receive a large number of women with APS due to obstetric complications.

The use of multiple visits per patient for some statistical calculations could have also biased our study. However, SRM values were not different when using 1 or more than 1 visit per woman, which is encouraging. Calculations of sensitivity, specificity, predictive values, and likelihood ratios may be influenced by the lack of independence of measurements. Thus, such results must be viewed as an intuitive and approximate guide of the usefulness and weakness of LAI-P, rather than as absolute values.

Almost three-quarters of women in our group were white. This may raise some doubts regarding the extension of this validation study to populations of other ethnic groups, specifically blacks, in whom lupus may be more severe. We believe that more florid lupus activity will be more easily detected by LAI-P without a concurrent decrease in specificity, thus this scale will probably be appropriate for black women with SLE.

Despite these limitations, this study represents a step forward in providing investigators a common and objective tool to unify observations regarding lupus activity during pregnancy. Indeed, 1 of our main aims was to recruit patients from several lupus-pregnancy clinics, so that the results would be more widely applicable than those from a single-center study. Although clinical management must be guided by observations beyond those recorded by any activity scale, the extended use of any of them will allow comparison of results between studies in a field in which heterogeneity in conclusions has been the rule in the past (7, 8).

Most authors today agree that women with less severe forms of lupus who have a good control of disease before conception are less likely to suffer SLE exacerbations during pregnancy and the puerperium (22). In fact, studies including unselected women with lupus (3–6) have usually found higher rates of flares than others in which women became pregnant with longstanding inactive SLE (23, 24). This was also an issue in this study, in which we observed a low proportion of women having flares (26%) as compared with older data from St. Thomas' Lupus Pregnancy Clinic (6). This reflected not only the incomplete followup of some of them (some women were enrolled late in pregnancy and those lost to followup were also included in the analysis), but also a better selection of the optimal time to become pregnant. However, determining the flare rate was not an objective of this work.

Although not validated yet, SLEPDAI has already been used in a study addressing the effect of hydroxychloroquine on SLE exacerbations in pregnant women (11). Future studies analyzing the role of other therapies, of clinical manifestations before pregnancy, of autoantibody profiles, or of the influence of hormonal changes will be able to measure SLE activity using LAI-P, which is supported by the results of this study.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES

The authors thank Stephanie Heaton, Lupus Specialist Research Nurse, for her contribution in the collection of data.

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. PATIENTS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. Acknowledgements
  8. REFERENCES