Systemic lupus erythematosus in three ethnic groups. XIX. Natural history of the accrual of the American College of Rheumatology criteria prior to the occurrence of criteria diagnosis

Authors


Abstract

Objective

To determine how the American College of Rheumatology (ACR) criteria for the classification of systemic lupus erythematosus (SLE) accrue in a multiethnic cohort of SLE patients.

Methods

SLE patients enrolled in a longitudinal study of outcome were analyzed (LUMINA; Lupus in Minorities: Nature versus nurture) for the manner in which ACR criteria manifestations occurred prior to the accrual of 4 of them. Time at which a criterion was said to be present was determined by review of all previously available medical records and interview. Univariable and multivariable Cox proportional hazard models were examined for the association with time to accrual of 4 ACR criteria; results were reported as hazard ratios.

Results

There were 103 Texas Hispanic (of Mexican or Central America ancestry) patients, 55 Puerto Rico Hispanics, 176 African Americans, and 137 Caucasians. The mean ± SD and median (range) time to accrual of 4 ACR criteria were 29.4 ± 52.0 months and 9.1 (0–328.7) months; time was shortest for the Texas Hispanics (18.4 ± 42.8 and 5.0 [0–248] months) and longest for the Caucasians (39.9 ± 59.3 months and 17.7 [0–324.6] months). Arthritis was the most frequent first criterion (34.5%); it was followed by photosensitivity (18.8%). When 2 criteria occurred from the outset, the most frequent combination was arthritis and antinuclear antibody positivity followed by malar rash and photosensitivity. A Cox-regression multivariable model identified Hispanic ethnicity (from Texas) and HLA–DRB1*0301 as predictors of short time to criteria accrual, whereas older age and married/living together were associated with long time to criteria accrual.

Conclusion

Significant variability in the evolution of ACR criteria manifestations does occur. Texas Hispanics are more likely to have a rapid evolution of criteria manifestations, but several years may elapse before ACR criteria are accrued.

INTRODUCTION

Systemic lupus erythematosus (SLE) is a disease affecting primarily women of reproductive age. It is a very heterogeneous disorder in terms of its presentation, course, and outcome. Patients from some ethnic groups (non-Caucasians by and large) not only develop SLE more frequently, but have a more aggressive course and poorer outcomes than Caucasians (1–5). However, the evolution of the disease from the onset of the first clinical manifestation clearly attributable to the disease to the time at which a patient may be considered to have SLE by defined and accepted criteria has not been determined in patients from different ethnic groups using longitudinal data (6, 7). We have taken advantage of a multiethnic cohort of patients with SLE of recent onset to examine the initial presentation of their disease.

PATIENTS AND METHODS

The LUMINA cohort (Lupus in Minorities: Nature versus nurture) is a multiethnic cohort of patients with SLE from the 3 main ethnic groups in the US: Caucasian, African American, and Hispanic. LUMINA patients were originally enrolled at 2 geographic areas (Alabama and Texas) at 3 institutions: the University of Alabama at Birmingham, the University of Texas Houston-Health Science Center, and the University of Texas Medical Branch at Galveston (2, 8). More recently, a third geographic area (Puerto Rico [PR]) has been added to LUMINA. Thus, LUMINA now includes a fourth institution (University of Puerto Rico Medical Sciences Campus) and a second Hispanic subgroup (PR Hispanics to distinguish them from the Hispanics from Texas [TX Hispanics]). The PR Hispanic patients have not been included in our previous full-length publications; these patients are, however, comparable to other Puerto Rican SLE patients included in other publications emanating from the island (9).

Patients with disease of ≤5 years' duration (from criteria diagnosis), of defined ethnicity (same as 4 grandparents), and living in the geographic area of the participating institutions or their associated practices were eligible to participate. Nearly all eligible patients were recruited into LUMINA. Prior to study enrollment, all available medical records were reviewed to confirm the patient's eligibility and to determine the precise time at which each criterion was first noted as per American College of Rheumatology (ACR) definitions (e.g., arthritis as documented by a physician). The enrollment visit included, in addition to the review of records, an interview, several self-administered questionnaires, a physical examination, phlebotomy, and urinalysis. Subsequent visits took place at 6 and 12 months, and yearly thereafter; in patients not returning for a followup visit, a review of all available medical records for the interval of interest was conducted.

Variables.

Variables obtained during all study visits were from the following domains: socioeconomic and demographic; clinical and immunologic; immunogenetic; and behavioral, psychological, and cultural. From the socioeconomic and demographic domain, the following variables were ascertained: age, sex, marital status, education, occupation, health insurance, income, unhealthy behaviors (including smoking, drinking, and not exercising), access to health care, and health care utilization. From the clinical and immunologic domain, the following variables were obtained: disease onset type; number and type of ACR criteria (10); clinical manifestations and their attribution (to lupus, to its treatment, or to neither); fatigue (per the Fatigue Severity Scale) (11); disease activity (per the Systemic Lupus Activity Measure) (12, 13); disease damage (per the Systemic Lupus International Collaborating Clinics Damage Index) (14); medications used; self-reported physical and mental functioning (per the Short Form 36) (15); and a panel of autoantibodies, including antinuclear antibodies (ANA), anti–double-stranded DNA, anti-Ro, anti-La, anti-Sm, and anti–U1 RNP. From the immunogenetic domain, HLA class II antigens were obtained. Finally, from the behavioral, psychological, and cultural domain, the following variables were obtained: social support (per the Interpersonal Support Evaluation List) (16), health-related behaviors (per the Illness Behavior Questionnaire) (17), and self efficacy for illness (per the corresponding section of the self-efficacy questionnaire) (18).

Statistics.

The number and type of ACR criteria as the initial manifestation of SLE; the time to the occurrence of 4 ACR criteria or time to diagnosis (TD); and the rate of 1, 2, or 3 criteria at initial presentation were examined for the entire cohort and then by ethnic group. Time to each criterion was determined from review of available records and interview. The relationship between the number and type of ACR criteria at initial presentation and different variables from the socioeconomic–demographic domain were then examined; differences between proportions were examined by chi-square analysis and differences between means were examined by analyses of variance or between medians by Kruskal-Wallis test for non-normally distributed data. Next, variables from the different domains were entered into Cox proportional hazard regression models in which the outcome variable was time to 4 ACR criteria. Variables with a P ≤ 0.20 in these univariable regressions were then entered into a multivariable Cox proportional hazard regression model to determine their independent contributions to time to accrue 4 ACR criteria. Results are reported as hazard ratios (HRs) where a value ≥1 indicates a shorter time to accrual of 4 ACR criteria and a value <1, a longer time.

RESULTS

A total of 471 LUMINA patients were included in these analyses. These patients included 103 TX Hispanics, 55 PR Hispanics, 176 African Americans, and 137 Caucasians. The socioeconomic–demographic and clinical features of the TX Hispanic, African American, and Caucasian patients included in these analyses are comparable to those of LUMINA patients included in previous publications (8, 19). The same applies to the PR Hispanics (9). Overall, the majority of the patients (90%) in all ethnic groups were middle age women (mean ± SD 37.6 ± 10.2 years) with a mean disease duration of about 18 months at the enrollment visit.

Number of ACR criteria as the initial manifestation of SLE.

More than half (55.4%) of the patients had only 1 ACR criterion as the initial manifestation of SLE; 20.0% had 2, 9.3% had 3, and 15.3% had 4 or more from the outset. As shown in Table 1, the distribution of the number of criteria at the outset was not uniform among patients from the 4 ethnic groups, with TX Hispanics having a greater proportion of patients with 4 or more criteria occurring concomitantly and for the first time at TD (25.2%). They were followed by the African Americans (15.3%). The PR Hispanics were those with the smallest proportion of 4 or more criteria present from the outset (5.5%). These differences were statistically significant (P = 0.0113).

Table 1. Frequency distribution of the number of initial ACR criteria as a function of ethnic group in LUMINA patients*
No. criteriaHispanicAfrican American (n = 176)Caucasian (n = 137)Total (n = 471)
Texas (n = 103)Puerto Rico (n = 55)
  • *

    Data given are number (%). ACR = American College of Rheumatology; LUMINA = Lupus in Minorities: Nature versus nurture. P = 0.0113 by chi-square analysis.

149 (47.6)35 (63.6)94 (53.4)83 (60.6)261 (55.4)
216 (15.5)9 (16.4)38 (21.6)31 (22.6)94 (20.0)
312 (11.7)8 (14.6)17 (9.7)7 (5.1)44 (9.3)
426 (25.2)3 (5.5)27 (15.3)16 (11.7)72 (15.3)

Time to 4 ACR criteria (TD).

The mean ± SD and the median (range) time to the development of 4 ACR criteria or TD were 29.4 ± 52 months and 9.1 (0–328.7) months for the entire cohort. However, the duration was shorter for the TX Hispanics (18.4 ± 42.8 months and 5.0 [0–248.0] months) compared with patients in the other ethnic groups, including African Americans, PR Hispanics, and Caucasians. Caucasians had the longest time to 4 ACR criteria (39.9 ± 59.3 months and 17.7 [0–324.6] months). The differences between the medians for these 4 ethnic groups were statistically significant (P < 0.0001). These data are shown in Table 2. As expected, the time to the occurrence of 4 ACR criteria was shorter for those patients who presented with 3 ACR criteria than those who initially presented with 2 or 1 criteria. These data are shown in Table 3.

Table 2. Time from initial ACR criteria to diagnosis as a function of ethnic group in LUMINA patients*
 HispanicAfrican American (n = 176)Caucasian (n = 137)Total (n = 471)
Texas (n = 103)Puerto Rico (n = 55)
  • *

    ACR = American College of Rheumatology; LUMINA = Lupus in Minorities: Nature versus nurture.

  • P < 0.0001 by Kruskal-Wallis test.

Time     
 Mean ± SD months18.4 ± 42.829.7 ± 40.627.6 ± 55.339.9 ± 59.329.4 ± 52.0
 Median (range) months5.0 (0–248.0)9.2 (0–183.6)7.1 (0–328.7)17.7 (0–324.6)9.1 (0–328.7)
Table 3. Time from initial ACR criteria to diagnosis as a function of the number of criteria present in LUMINA patients*
No. of initial ACR criterianTime to diagnosis (months)
Mean ± SDMedianRange
  • *

    For abbreviations, see Table 2.

126139.0 ± 58.812.20.9–328.7
29432.3 ± 57.210.20.9–324.6
34414.6 ± 57.25.11.0–81.1
472000

Most frequent initial ACR criteria.

Arthritis was the most frequently experienced initial manifestation of SLE prior to its criteria diagnosis (34.5%), followed by photosensitivity (18.8%) and ANA positivity (14.2%; data not shown). When 2 ACR criteria were present initially, the most common combination was arthritis and ANA positivity (20.2%), followed by malar rash and photosensitivity (16.0%; data not shown). No definite pattern was observed when 3 criteria were present from the outset.

Table 4 shows the distribution of the different ACR criteria by ethnic group as a function of the number of criteria present at the outset. For example, when only 1 criterion was present, arthritis was more likely in the TX Hispanics and African Americans than in the Caucasians and PR Hispanics (P = 0.0103), whereas discoid rash was more common in the African Americans than in patients from any of the other ethnic groups. When 2 criteria were present, malar rash as well as photosensitivity were distinctly more common among the Caucasians. When 3 criteria were present, again discoid rash was present only in the African Americans, whereas arthritis was more frequent among the TX Hispanics and Caucasians. Finally, when 4 criteria were present from the outset, renal, hematologic, and immunologic criteria were much more frequent among the TX Hispanics and African Americans.

Table 4. Frequency distribution of selected ACR criteria by ethnic group when only a given number of criteria are present*
Number and type of criteriaHispanicAfrican AmericanCaucasianP
TexasPuerto Rico
  • *

    Only those criteria achieving a P ≤ 0.05 are shown. ACR = American College of Rheumatology.

One criterion, no.49359483 
 Discoid rash2.02.96.400.0452
 Photosensitivity12.251.45.324.1< 0.0001
 Mucosal ulcers6.102.115.70.0007
 Arthritis42.922.943.624.10.0103
Two criteria, no.1693831 
 Malar rash25.0010.545.20.0011
 Photosensitivity12.533.313.254.90.0007
Three criteria, no.128177 
 Discoid rash0029.400.0144
 Arthritis7525.035.371.40.0449
Four criteria, no.2632716 
 Renal34.6051.96.30.0041
 Hematologic57.7059.325.00.0200
 Immunologic80.833.381.537.50.0064
Total number of patients10355176137 

In contrast, Table 5 shows the frequency distribution of each of the 11 ACR criteria as the initial criterion among the 4 ethnic groups regardless of how many criteria were present at the outset. Given this consideration, the sum of the percentages for each of the ethnic categories exceeds 100, as a patient may be counted up to 4 times depending on how many criteria were present concomitantly at the outset. The data in this Table are entirely consistent with those presented in Table 4; that is arthritis, renal, hematologic, and immunologic criteria occur more frequently initially among the TX Hispanics and African Americans, whereas mucosal ulcerations and malar rash occur more frequently at the outset among the Caucasians and photosensitivity among the PR Hispanics.

Table 5. Frequency distribution of each of the 11 ACR criteria as first criterion among patients from 4 ethnic groups*
CriteriaHispanicsAfrican American (n = 176)Caucasian (n = 137)P
Texas (n = 103)Puerto Rico (n = 55)
  • *

    Regardless of how many criteria are present. ACR = American College of Rheumatology; ANA = antinuclear antibodies.

  • Only P values ≤ 0.10 are shown.

Malar rash18.59.114.822.60.0910
Discoid rash1.95.510.22.20.0044
Photosensitivity17.550.911.932.9< 0.0001
Mucosal ulcerations14.67.35.716.10.0101
Arthritis54.425.546.632.90.002
Serositis20.45.514.811.70.0449
Seizures/psychosis2.903.43.7 
Renal13.6012.52.9< 0.0001
Hematologic17.57.316.57.30.0198
Immunologic27.218.221.68.80.0011
ANA45.634.640.930.70.0408

Univariable analyses.

Socioeconomic and demographic variables were examined for time to the accrual of 4 ACR criteria (Table 6). Hispanic (from Texas) ethnicity was associated with the shortest time to 4 ACR criteria (HR = 1.34, 95% confidence interval [95% CI] 1.063–1.696, P = 0.013), whereas Caucasian ethnicity was associated with the longest time (HR = 0.780, 95% CI 0.630–0.964, P = 0.022). Older age and being married/living together were negatively associated with the shortest time to 4 criteria (HR = 0.985, 95% CI 0.977–0.992, P < 0.001 and HR = 0.730, 95% CI 0.607–0.878, P = 0.001, respectively). Education and income were negatively associated with shortest time to criteria accrual, although these variables were less significant. Sex, unhealthy behaviors, and access to health care for lupus were not associated with a short time to the accrual of 4 ACR criteria. In addition to the sociodemographic and economic variables, autoantibodies not considered ACR criteria (anti-Ro, anti-La, and anti-RNP) and selected HLA–DR and DQ specificities were examined. Only HLA–DRB1*08 was found to be associated with short time to accrual of 4 ACR criteria.

Table 6. Time to accrual of 4 ACR criteria in LUMINA patients by univariate analyses*
VariableParameter estimateChi-squarePHazard ratio95% CI
  • *

    A hazard ratio ≥1 indicates a shorter time to accrual of 4 ACR criteria and a value <1 otherwise. ACR = American College of Rheumatology; LUMINA = Lupus in Minorities: Nature versus nurture; TX = Texas; 95% CI = 95% confidence interval.

  • Only P values ≤ 0.05 are shown.

Hispanic (TX) ethnicity0.2946.0920.0141.3421.063–1.696
Caucasian ethnicity−0.2495.2610.0220.7800.630–0.964
Age−0.01615.755< 0.0010.9850.977–0.992
Married/living together−0.31411.178< 0.0010.7300.607–0.878
Education−0.0365.8900.0150.9650.937–0.993
Income−0.0605.6040.0180.9480.908–0.991
Access to medical care0.2004.0450.0431.2211.005–1.483
HLA–DRB1*080.3074.130.0421.3601.011–1.829

Multivariable analyses.

As depicted in Table 7, TX Hispanic ethnicity (HR = 1.541, 95% CI 1.194–1.989, P < 0.001) and HLA–DRB1*0301 (HR = 1.330, 95% CI 1.027–1.723, P = 0.031) were positively associated with shortest time to the accrual of 4 criteria. Older age (HR = 0.990, 95% CI 0.981–0.999, P = 0.034) and married/living together (HR = 0.727, 95% CI 0.573–0.923, P = 0.031) were negatively associated with a short time to accrual of 4 ACR criteria.

Table 7. Time to accrual of 4 ACR criteria in LUMINA patients by multivariate analyses*
VariableParameter estimateChi-squarePHazard ratio95% CI
  • *

    A hazard ratio value ≥1 indicates an increased risk for a short time to accrual of 4 ACR criteria and a value <1, otherwise. TX = Texas; for additional abbreviations, see Table 6.

  • Only P values ≤ 0.05 are shown.

Hispanic (TX) ethnicity0.43211.0070.0011.5411.194–1.989
Age−0.0104.4940.0340.9900.981–0.999
Married/living together−0.3186.8870.0090.7270.573–0.923
HLA–DRB1*03010.2854.6640.0311.3301.027–1.723

DISCUSSION

As noted, SLE is a chronic disorder that is variable in its onset, course, and outcome. This heterogeneity extends to the type of clinical manifestations at disease onset and to the time it may take for these manifestations to evolve and make the diagnosis of lupus possible (20–22). The diagnosis of SLE may be delayed when the initial manifestations are atypical, few in number, or when the manifestations present are not the ones included in the ACR criteria, even though they occur frequently in patients with SLE (23, 24). In fact, patients who do not satisfy ACR criteria for the diagnosis of SLE are considered by some to have “incomplete” (25–27), “subclinical,” “mild,” “latent” (28), or “variant” (29) SLE. It is always possible, however, for these patients to develop SLE and satisfy the ACR criteria (25, 28, 30, 31), but this has not been systematically studied.

We have now studied the pattern of ACR criteria accrual in the LUMINA patients. Interesting results have emerged; the time to 4 ACR criteria, and hence to diagnosis of SLE, may extend to longer than 25 years, although on the average it extended to 2.5 years. The time was shorter for the TX Hispanics than for PR Hispanics and for patients in the other 2 ethnic groups. This may suggest that Amerindian genes may impact the rate of SLE development, if indeed the proportion of these genes is higher in the TX Hispanics than in the PR Hispanics (32). Given the fact that the original inhabitants of the island of Puerto Rico were decimated during the Spanish conquest, but not so the ones of Mexico, this is entirely possible; however, proof remains to be found (33). Although we found access to care, health care utilization, health insurance, and most other socioeconomic variables studied not to be associated with time to accrual of criteria, we cannot rule out the possibility that there is an unmeasured “center” effect. Thus, it is possible than the shorter time to the accrual of 4 ACR criteria among the TX Hispanics than among patients from the other groups relates to differences in the health care system in the continental US and the island of Puerto Rico. However, given that nearly all PR Hispanic patients but only 45% of TX Hispanics were insured, we would have expected the opposite results, that is PR Hispanics having a shorter time to criteria accrual than the TX Hispanics.

A relatively small proportion of our LUMINA patients developed 4 or more criteria within a very short time (a month or less); this proportion was higher among the TX Hispanics. These data suggest that if those patients who will develop SLE rapidly can be promptly identified, treatment can be provided even before criteria diagnosis has occurred or is feasible. Arthritis was the most frequent isolated criterion manifestation in all patients, although photosensitivity was clearly a “first” criterion more frequently in the PR Hispanics and Caucasians than among the TX Hispanics and African Americans. Discoid rash was more frequent among the African Americans. Renal, immunologic, and hematologic manifestations fulfilling their corresponding ACR criteria were more frequent among the TX Hispanics and the African Americans. So, from the outset, a pattern of disease emerges with more severe manifestations present in the African Americans and TX Hispanics than in the Caucasians and PR Hispanics. The pattern of these early events parallels the less favorable course and final outcome of lupus that we, and others, have observed among non-Caucasian patient groups (30, 34).

In trying to identify which patients are more likely to have a shorter time to the accrual of 4 ACR criteria, several potential variables were identified in univariable analyses. Hispanic (from Texas) ethnicity and HLA–DRB1*0301 remained significant and were positively associated with a short time to criteria accrual in the multivariable analyses, whereas age (older) and marital status (married/living together) were negatively associated with a short time to the accrual of 4 ACR criteria. HLA–DRB1*0301 has been found to be associated with the occurrence of SLE primarily in Caucasians (35–37) and to a lesser extent in TX Hispanics (38). Given that these 2 groups exhibited the longest and shortest time to criteria accrual, this association is difficult to interpret. Moreover, in contrast with all other variables examined in which the multivariate analyses are supported by the univariable analyses, that was not the case for this variable. In fact, in the univariable analyses, HLA–DRB1*08 was significantly associated with the shortest time of criteria accrual whereas HLA–DRB1*0301 was only of borderline significance.

Of course it can be argued that the timing of the different criteria, as recorded in our study, is a bit artificial because patients may have had manifestations that are not clinically apparent and that only get documented when other manifestations are more evident and patients seek medical care. Physicians seeing patients with some criteria manifestations may be more likely to order serologic tests to confirm or rule out the possibility of SLE in some clinical situations (new onset renal disease, for example) than in others. There is, however, no reason to believe that there has been a systematic error in the gathering of the prediagnostic data in relation to ethnicity, age, marital status, HLA–DRB1 status, and other patient characteristics.

In conclusion, the variability of SLE extends to the rapidity with which the disease manifests itself before its diagnosis can be made. The disease may appear to evolve over a period of a few days to many years. Arthritis and photosensitivity may precede the onset of other criteria manifestations of SLE by years, a fact of which clinicians need to be aware. On the other hand, young individuals, particularly those of TX Hispanic ethnicity, those not married/living together, and perhaps those possessing HLA–DRB1*0301 are more likely to rapidly develop lupus. Other patient characteristics (such as age, education, income) may also influence how the disease appears to evolve, but certainly these characteristics appear to be less influential. Health care providers caring for Hispanic patients (of Mexican/Central American descent) need to closely monitor patients with clinical manifestations suggestive of SLE.

Acknowledgements

Thanks to our patients without whom this study would not have been possible and to Ella Henderson, AA, for her most expert technical assistance in the preparation of this manuscript.

Current LUMINA investigators and staff:

At the University of Alabama at Birmingham: Graciela S. Alarcón, MD, MPH, Holly M. Bastian, MD, MSPH, Barri J. Fessler, MD, Gerald McGwin, Jr, MS, PhD, Jeffrey Roseman, MD, PhD, MPH, Martha L. Sanchez, MD, MPH, Ellen Sowell, AA, Sergio M. A. Toloza, MD, and América G. Uribe, MD.

At the University of Texas-Houston Health Science Center: John D. Reveille, MD, Alan W. Friedman, MD, Chul Ahn, PhD, Robert Sandoval, BA, and Li-Lun Wang, BS.

At the University of Texas Medical Branch at Galveston: Bruce A. Baethge, MD, and Sonia Hunnicutt, BS.

At the University of Puerto Rico Medical Sciences Campus: Luis M. Vilá, MD, Carmine Pinilla, BS, and William Borges, AA.

Ancillary