and rheumatism arthritis THE 1982 REVISED CRITERIA FOR THE CLASSIFICATION OF SYSTEMIC LUPUS ERYTHEMATOSUS SPECIAL ARTICLE

Transcription

1 arthritis and rheumatism * Official Journal of the American Rheumatism Association Section of the Arthritis Foundation SPECIAL ARTICLE THE 1982 REVISED CRITERIA F THE CLASSIFICATION OF SYSTEMIC LUPUS ERYTHEMATOSUS ENG M. TAN, ALAN S. COHEN, JAMES F. FRIES, ALFONSE T. MASI, DENNIS J. McSHANE, NAOMI F. ROTHFIELD, JANE GREEN SCHALLER, NMAN TALAL, and ROBERT J. WINCHESTER The 1971 preliminary criteria for the classification of systemic lupus erythematosus (SLE) were revised and updated to incorporate new immunologic knowledge and improve disease classification. The 1982 revised criteria include fluorescence antinuclear antibody and antibody to native DNA and Sm antigen. Some criteria involving the same organ systems were aggregated into single criteria. Raynaud s phenomenon and alopecia were not included in the 1982 revised criteria because of low sensitivity and specificity. The new criteria were 96% sensitive and 96% specific when tested with SLE and control patient data gathered from 18 participating clinics. When compared with the 1971 criteria, the 1982 revised criteria showed gains in sensitivity and specificity. In 1971 a subcommittee of the American Rheumatism Association (ARA) published a report on The From the Subcommittee for Systemic Lupus Erythematosus Criteria of the American Rheumatism Association Diagnostic and Therapeutic Criteria Committee. Developed with the assistance of Donald Young. Eng M. Tan, MD: Chairman of Subcommittee for Systemic Lupus Erythematosus Criteria, Scripps Clinic and Research Foundation, La Jolla; Alan S. Cohen, MD: Boston City Hospital, Boston; James F. Fries, MD: Stanford University School of Medicine, Stanford; Alfonse T. Masi, MD, Dr. PH: University of Illinois College of Medicine, Peoria; Dennis J. McShane, MD: Stanford University, Palo Alto; Naomi F. Rothfield, MD: University of Connecticut School of Medicine, Farmington; Jane Green Schaller, MD: University of Washington, Seattle; Norman Talal, MD: VA Medical Center, San Francisco; Robert J. Winchester, MD: Hospital for Joint Diseases, New York. Address reprint requests to American Rheumatism Association, 1314 Spring Street, Atlanta, GA Submitted for publication June 21, 1982; accepted June 22, Preliminary Criteria for the Classification of Systemic Lupus Erythematosus (1). These first criteria for the classification of patients with this disease were widely accepted and have had broad impact (2-6). They have been used as the basis for classification of patients in many clinical reports since that time. Since the data upon which these criteria were based were collected in the early 1960s, the preliminary criteria did not incorporate serologic tests currently widely used in diagnosis and management of systemic lupus erythematosus (SLE). These important disease markers include antibody to DNA, fluorescence antinuclear antibody (FANA), serum complement, and other serologic and immunopathologic assays. In addition, new computer-based techniques for criteria development now allow more versatility in comparison of alternatives, utilization of training and test populations (so the criteria are not tested against the same patients upon whom they were developed), and comparison against additional computer-stored patient data sets of high quality. With these considerations Dr. Daniel J. McCarty, president of the American Rheumatism Association in 1979, appointed a subcommittee of the Diagnostic and Therapeutic Criteria Committee. This subcommittee was charged with the updating and revision of the 1971 criteria. METHODS The ARA Subcommittee for Systemic Lupus Erythematosus Criteria consisted of the following individuals: Eng M. Tan (Chairman), Alan S. Cohen, James F. Fries, Alfonse T. Masi, Dennis J. McShane, Naomi F. Rothfield, Jane Green Schaller, Norman Talal, and Robert J. Winchester. The Subcommittee first met in 1979 with the task of Arthritis and Rheumatism, Vol. 25, No. 11 (November 1982) 1271

2 1272 TAN ET AL defining those individual variables which might later become part of a final criteria set. The original criteria were carefully reviewed, and where necessary, the definition of terms was modified to improve clarity and reduce ambiguity. New potential criteria variables, including the most widely used serologic tests and histologic examination of skin and kidney were added. Thirty potential criteria variables (including each of the original 1971 criteria) were subjected to critical study. The study design included a comparison of these variables in SLE patients and matched controls. Eighteen investigators representing major clinics were solicited to contribute patient report forms. The forms noted the presence or absence of each variable at the time of examination, or at any time in the past. Ten consecutive patients were to be reported by each investigator, enabling this SLE study population to be broadly representative of SLE patients at major institutions in the United States. For each SLE patient, the next age- (within 10 years), race-, and sexmatched patient with a nontraumatic, nondegenerative connective tissue disease seen at that hospital or clinic was to be similarly analyzed. The Subcommittee obtained reports on 177 patients with SLE and 162 control patients from 18 institutions.* SLE patients tended to be younger than the controls, but in general, a good match was obtained. Data were entered into ARAMIS (American Rheumatism Association Medical Information System) computer data banks at Stanford and stored on an IBM 370/3081 computer. Data accuracy was ensured by range checks, checks of original forms against computer data printouts and, when necessary, checks of outlying information against original data at the home institution. Data analysis was carried out using ARAMWTOD (Time-Oriented Databank) software. Cluster analysis and other multivariate techniques were used to analyze relationships among variables. The sensitivity and specificity of individual criteria were tabulated and analyzed for clinically and statistically significant differences, using chi square contingency table analysis. Approximately 30 combinations of variables (using aggregation techniques which clustered variables representing a single organ system) were created and tested for sensitivity and specificity. Combinations of individual and aggregated variables were then examined as potential criteria sets. Tests of potential criteria sets were performed on a training (exploratory) random subset of patients from case and control groups which consisted of two-thirds of the patients: 117 SLE patients and 108 con- *Contributing investigators and institutions were: Gene Ball, University of Alabama; Peter Barland, Montefiore Hospital; Eugene V. Barnett, University of California at Los Angeles; Carolyn Brunner, University of Virginia; John L. Decker, National Institutes of Health; James F. Fries, Stanford University; Marvin D. Fritzler, University of Calgary; Ellen M. Ginzler, SUNY Downstate Medical Center; Bevra Hahn, Washington University; Robert W. Lightfoot, Medical College of Wisconsin; Alfonse T. Masi, University of Illinois College of Medicine at Peoria; Naomi F. Rothfield, University of Connecticut; Gordon C. Sharp, University of Missouri; Jane Green Schaller, University of Washington; Mary Betty Stevens, Johns Hopkins University; Eng M. Tan, University of Colorado; John B. Winfield, University of North Carolina; and Nathan J. Zvaifler, University of California at San Diego. trols. Tests of the final contending criteria combinations were performed on the test (confirmatory) set of patients, made up of the remaining one-third of the patient data base, and against 172 patients with SLE, 299 with scleroderma, and 119 with dermatomyositis from the databank of the Scleroderma Criteria Cooperative Study (SCCS) (7). RESULTS The sensitivity and specificity of individual variables are shown in Table 1. At this stage, it was possible to reduce the criteria set by inspection and consensus, eliminating variables which were seldom performed or observed in the test and control populations, or which did not reach statistical significance at the level. Biopsies of skin and kidney were eliminated because they were seldom performed; Raynaud s phenomenon, because its combined sensitivityhpecificity was not statistically significant. Table 1 also presents the values reported for the same variables in the original study population used for development of the 1971 criteria. There is close agreement between the frequency of positivity of individual variables in the present population and in the earlier one. Certain additional features are of interest. The lupus erythematosus (LE) cell criterion had decreased in sensitivity from 92% (1971) to 73% (1982) but maintained a high degree of specificity, 96% (1982) and 98% (1971). It was also noted that of the total of 177 patients, the test was performed in 79 (45%), possibly reflecting the fact that the LE cell test is a laborintensive assay and that other serologic tests were being used in its place. This might be fairly representative of current practice, since the patient database was obtained from 18 participating clinics scattered across the country. The Subcommittee performed these analyses at the ARAMIS facilities at Stanford University. The final criteria set is presented in Table 2. This consists of 11 criteria; as with the 1971 criteria, a patient must fulfill 4 or more of the criteria to be classified as having systemic lupus erythematosus. Table 3 lists the sensitivity and specificity of each criterion in the final 1982 criteria set, when tested against the current patient database. It is noted that overall, there is 96% sensitivity and 96% specificity. Accuracy similar to that of the 1982 criteria (Table 2) could also be obtained by using fluorescent antinuclear antibody (FANA) as an entry criterion and a long list of 19 variables as individual criteria, but the resulting criteria set was deemed too cumbersome for

3 REVISED CRITERIA F SLE 1273 Table 1. Comparison of sensitivity and specificity of elements in the 1982 and 1971 patient database Sensitivity* Specificity? Element I Malar rash Discoid rash Alopecia Photosensitivity Oral ulcers Raynaud s Arthritis Proteinuria Urinary casts Dementia Seizures Coma Psychosis Focal neurologic Pleurisy Pericarditis Hemolytic anemia Leukopenia Thromboc ytopenia LE cells Sm antibody Serologic text for syphilis Renal biopsy Skin biopsy Antinuclear antibody DNA antibody CH50 c3 c4 c2 101/177 (57) 31/177 (18) 99/177 (56) 76/176 (43) 47/177 (27) 51/176 (29) 152/177 (86) 89/177 (50) (36) 11/177 (6) 21/177 (12) 8/177 (5) 22/176 (13) 21/177 (12) (52) 31/177 (18) 31/176 (18) 82/177 (46) 37/177 (21) 58/79 (73) 34/108 (31) 19/129 (15) 57/69 (83) 47/69 (68) 174/175 (99) 113/168 (67) 84/120 (70) (64) (64) 156/162 (96) 161/162 (99) (88) 155/162 (96) 155/162 (96) 132/162 (81) (37) 148/157 (94) (97) 160/162 (99) 160/162 (99) I62/162 ( 100) 161/162 (99) 15Y161 (96) (89) 155/162 (96) 160/161 (99) 144/161 (89) 160/161 (99) 46/48 (96) 59/62 (95) BOB0 (100) 10/10 (100) 13/16 (81) 68/139 (49) 84/91 (92) 33/47 (70) 69/76 (91) 33/51 (65) o/o o/o * Sensitivity was determined on the SLE population and is expressed as the number of patients who were positive or abnormal for the element over the number of patients in whom the test was determined. Numbers in parentheses indicate percentages. Data from 1971 were obtained from Cohen et al (I), and only information on percentages was available. t Specificity was determined on the control population and is expressed as the number of patients who were negative or normal for the element over the number of patients in whom the test was determined. Z NA = data not available from the 1971 criteria study. teaching or for practical application. All other criteria sets analyzed were less accurate. When designing criteria sets, it is possible to trade off false positives for false negatives by changing combinations of variables or by changing the number of individual criteria required. Analyses did not reveal any criteria sets which offered significant reductions in either false negatives or false positives over the chosen set, and we could determine no advantage to a different threshold level (for example, 3 or 5 criteria) than that selected. More than 300 separate computer analyses were performed, and more than a dozen final potential criteria sets were carefully analyzed. The strongest addition to the criteria set was clearly the fluorescence antinuclear antibody test. FANA was positive in some titer and at some point in the course of disease in 174 of 175 SLE patients in whom this test was done, emphasizing its nearly universal use in practice. One patient had never had a positive value on this test. On the other hand, this FANA criterion was relatively nonspecific, being positive in 51% of the controls. FANA is the only feature of SLE to be almost universally positive, and thus it could potentially operate as an entry or required criterion in a final criteria set. Application of this single criterion rejects nearly one-half of the control population while retaining essentially all SLE patients. One Subcommittee goal was to simplify the final criteria set as much as possible without losing accuracy. An approach to this objective was to aggregate criteria into organ system variables. For example, hemolytic anemia, leukopenia, lymphopenia, and thrombocytopenia were considered as 4 variables within the hematologic system, and the presence of

4 1274 TAN ET AL Table 2. The 1982 revised criteria for classification of systemic lupus erythematosus* Criterion 1. Malar rash 2. Discoid rash 3. Photosensitivity 4. Oral ulcers 5. Arthritis 6. Serositis 7. Renal disorder 8.Neurologic disorder 9. Hematologic disorder Definition Fixed erythema, flat or raised, over the malar eminences, tending to spare the nasolabial folds Erythematous raised patches with adherent keratotic scaling and follicular plugging; atrophic scarring may occur in older lesions Skin rash as a result of unusual reaction to sunlight, by patient history or physician observation Oral or nasopharyngeal ulceration, usually painless, observed by a physician Nonerosive arthritis involving 2 or more peripheral joints, characterized by tenderness, swelling, or effusion Pleuritis-convincing history of pleuritic pain or rub heard by a physician or evidence of pleural effusion Pericarditis-documented by ECG or rub or evidence of pericardial effusion Persistent proteinuria greater than 0.5 grams per day or greater than 3+ if quantitation not peformed Cellular casts-may be red cell, hemoglobin, granular, tubular, or mixed Seizures-in the absence of offending drugs or known metabolic derangements; e.g., uremia, ketoacidosis, or electrolyte imbalance Psychosis-in the absence of offending drugs or known metabolic derangements, e.g., uremia, ketoacidosis, or electrolyte imbalance Hemolytic anemia-with reticulocytosis Leukopenia-less than 4,000/mm3 total on 2 or more occasions Lymphopenia-less than 1,500/mm3 on 2 or more occasions Thrombocytopenia-less than 100,000/mm3 in the absence of offending drugs 10. Immunologic disorder 11. Antinuclear antibody Positive LE cell preparation Anti-DNA: antibody to native DNA in abnormal titer Anti-Sm: presence of antibody to Sm nuclear antigen False positive serologic test for syphilis known to be positive for at least 6 months and confirmed by Treponerna pallidurn immobilization or fluorescent treponemal antibody absorption test An abnormal titer of antinuclear antibody by immunofluorescence or an equivalent assay at any point in time and in the absence of drugs known to be associated with drug-induced lupus syndrome * The proposed classification is based on 11 criteria. For the purpose of identifying patients in clinical studies, a person shall be said to have systemic lupus erythematosus if any 4 or more of the 11 criteria are present, serially or simultaneously, during any interval of observation.

5 REVISED CRITERIA F SLE 1275 Table 3. Sensitivitv and sdecificitv of 1982 SLE criteria* Sensitivity Specificity SLE Control (177 ( criteria patients) patients) Malar rash 101/177 (57) 156/162 (96) Discoid rash 31/177 (18) 161/162 (99) Photosensitivity 76/176 (43) 155/162 (96) Oral ulcerations 47/177 (27) 155/162 (96) Arthritis 152/177 (86) 60/162 (37) Serositis 99/177 (56) 139/162 (86) Renal disorder 91/177 (51) 147/157 (94) Neurologic disorder 35/177 (20) 159/162 (98) Hematologic disorder 105/177 (59) 144/162 (89) Immunologic disorder 149/176 (85) 113/122 (93) Antinuclear antibody 174/175 (99) 68/139 (49) Application of 1982 criteria requirements 170/177 (96) 155/162 (96) * Definitions of sensitivity and specificity and explanations of data expressed in numbers are given in footnotes to Table 1. 1 or several variables was calculated as 1 criterion representing the hematologic system. This method of analysis was tested against the alternative analysis using each of the 4 variables as an independent criterion. After repeated computer analyses, the organ system approach was as accurate as the independent variable approach and could be applied to every organ system except the integument. Here, 4 variables (malar rash, discoid lupus, photosensitivity, and oral ulcers) each made an independent contribution to the accuracy of the final criteria set, and attempts to aggregate these variables resulted in loss of accuracy. Table 4 lists the diagnoses of patients in the control group. As in the control group for the 1971 criteria, the principal control diagnosis was rheumatoid arthritis (RA). False positives were similarly distributed between RA and non-ra controls. Because of the paucity of non-ra controls, it was deemed advisable to study additional populations of patients having other diagnoses. The Subcommittee for Scleroderma Criteria (7) of the American Rheumatism Association recently amassed a database of 590 patients having SLE, scleroderma, or dermatomyositis/polymyositis. These computer-stored data were available through ARA- MIS and had the advantage that each of the original 1971 criteria for SLE was contained in the data set. An acknowledged difference in data acquisition was that 2 variables in the 1982 revised criteria set (anti-dna and anti-sm) were not supplied by the individual cooperating clinics, but the analyses were performed on stored sera by selected laboratories. In spite of this, the Table 4. SLE criteria study-control population diagnosis No. of * patients Diagnosis 95 Rheumatoid arthritis 16 Scleroderma 7 Juvenile onset arthritis 6 Dermatomyositis 5 Ankylosing spondylitis 5 Psoriatic arthritis 4 Reiter s syndrome 3 Osteoarthritis 3 Mixed connective tissue disease 2 Diagnosis not specified 2 Sjogren s syndrome 2 Polyarthritis 2 Raynaud s syndrome 10 1 each of Behcet s syndrome; type B-I1 hyperlipidemia; reflex sympathetic dystrophy syndrome; Wegener s granulomatosis; vasculitis; regional enteritis; discoid lupus; undifferentiated connective tissue disease; chronic active hepatitis; multicentric reticulocytosis SCCS databank provided a unique resource of separately collected patient data which could be used to test the performance of the criteria. When the 1982 SLE criteria were tested against the scleroderma criteria patient population, sensitivity to the SLE patients was 83% and specificity against the combined scleroderma and dermato/polymyositis patients was 89%. When tested against the 1971 SLE criteria, sensitivity was 78% and specificity, 87%. These results are presented in Table 5. Thus with the limitations noted above, the 1982 criteria set showed gains in sensitivity and specificity over the 1971 preliminary criteria. The current patient database was examined for false positive patients, using both sets of criteria. As can be seen in Figure 1, the distributions of patients in the false positive category were very similar. There were 7 false positive patients using 1982 criteria and 8 using 1971 criteria. Six patients with diagnoses shown in Figure 1 were false positive by both criteria. One patient with RA was false positive by 1982 criteria Table 5. Comparison of sensitivity and specificity of 1982 and 1971 criteria using scleroderma criteria databank 1982 criteria 83% 1971 criteria 78% Specificity: Sensitivity: 299 scleroderma 172 SLE 119 dermato/polvmvositis 89% 87%

6 1276 TAN ET AL Figure 1. Venn diagram showing diagnoses of patients classified as false positive by the 1982 and 1971 criteria. Circle on right represents patients fulfilling 1982 criteria and that on left, those fulfilling 1971 criteria. PT = patient; PSS = progressive systemic sclerosis; PM = polymyositis, RA = rheumatoid arthritis, MCTD = mixed connective tissue disease. only. A different patient with RA and a patient with sclerodermdpolymyositis were false positive by the 1971 criteria only. A similar analysis was performed to analyze the false negative patients, i.e. patients considered to have SLE by the participating clinics but who did not fulfill classification criteria. There were 7 false negative patients by the 1982 criteria and 22 by the 1971 criteria (Table 6). These data make it apparent that the number of SLE patients who escape correct classification is greatly reduced by the 1982 criteria. DISCUSSION The Subcommittee looked closely at redundancy of variables, variables which had poor discriminating power between populations, and those which had higher frequencies in some control populations than in SLE patients. These analyses indicated that Raynaud s phenomenon should be deleted as a criterion. Alopecia also fell into this category, because it did not perform well in distinguishing SLE from scleroderma and dermato/polymyositis (data not shown). Additionally, it proved possible to group proteinuria and cellular casts into a single variable, representing clinical renal involvement, and to place false positive serologic test for syphilis into a serologic criterion which includes several immunologic tests with the greatest specificity (antibody to DNA, LE cell preparation, and antibody to Sm). There were also new serologic criteria added to the 1971 set. The FANA test, tests for antibody to DNA, and tests for antibody to Sm were not in the original criteria. Their inclusion was found essential, since the ability to use the greater sensitivity of the antinuclear antibody test and the considerable specificity of tests for antibody to DNA or Sm greatly improved performance of the 1982 criteria. The Subcommittee analyzed in detail whether serum complement determinations should be incorporated into the revised criteria set. After multiple computer analyses, we were unable to improve accuracy by using any combination of serum complement determinations, either as a separate criterion or by adding those determinations into one of the other combined variables. The diagnostic information contained in C3 determinations, therefore, appears to be largely redundant with that contained in other elements already in the criteria set. The clinical importance of serum complement determinations might reside in their ability to assist in judging new and impending clinical events rather than in the process of diagnosis or classification. The 1982 revised criteria show great continuity with those of Of the 21 elements aggregated to form the original 14 criteria, 16 elements remain. By elimination of Raynaud s phenomenon and alopecia, aggregation of granular casts and proteinuria into a single criterion, and aggregation of serologic tests into a single criterion, the number of criteria has been reduced from 14 to 11. The 1971 criteria have been simplified in the 1982 revised set; the 1982 criteria may therefore be a more valuable teaching tool in addition to being updated. SLE classification requires 4 positive criteria in both the 1971 preliminary criteria and the 1982 criteria. Thus, practical application of the 1982 set is also closely similar to that of the 1971 set. Table 6. Analysis of false positive and false negative cases Current patient database (177 SLE and 162 controls) 1982 criteria 1971 criteria False positive 7 (96%)* 8 (95%)* False negative 7 (96%)t 22 (88%)t ~ ~~ ~ ~ * Percentages denote overall specificity of the 1982 and 1971 criteria. t Percentages denote overall sensitivity of the 1982 and 1971 criteria.

7 REVISED CRITERIA F SLE 1277 As with the 1971 criteria, the 1982 revised criteria set should be used mainly for the purpose of classifying patients in reports relating to clinical, serologic, cellular, or pathogenetic studies of SLE. Although it has good discriminating power against RA, scleroderma, and dermato/polymyositis. its performance against other rheumatic diseases has not been tested. Its potential as a truly diagnostic criteria set in SLE should await the results of more extensive tests against a wider variety of diseases. ACKNOWLEDGMENT The assistance of Dr. Alan Rubinow of Boston University School of Medicine in portions of this work is gratefully acknowledged. REFERENCES 1. Cohen AS, Reynolds WE, Franklin EC, Kulka JP, Ropes MW, Shulman LE, Wallace SL: Preliminary criteria for the classification of systemic lupus erythematosus. Bull Rheum Dis 21: , Lie TH, Rothfield NF: An evaluation of the preliminary criteria for the diagnosis of systemic lupus erythematosus. Arthritis Rheum , Davis P, Atkins B, Josse RG, Hughes GR: Criteria for classification of SLE. Br Med J , Fries JF, Siege1 RC: Testing the preliminary criteria for classification of SLE. Ann Rheum Dis 32: , Trimble RB, Townes AS, Robinson H, Kaplan SB, Chandler RW, Hanissian AS, Masi AT: Preliminary criteria for the classification of systemic lupus erythematosus (SLE): evaluation in early diagnosed SLE and rheumatoid arthritis. Arthritis Rheum 17: , Canoso JJ, Cohen AS: A review of the use, evaluations, and criticisms of the preliminary criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 22: , Masi AT, Rodnan GP, Medsger TA, Altman RD, D Angelo WA, Fries JF, LeRoy EC, Kirsner AB, McShane DJ, Myers AR, Sharp GC: Preliminary criteria for the classification of systemic sclerosis (scleroderma). Bull Rheum Dis 31:l-6, 1981