Preliminary criteria for the classification of systemic lupus erythematosus SLE evaluation in early diagnosed SLE and rheumatoid arthritis.

BRIEF COMMUNICATION
Preliminary Criteria for the Classification of
Systemic Lupus Erythematosus (SLE)
Evaluation in Early Diagnosed SLE and Rheumatoid Arthritis
R. Bruce Trimble, Alexander S. Townes, Harry Robinson, Stanley B. Kaplan,
Robert W. Chandler, Aram S. Hanissian and Alfonse T. Masi
Preliminary criteria for the classificationof systemic lupus erythematosus
(SLE)were applied to patients with early diagnosed SLE and rheumatoid arthritis (RA) who are participants in a prospective community study
of arthritis of recent onset. Sensitivity of the criteria was 88 % , and specificity against RA was 97.5 ye. Sensitivity increased to 92 % without loss
of specificity when high titer antinuclear antibody (ANA) was used as an
alternative criteria manifestation to LE cells. These results show the efficacy of the proposed criteria for early diagnosed patients, and support
the inclusion of high titer ANA in the SLE criteria.
Recently published preliminary criteria of rheumatoid arthritis (RA)who met the
for the classification of systemic lupus ery- American Rheumatism Association (ARA)
thematosus (SLE) (1) were applied to 25 criteria for definite or classical disease (2).
patients with a clinical diagnosis of SLE, These groups comprised all patients of ages
and to 40 patients with a clinical diagnosis 16 to 45 years admitted consecutively from
1968 to 1973 to the Memphis and Shelby
From the Departments of Medicine and Community Medicine, University of Tennessee School County Arthritis Research Program, who
of Medicine, Memphis, Tennessee.
were so diagnosed on initial or subsequent
Supported in part by US Public Health Service annual evaluations. This continuing, comGrants AM 12049 and AM 05055.
prehensive, prospective study (3) recruits
R BRUCE TRIMBLE, MD': Fellow in Rheumatology,
supported by the Baptist Memorial Hospital, Mem- patients under age 45 years who are resiphis Tennessee; ALEXANDER s TOWNES, MD: Professor
dents of Memphis or Shelby County, and
of Medicine and Chief, Section of Rheumatology;
who are referred to the program within 6
HARRY ROBINSON, ScD: Associate Professor and Chief,
Section of Biostatistics, Dept of Community Medi- months of first diagnosis of arthritis. Thus,
cine; STANLEY B WPWN, MD: Professor of Medicine;
unlike the preliminary criteria report, this
ROWT
w CHANDLER, PhD: Assistant Professor of
Immunology in Medicine; ARAM s HANISSIAN, MD: study provides data on the applicability of
Associate Professor of Pediatrics and Chief, Section the proposed SLE criteria in a community
of Pediatric Rheumatology; ALFONSE T MASI, MD,
sample of early diagnosed SLE and RA paD P H Director, Division of Connective Tissue Distients. The value of high titer antinuclear
eases.
Address reprint requests to Dr A T Masi, MD, antibody (A-NA) as an alternate criteria
DPH. university of Tennessee, 858 Madison Avemanifestation to a positive LE cell test was
nue, Memphis, Tennessee 1 1 6 3 .
'Present address: Park Clinic, Mason City, Iowa also analyzed.
50401.
All ANA tests were performed in our
Submitted for publication July 24, 1973; accepted
laboratory using a standard indirect imOctober 29. 1973.
184
Arthritia and Rheumatism, Vol. 17, No. 2 (March-April 1974)
CLASSIFICATION OF SLE
munofluorescent technique with fresh frozen rat kidney sections and fluorescein conjugated polyvalent rabbit antihuman immunoglobulin prepared from lyophilized
aliquots of the same large pool of antiserum. High-titer ANA was arbitrarily defined prior to this analysis as a titer of 1:64
or greater on the basis of extensive previous experience in the clinical application
of this particular test system.
The proposed SLE criteria were tested
against the clinical diagnoses of SLE or RA
made by experienced rheumatologists on
patients entered into our study, and corroborated by follow-up diagnosis. Under
this assumption, a true positive diagnosis
in terms of the proposed SLE criteria (1)
corresponds to satisfying the criteria in a
patient with the clinical diagnosis of SLE.
A true negative in terms of the SLE criteria
indicates failure to satisfy the criteria in a
patient clinically diagnosed RA (non-SLE).
The term sensitivity may be considered
equivalent to the proportion of true positives among all clinically diagnosed SLE
patients, and the term specificity equivalent
to the proportion of true negatives among
the patients diagnosed RA.
Table 1 shows that 18 (72%) of our SLE
patients (23 women, 2 men, mean age at
entry 28 years) satisfied the proposed criteria for the classification of SLE (four or
more of the 14 specified manifestations) on
initial evaluation, and 22 (88%) did so over
the course of followup, which averaged 17.5
months. None of the 40 R A patients (31
women, 9 men, mean age at entry 30.7
years) met four or more criteria initially,
and only one (2.5y0) did so over a mean
follow-up of 37.5 months. If high titer
ANA is used as an alternate criteria manifestation to a positive LE cell test, then
88y0 of SLE patients satisfied four or more
criteria on initial evaluation, and 92y0 did
Table 1. Number of Patients with SLE and RA
who Satisfied Varying Numbers of
Proposed ARA Criteria for SLE
Number
of
SLE
Criteria
4 or
More
Initial Visit
25SLE 40RA
Any Visit
25SLE
40RA
18(21)
O(0)
22 (23)
l(1)
Percent
with 4
or More 72 (84)
0 (0)
88 (92)
2.5 (2.5)
‘Numbers in parentheses refer to number of pacritients fulfilling “alternate” SLE criteria-ie,
teria which allow a high titer ANA to be substituted for a positive LE cell test.
so overall. No additional RA patient met
criteria as a result of this modification.
Two RA patients had high titer ANA’s,
but had only one additional manifestation
(non-deforming arthritis); and the one RA
patient who developed four manifestations
did not have a high titer ANA at any time.
No RA patient had a positive LE cell test.
As Table 2 indicates, the prevalence or‘
various manifestations was similar in our
SLE patients and in those presented in the
analysis of the preliminary criteria (l),
with statistically significant differences by
the Chi-square test only in the frequencies
of nondeforming arthritis, Raynaud’s phenomenon, facial erythema, positive LE cell
tests, and cellular casts. Although some of
these differences might occur by chance because of our relatively small sample, other
Arthritis and Rheumatism, Vol. 17, No. 2 (March-April 1974)
185
TRIMBLE ET AL
Table 2. Percent Frequency of Manifestations
in Patients with SLE from Memphis and 245
Patients with SLE Included in the Prellminaty
Criteria Group
Manifestations
Criteria
Memphis Group
Facial erythema
40
Discoid lupus
32
Raynaud’s phenomenon
44
Alopecia
40
Photosensitivity
28
Oral or nasopharyngeal
ulceration
16
Arthritis without deformity
100’
LE cells
48
Chronic false positive STS
12
Profuse proteinuria
16
Cellular casts
16
Pleurisy/pericarditis
60
Psychosis/convulsions
20
Hemoiytic anemia/ leukopenia/
thrombocytopenia
52
ANA titer 1:64 or greater
&4
64
17
15
43
37
15
84
92
12
20
48
60t
19t
40t
-
‘Minimal decrease of ability to fully extend the
elbows and limitation of motion secondary to
effusion or pain were not considered deformity.
?The proposed criteria group had 60.4% pleuritis and 18.8% pericarditis; 19.2% psychosis
and 12.7% convulsions; and 39.6% leukopenia,
16.3% hemolytic anemia, and 11.4% thrombocytopenia. The most frequent single manifestation in each combination is cited since no data
are provided on the presence of one or the
other individual manifestations.
factors relating to patient evaluation and
duration of disease seem to be a more likely explanation. Since arthritis was a prerequisite for inclusion of patients in our
study, the difference in the frequency of
this manifestation in our patients as compared with the ARA series is anticipated.
The higher prevalence of Raynaud’s phenomenon in our patients may reflect the
interest in this manifestation at our center,
and the careful historical search for it incorporated in the detailed protocols of this
prospective study. The prevalence of facial
erythema and positive LE cell tests was
186
higher in the ARA criteria group than in
our patients, but the ARA criteria frequency was also higher than reported in two
large series (4,5) and in two other patient
groups to which the criteria have recently
been applied (6,7). This may be the result,
as suggested by Fries and Siege1 (7), of having selected patients with “classical” manifestations of SLE and RA for the study
groups from which the criteria were derived. The lower prevalence of cellular casts
in our patients probably reflects their early,
generally milder disease, compared with
the patients used in formulating the ARA
criteria.
Arthritis without deformity did not discriminate between these early diagnosed
RA and SLE patients, since only one RA
patient had deforming arthritis. No significant difference was found in the prevalence
of mouth ulcers between the SLE (4 of 25)
and RA (3 of 40) patients, and omission of
this manifestation did not change the sensitivity or specificity of the criteria in our
patients. Also, the Chi-square test (with
Yate’s correction when appropriate) confirmed the highly discriminating value of
high titer ANA which was positive in 21
of 25 SLE and only 2 of 40 RA patients
(P < 0.001).
I n order to rank the independent contributions of the 14 individual SLE criteria
variables, plus high titer ANA, to the variability of the clinical diagnosis of SLE or
RA, the data were further analyzed using
the technique of stepwise multiple regression (8). In brief, the technique quantitates
the independent contributions of variables
- e g , each of the SLE criteria manifestations, to the total variance of a dependent
variable--eg. the clinical diagnosis of SLE
or RA, after allowing for any intercorrelations which may exist in the independent
variables. I n Table 3, the multiple correla-
Arthritis and Rheumatism, Vol. 17, No. 2 (March-April 1974)
CLASSIFICATION OF SLE
Table 3. Multiple Regression Analysis of SLE Criteria Manifestations
Plus Positive ANA In Relation to the Clinical Diagnosis of SLE Versus RA
SLE Criteria Manifestations.
>
ANA (titer
1:64)
Hemolytic anemia, leukopenia
thrombocytopenia
Pleuritis, pericarditis
Photosensitivity
LE cells
Raynaud’s phenomenon
*Manifestations contributing significantly (P
sus RA.
Cumulative
Percent
Variability
Accounted for
(RP)
Increase
in R’ due to
Variable
Specified
(% 1
0.8038
64.61
64.61
0.8631
0.8809
0.8955
0.9046
0.9118
74.50
77.61
80.19
81.83
83.13
9.87
3.11
2.58
1.64
1.30
Multiple
Correlation
Coefficient
(R)
4
.a15) to the
variance of clinical diagnosis of SLE ver-
tion coefficient squared (RZ)
has the same pendently to the variability in diagnosis at
interpretation as the squared coefficient (r2) a 5 percent significance level.
in univariate analysis. In both cases, the
This study shows the effectiveness of the
squared coefficient corresponds to the pro- preliminary SLE criteria when applied to
portion of variability in the dependent early diagnosed SLE and RA patients, esvariable accounted for by the “independ- pecially if high titer ANA is used as an
ent” variables. All variables were analyzed alternative manifestation for a positive LE
as dichotomies although the technique was cell test. Such substitution improved the
devised basically for multinormal distribu- sensitivity of the criteria, especially at initions. The modification was utilized as an tial evaluation, without sacrificing specificavailable method to approximately rank ity against RA. Thus, an ANA titer of 1:64
the various criteria in terms of their rela- or greater had a sensitivity of 84y0 for SLE
tive independent contributions to diagnos- patients, and a specificity of 95y0 against
tic variability, and is not intended as a RA patients.
definitive statement which would require a
It should be realized, however, that while
technique specifically designed for discrete high titer ANA was the most discriminatvariables, and a much larger population of ing factor in this reported series, standardpatients.
ization of technique such as obtained here
As shown in Table 3, high titer ANA is not always possible, and that there is no
accounted for by far the largest amount of standard for comparison of results between
the variance, supporting the widespread different laboratories. Although the LE cell
clinical impression of its diagnostic value. test has advantages regarding availability
It is probable, of course, that a positive LE and standardization, the ANA as in our
cell test would have contributed much series is a more sensitive test for antinuclear
more had ANA been omitted from the antibodies. Another recent study (9) has
analysis, since the ANA and LE cell tests confirmed a high degree of correlation
were highly significantly correlated in the (92%) of proposed SLE criteria in clinictotal group of patients (P < 0.001). Only ally diagnosed SLE patients with raised
the six variables shown contributed inde- levels of antibodies to a specific nuclear
Arthritis and Rheumatism, Vol. 17, No. 2 (March-April 1974)
187
TRIMBLE ET AL
antigen, DNA. These considerations urge
that standardization of ANA tests be developed, and that further testing of the value of ANA as a criteria manifestation for
SLE classification be pursued in comparison of SLE patients with other groups, including patients with RA more advanced
than in this series, and patients with other
connective tissue diseases.
Both SLE and RA patients in our Arthritis Research Program met additional
SLE criteria with time. Thus, the cumulative proportion of true positives (sensitivity) of 92 percent for the proposed but
modified criteria (including ANA) may increase even further with follow-up of our
early diagnosed SLE patients. However, it
is also possible that the proportion of true
negatives (specificity) may decrease in our
RA patients as they continue to be followed, and only later analysis will reveal
whether the present proportion of 97.5 percent will be maintained.
Although arthritis of recent onset was a
necessary prerequisite for referral of patients to this study, these patients presented
a typically broad range of clinical problems
due to SLE (Table 2) during the course of
subsequent follow-up, from mild disease requiring no specific treatment to fatal termination due to renal disease (3 cases) or central nervous system involvement (1 case).
Thus, the high degree of sensitivity and
specificity of the preliminary ARA criteria
for SLE in these patients with early diag-
188
nosed disease even at the time of initial
evaluation strongly supports the efficacy of
these criteria in the classification of patients
with SLE.
REFERENCES
1. Cohen AS, Reynolds WE, Franklin EC, et al:
Preliminary criteria for the classification of
systemic lupus erythematosus. Bull Rheum
Dis 21:643-648, 1971
2. Ropes MW, Bennett GA, Cobb S, et al: 1958
revision of diagnostic criteria for rheumatoid
arthritis. Bull Rheum Dis 9:175-176, 1958
3. Masi AT, Robinson H, Hughen M: The
Arthritis Research Program survey of newly
diagnosed arthritis among residents of Memphis and Shelby County. Memphis Mid-South
Med J 43:359-364, 1968
4. Tuffanelli DL, Dubois EL: Cutaneous manifestations of systemic lupus erythematosus.
Arch Dermatol 90:377-386, 1964
5. Estes D, Christian CL: The natural history
of systemic lupus erythematosus by prospective analysis. Medicine 50:85-95, 1971
6. Lie T H , Rothfield NF: An evaluation of the
preliminary criteria for the diagnosis of systemic lupus erythematosus. Arth Rheum 15:
532-534, 1972
7. Fries JF, Siege1 RC: Testing the ‘Preliminary
Criteria for Classification of SLE.’ Ann
Rheum Dis 32171-177, 1973
8. Draper NR, Smith H: Applied regression
analysis. New York, John Wiley & Sons, Inc,
1966, pp 163-216
9. Davis P, Atkins B, Josse RG, et al: Criteria
for classification of SLE. Br Med J 2:88-89,
1973
Arthritis and Rheumatism, Vol. 17, No. 2 (March-April 1974)