Diagnostic tests for rheumatic disease: clinical utility revisited.

Abstract: Establishing a diagnosis of systemic rheumatic disease requires an integration of a patient's symptoms, physical examination findings, and the results of diagnostic testing. There is often a temptation by clinicians to rely heavily on objective measures such as the presence or absence of an autoantibody. Medical textbooks and the medical literature may overestimate the diagnostic utility of many commonly ordered tests for rheumatic disease because the tests are usually analyzed among patients with established rheumatic disease rather than among patients with an uncertain cause of symptoms as is common in practice. Few diagnostic tests are highly sensitive, though the antinuclear antibody in systemic lupus erythematosus (SLE) and the erythrocyte sedimentation rate in temporal arteritis are notable exceptions. Conversely, few diagnostic tests are highly specific; anti-proteinase-3 and antimyeloperoxidase antibodies (types of antineutrophilic cytoplasmic antibodies) among patients with Wegener granulomatosis (and related vasculitides) and anti-double-stranded and anti-Smith antibodies among patients with SLE may be particularly helpful in the proper clinical settings due to their high specificity. Anticitrullinated cyclic protein (anti-CCP), a newly described autoantibody that may be highly specific for rheumatoid arthritis, requires additional study as its utility in clinical practice is uncertain.

Key Words: diagnostic tests, predictive value, rheumatic disease, sensitivity, specificity

**********

The diagnosis of systemic rheumatic disease requires an integration of a patient's symptoms, physical examination findings, and the results of diagnostic testing. However, a number of questions face any clinician seeing a patient with rheumatic complaints such as arthralgia, myalgia, or other symptoms suggestive of an inflammatory process: should any tests be ordered right away, or would it be better to wait? Which tests should be ordered? How should the results be interpreted? Because patients with rheumatic disease often have no specific findings for a single diagnosis (and perhaps because many clinicians do not feel confident about their musculoskeletal examinations), there is a temptation to rely heavily on objective measures such as the presence or absence of an autoantibody or its titer.

However, as is true for any imperfect test, the pretest probability of rheumatic disease dramatically affects the predictive value of tests intended to identify them. That is particularly important because musculoskeletal symptoms are exceedingly common compared with the prevalence of systemic rheumatic disease. That is, the pretest probability of systemic rheumatic disease in the population is rather low compared with musculoskeletal symptoms that are nearly ubiquitous.

Establishing the diagnosis of a rheumatic disease may require exclusion of other conditions that present in a similar fashion, yet the specific conditions to exclude and how long to continue efforts to rule out other diseases are matters of considerable judgment. Even the disease criteria, which are often used by clinicians as a guide, perform poorly in the evaluation of an individual patient presenting with new polyarthritis. (1) For example, the "Classification Criteria" for rheumatoid arthritis (2)--so named to distinguish them from diagnostic criteria--were developed to ensure some measure of comparability among patients enrolled in studies of rheumatoid arthritis (RA), not to diagnose RA in the office. And despite their use as inclusion criteria for studies, it is not rare for patients to meet the Classification Criteria for RA and have another rheumatic disease; conversely, clinical rheumatologists routinely diagnose RA in patients who do not meet these criteria.

Other problems make studies of these tests problematic. Many rheumatic disease tests, such as the antinuclear antibody (ANA), have no standardized methodology for measurement. Different laboratories vary in how they perform tests, report results, and choose cut-offs for positive or negative. And it is important to recognize disease severity and spectrum bias in published reports of test utility. Studies of rheumatic disease nearly always require patients to meet well-established clinical criteria for inclusion, and treatment protocols typically require at least moderately severe disease to be present (eg, patients with RA must have a minimum number of tender joints). Diagnostic studies in such patients, for example erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), or rheumatoid factor (RF), are more likely to be abnormal than those with milder (and perhaps more representative) disease seen in clinical practice. In fact, one recent analysis (3) suggested that only 15 to 35% of patients with RA seen in clinical practice meet common trial criteria. Most patients in these practices would have been excluded because their disease was not active enough. For all of these reasons, assessing the true utility of diagnostic studies in rheumatic disease is problematic. And it is likely that the medical literature overestimates diagnostic test utility in the rheumatic diseases as a result.

The diagnostic strengths and shortcomings of a number of commonly ordered tests will now be considered. This review will focus on primary, systemic rheumatic diseases rather than rheumatologic manifestations of other disorders. For example, diagnostic studies to detect an infectious or neoplastic process that might be causing musculoskeletal symptoms will not be considered here.

Erythrocyte Sedimentation Rate

A number of reviews have evaluated the usefulness of the ESR in clinical practice. (4,5) In general, the test has been deemed highly sensitive for systemic inflammation; however, this high sensitivity comes at the price of low specificity. Even so, a sizeable minority of patients with systemic rheumatic disease, including systemic lupus erythematosus (SLE) and systemic vasculitis, have normal ESRs. In fact, sensitivity high enough to be clinically useful is limited to temporal arteritis and, perhaps, polymyalgia rheumatica (PMR). According to older research studies, the ESR in PMR is elevated in up to 95% of cases, but some newer studies suggest that its sensitivity is considerably lower, (6,7) closer to 80%. The finding of a normal ESR argues strongly against the diagnosis of temporal arteritis (8) but is not enough to discount the possibility of RA, SLE, vasculitis, inflammatory muscle disease or any other rheumatic disease. The ESR may be useful to follow patients with nonrheumatologic illness, including infections or other inflammatory conditions that are difficult to monitor noninvasively; examples include subacute bacterial endocarditis, abscess, osteomyelitis, and inflammatory bowel disease. The ESR may also be helpful to monitor (rather than diagnose) patients with rheumatic disease. However, overreliance on this test is fraught with hazard, and it should be considered an adjunct to a patient's overall clinical status rather than an absolute reflection of disease activity.

Another limitation of the ESR worth emphasizing is that it may rise with conditions that are not clearly associated with inflammation, such as advancing age or the presence of serum paraproteinemia. To adjust the Westergren ESR for age, allow its upper limit of normal to equal the patient's age divided in half. For the patient with the "sed rate of unknown origin," infectious, neoplastic, and rheumatologic conditions should be strongly considered, but if these are ruled out, it may be useful to order a serum protein electrophoresis. Gammopathies, whether benign or associated with hematologic or oncologic disease, may explain the elevated ESR and obviate additional evaluation.

Finally, the degree of elevation of the ESR has some predictive value--the higher the value, the more likely a truly systemic, inflammatory process is present. According to one review, (5) an ESR of 100 mm/h or greater has a positive predictive value of 90% for serious illness, especially metastatic neoplasm, infection, or rheumatic disease. However, it is not true that marked elevations have a limited number of diagnostic possibilities. Any acute and many chronic inflammatory conditions can increase the ESR over 100 mm/min. Examples include gout, necrotizing pneumonia, occult intra-abdominal abscess, and monoclonal gammopathy of uncertain significance.

C-reactive Protein

The CRP is also a nonspecific marker of inflammation. It has gained renewed popularity as an assessment of risk for cardiovascular disease. However, as a diagnostic test for patients with suspected rheumatic disease, there is no clear advantage of CRP over ESR. It does change more quickly than the ESR (within hours rather than days), and it does not increase with age as may occur with the ESR. However, the CRP tends to be higher among persons with obesity or taking oral contraceptive pills. (9)

There is no consensus regarding the relative usefulness of ESR and CRP and it is uncertain how to interpret discordant values (eg, ESR is high, CRP is normal or vice versa). In one study of patients with RA, (10) 28% of results were discordant. The study author suggested that because ESR is affected by immunoglobulins, RF, and hemoglobulin, it may not be as accurate a measure of inflammation as the CRP among patients with RA. Rather than a measure of disease activity, the ESR may be a better predictor of disease severity, correlating as it does with the presence of RF and depressed hemoglobin (both of which correlate directly with disease severity). Other studies comparing these tests have come to mixed conclusions. For example, some have found that CRP is a more accurate measure of current disease activity of PMR while ESR is a more accurate predictor of relapse. (11) Others (12) suggest CRP is the more sensitive diagnostic test and a better predictor of relapse than ESR. For ankylosing spondylitis, neither test may be particularly accurate as a measure of disease activity. (13) When the CRP is markedly elevated in SLE, infection, rather than active lupus, is more likely. (14)

Rheumatoid Factor

Since its discovery in the 1950s, the RF has become a routine diagnostic test in evaluating patients with possible RA or other rheumatic disease. While its sensitivity is estimated at 70 to 90% and its specificity is reportedly in the range of 95%, these estimates have been derived from patient populations that may not reflect those in whom the test is most commonly ordered. In fact, the sensitivity of the RF may be much lower than is commonly quoted. (15,16) In addition, the positive predictive value of the RF for RA (that is the chance that a patient with a positive RF has RA) may be remarkably low when a patient's pretest probability of RA (or the prevalence of the disease in a group) is low. (17) Conversely, high negative predictive value in settings of low pretest probability is not as helpful as it might seem: a negative RF might have a negative predictive value of 99.5%, but if the pretest probability was only 1%, the probability of disease has only fallen from 1% to 0.5%, not a very useful result.

The effect of pretest probability on the utility of the RF is demonstrated in Table 1. Clearly, when the pretest probability of RA is low, testing for RF is more likely to provide an unhelpful negative result or a false-positive result than it is likely to provide diagnostically useful information. However, in the proper clinical setting--when the pretest probability is at least moderate--predictive value rises dramatically.

A number of other rheumatologic and nonrheumatologic conditions may be associated with RF, but it is clinically useful primarily in patients with features suggestive of RA, Sjogren syndrome, mixed connective tissue disease, or cryoglobulinemia. A highly positive test might lead a clinician logically to one of these diagnoses. While the RF may be positive in up to 35% of patients with SLE (Table 2), a positive RF does not usually move SLE to the top of the list of diagnostic possibilities. Even more misleading is the patient with a nonrheumatologic disease associated with arthralgia or arthritis in whom a RF may be positive, such as viral infection, liver disease, or endocarditis. (16) For example, it is not unusual for a patient with a viral infection to have polyarthralgia and a positive RF; such patients may be misdiagnosed with and treated for RA (or another RF-associated rheumatic disease). So despite the number of rheumatologic and nonrheumatologic conditions reportedly associated with a positive RF, many of these are not clinically useful or relevant--no clinician should rely on a RF to diagnose viral infection, endocarditis, or SLE. While RF may have good predictive value in a patient with a presentation suggestive of RA (eg, chronic, symmetric polyarthritis of small joints), it is not good as a screening test for rheumatic diseases or as a first diagnostic study to evaluate arthralgia.

The titer of the RF does affect the operating characteristics of the test. In general the higher the titer, the higher the specificity and positive predictive value but the lower the sensitivity.

Anticitrullinated Cyclic Protein

Over the past decade, a number of novel autoantibodies have been detected in association with RA. While several names were initially proposed, including antiperinuclear factor, antikeratin and antifillagrin antibodies, each of these antibodies targets a posttranslationally modified arginine residue, citrulline. The cyclic structure of this antigen has led to a consensus regarding its name, anticitrullinated cyclic protein (anti-CCP). Its high specificity (up to 95%) reported in early studies has led to enthusiasm about its potential clinical usefulness in the diagnosis (or perhaps prognosis) of RA. Early researchers speculated that this antibody might actually be pathogenic--an understandable theory given that the antigen, CCP, can be detected in the inflamed synovium of patients with RA and because the antibody appears to be highly specific for the disease. However, given the low sensitivity of the antibody (40-60%) and the finding of the same antigen in the synovium of patients with other inflammatory (and even some noninflammatory) conditions, (18) the pathogenic potential of this antibody seems low.

Early clinical research has suggested that the anti-CCP antibody may be particularly helpful in three situations:

1. Early arthritis, when RA is under consideration but cannot be unequivocally diagnosed. In fact, because anti-CCP antibody may be present before symptoms of RA (19) and because it appears to be highly specific, a patient with early arthritis and a positive anti-CCP antibody has a moderately high chance of eventually developing RA. (20,21)

2. Seronegative RA in a patient with mild symptoms and equivocal findings of polyarthritis (including elderly onset disease that may overlap clinically with PMR). A positive anti-CCP may increase confidence in the diagnosis of RA. (22,23)

3. Possible RA but in a clinical setting that makes the RF unreliable. The best example may be arthralgia associated with hepatitis C and cryoglobulinemia, (24) in which a positive RF is common regardless of whether concomitant RA is present.

However, before relying on the anti-CCP antibody to establish a diagnosis or drive therapeutic decision making, there are reasons to be cautious. Many of the studies evaluating the sensitivity and specificity of the anti-CCP have been relatively small, did not include a wide variety of potential mimics of RA in the comparison group, and did not include patients without a known diagnosis. That is, utility has not been established in patients with arthralgia or in other clinical settings in which RF is commonly ordered.

In addition, measurement of the anti-CCP antibody is unlikely to be helpful when a patient clearly has RA or clearly does not have the disease. Even if testing for the presence of anti-CCP helps predict the eventual diagnosis of RA among patients with early arthritis, it is unclear whether these results should alter management. In sum, the marginal benefit of anti-CCP measurement above and beyond standard clinical evaluation (including the history, examination, routine blood tests, and RF) is currently unknown and may be relatively small.

Antinuclear Antibody

As a diagnostic test, the ANA is highly associated with SLE, (present in up to 99% of cases) and drug-induced lupus (which, by definition, is always ANA positive), but there are a number of other ANA-associated illnesses, including Sjogren syndrome, progressive systemic sclerosis, mixed connective tissue disease, and RA (Table 3). Certain nonrheumatic conditions are also associated with the presence of ANA however, including Hashimoto thyroiditis, Graves disease, autoimmune hepatitis, primary biliary cirrhosis, and a number of chronic infections (such as hepatitis C and human immunodeficiency virus). These other conditions may be associated with arthralgia, rash, and other clinical features that may mimic rheumatic illness, posing a real risk of misdiagnosis. In general, the higher the titer, the more likely an ANA-associated disease is present, but exceptions are common. It is not rare for patients with SLE to have a low titer ANA, and some ANAs found in healthy people are of high titer. Some of these patients with clinical false-positive results may eventually develop SLE, (25) but there is currently no reliable way to identify such patients. In settings of low pretest likelihood of ANA-associated rheumatic disease, testing may produce more false-positive than true-positive results. (26) Complicating the assessment of ANA utility is that there is no standardized assay for this antibody, (27) and a number of factors can affect sensitivity of this test, including quality control, choice of the cut-off considered positive, and choice of antigen substrate.

ANAs may be directed against a number of antigens, and this can cause the ANA to be positive in a variety of patterns (Table 4), including peripheral, diffuse, and speckled. However, the patterns themselves lack specificity and may be prone to interobserver variability. For example, while a peripheral (or rim) pattern of the ANA correlates with anti-double-stranded DNA, actual measurement of anti-double-stranded DNA antibody is more reliable. The specificity of these ANA subtypes tend to be much higher than ANA testing alone. For example, a positive anti-double-stranded DNA or anti-Smith antibody is highly specific for SLE (Table 4).

Antiribosomal P

Studies from the late 1980s (28) raised the hope that antiribosomal P protein would allow clinicians to distinguish patients with neuropsychiatric lupus (especially psychosis) from those who had neuropsychiatric symptoms without SLE or those with SLE and an unrelated cause of neuropsychiatric symptoms. However, the findings of follow-up studies (29) have called the diagnostic utility of this test into question. Its sensitivity for SLE-related neuropsychiatric disease has been estimated to be as high as 90% though some studies have found lower sensitivity. Its specificity has also been questioned--antibody to ribosomal P protein has been noted in a number of patients with SLE even without neuropsychiatric problems, including lupus-associated hepatitis, and lupus nephritis. While some clinicians find that the presence of this antibody increases their confidence about the presence of central nervous system lupus, the diagnostic utility of this antibody test is uncertain.

Antineutrophilic Cytoplasmic Antibody

Antineutrophilic cytoplasmic antibodies (ANCAs) directed against proteinase-3 (pr-3) and myeloperoxidase (MPO) are highly associated with Wegener granulomatosis and related vasculitides (Table 5) and may even be pathogenic. (30) Because the test may be negative in the presence of an ANCA-associated illness (that is, it is far from a perfectly sensitive test), its diagnostic utility largely relates to its high specificity, estimated to be well above 90% in most studies. However, it is important to emphasize that this high specificity does not apply only to the presence or absence of ANCA. It applies to the presence or absence of antibodies directed against pr-3 or MPO. These antigen specificities correlate with the two main patterns of immunofluorescent staining noted in patients with a positive ANCA: anti-MPO specificity typically causes a perinuclear ANCA (p-ANCA) pattern while anti-pr-3 produces a cytoplasmic ANCA (c-ANCA) pattern. Atypical p-ANCAs are relatively common, especially among patients with a positive ANA or in patients with ulcerative colitis (see below). In addition, non-MPO p-ANCAs are more nonspecific, as they have been described in a number of inflammatory and noninflammatory conditions.

In a study of nearly 500 consecutive ANCA tests ordered at our institution, (31) only 3% had an ANCA-associated vasculitis; the positive predictive value of an ANCA directed against pr-3 or MPO was only 54%. And despite the test's high specificity (98% in our study), the relatively high frequency of test ordering, the rarity of ANCA-associated vasculitis, and the occasional false-positive result may limit its diagnostic utility even when positive.

ANCA testing is also potentially useful for patients with suspected inflammatory bowel disease, particularly ulcerative colitis. Up to 80% of patients will have a positive ANCA, usually with a p-ANCA or other nonspecific immunofluorescent pattern that lacks anti-MPO specificity.

Myositis-specific Antibodies: Anti-Jo-1, Anti-Mi-2, Anti-SRP

For patients with suspected idiopathic inflammatory myopathy, autoantibody testing may be useful, as several autoantibodies have been reported to have high specificity. (32) These myositis-specific antibodies (MSAs) include anti-Jo-1, anti-Mi-2, and anti-signal recognition protein (anti-SRP). However, because they are not particularly sensitive--approximately half of patients with polymyositis or dermatomyositis have one of these antibodies--a negative result is not helpful.

It is worth noting that the presence of one of these autoantibodies may correlate with clinical features of the myositis. For example, patients with anti-SRP antibodies tend to have acute onset of severe weakness during the fall, cardiomyopathy, and a poorer prognosis than those with other MSAs according to some reports. Patients with anti-Jo-1, on the other hand, tend to have fever, arthritis, Raynaud phenomenon, and interstitial lung disease. Those with anti-Mi-2 antibodies typically have dermatomyositis and a good prognosis.

Although these tests are called myositis specific, studies of large numbers of patients with other conditions that might mimic polymyositis and dermatomyositis have not been published and at least one recent study (33) has questioned the high specificity of the anti-SRP. It is uncertain, for example, whether a patient with one of the many muscular dystrophies or sarcoid myopathy might produce these antibodies.

Anti-Scl-70 and Anticentromere antibodies

Anti-Scl-70 (due to antibodies directed against topoisomerase) and ANAs with an anticentromere pattern are highly associated with progressive systemic sclerosis (the diffuse form of scleroderma) and CREST syndrome, respectively. (34) In the proper clinical setting--that is, when one of these conditions is suggested and suspected--these tests have a high positive predictive value. They are not particularly sensitive, however, and despite high specificity, other conditions may be associated with these antibodies, particularly primary biliary cirrhosis (associated with anticentromere antibodies) and overlap syndromes. The presence of these antibodies may also provide prognostic information: anticentromere antibodies tend to be associated with a lower incidence of interstitial lung fibrosis, a higher rate of pulmonary hypertension and lower mortality, while the presence of anti-Scl-70 increases the risk of pulmonary fibrosis and death.

Human Leukocyte Antigen (HLA) B27

The strong association of HLA-B27 with ankylosing spondylitis (AS) and other spondyloarthropathies may suggest that it would have high diagnostic utility in the evaluation of patients with back pain or oligoarthritis. However, the background rate of HLA-B27 (6-8% in Caucasian populations), the high prevalence of low back pain and the relative rarity of HLA-B27-positive spondyloarthropathy limit the diagnostic utility of this genetic marker. In addition, its sensitivity in spondyloarthropathies other than AS is remarkably lower than it is in AS particularly when sacroiliitis is absent (Table 6).

Uric Acid

Hyperuricemia is common among patients with gout, however, it is also common in patients with hypertension, renal insufficiency, obesity, alcohol abuse, advancing age, and a number of other common conditions and diseases. Perhaps the greatest diagnostic utility of measuring uric acid is when the result is low. When the serum uric acid level is less than 5.5 mg/dL, the diagnosis of gout should be called into question. In other words, a low uric acid has high negative predictive value. The serum uric acid is also useful to monitor serum uric acid among patients treated with urate-lowering medications. The dose of allopurinol, for example, may need to be increased if uric acid suppression is inadequate (eg, still above 5.5-6.0 mg/dL).

Measuring uric acid in the midst of a gouty flare has been considered unreliable because the level may fall, providing a false-negative result. This phenomenon may be overstated; in most patients, the drop in uric acid during an acute attack of gout is not profound. The finding of high-normal or elevated serum uric acid results is not particularly helpful as a diagnostic test, even when gout is highly suspected. The patient with a known history of gout could have a new or separate problem, and little is learned about the diagnostic possibilities for a patient without a prior diagnosis of gout. The clinical presentation and, ideally, the results of synovial fluid analysis are diagnostically much more useful.

Conclusion

Few diagnostic tests for suspected rheumatic disease are truly diagnostic. Very few have high enough specificity to stand alone. However, the presence of anti-double-stranded DNA or anti-Smith antibody is highly suggestive of SLE and anti-MPO and anti-pr-3 are highly suggestive of Wegener granulomatosis or related vasculitides. Even so, clinical judgment and skepticism should accompany any test result. Clinicians diagnose rheumatic disease by the clinical presentation, not by blood tests. Conversely, few tests are sensitive enough to argue strongly against the presence of the disease when negative or normal. The ANA among patients with SLE or DILE and the ESR among patients with temporal arteritis are exceptions.

In nearly every clinical scenario, the history, physical examination, and non-exotic test results (such as blood counts, renal function tests, and urinalyses) provide more helpful information than panels of rheumatologic blood tests. In fact, a number of mimics of rheumatic disease commonly lead rheumatologists to order iron studies (for possible hemochromatosis), thyroid-stimulating hormone (for possible hypothyroidism), immunoglobulin M antibody to parvovirus B19 (for possible polyarthritis associated with acute infection with this virus), and liver function tests and serologies (for rheumatic symptoms that may be associated with hepatitis B and C). These are often more enlightening than the ANA or RF and have profound implications for diagnosis, prognosis, and treatment.

Finally, the passage of time is frequently the most helpful test of all, as many patients with musculoskeletal symptoms have self-limited complaints and cannot be specifically diagnosed no matter how many tests are ordered.

The utility of many of the tests to diagnose or rule out rheumatic disease may be overestimated by medical textbooks and the medical literature. However, the utility of these tests might rise if they were ordered more selectively. The cost of the tests may be high when one considers the direct financial costs associated with test performance in addition to indirect costs (such as the costs of follow-up evaluation and the psychological impact) that follow false-positive and false-negative results. Most of the medical literature quoting sensitivity and specificity on commonly ordered tests for possible rheumatic disease have evaluated these tests only in a highly selected group of patients. To understand their real-world utility, real-world study is essential. This is particularly important for tests that are new or unfamiliar to the ordering clinician.

References

1. Harrison BJ, Symmons DP, Barrett EM, et al. The performance of the 1987 ARA classification criteria for rheumatoid arthritis in a population based cohort of patientswith early inflammatory polyarthritis. J Rheumatol 1998;25:2324-2330.

2. Arnett FC, Edworthy SM, Bloch DA, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315-324.

3. Sokka T, Pincus T. Most patients receiving routine care for rheumatoid arthritis in 2001 did not meet inclusion criteria for most recent clinical trials or american college of rheumatology criteria for remission. J Rheumatol 2003;30:1138-1146.

4. Sox HC, Liang MH. The erythrocyte sedimentation rate. Guidelines for rational use. Ann Intern Med 1986;104:515-523.

5. Brigden M. The erythrocyte sedimentation rate. Still a helpful test when used judiciously. Postgrad Med 1998;103:257-262, 272-274.

6. Helfgott SM, Kieval RI. Polymyalgia rheumatica in patients with a normal erythrocyte sedimentation rate. Arthritis Rheum 1996;39:304-307.

7. Gonzalez-Gay MA, Rodriguez-Valverde V, Blanco R, et al. Polymyalgia rheumatica without significantly increased erythrocyte sedimentation rate. A more benign syndrome. Arch Intern Med 1997;157:317-320.

8. Smetana GW, Shmerling RH. Does this patient have temporal arteritis? JAMA 2002;287:92-101.

9. Albert MA, Glynn RJ, Buring J, et al. C-reactive protein levels among women of various ethnic groups living in the United States (from the Women's Health Study). Am J Cardiol 2004;93:1238-1242.

10. Wolfe F. Comparative usefulness of C-reactive protein and erythrocyte sedimentation rate in patients with rheumatoid arthritis. J Rheumatol 1997;24:1477-1485.

11. Cantini F, Salvarani C, Olivieri I, et al. Erythrocyte sedimentation rate and C-reactive protein in the evaluation of disease activity and severity in polymyalgia rheumatica: a prospective follow-up study. Semin Arthritis Rheum 2000;30:17-24.

12. Salvarani C, Cantini F, Niccoli L, et al. Acute-phase reactants and the risk of relapse/recurrence in polymyalgia rheumatica: a prospective followup study. Arthritis Rheum 2005;53:33-38.

13. Spoorenberg A, van der Heijde D, de Klerk E, et al. Relative value of erythrocyte sedimentation rate and C-reactive protein in assessment of disease activity in ankylosing spondylitis. Rheumatol 1999;26:980-984.

14. ter Borg, EJ, Horst G, Limburg PC, et al. C-reactive protein levels during disease exacerbations and infections in systemic lupus erythematosus: a prospective longitudinal study. J Rheumatol 1990;17:1642-1648.

15. Lichtenstein MJ, Pincus T. Rheumatoid arthritis identified in population based cross sectional studies: low prevalence of rheumatoid factor. J Rheumatol 18:989-993.

16. Shmerling RH, Delbanco TL. The rheumatoid factor: an analysis of clinical utility. Am J Med 1991;91:528-534.

17. Shmerling RH, Delbanco TL. How useful is the rheumatoid factor? An analysis of sensitivity, specificity, and predictive value. Arch Intern Med 1992;152:2417-2420.

18. Vossenaar ER, Smeets TJM, Kraan MC, et al. The presence of citrullinated proteins is not specific for rheumatoid synovial tissue. Arthritis Rheum 2004;50:3485-3494.

19. Nielen MM, van Schaardenburg D, Reesink HW, et al. Specific autoantibodies precede the symptoms of rheumatoid arthritis: a study of serial measurements in blood donors. Arthritis Rheum 2004;50:380-386.

20. Saraux A, Berthelot JM, Chales G, et al. Ability of the American College of Rheumatology 1987 criteria to predict rheumatoid arthritis in patients with early arthritis and classification of these patients two years later. Arthritis Rheum 2001;44:2485-2491.

21. van Gaalen FA, Linn-Rasker SP, van Venrooij WJ, et al. Autoantibodies to cyclic citrullinated peptides predict progression to rheumatoid arthritis in patients with undifferentiated arthritis: a prospective cohort study. Arthritis Rheum 2004;50:709-715.

22. Lopez-Hoyos M, Ruiz de Alegria C, Blanco R, et al. Clinical utility of anti-CCP antibodies in the differential diagnosis of elderly-onset rheumatoid arthritis and polymyalgia rheumatica. Rheumatology (Oxford) 2004;43:655-657.

23. Lee DM, Schur PH. Clinical utility of the anti-CCP assay in patients with rheumatic diseases. Ann Rheum Dis 2003;62:870-874.

24. Bombardieri M, Alessandri C, Labbadia G, et al. Role of anti-cyclic citrullinated peptide antibodies in discriminating patients with rheumatoid arthritis from patients with chronic hepatitis C infection-associated polyarticular involvement. Arthritis Res Ther 2004;6:R137-R141.

25. Arbuckle MR, McClain MT, Rubertone MV, et al. Development of autoantibodies before the clinical onset of systemic lupus erythematosus. N Engl J Med 2003;349:1526-1533.

26. Slater CA, Davis RB, Shmerling RH. Antinuclear antibody (ANA) testing: A study of clinical utility. Arch Int Med 1996;56:1421-1425.

27. Tan EM, Feltkamp TE, Smolen JS, et al. Range of antinuclear antibodies in "healthy" individuals. Arthritis Rheum 1997;40:1601-1611.

28. Bonfa E, Golombek SJ, Kaufman LD, et al. Association between lupus psychosis and anti-ribosomal P protein antibodies. N Engl J Med 1987;317:265-271.

29. Gerli R, Caponi L, Tincani A, et al. Clinical and serological associations of ribosomal P autoantibodies in systemic lupus erythematosus: prospective evaluation in a large cohort of Italian patients. Rheumatology (Oxford) 2002;41:1357-1366.

30. Bartunkova J, Tesar V, Sediva A. Diagnostic and pathogenetic role of antineutrophil cytoplasmic autoantibodies. Clin Immunol 2003;106:73-82.

31. Mandl LA, Solomon DH, Smith EL, et al. Using antineutrophil cytoplasmic antibody testing to diagnose vasculitis: can test-ordering guidelines improve diagnostic accuracy? Arch Intern Med 2002;162:1509-1514.

32. Miller F, Myositis-specific autoantibodies. Touchstones for understanding the inflammatory myopathies (clinical conference). JAMA 1993;270:1846-1849.

33. Kao AH, Lacomis D, Lucas M, et al. Anti-signal recognition particle autoantibody in patients with and patients without idiopathic inflammatory myopathy. Arthritis Rheum 2004;50:209-215.

34. Cepeda EJ, Reveille JD. Autoantibodies in systemic sclerosis and fibrosing syndromes: clinical indications and relevance. Curr Opin Rheumatol 2004;16:723-732.

You will find that the mere resolve not to be useless, and the honest
desire to help other people, will, in the quickest and delicatest ways,
improve yourself.
--John Ruskin

Robert H. Shmerling, MD

From the Division of Rheumatology, Department of Medicine, Beth Israel Deaconess Medical Center, Boston MA.

Dr. Shmerling receives financial support from Beth Israel Deaconess Medical Center, Harvard Medical Faculty Physicians, Harvard Medical School, Harvard Health Publications.

Reprint requests to 110 Francis Street, Suite 4B Boston, MA 02215. Email: [e-mail]

Accepted April 17, 2005.

RELATED ARTICLE: Key Points

* Most tests ordered in the evaluation of patients with suspected rheumatic disease are not actually diagnostic.

* Diagnostic utility of many commonly ordered tests, such as the rheumatoid factor and antinuclear antibody, may be overestimated.

* The passage of time is one of most useful diagnostic tests as many patients with musculoskeletal symptoms improve over time without a clear diagnosis.

Table 1. Posttest probabilities of rheumatoid arthritis vary by pretest
probability and RF result, based on estimated RF sensitivity of 80% and
specificity of 95% (a,b)

Pretest      Posttest probability,  Posttest probability,
probability  RF-positive            RF-negative

 1% (c)          16%                     0.2%
15%              74%                     4%
25%              84%                     7%
50%              94%                    17%
75%              98%                    39%
90%              99%                    65%

(a) Reprinted from Shmerling RH, Delbanco TL. The rheumatoid factor: an
analysis of clinical utility. The American Journal of Medicine, 1991;91:
528-534. With permission from Excerpta Medica, Inc.
(b) RF, rheumatoid factor.
(c) The estimated prevalence of RA in the United States is 0.5% to 3%.

Table 2. Rheumatic disease associated with a positive rheumatoid
factor (a)

Disease                               Frequency

Rheumatoid arthritis                  50-90%
Systemic lupus erythematosus          15-35%
Sjogren's syndrome                    75-95%
Systemic sclerosis                    20-30%
Polymyositis/dermatomyositis           5-10%
Cryoglobulinemia                      40-100%
Mixed connective tissue disease       50-60%

(a) Reprinted from Shmerling RH, Delbanco TL. The rheumatoid factor: an
analysis of clinical utility. The American Journal of Medicine, 1991;91:
528-534. With permission from Excerpta Medica, Inc

Table 3. Selected diseases associated with a positive ANA (a)

Rheumatologic diseases
  Systemic lupus erythematosus
  DILE
  Sjogren's syndrome
  Scleroderma
  Mixed connective tissue disease
  Rheumatoid arthritis
Non-rheumatologic diseases
  Hashimoto's thyroiditis
  Graves' disease
  Autoimmune hepatitis
  Primary biliary cirrhosis
  Acute or chronic infection (eg, HIV, Hepatitis B, Hepatitis C)

(a) ANA, antinuclear antibody; DILE, drug-induced lupus eythematosus;
HIV, human immunodeficiency virus.

Table 4. ANA subtypes (a)

Patterns                  Disease association

Anti-double-stranded DNA  Highly specific, but not sensitive for SLE
  (peripheral)
Anti-Smith (speckled)     Highly specific, but not sensitive for SLE
Anti-Ro (speckled)        SLE, SS, neonatal SLE
Anti-La (speckled)        SLE, SS, neonatal SLE
Antihistone (diffuse)     SLE, RA, aging, DILE
Anti-RNP (speckled)       SLE, MCTD
Anticentromere            Limited scleroderma (CREST syndrome)

(a) ANA, antinuclear antibody; SLE, systemic lupus erythematosus; SS,
Sjogrens syndrome; RA, rheumatoid arthritis; DILE, drug-induced lupus
erythematosus; RNP, ribo-nuclear protein; MCTD, mixed connective tissue
disease; CREST, calcinosis, Raynauds, esophageal dysfunction,
sclerodactyly, telangiectasia.

Table 5. Sensitivity of ANCA in selected diseases (a)

Disease                   c-ANCA/pr-3  p-ANCA/MPO

Wegener granulomatosis    80-90%        5-10%
Microscopic polyangiitis  40-50%       40-50%
Polyarteritis nodosa       5-10%        5-10%
Churg-Strauss vasculitis  10%          70-80%
Idiopathic pauci-immune    5-10%       65-75%
  glomerulonephritis
Goodpasture syndrome                   20-30%
  (anti-GBM disease)

(a) ANCA, Antineutrophilic cytoplasmic antibody; c-ANCA, cytoplasmic
ANCA; pr-3, proteinase-3; p-ANCA, perinuclear ANCA; MPO,
myeloperoxidase; GBM, glomerular basement membrane.

Table 6. Sensitivity of HLA-B27 in selected diseases

Disease                                   HLA-B27

Ankylosing spondylitis                    90-95%
Enteropathic arthritis                    30%
  with sacroiliitis                       60%
Psoriatic arthritis                       20-40%
  with spondylitis                        50-70%
Reiters disease                           60%
  with spondylitis                        80%
Juvenile spondylitis                      60-90%
Acute anterior uveitis                    50%

HLA-B27, human leukocyte antigen B27.