INTRODUCTION — Rheumatoid factors (RFs) are antibodies directed against the Fc portion of immunoglobulin (Ig) G. The RF, as initially described by Waaler and Rose in 1940 and as commonly measured in clinical practice, is an IgM RF, although other immunoglobulin types, including IgG and IgA, can exhibit this characteristic.
Testing for RF is primarily used for the diagnosis and establishing prognosis of rheumatoid arthritis (RA); however, RF may also be present in other rheumatic diseases and other conditions, including acute and chronic infections and neoplastic disease.
The biology of RFs and the clinical utility of their measurement are reviewed here. Discussions of clinically useful biologic markers in RA, including RF, and of the diagnosis of RA are presented separately. (See "Biologic markers in the diagnosis and assessment of rheumatoid arthritis" and "Diagnosis and differential diagnosis of rheumatoid arthritis".)
PATHOPHYSIOLOGY
Origins of rheumatoid factor — The origin of rheumatoid factor (RF) is incompletely understood [1,2]. An abnormal immune response appears to select, via antigenic stimulation, high-affinity RF from the host's natural antibody repertoire [3]. This may occur in rheumatic diseases, such as rheumatoid arthritis (RA), and in a number of inflammatory disorders characterized by chronic antigen exposure, such as infective endocarditis (IE) and chronic hepatitis C virus infection. The development of RF after such infections has suggested that they represent an antibody response to antibodies that have reacted with microbes. This possibility is supported by experimental evidence showing that mice immunized with IgM-coated vesicular stomatitis virus (VSV) develop RFs [4].
Normal human lymphoid tissue commonly possesses B lymphocytes with RF expression on the cell surface. However, RF is not routinely detectable in the circulation in the absence of an antigenic stimulus. Modified IgG, release of heavy chains from apoptotic B cells [5], and components of microbial organisms (such as Epstein-Barr virus [6]) may be triggers for RF production and could be important components of RA pathogenesis [6].
Costimulation of B cells, perhaps initiated by chronic infection and mediated by toll-like receptors (TLRs), may allow B cells with low-affinity receptors for IgG to become activated. TLRs are components of the innate immune system and provide signals after engaging various bacterial and viral products [7,8]. (See "Toll-like receptors: Roles in disease and therapy".)
RFs possess significant heterogeneity related to mutations within heavy and light chain genes [9]. While IgM RFs from patients with RA react with a variety of antigenic sites on autologous IgG [10], certain binding specificities appear to be more closely associated with seropositive RA than other specificities [11]. These include specificity for IgG1, IgG2, and IgG4 more than IgG3, and to the Fc elbow and tail regions more than the Fc receptor binding region.
RFs may also react against a variety of cellular and tissue antigens but may have different biologic activity in different hosts and anatomic locations. As an example, one report of RF derived from synovial tissue lymphocytes in a patient with RA found specificity for gastric glands, nuclei, and smooth muscle; by contrast, RF derived from a control patient's peripheral blood did not show this pattern of reactivity [12].
Possible functions — The function of RF is poorly understood. Possible functions include:
●Binding and processing of antigens embedded in immune complexes
●Presentation of antigens to T lymphocytes in the presence of human leukocyte antigen (HLA) molecules
●Immune tolerance
●Amplification of the humoral response to bacterial or parasitic infection
●Immune complex clearance
The role of RFs in the pathogenesis and perpetuation of RA or other rheumatic diseases is also unknown. One model of RA pathogenesis places RF in a key etiopathogenic role as an antibody to immune complexes that activates the complement cascade and promotes production of inflammatory mediators, thereby facilitating a state of ongoing inflammation [6]. There is in vitro evidence that RF may promote inflammatory cytokine production through interactions with anti-citrullinated peptide antibodies (ACPA) immune complexes in patients with RA [13].
While genetic factors appear to be important in the development of RA, the interplay between genetic predisposition, autoantibody production, and disease development is complex and incompletely understood. A number of studies have identified genetic differences between individuals with RF-negative and RF-positive RA [14-17]. In addition, a high correlation for RF has been noted among identical twins with RA [18]. However, other studies have demonstrated that patients with RF-negative and RF-positive RA may have similar HLA susceptibility alleles [19-21], including the shared epitope, suggesting that there may be a similar immunogenetic predisposition to RA in some patients that is independent of RF. (See "HLA and other susceptibility genes in rheumatoid arthritis", section on 'Seronegative rheumatoid arthritis'.)
The pathogenesis of RA is discussed in detail separately. (See "Pathogenesis of rheumatoid arthritis".)
CLINICAL DISORDERS ASSOCIATED WITH RHEUMATOID FACTOR POSITIVITY
Rheumatic disorders — Patients may have detectable serum rheumatoid factor (RF) in a variety of rheumatic disorders, many of which share similar features, such as symmetric polyarthritis and constitutional symptoms [22]. A positive RF test can also be found in nonrheumatic disorders and in healthy subjects [22]. The rheumatic disorders associated with a positive RF include:
●Rheumatoid arthritis (RA) – 26 to 90 percent (see 'Clinical uses of rheumatoid factor testing' below)
●Sjögren's syndrome – 75 to 95 percent (see "Diagnosis and classification of Sjögren's syndrome")
●Mixed connective tissue disease – 50 to 60 percent (see "Clinical manifestations and diagnosis of mixed connective tissue disease")
●Mixed cryoglobulinemia (types II and III) – 40 to 100 percent (see "Mixed cryoglobulinemia syndrome: Clinical manifestations and diagnosis")
●Systemic lupus erythematosus – 15 to 35 percent (see "Clinical manifestations and diagnosis of systemic lupus erythematosus in adults")
●Polymyositis or dermatomyositis – 5 to 10 percent (see "Clinical manifestations of dermatomyositis and polymyositis in adults")
The reported sensitivity of the RF test in RA has been as high as 90 percent. However, population-based studies, which include patients with mild disease, have found much lower rates of RF-positive RA (26 to 60 percent) [23-26]. This difference may reflect classification criteria that led published series of patients with RA to be biased toward more severe (and more seropositive) disease, thereby overestimating the sensitivity of RF in RA.
Other autoantibodies, including anti-citrullinated peptide antibodies (ACPA, which include anti-cyclic citrullinated peptides [anti-CCP]), may be present in patients with suspected or established RA who are RF negative, as well as those who are positive for RF [27]. The optimal clinical use of ACPA testing and its relationship to RF testing remain uncertain [28-32]. Although ACPA and RF have similar sensitivity for the diagnosis of RA, ACPA is a more specific marker for RA [33]. Of note, the 2010 revised classification criteria for RA include both RF and ACPA [34]. ACPA testing is discussed in more detail elsewhere. (See "Diagnosis and differential diagnosis of rheumatoid arthritis", section on '2010 ACR/EULAR criteria' and "Biologic markers in the diagnosis and assessment of rheumatoid arthritis", section on 'Anti-citrullinated peptide antibodies'.)
IgG and IgA RFs are occasionally present in patients with RA in the absence of IgM RF [35,36]. Measurement of these non-IgM RFs is not widely available in the United States. However, they may be of prognostic value since there is evidence suggesting that IgG, IgA, and 7S IgM RFs are associated with more severe disease [37-41]. This risk appears to be independent of human leukocyte antigen (HLA) alleles associated with severe disease [21].
Nonrheumatic disorders — Nonrheumatic disorders characterized by chronic antigenic stimulation (especially with circulating immune complexes or polyclonal B lymphocyte activation) commonly induce RF production (table 1). Included in this group are [22]:
●Indolent or chronic infection, as with infective endocarditis (IE) or with hepatitis B or C virus infection. As an example, studies have demonstrated that hepatitis C virus infection is associated with a positive RF in 54 to 76 percent of cases [42-45] and is even more common in patients with hepatitis C virus and mixed cryoglobulinemia. RF production typically ceases with resolution of the infection in these disorders. These molecules may be produced by activated hepatic lymphocytes [46]. (See "Mixed cryoglobulinemia syndrome: Clinical manifestations and diagnosis".)
●Inflammatory or fibrosing pulmonary disorders, such as sarcoidosis [47-49].
●Malignancy, particularly B-cell neoplasms [50].
●Primary biliary cholangitis (previously referred to as primary biliary cirrhosis).
Healthy individuals — RFs have been found in up to 4 percent of young, healthy individuals [51]. The reported incidence may be higher in older subjects without rheumatic disease, ranging from 3 to 25 percent [52,53]. Part of this wide range may be explained by a higher incidence of RF among chronically ill older adults as compared with healthy older patients [54]. However, other studies have reported that the prevalence of RF among healthy older individuals is low and does not increase with aging [55,56]. In a large representative population study in the United States, RF was present in 5 percent of the older population [57]. When present, RF is typically found in low to moderate amounts (eg, titers of 1:40 to 1:160) in individuals with no demonstrable rheumatic or inflammatory disease.
Cigarette smoking is associated with the presence of RF [58,59] as well as seropositive RA; whether this association is causal remains uncertain.
CLINICAL USES OF RHEUMATOID FACTOR TESTING
Rheumatoid factor is not useful as a screening test — Although rheumatoid factor (RF) may be found in the circulation prior to the development of rheumatoid arthritis (RA), most asymptomatic persons with a positive RF do not progress to RA; as a result, measurement of RF is a poor screening test [55]. Population-based studies have shown that some healthy people with a positive RF develop RA over time, especially if more than one isotype is persistently elevated and if there are high levels of RF [55,60]. Retrospective study of stored blood samples collected as part of routine blood donation has demonstrated that nearly 30 percent of those who later develop RA have serum RF present for a year or more prior to diagnosis (median 4.5 years) [61,62].
Predictive and diagnostic value — The predictive and diagnostic value of RF testing depend upon the sensitivity and specificity of the test, prevalence of RA in the population under study, and the clinical features of the patient being tested:
●The sensitivity and specificity of RF for the diagnosis of RA are moderately high – Estimates of the sensitivity and specificity of RF vary depending upon the populations being examined, and, of course, these estimates affect the calculated predictive value. As noted, the published sensitivity of RF in RA (ie, the proportion of patients with RA who are RF positive) ranges from 26 to 90 percent. A meta-analysis reported the overall sensitivity to be 69 percent (95% CI 65-73 percent) [63]. It is important to note that such data come from adults with RA; the operating characteristics of RF as a diagnostic test in individuals with suspected juvenile idiopathic arthritis (JIA) are quite different and depend upon the disease subset [64]. For example, most children with JIA are RF negative, especially if younger than age 10, while RF positivity is more commonly observed among older children with polyarticular disease. (See 'Clinical disorders associated with rheumatoid factor positivity' above and "Polyarticular juvenile idiopathic arthritis: Clinical manifestations, diagnosis, and complications" and "Classification of juvenile idiopathic arthritis".)
The specificity (ie, the proportion of a control population without RF) depends substantially upon the choice of the control group. The overall specificity of RF reported in heterogeneous publications analyzed as part of a meta-analysis was 85 percent [63]. As described above, the prevalence of RF in a young healthy population is approximately 4 percent, consistent with a specificity of 96 percent in this group. The specificity with respect to disease control populations is substantially lower, especially if the disease control populations include patients with rheumatic and other diseases associated with RF.
●Use in patients with a moderate to high pretest probability of RA:
•For patients presenting with inflammatory arthritis and with a moderate to high pretest probability of RA, RF testing is often useful – RF testing in a rheumatology clinic practice with a prevalence of RA of 16.4 percent demonstrated a specificity of 97 to 98 percent in patients with noninflammatory disorders and of 95 to 97 percent in all patients, yielding a positive predictive value of approximately 80 percent [56]. In patients with "undifferentiated" inflammatory arthritis, the presence of RF was helpful in predicting the ultimate diagnosis of RA, with a positive odds ratio of approximately 30 [65]. While an earlier study found no predictive value for RF testing in such patients [66], most studies of patients in early arthritis clinics demonstrate that RF, as well as anti-citrullinated peptide antibodies (ACPA), contributes to the ability to predict persistent synovitis and/or RA [67].
•In patients with a high pretest likelihood of RA, the negative predictive value of RF is limited by the relatively high frequency of seronegative RA – For example, if the pretest probability is 50 percent and the sensitivity is 70 percent, the post-test probability after a negative result is 27 percent (which may not be low enough to rule out RA). Since RF-negative RA makes up approximately one-third of all cases, a negative RF never rules out RA, regardless of pretest probability.
●Use in patients with a low pretest probability of RA:
•Among patients with a low prevalence of RF-associated rheumatic disease and in those with few clinical features of RA, the positive predictive value of RF is low – As with any diagnostic test, the predictive value of the RF is directly affected by the estimated likelihood of disease prior to ordering the test (ie, the pretest probability) and, with RF, by the proportion of individuals being tested with a nonrheumatic disorder associated with RF production (table 1).
In a study of consecutive tests ordered by health care providers in a large academic medical center in the United States, the prevalence of RA was approximately 13 percent [68]. The positive predictive value of RF (the likelihood of having disease if the RF is positive) was only 24 percent for RA and 34 percent for any rheumatic disease.
•In a population with a low pretest probability of RA, including those without inflammatory arthritis, the positive predictive value of RF testing is poor – Testing patients with nonspecific arthralgia, fibromyalgia, or osteoarthritis is not recommended because a positive test result is likely to represent a false-positive result. By contrast, RF testing for patients with inflammatory arthritis (which is much less common than the noninflammatory conditions) has high positive predictive value and clinical utility [69].
•In a population with a low pretest probability of RA, the calculated negative predictive value of the RF (the likelihood of not having disease if the RF is negative) tends to be high – Even so, a negative RF in this setting may not be particularly useful clinically because the likelihood of disease based on RF results may not change much. Suppose, for example, that a patient has an estimated 10 percent chance of RA; a negative RF test (assuming a sensitivity of 70 percent and specificity of 85 percent) will decrease the likelihood of RA from 10 to 4 percent. This small reduction in likelihood may not justify performing the test.
Rheumatoid factor quantification — The quantified level of RF should be considered when analyzing its utility. The higher the amount, the greater the likelihood that the patient has a rheumatic disease. There are, however, frequent exceptions to this rule, particularly among patients with one of the chronic inflammatory disorders noted above. Furthermore, the use of a higher cutoff for diagnosis decreases the sensitivity of the test as it simultaneously increases the specificity (by decreasing the incidence of false-positive results). In a study by the author and colleagues, for example, an RF titer of 1:40 or greater was 28 percent sensitive and 87 percent specific for RA; by comparison, a titer of 1:640 or greater increased the specificity to 99 percent (ie, almost no false-positive results) but reduced the sensitivity to 8 percent [68].
Treatment for RA can lower the level of RF [70], but such fluctuations do not reliably reflect disease activity.
Prognostic value — The presence of RF may predict a worse prognosis and the degree of responsiveness to particular drugs. RF-positive patients with RA may experience more aggressive and erosive joint disease and extraarticular manifestations than those who are RF negative [71-75]. Similar findings have been observed in JIA [76]. These general observations, however, are of limited utility in an individual patient because of wide interpatient variability. In addition, the effect of RF status on radiographic progression may be decreasing in the era of earlier and more effective therapy [77]. Accurate prediction of the disease course is not possible from the RF alone. (See "HLA and other susceptibility genes in rheumatoid arthritis".)
RF status may be useful in combination with other indicators, including ACPA status, C-reactive protein and erythrocyte sedimentation rate levels, and severity of synovitis on physical exam, to predict progression of radiographic changes in patients with RA and to guide treatment [78-80]. (See "General principles and overview of management of rheumatoid arthritis in adults", section on 'Prognosis'.)
RF status may have value in predicting response to treatment and comorbidity. As an example, the anti-CD20 B-cell depleting monoclonal antibody, rituximab, may be less effective for patients with seronegative RA than for those with seropositive RA [81]. Conversely, anti-tumor necrosis factor (TNF) therapy may be less effective in RF-positive RA than in seronegative disease [82]. In addition, the presence of RF among patients with RA may be associated with a higher risk of cardiovascular disease [83].
Role of repeat testing — Repeat testing of RF may be useful if a patient's diagnosis remains uncertain. However, there is usually no clear benefit to serial testing [84], particularly in a patient with established RA. In Sjögren's syndrome, the disappearance of a previously positive RF may herald the onset of lymphoma [85], so some clinicians check RF repeatedly in their patients with Sjögren's syndrome. The clinical utility of this practice, however, has not been critically assessed. (See "Clinical manifestations of Sjögren's syndrome: Extraglandular disease", section on 'Prevalence and risk' and "Clinical manifestations of Sjögren's syndrome: Extraglandular disease", section on 'Autoantibodies'.)
RF titer may fall with effective treatment of RA in patients who are originally RF positive [86]. Nevertheless, the fluctuation in level of RF does not correlate closely with disease activity. RF should not be used routinely to monitor RA disease activity in clinical practice.
Analytical considerations — The presence of RF can be detected by a variety of techniques. These include agglutination of IgG-sensitized sheep red cells or of bentonite or latex particles coated with human IgG, with results typically reported as titers based on manual dilution of the serum tested. Assays performed by immunoturbidimetry and nephelometry (two methods that measure light scatter caused by antigen-antibody interactions and widely used in large clinical laboratories) and enzyme-linked immunosorbent assay (ELISA) are usually quantified in terms of international units per mL; these are typically better standardized than manual methods [87-92]. Unfortunately, measurement of RF is imperfectly standardized; therefore, the same patient may have different results in different laboratories. Although no one technique has a clear advantage over others, automated methods, such as nephelometry, immunoturbidimetry, and ELISA, tend to be more reproducible than manual methods.
RFs in patient specimens sometimes interfere with measurements of other plasma components, including IgM antibodies to infectious agents. For example, IgM RFs can cause false-positive results in point-of-care tests for IgM antibodies to severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the agent of coronavirus disease 2019 (COVID-19) [93], or in assays for inflammatory cytokines [94].
Cost — Measurement of the RF titer is generally inexpensive. However, the cost per true-positive result may be high if the test is ordered in patients with a low prevalence of disease. In a 1992 analysis of 563 patients studied over a six-month period, for example, the cost per true-positive RF result was USD $563 [68].
SUMMARY AND RECOMMENDATIONS
●Rheumatoid factors (RFs) are antibodies directed against the Fc portion of immunoglobulin (Ig) G. Normal human lymphoid tissue commonly possesses B lymphocytes with RF expression on the cell surface. However, RF is not routinely detectable in the circulation in the absence of an antigenic stimulus. (See 'Pathophysiology' above.)
●How chronic infections and rheumatic diseases lead to increased RF in serum is uncertain. (See 'Pathophysiology' above.)
●Whether RF has a physiologic function is uncertain, though some potentially pathogenic and other potentially beneficial activities have been suggested. (See 'Possible functions' above.)
●RF is detected in the setting of various rheumatic diseases, in infections, in other inflammatory diseases, and in some healthy people. (See 'Clinical disorders associated with rheumatoid factor positivity' above.)
●There is no clear consensus regarding the indications for ordering the RF. The overall utility of this test may be historically overestimated, and the pretest probability of RF-associated disease as well as confounding inflammatory disorders should be considered. (See 'Clinical uses of rheumatoid factor testing' above.)
●We recommend not screening healthy individuals by testing for RF (Grade 1B). Measurement of RF has little value as a screening test to diagnose or exclude rheumatic disease in either healthy populations or in those with arthralgias. (See 'Clinical uses of rheumatoid factor testing' above.)
●We suggest not using the RF as a diagnostic test in patients who have arthralgias but no other symptoms or signs of a rheumatic disease. (See 'Clinical uses of rheumatoid factor testing' above.)
●The RF has a higher positive predictive value if ordered selectively in patients with a moderate or higher chance of having an RF-associated rheumatic disease such as rheumatoid arthritis (RA), Sjögren's syndrome, or mixed cryoglobulinemia syndrome. Included in this group are patients with prominent morning stiffness, sicca symptoms, or arthralgia or arthritis in a rheumatoid distribution (ie, symmetric polyarthritis involving small joints). (See 'Predictive and diagnostic value' above.)
●The negative predictive value of RF is limited by the significant prevalence of RF-negative RA and by the low clinical utility of RF testing in patients with nonspecific arthralgia (since RF-associated rheumatic disease is relatively uncommon in that population). (See 'Predictive and diagnostic value' above.)
●Higher levels of RF tend to have a higher positive predictive value for RA. (See 'Rheumatoid factor quantification' above.)
●Although, in aggregate, seropositive disease and higher titers of RF are associated with more severe RA, measurement of RF has limited prognostic value in the individual patient with RA. When RF is combined with other clinical data (such as joint count, anti-citrullinated peptide antibodies [ACPA] results, and C-reactive protein measurement), prediction can be improved but is still limited. (See 'Prognostic value' above.)