Previous diagnosis of Sjögren’s Syndrome as rheumatoid arthritis or systemic lupus erythematosus

Abstract

Objective. The diagnosis of SS is often difficult and many patients are symptomatic for years with other diagnoses before confirmation of SS. Our aim was to determine whether overlapping clinical and serologic features with RA and SLE may in part drive the misdiagnoses.

Methods. A total of 1175 sicca patients were evaluated in a multidisciplinary clinic and classified as having SS based on the American–European Consensus Group Criteria. They were interrogated for a past history of suspicion or diagnosis of RA, SLE or SSc. These diseases were confirmed or ruled out by applying the corresponding classification criteria if the patients responded affirmatively.

Results. Of these, 524 (44.6%) subjects reported previous diagnosis or suspicion of RA, SLE or SSc, which was confirmed in 130 (24.8%) but excluded in 394 (75.2%) subjects. Of those previously diagnosed with another illness, 183 (34.9%) met the criteria for primary SS. RF was present in 70/191 patients with previous diagnosis of RA compared with 445/845 without a prior RA diagnosis (P = 3.38E-05), while 128/146 with a diagnosis of SLE had positive ANA compared with 622/881 without the diagnosis (P = 8.77E-06). Age also influenced former diagnoses: people with suspected RA were older than those without the diagnosis (P = 5.89E-06), while patients with SLE suspicion were younger (P = 0.0003). Interestingly, the previous diagnoses did not significantly delay a final classification of SS.

Conclusion. Among subjects classified as SS, the presence of a positive ANA or RF was associated with a previous, apparently erroneous diagnosis of SLE or RA, respectively.

Rheumatology key messages

• Diagnosis of SS is difficult and often preceded by presumptive diagnoses of other connective tissue disorders.

• The presence of ANA and/or RF increases the risk of misdiagnosing SS as SLE or RA, respectively.

• Advances in biomarker and physiopathology discovery should result in more sensitive and specific tests for Sjögren’s syndrome.

Introduction

SS is a chronic autoimmune disease (AID) characterized by exocrine gland damage and dysfunction mediated by autoantibodies and lymphocytic infiltrates, resulting in xerostomia and KCS. A significant proportion of patients with SS have systemic manifestations, including arthritis, fatigue, haematological abnormalities, pulmonary, renal and peripheral nervous system involvement, and lymphoma. The disease may occur in isolation [primary SS (pSS)] or in conjunction with other AIDs, most commonly RA or SLE (secondary SS or overlap syndromes).

The diagnosis of SS is difficult to establish because there is no single diagnostic gold standard test, and often a multidisciplinary team and invasive procedures are required. For research purposes, several classification methods have been described, and the most widely used are the 2002 revised American European Consensus Group (AECG) classification criteria [ 1 ]. These are currently being revised anew with the goal of creating an updated version supported by both the ACR and the EULAR. Consequently, SS is frequently misdiagnosed, underdiagnosed or diagnosed at late stages of the disease. Clinical diagnosis often takes 6–10 years [ 2 ], leading to a lag in potential preventive and therapeutic strategies and likely contributing to damage accrual.

In a cohort of patients with sicca symptoms who underwent detailed evaluation for SS, we compared the characteristics of those who described a prior diagnosis or suspected diagnosis of RA, SLE or SSc with those who did not have such prior diagnoses. Our aim was to identify factors that may delay or confuse the diagnosis of SS.

Patients and methods

Subjects

The majority of patients were examined in the Oklahoma Medical Research Foundation (OMRF) Sjögren’s Research Clinic (SRC); alternatively, some subjects underwent a similar evaluation at the University of Minnesota (UMN) in Minneapolis, MN or at the Cedars–Sinai Medical Center (CSMC) in Los Angeles, CA. The OMRF and UMN clinics were established as research-recruitment sites for patients with dry eyes and dry mouth and do not provide routine clinical care or longitudinal follow-up. Potential participants were referred by a variety of health care providers or they may have been self-referred [ 3 ]. The CSMC clinic is a rheumatology practice where patients receive regular clinical care and follow-up and are invited by their attending rheumatologist to participate in the study. Each participant provided written informed consent prior to entering the study in accordance with the Declaration of Helsinki, and the Institutional Review Board of the University of Minnesota, the Institutional Review Board, Oklahoma Medical Research Foundation and the CSMC Institutional Review Board approved the study and all procedures.

The SRC participation involved a single-visit comprehensive evaluation using standardized protocols across the three sites for research classification and sample/data procurement [ 3 ]. In order to participate in the SRC, potential participants underwent a phone interview by trained personnel who verified that they met study inclusion criteria and did not have exclusion criteria [ 1 , 3 ]. As part of the initial interview, all potential participants were asked whether they had ever been diagnosed with RA, SLE or SSc and the response options were no, unsure or yes. For the present study, we considered that a participant had a previous or suspected diagnosis of RA, SLE or SSc if they answered yes or unsure, while those who answered no were considered to have no previous diagnosis of the corresponding AID.

Once the participants were accepted in the study, they were evaluated at the SRC as previously described in detail [ 3 ]. Briefly, the evaluation included an oral exam consisting of measurement of stimulated and timed whole unstimulated salivary flow, a minor salivary gland lip biopsy and collection and storage of saliva. The ocular specialist performed ocular surface staining with lissamine green and fluorescein, unanaesthetized Schirmer’s I test, and collection and storage of tears. Blood samples were collected for general laboratory tests, determination of 27 autoantibody specificities (including ANA, RF, anti-Ro/SSA and anti-La/SSB) and for extraction of DNA, RNA and serum. A physician completed a detailed history and physical examination, including general medical, rheumatological and neurological evaluations. If patients gave a history of a past diagnosis of other AIDs (RA, SLE, SSc, MCTD, myositis, primary biliary cirrhosis or multiple sclerosis), a full review of previous history and medical records was included as part of their evaluation.

A final determination was made about whether subjects met classification criteria for SS [ 1 ], RA [ 4 ], SLE [ 5 , 6 ] or SSc [ 7 , 8 ] by analysing the results of the SRC procedures, a detailed review of their past medical records and laboratory results.

The possible resulting classifications were: pSS, patients that met any four of the six AECG items, as long as either histopathology or serology was positive or presence of any three of the four objective items [ 1 ]; secondary SS, patients with another systemic AID who have at least three AECG items present (one subjective plus any two objective criteria except autoantibodies) [ 1 ]; non-SS AID, patients with another AID (in this case RA, SLE and/or SSc), but without the criteria for SS; and non-SS sicca, patients with sicca who did not meet the criteria for SS or other AID serology. Anti-Ro, anti-La, ANA, anti-dsDNA, anti-Sm and RF were measured as previously described [ 3 ].

Statistics

Categorical data were assessed by Pearson’s χ ² or Fisher’s exact test with 95% CIs, while continuous data were assessed by Student’s t test (GraphPad Prism version 6.0f for Mac OS X, GraphPad Software, La Jolla, CA, USA, www.graphpad.com ). A P ⩽ 0.05 was considered statistically significant.

Results

A total of 1175 subjects with symptomatic dry eyes and dry mouth had full responses to interview questions regarding a prior diagnosis or diagnostic suspicion of RA, SLE or SSc. A negative response was obtained from 651 participants (55.4%), while 524 (44.6%) responded affirmatively or unsure. Of those who responded yes/unsure, 191 had previously been told they had RA, 146 SLE, 29 SSc and 158 two or more of these diseases. While there were many combinations of diseases, the most common one was concurrent RA and SLE (n = 89).

After detailed evaluation of all these patients, 8 (1.2%) in the no response group were confirmed to meet, or at some point to have met, classification criteria for one or more non-SS AIDs in isolation (n = 2; 25%) or in conjunction with SS (n = 6; 75%); while 130 (24.5%) in the yes/unsure group had criteria for one or more non-SS AIDs confirmed (69.2% overlapping with SS and 30.8% in isolation) ( Fig. 1 ). Interestingly, in some cases the confirmed non-SS AID was different from the AID the patient had as a previous diagnosis.

F ig . 1 .

Distribution of study participants based on prior diagnosis or suspicion of SLE, RA or SSc

( A ) At the screening interview, the investigators assessed the possibility and level of certainty of prior alternative diagnoses of autoimmune diseases; these were confirmed or ruled out after complete evaluation for SS and other autoimmune diseases. ( B ) Discordance in pre- and post-study diagnoses for participants reporting previous diagnoses of other autoimmune diseases. ( C ) Discordance in pre- and post-study diagnoses for participants responding negatively to such previous diagnoses.

The prior diagnoses were ruled out in 394 (75.2%) of patients in the yes/unsure group, and of these, 183 (46.5%) met criteria for pSS while 211 (53.6%) did not meet the criteria for pSS. When comparing subjects who answered yes/unsure and had the non-SS AID confirmed with those who answered yes/unsure but in whom the non-SS AID was ruled out, the confirmed group had a significantly higher likelihood of also meeting the criteria for SS (69.2% vs 46.4%; P = 3.4E-06; OR = 2.59, 95% CI: 1.70, 3.96). Finally, among the no responders, the proportion of pSS was 43.8% (n = 285), while 55% (n = 358) were classified as non-SS sicca. The distribution of SS and non-SS sicca did not differ significantly from that observed in the yes/unsure group as a whole (P = 0.521, OR = 1.089, 95% CI: 0.85, 1.4) ( Fig. 1 ).

Participants who reported a previous diagnosis of RA were significantly older [58 (interquartile range: IQR) 50–68 vs 54 (IQR) 43–63, P = 5.89E-06] and more likely to be non-Caucasian (P = 0.03) than participants who did not have a prior suspicion of RA ( Table 1 ). There was no difference in the distribution of gender or ethnicity between the two groups. The difference in age was reversed when those who had a prior diagnosis of RA were split into two groups: confirmed RA and ruled-out RA. People with confirmed RA were significantly younger than their counterparts who did not meet the criteria for RA [54 (IQR) 46.5–62.5 vs 59 (IQR) 51–69, P = 0.002] ( Table 1 ). A positive rheumatoid factor was highly associated with both a prior diagnosis of RA (P = 3.38E-05, OR = 1.92, 95%CI: 1.39–2.66) and a later confirmation of such diagnosis (P = 4.64E-06, OR = 4.29, 95% CI: 2.21–8.33); there were no significant differences in the rate of ANA, anti-Ro/SSA or anti-La/SSB ( Table 1 ). Similar results were found when the comparison was done between confirmed RA and the rest of the cohort (data not shown). Results of anti-CCP tests from medical records were only available for 14 individuals (12 positive and 2 negative), of whom 11 had a previous diagnosis of RA (9 anti-CCP positive and 2 negative); we were able to confirm classification of RA in the 9 anti-CCP-positive patients.

T able 1 .

Demographic and serologic characteristics of participants with and without prior diagnosis of RA ^a

	RA
	Previous diagnosis			Confirmed diagnosis
	Yes n = 191	No n = 845	P-value	Yes n = 53	No n = 146	P-value
Age, median (IQR)	58 (50–68)	54 (43–63)	0.014	54 (46.5–62.5)	59 (51–69)	0.002
Female, n (%)	174 (91.1)	786 (93.0)	0.358	50 (94.3)	131 (89.7)	0.605
Race, n
White b	164	771	0.030	40	127	0.079
Non-White c	27	74		13	19
Ethnicity, n (%)
Non-Hispanic	181 (95.4)	815 (96.5)	0.405	48 (90.1)	141 (96.6)	0.059
Hispanic	9 (4.6)	30 (3.5)		5 (9.4)	4 (2.7)
Serology, n
ANA (+)	136	621	0.528	37	105	0.859
RF (+)	70	445	3.38E-05	34	43	4.64E-06
Anti-Ro/SSA (+)	59	273	0.732	22	43	0.125
Anti-La/SSB (+)	40	177	1.0	9	44	0.438

^a For this analysis, only participants who responded yes to a single prior diagnosis of AID were included; all participants with two or more prior or final diagnoses were excluded.

^b White: Subjects self-identified as White/Caucasian or two or more races, one of which is White.

^c Non-White: Subjects self-identified as African American, Asian, Native American or Alaska Native, Pacific-Islander or two or more races not including White. Bold text denotes statistically significant P-values.

n: number.

Suspected SLE was more common in younger [52 (IQR) 41–59 vs 56 (IQR) 46–64, P = 0.0003] and female (P = 0.017) participants than in those without a prior diagnosis of SLE. These differences disappeared when comparing confirmed SLE vs ruled-out SLE, and there were no racial or ethnic differences between these groups either ( Table 2 ). Presence of ANA was associated both with a prior diagnosis of SLE (P = 8.77E-06, OR = 2.96, 95% CI: 1.77, 4.96) and a confirmation of SLE (P = 0.013, OR = 9.03, 95% CI: 1.17, 69.85); positive anti-Ro/SSA and positive anti-La/SSB were more frequent with confirmed SLE compared with those in which SLE classification could not be made (P = 2.42E-05 and P = 0.023, respectively) ( Table 2 ). Anti-Sm and anti-dsDNA antibodies were measured in all participants: 16 had positive anti-dsDNA antibodies by BioPlex 2200 (Bio-Rad BioPlex 2200 ANA, Bio-Rad, Hercules, CA, USA), 2 of whom had a previous diagnosis of SLE, which we later confirmed; 7 were positive by Chritidia luciliae immunofluorescence (Inova Diagnostics, San Diego, CA, USA), 5 of them had a prior and later confirmed diagnosis of SLE. Anti-Sm antibodies were negative by immunoprecipitation in all participants, but four had positive anti-Sm by BioPlex 2200, of whom three had a previous diagnosis of SLE that we were unable to confirm.

T able 2 .

Demographic and serologic characteristics of participants with and without prior diagnosis of SLE ^a

	SLE
	Previous diagnosis			Confirmed diagnosis
	Yes n = 146	No n = 881	P-value	Yes n = 52	No n = 113	P-value
Age, median (IQR)	52 (41–59)	56 (46–64)	0.0003	51.5 (41.3–58)	53 (41.5–58.5)	0.380
Female, n (%)	142 (97.3)	808 (91.7)	0.017	50 (96.2)	109 (96.5)	0.995
Race, n
White b	130	790	0.771	45	99	0.807
Non-White c	16	91		7	14
Ethnicity, n (%)
Non-Hispanic	138 (94.5)	847 (96.1)	0.366	51 (98.1)	105 (92.9)	0.275
Hispanic	8 (5.5)	34 (3.9)		1 (1.9)	8 (7.1)
Serology, n
ANA (+)	128	622	8.77E-06	51	96	0.013
RF (+)	21	236	0.0009	11	14	0.164
Anti-Ro/SSA (+)	63	263	0.387	30	41	2.42E-05
Anti-La/SSB (+)	28	187	0.661	16	17	0.023

^a For this analysis, only participants who responded yes to a single prior diagnosis of AID were included; all participants with two or more prior or final diagnoses were excluded.

^b White: Subjects self-identified as White/Caucasian or two or more races, one of which is White.

^c Non-White: Subjects self-identified as African American, Asian, Native American or Alaska Native, Pacific-Islander or two or more races not including White. Bold text denotes statistically significant P-values.

n: number.

The number of participants with a suspicion of SSc was much smaller: 30 responded yes/unsure and 14 had the diagnosis confirmed ( Fig. 2 ). There were no significant differences in the sociodemographic or serologic characteristics between suspected and confirmed SSc cases.

F ig . 2 .

Final diagnoses of participants with previous diagnosis of other non-Sjögren’s autoimmune diseases

Distribution of final diagnoses of participants who reported a previous diagnosis or suspicion of: ( A ) RA; ( B ) SLE; ( C ) SS; and ( D ) two or more autoimmune diseases.

Anti-Ro/SSA and anti-La/SSB autoantibodies were most frequently present in subjects with pSS (60.9 and 38.9%, respectively), irrespective of whether they had a previous diagnosis of another disease or not (P = 0.78 and P = 0.23, respectively, for anti-Ro/SSA and anti-La/SSB), and in patients with secondary SS: 60.4 and 31.3%, respectively. An intermediate level was observed in patients with confirmed RA, SLE or SSc but without SS: they were anti-Ro (+) in 14.3% and anti-La (+) in 4.8% of cases. The lowest proportion of anti-Ro and anti-La autoantibodies was identified in non-SS sicca participants: 4.6 and 4.4%, respectively. Thus, seropositivity for these antibodies does not predict whether a prior diagnosis of RA, SLE or SSc will be confirmed or if the patient will meet the criteria for pSS vs secondary SS.

A past diagnosis of RA, SLE, SSc or more than one disease did not predict whether patients would meet the criteria for SS or not ( Fig. 2 ). Among patients with a past diagnosis of RA, 72 were classified as pSS and 74 as non-SS sicca, while 396 without past diagnosis of RA were pSS and 495 were non-SS sicca (P = 0.30, OR = 0.82, 95% CI: 0.58, 1.17) ( Fig. 2 A). Similarly, patients with a past diagnosis of SLE were classified as pSS in 51 cases and non-SS sicca in 62 instances, while those without prior SLE suspicion were pSS in 417 cases and non-SS sicca in 507 (P = 0.92, OR = 1.0, 95% CI: 0.68, 1.48) ( Fig. 2 B). Among those with prior suspicion of two or more AIDs, only 7.14% had confirmation of any two or more AIDs, while 42.9 and 50.0% were classified as pSS and non-SS sicca, respectively (P = 0.056, OR = 1.05, 95% CI: 0.71, 1.54) ( Fig. 2 D).

We had sufficient data to calculate the diagnostic delay, for example, the time between onset of symptoms and diagnosis of SS, for 323 participants: 103 with a prior diagnosis of SLE, RA or SSc and 220 without other diagnoses. Subjects with prior non-SS diagnoses had slightly longer diagnostic delays than those without other suspected AIDs, but the difference was not statistically significant [7.15 (10.41) vs 5.91 (9.50) years, respectively, P = 0.289].

Discussion

Diagnostic accuracy is of the greatest importance both in the clinical setting and for research purposes; unfortunately, diffuse connective tissue syndromes such as SLE, RA and SS often mimic other conditions and do not have a single gold standard test for diagnosis except expert opinion. Thus, multiple versions of classification criteria have been developed for each disease and their main focus is to standardize definitions for use in research, albeit the various criteria are often also used as a guide for clinical diagnosis. Further complicating the matter, these three diseases have overlapping clinical and serologic features, and some individuals with one AID may develop additional autoimmune disorders, either concurrently or at a later stage of their ailment. This results in diagnostic delay, increased morbidity for patients and misclassification in research studies.

In our cohort of patients with sicca, we identified a significant proportion of subjects eventually classified as pSS who had been told they had RA, SLE, SSc or two or more other diseases. Serological abnormalities, namely the presence of ANAs and/or RF, seem to be the drivers of such misdiagnoses. ANAs are present in >95% of SLE patients [ 5 ], but they are also highly prevalent in patients with SS (59–85%) [ 9 ]; similarly, positive RF is identified in 70–90% of RA cases [ 10 ], and 36–74% of SS patients [ 9 ]. Thus, neither ANA nor RF is highly specific for the diagnosis of SLE or RA, and these tests are not sufficient to support the differential diagnosis with SS. Other highly disease-specific biomarkers, such as anti-CCP for RA and anti-Sm and anti-dsDNA for SLE, should be helpful in achieving accurate early differential diagnoses. Unfortunately, in our cohort, these markers were only available or present in a small subset of participants and did not allow for detailed analysis. Other factors that influenced alternate diagnoses were related to the socio-demographic characteristics classically associated with each disease: being younger and female increased the likelihood of a suspicion of SLE, while being older and non-white increased the suspicion of RA.

While the time from symptom onset to diagnosis was slightly longer in subjects with previous diagnoses of other AIDs than in patients without previous non-SS diagnoses, the difference did not reach statistical significance and did not support our initial hypothesis that erroneous AID diagnoses were central causes of diagnostic delay. One potential reason for the significant discrepancy between the a priori diagnosis and the conclusion reached after our clinical evaluation is that the initial diagnosis was in almost all cases based on a clinician’s impression and in many cases was at best a diagnostic suspicion (when participants responded unsure to having a prior diagnosis of another AID), while we applied the more rigorous and research-oriented classification criteria. Furthermore, the source of referral of our study participants is varied: self-referral, rheumatologists, ocular specialists, dental practitioners and primary care health professionals. The only subset with an homogeneous and well-documented referral source was the CSMC cohort (n = 34), where all the participants were pre-evaluated and invited to join the study by an experienced rheumatologist. Their rate of prior non-SS AID diagnoses was significantly lower than among the OMRF and UMN participants (17.6% vs 45.4%; P = 0.001). It would have been interesting to know whether the specialty of the physician suspecting a connective tissue disease influenced the diagnosis given to the patient in the complete study population; unfortunately, in the case of the OMRF and UMN cohorts, we did not have enough details about the circumstances of the alternative diagnosis to arrive at any conclusions. However, the limited data we have suggest that increased awareness about SS in primary and other less specialized health care providers would improve the accuracy and timeliness of the SS diagnosis.

We have assumed that the previous diagnoses were erroneous, but it is possible that a few subjects had either RA or SLE at a previous point that was not documented in the available medical records, and that at the time of our evaluation they were best classified as pSS. It is also possible and even likely that some of them will go on to develop additional systemic AIDs. The coexistence of SS and SLE has ranged between 9 and 19% in several studies; up to 34% of SLE patients report sicca symptoms, and SLE is the most common systemic AID reported in SS patients [ 11 , 12 ]. The prevalence of sicca symptoms among RA patients is 30–50%, and 4–31% also meet the criteria for pSS [ 11–13 ]. Furthermore, patients with pSS may meet several of the classification criteria for SLE (i.e. ANA, leukopenia, photosensitivity and arthritis) or for RA, such as RF, arthritis and morning stiffness; while individuals with Raynaud’s, sicca symptoms and anticentromere antibodies may have criteria for both SSc and SS. Therefore, it has been proposed that the term secondary SS be dropped from revised classification systems for SS, given that its temporal relationship is subjective and hard to establish [ 11 , 14 ]. Terms such as concurrent AID, polyautoimmunity or overlap syndrome, none of which make this distinction, may be more appropriate [ 15 ]. Our findings further highlight the tenuous distinction in clinical practice between these related disorders.

Long-term prognosis of rheumatic disease depends on accurate and timely diagnosis to determine the optimal therapeutic interventions. Further analysis into the potential causes of delayed, incorrect or incomplete characterization of each patient into diagnostic and prognostic categories is crucial for improving quality of life and outcomes in SS. Perhaps a deep and broad understanding of the fundamental pathophysiological elements leading to the disease, including underlying genotypic variation, will someday result in highly sensitive and specific testing. Until such time, the field is left with expert opinion and criteria.

Funding : This work was supported in part by National Institutes of Health (grant numbers AR053483, AR050782, DE018209, DE015223, AI082714, GM104938 and 1P50 AR060804), the Oklahoma Medical Research Foundation, the Phileona Foundation and the Sjögren’s Syndrome Foundation.

Disclosure statement : R.H.S. has been a paid consultant to UCB in regards to Sjögren’s syndrome and received honoraria from Lilly for speaking to a Lilly corporate function about Sjögren’s syndrome. All other authors have declared no conflicts of interest.