Verification of Berlin algorithm for diagnosing undifferentiated spondyloarthropathy patients in Chinese population

Available online at

www.sciencedirect.com Joint Bone Spine 76 (2009) 146e149

Original article

Verification of Berlin algorithm for diagnosing undifferentiated spondyloarthropathy patients in Chinese population Zetao Liao a, Jieruo Gu a,*, Feng Huang b, Zhimin Lin a, Like Zhao a, Buyun Yu a a

Department of Rheumatology, Third Affiliated Hospital of Sun Yat-Sen University, 600 Tianhe Road, Guangzhou 510630, China b Department of Rheumatology, Chinese PLA General Hospital, Beijing 100853, China Accepted 9 June 2008 Available online 10 January 2009

Abstract Objective: To evaluate the diagnosing value of the Berlin algorithm, comparing to that of the established ESSG and Amor criteria in Chinese patients with undifferentiated spondyloarthropathy. Methods: A total of 92 clinically diagnosed undifferentiated spondyloarthropathy patients with axial involvement were compared to 123 patients with other kinds of rheumatic diseases by using the parameters listed in the Berlin algorithm, ESSG criteria, and Amor criteria. Results: In the 92 undifferentiated spondyloarthropathy patients with axial involvement, the prevalence rate of HLA-B27 was 71.76% (61/85). Elevated ESR and/or CRP was found in 40.96% (34/83) SpA patients and abnormal MRI manifestation of sacroiliac joint was found in 91% (39/ 43) SpA patients. The specificity of HLA-B27 was 78% and similar with the Berlin study. The sensitivity/specificity of ESSG, Amor criteria and Berlin algorithm on diagnosing USpA was 72.83%/92.68%, 64.13%/93.50% and 67.39%/95.93%, respectively. The coincidence between the three diagnosing criteria and the rheumatologist’s opinion was moderate. Conclusion: Our study showed the new Berlin algorithm has important value of diagnosing undifferentiated spondyloarthropathy in China, which has the similar diagnosing capacity comparing to the traditional criteria ESSG and Amor criteria. Ó 2008 Elsevier Masson SAS. All rights reserved. Keywords: Spondyloarthropathy, undifferentiated; Diagnosis; Berlin algorithm; Chinese

1. Introduction Undifferentiated spondyloarthropathy (USpA) is one of the commonest subtypes of spondyloarthropathy (SpA). An estimated prevalence was 0.67% [1] and more than 50% of them would develop to ankylosing spondylitis (AS) 5e10 years later. The most commonly criteria used for diagnosing SpA are the European Spondylarthropathy Study Group (ESSG) criteria and Amor criteria [2,3]. The two sets of criteria have been verifying in some clinical studies and showed important value in diagnosing of SpA [4e6]. In the two criteria diagnosis of SpA rely on the presence of radiographic sacroiliitis, which

* Corresponding author. Tel.: þ86 2085252055/þ86 13922280820; fax: þ86 2085252701. E-mail address: [email protected] (J. Gu). 1297-319X/$ - see front matter Ó 2008 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.jbspin.2008.06.015

frequently appears late in the course of the disease. Thus, a long delay may exist between onset of symptoms and establishing a diagnosis [7,8]. In the ESSG criteria, USpA is defined as a group of patients with clinical or radiographic manifestations suggesting SpA but not fulfilling any criteria for the currently established diseases in this group (such as AS, reactive arthritis, psoriatic arthritis, inflammatory bowel disease associated arthritis, and so on) [2]. However a recent study from Spain showed limited value of ESSG criteria in diagnosing patients with suspectable SpA [9]. To make better early diagnosis of AS and axial SpA, Rudwaleit et al. proposed an algorithm for diagnosing SpA with axial involvement, and their results showed the diagnosing probability of 90% using this approach [10]. But so far we do not have the evidences to verify the diagnosing value of the algorithm in other countries or regions. Here we initially evaluated the feasibility and value of the Berlin algorithm

Z. Liao et al. / Joint Bone Spine 76 (2009) 146e149

comparing to the conventional ESSG criteria and Amor criteria for diagnosing the USpA patients in a Chinese population. 2. Methods All subjects were from the inpatients and outpatients of the Third Affiliated Hospital of Sun Yet-sen University from January 2004 to May 2005. Patients consist of axial USpA and other rheumatic diseases were eligible. ‘‘Axial USpA’’ was defined as USpA patients with clinically predominant axial involvement but without radiographic sacroiliitis. The clinical diagnosis of USpA was made by the experienced rheumatologists. Then the clinical data, including the characteristics of back pain, peripheral arthritis, iritis, inflammatory bowel disease, psoriasis, urethritis, alternating buttock pain, enthesitis, family history, good response to NSAIDs, raised acute reactant, HLA-B27 test and the X-ray plain and MRI of sacroiliac joint (SIJ), was collected by a specialist. The inflammatory back pain (IBP) was defined referring to Calin criteria [11]. The control groups were investigated in the same process. The diagnosis of each patient with other rheumatic diseases fulfilled the corresponding criteria. Patients were classified according to the ESSG criteria, Amor criteria and Berlin algorithm for SpA. The diagnosing coincidence between the three sets of criteria and expert’s opinion was calculated by using the chi-square test. All statistics were performed by SPSS 11.0 software. 3. Results Ninety-two axial USpA patients, consisting of 60 males and 32 females, were investigated. The average age was 25.4 years and average onset age was 22.9 years. The disease duration range from 3 month to 15 years and 29.8 months of average. 89 patients of them (96.7%) were onset before 40 years. 85 of them had HLA-B27 test and 61 (71.7%) were B27 positive. 41% (34/83) had raised CRP. And none of them had positive radiographic sacroiliitis on the X-ray plain. 43 patients had MRI exam of SIJ and 39 showed abnormal manifestation (90.7%). The control groups, including 123 axial involved patients with other rheumatic diseases, were consisting of 39 males and 84 females. The age ranged from 13 to 86 years and the average age was 54.4 years. 27 patients (21.9%) were HLAB27 positive. The control group consisted of patients with rheumatoid arthritis (n ¼ 42), osteoarthritis (n ¼ 26), systemic lupus erythematosus (n ¼ 23), gout (n ¼ 17), rheumatic fever (n ¼ 11), inflammatory myopathies (n ¼ 10), undifferentiated connective tissue disease (n ¼ 8), Sjogren syndrome (n ¼ 4) and other infrequent rheumatic diseases (n ¼ 36). Additional 146 health volunteers had HLA-B27 examination and 9 of them were positive and the prevalence was 6.16%. 1. Response of parameters of the Berlin algorithm and ESSG or Amor criteria for diagnosing USpA and the

147

Concordance of diagnosis between Berlin algorithm and ESSG or Amor criteria in the 92 axial USpA: The sensitivity, specificity and the LR value of every parameter are listed in Table 1. 2. The coincidence of diagnosis between the rheumatologists’ diagnosis and the Berlin algorithm, ESSG or Amor criteria in axial USpA patients are showed in Table 2. This table shows concordance of diagnosis between the new Berlin algorithm and the traditional ESSG and Amor criteria. It shows that the Berlin algorithm has similar diagnosing capability with the other two established criteria sets. 3. Comparison of established studies on verification of ESSG criteria, Amor criteria or Berlin algorithm with the present study are showed in Table 3 [5,6,10,12e14]. Comparing to the previous reports, the present study had showed lower sensitivity and similar specificity of ESSG Table 1 List of sensitivity and specificity of every parameter of ESSG criteria. Parameters

Number and sensitivity (%)

Specificity (%)

LR

Inflammatory back pain Asymmetrical arthritis of lower limbs History of psoriasis History of ulcerative colitis or Crohn’s disease Previous inflammation before arthritis in one month (non-gonococcus urethritis, acute diarrhea, glansitis) Alternating buttock pain Sausage digit Heel pain Iritis Enthesitis Family history Positive HLA-B27 Good response to NSAIDs (relief within 48 h and relapse after withdrawal) Raised acute phase reactants (CRP, ESR) Defined radiographic sacroiliitis MRI abnormality of SIJ Fulfill the ESSG criteria Fulfill Amor criteria Fulfill Berlin algorithm

73 (79.3) 34(36.9)

86.2 68.3

5.7 1.2

0 (0) 2 (2.2)

100 100

a

4 (4.3)

99.2

5.3

61 (66.3) 4 (4.3) 16(17.3) 3 (3.2) 31(33.7) 17(18.5) 61/(n ¼ 85) (71.7) 72 (78.2)

95.1 100 98.4 100 98.4 100 78.0 48

13.6

3.3 1.5

34/(n ¼ 83) (40.9)

76.4

1.1

0 (0) 39(n ¼ 43) (90.7)b 67 (72.8) 59 (64.1) 62 (67.4)

98.4 / 92.7 93.5 95.9

b

a

a

10.7 a

20.7 a

9.9 9.9 16.5

LR ¼ sensitivity/(1specificity). ESSG, European Spondylarthropathy Study Group; HLA-B27, human leucocyte antigen B27; NSAIDs, non-steroid anti-inflammatory drugs; SIJ, sacroiliac joint; ESR, erythrocyte sedimentation rate; CRP, C reaction protein; LR, likelihood ratio. a Specificity ¼ 100%, so the LR could not be calculated. b In the study, only 43 USpA patients to check SIJ MRI. 39 of 43 showed positive sacroiliitis on MRI.

Z. Liao et al. / Joint Bone Spine 76 (2009) 146e149

148

Table 2 The diagnosing coincidence between the rheumatologist’s personal opinion and the ESSG criteria, Amor criteria or Berlin algorithm on diagnosing axial cases. Axial USpA patients

Control

Total

ESSG criteria Fulfill Not fulfill Total

67 25 92

9 114 123

76 139 215

Amor criteria Fulfill Not fulfill Total

59 33 92

8 115 123

67 148 215

Berlin algorithm Fulfill Not fulfill total

62 30 92

5 118 123

67 148 215

Chi-square test: ESSG criteria: Pearson Chi-square ¼ 98.834, p < 0.001; McNemar test: p ¼ 0.009; Kappa test: p < 0.001, Kappa ¼ 0.670; Amor criteria: Pearson Chi-square ¼ 81.477, p < 0.001; McNemar test: p < 0.001; Kappa test: p < 0.001, Kappa ¼ 0.597; Berlin algorithm: Pearson Chisquare ¼ 98.393, p < 0.001; McNemar test: p < 0.001; Kappa test: p < 0.001, Kappa ¼ 0.656. Pearson Chi-square tests the correlation between the result of diagnoses criteria and the rheumatologists’ opinion. P < 0.05 shows the existence of correlation. McNemar test shows the diagnosing coincidence between a criteria and the expert. P < 0.05 shows that there are some differences and we can see that the diagnosing sensitivity is lower comparing to the rheumatologists’ opinion. Kappa value also shows the coincidence between the two methods. The value between 0.4 and 0.75 shows a medium coincidence.

criteria, Amor criteria or Berlin algorithm on diagnosing axial USpA [12,13]. 4. Discussion There is no definite criterion for diagnosis of USpA yet. Generally people make the diagnosis depending on exclusion of the other subsets of SpA such as AS, PsA and ReA, etc. The most often cited classification criteria for USpA are the ESSG and Amor criteria which have been evaluated in other clinical studies. For the specificity depends on the choice of control group, the specificity may be different between different studies. But the sensitivities of the two criteria were similar in different studies of different country. The sensitivity of our Table 3 List of the sensitivity and specificity of ESSG criteria, Amor criteria or Berlin algorithm of SpA or USpA in various studies.

Spain Lebanon Turkey Brazil This study (China) Berlin paper Netherland

Berlin algorithm

ESSG criteria

Amor criteria

Sen

Spe

Sen

Spe

Sen

Spe

/ / / / 67.4 90 64.7

/ / / / 95.9 90 /

83.5 91.2 86.6 98.5 72.8 / 83.8

95.2 100 91.1 88.7 92.7 / /

90.8 77.2 88.5 / 64.1 / 70.6

96.2 97.5 91.9 / 93.5 / /

Sen ¼ sensitivity; Spe ¼ specificity.

studies was a little bit lower than other reports but the specificity was similar. In order to make early diagnose for axial SpA, a diagnosing algorithm was proposed at 2004 Europe Berlin annual meeting basing on a systemic analysis of a number of clinical research data. In this algorithm diagnosis is made basing on inflammatory spinal pain together with three clinical features of SpA, plus HLA-B27 and SIJ MRI if the patient has less than three clinical features [10]. Our study showed that for diagnosing USpA in a Chinese population the Berlin algorithm had the sensitivity between the ESSG criteria and Amor criteria. Chisquare test showed that the algorithm had close diagnosing capability with Berlin algorithm and ESSG or Amor criteria, but lower than that of experts’ personal opinion. In the Berlin algorithm, IBP was the basic parameter for diagnosis [10]. About the prevalence of IBP, the Berlin study showed 38e95% and an average value of 75% in SpA. But in USpA the prevalence of IBP were only 54.9%, 49% and 68% in Spain, Brazil and our study, respectively [5,13]. So it seemed too strict to use IBP, defined by Calin criteria, as a screening parameter to make a diagnosis of USpA because it would lead to more than one third of omitted diagnosis. In the Berlin algorithm the SpA clinical features were important parameters for diagnosing early SpA except for IBP, such as asymmetrical arthritis, iritis, inflammatory bowel disease, psoriasis, urethritis, alternating buttock pain, enthesitis and response to NSAIDs treatment. More than 50% of the cases could be diagnosed according to IBP plus three or more parameters of clinical features of SpA. HLA-B27 was another very important parameter with the sensitivity and specificity of 90% and LR value of 9.0 [11]. Furthermore, HLA-B27 was the most valuable single parameter in the distinction of SpA and non-SpA in the ESSG criteria. But because ESSG criteria had to be applicable to population in field studies without needs of laboratory, HLA-B27 was removed from the criteria set [2,15]. For those patients who had not enough clinical features making diagnosis was basing on 1e2 SpA features plus positive HLAB27. In that case, HLA-B27 showed significant value that was not influenced by doctor’s or patient’s opinion. All studies showed the prevalence of HLA-B27 in population (which were 6.16% in our research, 10% in Berlin algorithm data and 9.3% in another Berlin research of population [1] respectively) was significantly different from that in USpA patients (which the percentage were 72% in our research, 90% in Berlin algorithm data or 54% in another Berlin study of USpA respectively). Thus we could clearly consider that HLA-B27 was of great importance to make the diagnosis of USpA. In the ESSG and Amor criteria sets, sacroiliitis on the Xray plain is one of the most important parameters, but it usually takes several years to develop definite radiographic sacroiliitis from onset of symptom [16]. In recent years, many studies have reported the value of SIJ MRI to diagnose sacroiliitis. A systemic analysis had showed advantage of the value of MRI to detect early sacroiliitis comparing to X-ray plain and CT [17]. These studies of MRI showed the sensitivity and specificity reached more than 90% for diagnosing sacroiliitis [18e20] and was one of the strongest parameters to

Z. Liao et al. / Joint Bone Spine 76 (2009) 146e149

diagnose early USpA. In the Berlin algorithm it was the first time that MRI was regarded as a diagnosis parameter in early SpA. But the high cost of MRI limits the widespread use of this technology and in Berlin algorithm it was also considered as the ‘‘final’’ diagnosing step. In our study, 43 USpA patients had MRI test of SIJ and sacroiliitis was detected in 91% of patients, and this was similar with the results of other countries [10]. In summary, our study showed that the diagnosing capacity of Berlin algorithm in axial USpA patients in Chinese population was similar to ESSG and Amor criteria. And our study shared the same point of view as the Berlin algorithm did on the SpA clinical features and HLA-B27 that had important diagnosing value for USpA patients. In this study if we define IBP base on Calin criteria, the omitted diagnosis rate would be rather high. But the Berlin algorithm had not given out any opinion for the USpA patients without inflammatory back pain. So the diagnosing experience for such patients should be thought highly of. Acknowledgement

[6]

[7] [8]

[9]

[10] [11] [12]

[13]

[14]

The work of Dr. Gu Jieruo’s work is supported by research Grant 30325019 of National Science in China. [15]

References [16] [1] Braun J, Bollow M, Remlinger G, et al. Prevalence of spondylarthropathies in HLA-B27 positive and negative blood donors. Arthritis Rheum 1998;41:58e67. [2] Dougados M, van der Linden S, Juhlin R, et al. The European Spondyloarthropathy Study Group preliminary criteria for the classification spondyloarthropathy. Arthritis Rheum 1991;34:1218e27. [3] Amor B, Dougados M, Mijiyawa M. Criteria de classification des spondyloarthropathies. Rev Rhum 1990;57:85e9. [4] Boyer GS, Templin DW, Goring WP. Evaluation of the European Spondyloarthropathy Study Group preliminary classification criteria in Alaskan Eskimo populations. Arthritis Rheum 1993;36:534e8. [5] Collantes-Estevez E, Cisnal del Mazo A, et al. Assessment of 2 systems of spondloarthropathy diagnostic and classification (Amor and ESSG) by

[17] [18]

[19]

[20]

149

a Spanish multicenter study. European Spondyloarthropathy Study Group. J Rheumatol 1995;22:246e51. Baddoura R, Awada H, Okais J, et al. Validation of the European Spondyloarthropathy Study Group and B. Amor criteria for spondyloarthropathied in Lebanon. Rev Rhum Engl Ed 1997;64:459e64. Koh WH, Boey ML. Ankylosing spondylitis in Singapore: a study of 150 patients and a local update. Ann Acad Med Singap 1998;27:3e6. Mau W, Zeidler H, Mau R, et al. Clinical features and prognosis of patients with possible ankylosing spondylitis. Results of a 10-year followup. J Rheumatol 1988;15:1109e14. Collantes E, Veroz R, Escudero A, et al. Can some cases of ‘‘possible’’ spondyloarthropathy be classified as ‘‘definite’’ or ‘‘undifferentiated’’ spondyloarthropathy? Value of criteria for spondyloarthropathies. Spanish Spondyloarthropathy Study Group. Joint Bone Spine 2000;67: 516e20. Rudwaleit M, van der Heijde D, Khan MA, et al. How to diagnose axial spondyloarthropathies early. Ann Rheum Dis 2004;63:535e43. Calin A, Porta J, Fries JS, et al. Clinical history as a screening test for ankylosing spondylitis. JAMA 1977;237:2613e4. Erturk M, Alaca R, Tosun E, et al. Evaluation of the Amor and ESSG classification criteria for spondylarthropathies in a Turkish population. Rev Rhum Engl Ed 1997;64:293e300. Cury SE, Vilar MJ, Ciconelli RM, et al. Evaluation of the European Spondylarthropathy Study Group (ESSG) preliminary classification criteria in Brazilian patients. Clin Exp Rheumatol 1997;15:79e82. Heuft-Dorenbosch L, Landewe´ R, Weijers R, et al. Performance of various criteria sets in patients with inflammatory back pain of short duration; the Maastricht early spondyloarthritis clinic. Ann Rheum Dis 2006;66:92e8. Hawkins BR, Dawkins RL, Christiansen FT, et al. Use of the HLA-B27 test in the diagnosis of ankylosing spondylitis: a statistical evaluation. Arthritis Rheum 1981;24:743e6. Khan MA. Ankylosing spondylitis: introductory comments on its diagnosis and treatment. Ann Rheum Dis 2002;61(Suppl. 3):iii3e7. Braun J, van der Heijde D. Imaging and scoring in ankylosing spondylitis. Best Pract Res Clin Rheumatol 2002;16:573e604. Feng F, Yu W, Meng CL. Fat at sacroiliac joints: MR imaging findings (comparison and analysis of 18 volunteers and 52 patients with Ankylosing Spondylitis). Zhongguo Yi Xue Ke Xue Yuan Xue Bao 2002;24: 79e83. Braun J, Bollow M, Eggens U, et al. Use of dynamic magnetic resonance imagine with fast imaging in the detection of early and advanced sacroiliitis in spondiloarthropathy patients. Arthritis Rheum 1994;37:1039e45. Puhakka KB, Jurik AG, Egund N, et al. Imaging of sacroiliitis in early seronegative spondylarthropathy. Assessment of abnormalities by MRI in comparison with radiography and CT. Acta Radiol 2003;44:218e29.