C l AS sification criteria for the diagnosis of P soriatic AR thritis: The CASPAR Study

Rehabilitation Teaching & Research Unit, Wellington School of Medicine & Health Sciences, University of Otago

ClASsification criteria for the diagnosis of Psoriatic ARthritis: The CASPAR Study

Taylor W4, Helliwell P1, Gladman D3, Mease P5, Mielants H6, Marchesoni A2, for the CASPAR

investigators .1Academic Unit of Musculoskeletal Medicine, University of Leeds, UK; 2Istitutio Ortopedico G. Pini, Milan, Italy;

3University of Toronto, Canada; 4University of Otago, Wellington, New Zealand; 5Seattle Rheumatology Associates, USA; 6University of Ghent, Belgium.

Wellington Regional Rheumatology Unit, Hutt Hospital, Wellington, New Zealand


The CASPAR investigators

Australia: M Lassere; Belgium: H Mielants, M Van de Berghe, H Zmierczak, K de Vlam; Canada: A Russell, D Gladman; France: B Fournie, M Dougados, E Dernis, L Gossec, D Zerkak; Ireland: D Veale, O Fitzgerald, M O’Rouke; Morocco: N Hajjaj-Hassouni, N Bentalha; New Zealand: W Taylor, P Healy; Italy: A Marchesoni, C Salvarani, P Macchioni, E Lubrano, I Olivieri; South Africa: A Kalla, J Potts, G Mody, N Patel; Spain: J Torre Alonso; Sweden: B Svensson, U Lindqvist, G Holmstrom, E Theander, S Dahlqvist, G Alenius, K Ek; United Kingdom: A Isdale, D McGonagle, J Holdsworth, H Sharlala, A Adebajo, L Kay, N McHugh, J Lewis, P Owen, N Barkham, V Bejarano, P Emery, P Helliwell, G Ibrahim; United States: C Ritchlin, L Espinoza, L Candia, P Mease, L Wang, L Gunter.


Background

• Classification criteria for distinguishing between groups of PsA and non-PsA

• There are 7 proposed classification criteria for the diagnosis of PsA

• Only 1 has been derived from observed data

• None have been adequately validated


• Moll & Wright (1973)

• Bennett (1979)

• Vasey & Espinoza (1984)

• Gladman (1987)

• ESSG (1991)

• McGonagle (1999)

• Fournie (1999)


RF

negative

Inflamm

atory arthritis

Clinical sacroiliitis

Clinical spondylitis

Clinical enthesitis

Dactylitis

Xray features

HL

A

Evidence of skin

disease

Family history

Other features

Moll and Wright

Bennett

DIP, nodules, asymmetry, synovial tissue and fluid analysis

Vasey and Espinoza

DIP disease

Gladman Excluding other defined diseases

ESSG Asymmetrical lower limb

McGonagle DIP, rare associated conditions

Fournie DIP involvement


Background

• There are 7 proposed classification criteria for the diagnosis of PsA

• Only 1 has been derived from observed data

• None have been adequately validated

• In 2001, CASPAR arose from a collaborative effort determined to address this problem


Aim

• To compare the test performance characteristics of existing classification criteria

• To determine whether new criteria derived from observed data would be more accurate than existing criteria


Design

• Prospective, observational study of consecutive clinic patients with PsA and other inflammatory arthritis

• Target sample size of 1012 in total• 30 clinics in 13 countries• Gold-standard of diagnosis based on

physician’s opinion• Data collected between Feb 02 to Mar 04


Methods

• Data collected included:– Clinical and examination features– Xrays of spine, sacroiliac joints, hands and feet– Rheumatoid factor, [HLA], anti-CCP, stored blood

• Xrays were read centrally by 2 readers in tandem, blinded to diagnosis

• Clinical gold-standard validated by quality control and Latent Class Analysis (statistical modelling)

• New criteria developed using CART and logistic regression


Characteristics of the cohort

PsA (n=588) Controls (n=536)

Disease (%)RA (70), AS (13), UA (7),

CTD (3), other (5)

Age, yrs (mean, SE) 50.3 (0.54) 55.2 (0.62)*

Disease duration, yrs (mean, SE)

12.5 (0.40) 13.3 (0.46)

Male (%) 52.0 37.0*

RF positive (%) 4.6 57.3*

Anti-CCP positive (%) 7.6 54.5*

PASI (median, range) 2.15 (0 to 54)

* p<0.001


327

495

298

29

495361

776

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Gladman McGonagle Fournie ESSG Moll and Wright Bennett Vasey andEspinoza

Prop

ortio

ns

1

10

100

1000

DO

R (log scale)

Not able to be classified Sensitivity Specificity DOR


0

1

2

3

4

5

6

7

8

Gladman McGonagle ESSGMoll &Wright Bennett

Vasey &Espinoza

Fa

lse

po

sit

ive

ra

te (

Od

ds

ra

tio

)

Worse specificity

Better specificity


0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

Tru

e p

os

itiv

e r

ate

(O

dd

s r

ati

o)

0

1

2

3

4

5

6

7

8

Gladman McGonagle ESSGMoll &Wright Bennett

Vasey &Espinoza

Fa

lse

po

sit

ive

ra

te (

Od

ds

ra

tio

)

Worse specificity

Better specificity

Better sensitivity

Worse sensitivity


Validation of ‘gold-standard’

• Data control committee closely examined the data for 124 randomly selected subjects (only 1 control reclassified as PsA)

• Latent Class Anlaysis used to reclassify subjects on the basis of modelling using the agreement pattern between existing criteria: very close agreement with clinical diagnosis (kappa >0.9)


Latent class model fitted to the data (n=949)

Criteria Sensitivity Specificity Probability of a positive response, given

subject is in class-3

McGonagle 0.99 1.00 0.93

Vasey & Espinoza

1.00 1.00 0.24

Moll & Wright 0.95 1.00 0.003

ESSG 0.76 1.00 0.69

Gladman 0.95 1.00 0.08

Bennett 0.53 1.00 0.002


Vasey & EspinozaIn: Calin A, editor. Spondyloarthropathies. Orlando, Florida: Grune & Stratton; 1984. p. 151-185

Psoriatic skin or nail involvement [current psoriasis, history of psoriasis, or nail disease]

PLUS One of these 2

(a) Peripheral pattern (any of):

1/ DIP involvement [finger DIP swollen]

2/ Asymmetry or dactylitis

3/ Symmetry in absence of RF and nodules

4/ Pencil-in-cup deformity, whittling of terminal phalanges, fluffy periostitis and bony ankylosis [radiographic osteolysis, tuft erosion, ankylosis, or juxta-articular new bone formation]

(b) Axial pattern (any of):

1/ Spinal pain and stiffness with the restriction of motion present for over 4 weeks

2/ Grade 2 symmetric sacroiliitis according to the New York criteria

3/ Grade 3 or 4 unilateral sacroiliitis


CART analysis

• Iterative partitioning of the dataset using most discriminating variables to produce the most ‘pure’ groups; progressive pruning back of the tree to balance complexity with accuracy

• CART selected only 2 variables: – history of psoriasis and current psoriasis

– surrogates: nail dystrophy, family history, dactylitis, RF

– specificity 96.8%; sensitivity 96.1%


Discrimination of clinical, laboratory and xray items

Univariate analysis Multivariate analysis

Sensitivity Specificity OR (95% CI) p-value

RF negative 95.1 60.2 27.8 (6.5-125.0) <0.001

Current dactylitis

History of dactylitis

53.6 94.617.9 (5.2-62.1)

6.0 (1.9-19.0)<0.001

History of psoriasis 93.6 97.6 102.6 (37.8-278.5) <0.001

Family history of psoriasis

46.7 91.3 5.6 (2.3-13.3) <0.001

Current psoriasis 88.3 97.8 22.5 (8.5-59.1) <0.001

Juxta-articular new bone formation

18.7 95.4 4.6 (1.3-16.9 <0.001


0.0 0.2 0.4 0.6 0.8 1.0

1 - Specificity

0.0

0.2

0.4

0.6

0.8

1.0S

ensi

tiv

ity

Selected cutpoint, sens 0.93 spec 0.987

ROC Curve

Area under the curve 0.990 (95% CI 0.984 to 0.995)

1. Current psoriasis

2. History of psoriasis (if current psoriasis not present)

3. Family history of psoriasis (if neither history nor current psoriasis is present)

4. Psoriatic nail dystrophy

5. RF negative

6. Current dactylitis

7. History of dactylitis (if current dactylitis not present)

8. Xray signs of juxta-articular new bone formation

(3 or more items)


CASPAR criteriaInflammatory musculoskeletal disease (joint, spine, or entheseal)

With 3 or more of the following:

1. Current psoriasis Psoriatic skin or scalp disease present today as judged by a rheumatologist

2. Personal history of psoriasis (if current psoriasis not present)

A history of psoriasis that may be obtained from patient, family doctor, dermatologist or rheumatologist

3. Family history of psoriasis (if personal history of psoriasis or current psoriasis is not present)

A history of psoriasis in a first or second degree relative according to patient report

4. Psoriatic nail dystrophyTypical psoriatic nail dystrophy including onycholysis, pitting and hyperkeratosis observed on current physical examination

5. A negative test for rheumatoid factorBy any method except latex but preferably by ELISA or nephlemetry, according to the local laboratory reference range

6. Current dactylitis Swelling of an entire digit

7. History of dactylitis (if current dactylitis is not present)

A history of dactylitis recorded by a rheumatologist

8. Radiological evidence of juxta-articular new bone formation

Ill-defined ossification near joint margins (but excluding osteophyte formation) on plain xrays of hand or foot


Conclusion

• Robust design (large sample, many centres, unselected subjects, validation of gold-standard, large number of items examined, convergence of separate statistical approaches)

• Results not applicable to early disease or to non-rheumatic populations (e.g. general population)

• CASPAR criteria are simple, highly specific, and derived from observed patient data: a new standard for the case-definition of PsA?


Acknowledgements

Funding: EULAR, Barnsley District NHS Trust, Groote Schuur Hospital (Cape Town), Department of Medical Sciences (University Hospital, Uppsala), Krembil Foundation, St. Vincent’s University Hospital Radiology Department (Dublin), Inkosi Albert Luthuli Central Hospital (Durban), El Ayachi Hospital (Morocco), National Psoriasis Foundation (USA), The Foundation for Scientific Research of the Belgian Society of Rhumatology, Arthritis New Zealand.

Data Quality Committee: Dennis McGonagle, Philip Helliwell, Mike Green, Leeds; Deborah Symmons, Manchester, UK

Radiology: Guy Porter, Keighly, UKCCP analysis: Neil McHugh, Pat Owen, Bath, UKStatistical analysis: John Horwood, Christchurch, NZ


Discussion• Is better specificity worth the trade-off with

sensitivity?• Construction of CASPAR criteria

– Dependent nature of some items– Meaning of the initial mandatory criterion (note that

1.6% of dataset had no involved joints)?– Should some degree of chronicity be required?

• Practical application of CASPAR criteria– Should dactylitis be excluded from the initial

mandatory criterion (“double-dipping”)?– Hands and feet xrays required


Statistical comparison of sensitivity and specificity.

OR (95% CI) for classifying cases as PsA (true positive rate) and classifying controls as PsA (false positive rate) compared to Vasey

& Espinoza (red means significantly different)

Rule True positive rate (value of more than 1 means

better sensitivity)

False positive rate

(value of less than 1 means better specificity)

Gladman 0.36 (0.20-0.64) 0.72 (0.29-1.81)

McGonagle 1.72 (0.8-3.81) 3.65 (1.85-7.23)

ESSG 0.11 (0.07-0.19) 3.14 (1.57-6.28)

Moll & Wright 0.38 (0.21-0.68) 0.36 (0.11-1.13)

Bennett 0.04 (0.02-0.06) 0.04 (0.003-0.76)


Anti-CCP performance

0.0 0.2 0.4 0.6 0.8 1.0

1 - Specificity

0.0

0.2

0.4

0.6

0.8

1.0

Sen

siti

vity

Sens 0.51, spec 0.96 (>40)

Sens 0.54, spec 0.92 (>20)

ROC Curve

AUC 0.76 (95% CI 0.72 to 0.79)


Item (frequency, N) Sens. Spec.

Normal acute phase reactant (279) 31.8 80.3

RF negative (727) 95.1 60.2

Anti-CCP negative (483) 90.4 56.8

Any tender enthesis (436) 53.2 76.8

Chest wall pain (239) 25.2 83.0

Diffuse enthesis pain (122) 14.0 92.5

Inflammatory heel pain (335) 39.4 80.7

Inflammatory LBP (408) 46.0 74.4

Lumbar stiffness (481) 53.0 51.0

Inflammatory neck pain (407) 38.4 66.2

Neck stiffness (695) 69.0 31.0

Inflammatory thoracic spinal pain (180) 19.6 87.9

Thoracic stiffness (176) 17.0 84.0

Clinical sacroiliitis (263) 30.6 84.4

Absence of subcutaneous nodules (1039) 99.8 15.7

Iritis (62) 4.6 93.4

Dactylitis (338) 53.6 94.6

Nail dystrophy (352) 57.9 97.7

History of psoriasis (538) 93.6 97.6

Item (frequency, N) Sens. Spec.

Family history of psoriasis (318) 46.7 91.3

Current psoriasis on exam (532) 88.3 97.8

Less than 4 MCP involved (628) 65.7 55.0

Any DIP involved (202) 28.9 94.0

Any toe DIP involved (168) 23.3 94.2

Symmetry of joint involvement (924) 79.8 15.1

All joints of a single ray involved (90) 11.2 95.5

Interphalangeal bony ankylosis (74) 11.9 97.0

Bilateral sacroiliitis (111) 11.0 87.0

DIP erosive disease (170) 61.9 89.0

Entheseal erosion (76) 6.7 90.0

Entheseal bony proliferation (143) 15.8 85.0

Juxtaarticular new bone formatn (116) 18.7 95.4

Marginal syndesmophytes (56) 4.5 91.4

Non-marginal syndesmophytes (87) 10.0 90.2

Joint osteolysis (102) 12.6 91.5

Ray involvement (37) 6.1 98.6

Tuft osteolysis (22) 4.3 100.0

Unilateral sacroiliitis (34) 5.38 98.4

Documents

C l AS sification criteria for the diagnosis of P soriatic AR thritis: The CASPAR Study