Additional Information from CT or MRI Imaging Can Increase Rheumatologists’ Consensus on Grading Sacroiliitis By Radiography: Results of the Trimage Project

Nigil Haroon¹, Xenofon Baraliakos², Anne Grethe Jurik³, Gercek Can⁴, Ali Balcı⁵, Muhammet Cinar⁶, Ediz Dalkilic⁷, Salim Donmez⁸, Omer Nuri Pamuk⁹, Yavuz Pehlivan¹⁰, Salih Pay¹¹, Handan Yarkan¹², Gokce Kenar¹³ and Nurullah Akkoc¹³, ¹Rheumatology, Toronto Western Hospital, University of Toronto, Spondylitis Clinic, Toronto, ON, Canada, ²Rheumazentrum Ruhrgebiet Herne, Ruhr-University Bochum, Herne, Germany, ³Department of Radiology, Aarhus University Hospital, Aarhus, Denmark, ⁴Department of Rheumatology, Dokuz Eylul University Faculty of Medicine, Izmir, Turkey, ⁵Radiology, Dokuz Eylul University, Izmir, Turkey, ⁶Division of Rheumatology, Gülhane Military Medical Academy, School of Medicine, Ankara, Turkey, ⁷Department of Rhematology, Uludag University Faculty of Medicine, Bursa, Turkey, ⁸Department of Rheumatology, Trakya University Faculty of Medicine, Edirne, Turkey, ⁹Rheumatology, Trakya University Faculty of Medicine, Edirne, Turkey, ¹⁰Rheumatology, Uludag University Medcal Faculty, Bursa, Turkey, ¹¹Rheumatology, Yüksek Ihtisas University, Special Koru Hospital, Ankara, Turkey, ¹²Rheumatology, Dokuz Eylul University Faculty of Medicine, İzmir, Turkey, ¹³Rheumatology, Dokuz Eylul University Faculty of Medicine, Izmir, Turkey

Meeting: 2016 ACR/ARHP Annual Meeting

Date of first publication: September 28, 2016

Keywords: Computed tomography (CT), Imaging, MRI, radiography and spondylarthritis

Session Information

Date: Sunday, November 13, 2016

Session Title: Spondylarthropathies and Psoriatic Arthritis – Clinical Aspects and Treatment - Poster I: Axial and Peripheral Spondyloarthritis – Clinical Aspects, Imaging and Treatment

Session Type: ACR Poster Session A

Session Time: 9:00AM-11:00AM

Background/Purpose: The diagnosis of ankylosing spondylitis (AS) is anchored on definitive changes of sacroiliits by radiography. This is relevant not only to clinical diagnosis, but also for reimbursement and access to treatment modalities such as TNF inhibitors. However, radiographic grading of sacroiliitis is plagued by lack of interrater reliability. This study evaluated the impact of additional information from sacroiliac MRI or CT images on rheumatologists correctly identifying definite sacroiliitis by radiography.

Methods: Pelvis images from all three modalities (X-ray, CT and MRI) performed within a period of 1 year were available , from 50 patients with chronic back pain. Consensus reads for radiographic sacroiliitis, based on the interpretation of all three imaging modalities by four spondyloarthritis experts (one radiologist and 3 rheumatologists) was the gold standard grading. In the TRIMAGE project 6 rheumatologists and 2 rheumatology fellows scored radiographs online in a random order for the presence of radiographic sacroiliitis, in the first round before and after viewing the MRI images and in the second round before and after viewing the CT images. The same set of images were scored again one month later, following a two day training meeting, which included lectures for each imaging modality and supervised scoring exercises on a training set of 15 patients. Intrarater agreement, was computed for each rheumatologist before and after training. Kappa statistic and percent agreement were used to compare agreement between the expert consensus scoring and those of the rheumatologists/fellows. Sensitivity and specificity of the radiographic ratings by the rheumatologists/fellows were calculated using the final expert consensus classification as the gold standard.

Results: Based on expert consensus grading, 32 of the 50 patients (64%) had radiographic sacroiliitis. The results are summarized in Table 1 and Table 2.

Table 1. Performance of rheumatologists/fellows before and after training for diagnosing radiographic sacroiliitis when the expert consensus scoring is taken as the gold standard
	BEFORE TRAINING			AFTER TRAINING
	Assessment with X-ray	Assessment with X-ray & MRI	Assessment with X-ray & CT	Assessment with X-ray	Assessment with X-ray & MRI	Assessment with X-ray & CT
Intrarater agreement, mean/median	0.412 / 0.422	–	–	0.512 / 0.521	–	–
Kappa score, mean/median	0.228 / 0.253	0.305 / 0.296	0.556 / 0.554	0.363 / 0.422	0.465 / 0.480	0.533 / 0.565
Overall agreement (%), mean/median	56.5 / 58.0	62.0 / 63.0	78.3 / 78.0	67.0 / 70.0	73.8 / 75.0	78.3 / 78.0
Positive agreement (%), mean/median	51.3 / 54.3	59.6 / 61.6	81.2 / 81.7	67.1 / 72.8	75.6 / 79.0	82.1 / 82.1
Negative agreement (%), mean/median	59.8 / 60.6	62.5 / 62.6	73.3 / 73.7	63.6 / 67.5	68.3 / 68.3	69.2 / 74.7
Sensitivity, mean/median	37.9 / 40.6	48.1 / 46.9	75.4 / 75.0	59.8 / 65.6	71.5 / 73.5	80.1 / 76.6
Specificity, mean/median	89.6 / 91.7	86.8 / 91.7	83.3 / 88.9	79.9 / 83.4	77.8 / 86.1	75.0 / 86.1

Table 2. Kappa scores before and after training for the agreement between the rheumatologists/fellows and experts at different rounds of scoring
	BEFORE TRAINING			AFTER TRAINING
	Assessment with X-ray	Assessment with X-ray & MRI	Assessment with X-ray & CT	Assessment with X-ray	Assessment with X-ray & MRI	Assessment with X-ray & CT
Rheumatologist 1	0.178	0.205	0.503	0.491	0.491	0.327
Rheumatologist 2	0.288	0.363	0.707	0.468	0.525	0.55
Rheumatologist 3	0.11	0.492	0.399	0.045	0.478	0.33
Rheumatologist 4	0.258	0.288	0.684	0.528	0.532	0.769
Rheumatologist 5	0.363	0.413	0.826	0.376	0.48	0.579
Rheumatologist 6	0.288	0.304	0.32	0.335	0.443	0.595
Fellow 1	0.247	0.247	0.604	0.595	0.669	0.632
Fellow 2	0.093	0.126	0.407	0.069	0.168	0.481

Conclusion: Compared to expert consensus, rheumatologists/fellows had poor to fair performance in identifying radiographic sacroiliitis on plain radiographs, but most of them improved with training. Availability of CT images improved their diagnostic performance even before training while MRI improved it only after training.

Disclosure: N. Haroon, None; X. Baraliakos, None; A. G. Jurik, None; G. Can, None; A. Balcı, None; M. Cinar, None; E. Dalkilic, None; S. Donmez, None; O. N. Pamuk, None; Y. Pehlivan, None; S. Pay, None; H. Yarkan, None; G. Kenar, None; N. Akkoc, None.

To cite this abstract in AMA style:

Haroon N, Baraliakos X, Jurik AG, Can G, Balcı A, Cinar M, Dalkilic E, Donmez S, Pamuk ON, Pehlivan Y, Pay S, Yarkan H, Kenar G, Akkoc N. Additional Information from CT or MRI Imaging Can Increase Rheumatologists’ Consensus on Grading Sacroiliitis By Radiography: Results of the Trimage Project [abstract]. Arthritis Rheumatol. 2016; 68 (suppl 10). https://acrabstracts.org/abstract/additional-information-from-ct-or-mri-imaging-can-increase-rheumatologists-consensus-on-grading-sacroiliitis-by-radiography-results-of-the-trimage-project/. Accessed March 28, 2023.

« Back to 2016 ACR/ARHP Annual Meeting

ACR Meeting Abstracts - https://acrabstracts.org/abstract/additional-information-from-ct-or-mri-imaging-can-increase-rheumatologists-consensus-on-grading-sacroiliitis-by-radiography-results-of-the-trimage-project/