The research leading to these results has received support from the Innovative Medicines Initiative Joint Undertaking under grant Agreement n° 115770, resources of which are composed of financial contribution from the European Union's Seventh Framework Programme (FP7/2007-2013) and EFPIA companies’ in kind contribution. See IMI Europe: www.imi.europa.eu This communication reflects the views of the authors and neither IMI nor the European Union and EFPIA are liable for any use that may be made of the information contained herein.

AUTHORS:

www.approachproject.eu

INTERSITE COMPARISON AND TEST-RETEST RELIABILITY OF CARTILAGE THICKNESS AND COMPOSITIONAL ANALYSIS IN THE APPROACH STUDY - A 2-YEAR MULTICENTER EUROPEAN

EXPLORATORY STUDY FOR PHENOTYPE CHARACTERIZATON OF KNEE OSTEOARTHRITIS

F. Eckstein1,2, S. Maschek2,1, A. Bay-Jensen3, J. Boere4, L. Hussaarts4, C. Ladel5, A. Lalande6, J. Larkin7, J. Loughlin8, A. Mobasheri9, L. Paolozzi6, H. Weinans10, F. P. Lafeber10, F. Berenbaum11, R. Soler12, F. J. Blanco12, T. Isaksen13, I. K. Haugen14,

M. Loef15, M. Kloppenburg15, A. C. Marijnissen10, E. van Helvoort10, W. E. van Spil10, F. Roemer16,17, W. Wirth1,2 1Paracelsus Med. Univ., Salzburg, Austria, 2Chondrometrics GmbH, Ainring, Germany, 3Nordic BioSci., Herlev, Denmark, 4Lygature, Utrecht, Netherlands, 5Merck KGaA, Darmstadt, Germany, 6Servier, Suresnes, France, 7Glaxo Smith Kline, Collegeville, PA, 8Newcastle Univ., Newcastle, United Kingdom, 9State Res. Inst. Ctr. for Innovative Med., Vilnius, Lithuania, 10Univ. Med. Center, Utrecht, Netherlands, 11Hôpital Saint-Antoine, APHP Sorbonne Univ. INSERM, Paris, France, 12INIBIC-Hosp. Univ., A Coruña, Spain, 13Volvat Med. Ctr., Oslo, Norway, 14Diakonhjemmet Hosp., Oslo, Norway, 15Leids Univ. Med. Center, Leiden, Netherlands, 16Boston Imaging Core Lab (BICL), Boston, MA, 17Univ. of Erlangen, Germany

Disease modifying therapy of osteoarthritis (OA) represents an unmet clinical need. Appropriate outcome measures are required to robustly identify patients with different OA phenotypes who may benefit from specific therapies. APPROACH is an exploratory, European, 5-centre, 2-year prospective follow-up cohort study. It entails extensive clinical, imaging, biomechanical and biochemical para-meters, in a cohort recruited using machine learning models based on retrospective patient data to display a high likelihood of radiographic JSW loss and/or knee pain over the course of the study.

Purpose: (1) To report baseline test-retest

reliability for cartilage thickness and composition (T2) by MRI, and (2) to identify potential differences between the participating sites.

Methods: 3 volunteers had both knees imaged at 4 of 5

sites (2 with 1.5T MRI: A Coruña and Oslo; 2 with 3.0T MRI: Leiden and Utrecht; Table 1).

26 participants across the 5 sites had test-retest MRI with repositioning of the knee (Table 2).

Sagittal 3D WATS, 3D FFE, or FLASH MRIs were acquired for determining cartilage thickness (57-120 1.5 mm slices; 0.31mm in-plane resolution; TR 17ms, TE 7ms, FA 12-15°; Fig. 1)

Sagittal multi echo spin echo (MESE) were acquired for generating T2 maps (21 3.0 mm slices with 3.5 mm gap; 0.27-0.31 mm in-plane; TR 2700ms, 5-7 echos between 0-70ms; Fig. 1).

Segmentation of the total weight-bearing femoro-tibial joint was performed using Works 3.0 software (Chondrometrics GmbH, Ainring, Germany) to measure cartilage thickness (1)

3 central MRI slices were analyzed in the medial (MFTC; Fig. 1) and lateral femorotibial compart-ments (LFTC) for determining cartilage T2 (2), given lack of full anatomical coverage.

Sets of inter-site comparison images and test-retest pairs, were read by the same reader, in 1 session, with reference to each other.

The root-mean square (RMS) standard deviation (SD) and coefficient of variation (CV%) were used to estimate measurement variability (3).

Results: 1 volunteer of the inter-site calibration study

displayed motion in most images, so that only 4 knees of 2 participants were available (Table 1).

The mean cartilage thickness in the MFTC was 3.00-3.11 mm across 4 sites, and that in the LFTC 3.33-3.47 mm (Table 1).

The test-retest reliability was 45µm (RMS SD) or 1.5% (RMS CV%) in the MFTC (range 1.2% to 2.9%) across the 3 sites. Test-retest errors in the LFTC were somewhat lower (Table 2).

The inter-site variability (RMS SD and CV%) for cartilage thickness was somewhat greater than the average test-retest errors (Tables 1 & 2).

The cartilage T2 was consistently longer in the superficial than deep layer, and consistently longer in the LFTC than in the MFTC (Table 1).

There were differences in the T2 values between sites, exceeding the variability of the 4 knees.

Longer TEs were apparent at 1.5T (A Coruña) compared with 3.0T (Leiden & Utrecht; Table 1).

The inter-site variability of cartilage T2 was substantially greater than the T2 test-retest errors, and T2 test-retest errors substantially larger than for cartilage thickness (Table 2).

Conclusions: In this pilot study, we have explored inter-site

differences and test-retest reliability of cartilage thickness and composition (T2) analysis by MRI.

A limitation is the small sample, but few multi-center studies have thus far made an effort to ascertain measurement variability across sites.

Cartilage thickness measures displayed small differences between sites and were well repro-duced with repositioning.

Cartilage T2 also showed reasonable test-retest reliability, but notable differences between sites, particularly between 1.5 and 3.0T.

Hence, in contrast to cartilage thickness measurement, T2 should not be pooled across sites for cross-sectional statistical analysis.

References: 1. Wirth W, Eckstein F. IEEE Trans Med Imaging. 2008 Jun;27(6):737-44.

2. Wirth W, et al. Osteoarthritis Cartilage. 2017 Aug;25(8):1313-1323.

3. Glüer CC et al. Osteoporos Int. 1995;5(4):262-70.

Figure 1: Top row: Sagittal gradient echo MRI with water excitation (or fat suppression) acquired for determining cartilage thickness; Bottom row: Sagittal multi echo spin echo (MESE) acquired for generating the cartilage transverse relaxation time T2

Table 1: Inter-site Comparison: Mean+/-SD at each site, and RMS SD

(and CV%) across 4 of the 5 sites Thickness /

T2. n= Cartilage Thickness

Cartilage Superficial T2

Cartilage Deep T2

Medial femoro-tibial compartment (MFTC)

A Coruña [1.5 T] 4 / 4 3.10 +/- 0.09 mm 50.7 +/- 3.2 ms 40.6 +/- 0.7 ms

Oslo [1.5 T] 4 / 0 3.11+/- 0.07 mm T2 not acquired

Leiden [3.0 T] 4 / 4 3.00 +/- 0.07 mm 43.7 +/- 0.8 ms 28.8 +/- 1.5 ms

Utrecht [3.0 T] 4 / 4 3.08 +/- 0.11 mm 47.3 +/- 3.3 ms 28.0 +/- 1.1 ms

RMS SD (CV %) 16 / 12 79 um (2.6%) 4.5 ms (9.4%) 7.1 ms (22%)

Lateral femoro-tibial compartment (LFTC)

A Coruña 3 / 4 3.33 +/- 0.13 mm 54.6 +/- 4.1 ms 42.3 +/- 2.1 ms

Oslo 4 / 0 3.42 +/- 0.14 mm T2 not acquired

Leiden 4 / 4 3.40 +/ - 0.10 mm 48.2 +/- 0.8 ms 33.4 +/- 2.6 ms

Utrecht 4 / 4 3.47 +/- 0.10 mm 49.1 +/- 3.6 ms 30.9 +/- 2.6 ms

RMS SD (CV%) 16 / 12 78 um (2.3%) 4.1 ms (7.9%) 6.2 ms (17%)

Table 2: RMS SD (and CV%) of test retest acquisitions at each site, and across all 5 sites

Thickness / T2: n=

Cartilage Thickness

Cartilage Superficial T2

Cartilage Deep T2

Medial femoro-tibial compartment (MFTC)

A Coruña [1.5 T] 8 / 6 37 um (1.2%) 2.05 ms (4.3%) 2.08 ms (5.0%)

Oslo [1.5 T] 6 / 0 35 µm (1.4%) T2 not acquired

Leiden [3.0 T] 6 / 6 50 um (1.7%) 3.74 ms (6.6%) 1.40 ms (4.1%)

Utrecht [3.0 T] 5 / 5 43 um (1.4%) 1.24 ms (2.5%) 1.97 ms (5.5%)

Paris [3.0 T] 1 / 1 97 µm (2.9%) 0.12 ms (0.3%) 2.42 ms (8.4%)

ALL 26 / 18 45 um (1.5%) 2.55 ms (4.7%) 1.87 ms (5.1%)

Lateral femoro-tibial compartment (LFTC)

A Coruña 8 / 6 39 um (1.4%) 2.85 ms (5.8%) 2.46 ms (5.6%)

Oslo 6 / 0 29 µm (1.1%) T2 not acquired

Leiden 6 / 6 31 um (1.0%) 0.90 ms (2.0%) 0.49 ms (1.4%)

Utrecht 5 / 5 11 um (0.3%) 1.33 ms (3.1%) 1.63 ms (5.4%)

Paris 1 / 1 23 µm (0.6%) 1.75 ms (4.0%) 1.11 ms (4.1%)

ALL 26 / 18 30 um (1.1%) 1.91 ms (4.0%) 1.71 ms (4.5%)