Abstract
Objective. The Tight Control of inflammation in Psoriatic arthritis (TICOPA; isrctn.com: ISRCTN30147736) trial compared standard care (StdC) and tight control (TC) in early psoriatic arthritis (PsA), demonstrating better outcomes for TC. This substudy evaluated the performance metrics of modern imaging outcomes and compared them to the clinical data.
Methods. Non-contrast 0.2T magnetic resonance imaging (MRI; single hand) was assessed using the Outcomes in Rheumatology (OMERACT) PsA MRI Scoring System (PsAMRIS) with an additional global inflammation score. Ultrasound (US; same hand) was scored for greyscale, power Doppler, and erosions at the metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints and scores summated.
Results. Seventy-eight patients had paired (baseline and 48 weeks) US data and 61 paired MRI data; 50 had matched clinical, MR, and US data. Significant within-group changes were seen for the inflammatory PsAMRIS components at MCP level: MRI global inflammation [median difference (range), standardized response mean (SRM)]: 3.25 (−5.0 to 12.0), 0.68; 1.0 (−4.5 to 17.5), 0.45 for TC and StdC, respectively. Similar within-group differences were obtained for US: 1.0 (−13.0 to 23.0), 0.45; 3.0 (−6.0 to 21.0), 0.77 for TC and StdC, respectively. No differences were seen between treatment groups. Significant correlations were found between baseline and change MRI and US scores. A significant correlation was found between baseline PsA disease activity scores and MRI global inflammation scores (Spearman ρ for MCP, PIP: 0.46, 0.63, respectively). No differences in erosion progression were observed.
Conclusion. The PsAMRIS and US inflammation scores demonstrated good responsiveness. No between-group differences were demonstrated, but this substudy was likely underpowered to determine differences between the 2 treatment strategies.
Treating inflammatory arthritis as early as possible to minimize damage and functional disability has been shown to be effective in rheumatoid arthritis (RA)1, and the concept has been extended to other inflammatory arthritides such as psoriatic arthritis (PsA). The Tight Control of inflammation in Psoriatic arthritis (TICOPA) trial targeted early, treatment-naive patients and demonstrated improved clinical outcomes above usual care but was unable to demonstrate an advantage regarding radiographic progression in hands and feet2.
Modern imaging modalities such as magnetic resonance imaging (MRI) or ultrasound (US) provide sensitive tools to explore both objective inflammation and damage responses, though there are few PsA studies using these modalities3. It is also unclear, given the known patterns of PsA joint involvement, whether imaging a single hand (as is typically done in RA trials using MRI) will provide a responsive tool.
The aim of our study was therefore to describe and compare the performance metrics of commonly used MRI and US scores in an imaging substudy of the TICOPA study, and to compare these imaging outcomes with the clinical data obtained in this randomized trial.
MATERIALS AND METHODS
The full trial protocol and clinical results of the TICOPA study have been previously reported (isrctn.com: ISRCTN30147736)2,4. In brief, this randomized, controlled, parallel group, open-label, multicenter clinical trial recruited people with early (less than 2 yrs) treatment-naive PsA. The trial had ethical approval from North East York Ethics Committee (14/NE/1090) and all participants gave written informed consent. The primary objective of the main trial was to compare tight control (TC) with standard care (StdC), using minimal disease activity (MDA)5 as the treatment target. Participants received either TC or StdC for a period of 48 weeks. Participants randomized to TC were seen every 4 weeks by the study physician and treated according to a predefined treatment protocol. Participants randomized to the StdC arm were treated in a general rheumatology outpatient clinic supervised by a consultant rheumatologist. These patients were generally reviewed every 12 weeks but were seen more often if clinically indicated, with no formal measures of disease activity used in clinical decision making. A blinded assessor collected clinical assessments and patient-reported outcomes every 12 weeks. Disease activity was measured using the PsA Disease Activity Score (PASDAS), which assesses patient and physician’s global assessment of disease, tender and swollen joint counts, dactylitis and enthesitis, C-reactive protein (CRP), and the physical summary subscale of the Medical Outcomes Study Short-Form 36 health-related quality of life survey6.
MRI
At the Leeds site, consenting patients were invited to participate in both MRI and US substudies, imaging the hand (the most affected hand, or the dominant hand if both were asymptomatic). Imaging was performed with both a non-contrast low field extremity MRI (0.2T C-scan, Esaote) and an US scan at baseline and 48 weeks.
MRI imaging
For the MRI scan, the imaging sequences and details of scoring are as follows:
Scout. Whole hand FOV 140*140 TR 140 ms. Matrix 192*128.
STIR coronal. TR 2620 ms. 160*160 matrix 192*144. 3 slices. 24 echoes.
STIR sagittal. TR 2840 ms. 190*190. 192*144. 4 slices. 25 echoes.
T1 3-D coronal. T3D T1. TR 35 ms. 140*140 80 matrix 192*160 72. 2 slices. 88 echoes.
Images were scored for the second to fifth fingers at each level in the hand [metacarpophalangeal (MCP) joint, proximal interphalangeal joint (PIP), distal interphalangeal joint (DIP)] for the following features based on the Outcomes in Rheumatology (OMERACT) PsA MRI Scoring System (PsAMRIS) score7:
Synovitis: Grading scale: 0 is normal, while 1–3 is mild, moderate, severe, by thirds of the maximum potential volume of tissue. Score range at each level for each finger, 0–36.
Tenosynovitis: Grading scale: the maximal thickness of signal as follows: 0: none; 1: < 1/2 tendon thickness; 2: ≥ 1/2 and < 1 tendon thickness; 3: ≥ 1 tendon thickness. Score range at each level for each finger, 0–36.
Periarticular inflammation (distal and proximal): Grading scale: 0 absent, 1 present on both dorsal and volar aspects. Score range at each level for each finger, 0–24.
Bone edema (distal and proximal): Grading scale: the scale is based on the proportion of bone with edema, compared to the “assessed bone volume,” judged on all available images: 0: no edema; 1: 1–33% of bone edematous; 2: 34–66%; 3: 67–100% scored on either side of the joint. Score range at each level for each finger, 0–72.
Scores for synovitis, tenosynovitis, periarticular inflammation and bone edema were summed to give a “global inflammation” score at each level, for each finger, score range 0–168. The aggregate scores at each level were retained to examine the responsiveness of “global inflammation” in very small joints such as the DIP joints, and to determine which joints demonstrated most change regarding this feature.
Bone erosion (distal and proximal): Grading scale: the scale is 0–10, based on the proportion of eroded bone compared to the “assessed bone volume,” judged on all available images: 0: no erosion; 1: 1–10% of bone eroded; 2: 11–20%, and so on. Scored at either side of the joint. Score range at each level for all 4 fingers, 0–240.
Bone proliferation: Grading scale: 0 absent, 1 present. Score range at each level for all 4 fingers, 0–12.
The images were read by 2 independent readers (NC, GL), anonymized to patient demographics, treatment group, and time order. Interrater reliability for domain scores at each joint level was calculated by intraclass correlation coefficients.
US imaging
One of 2 ultrasonographers (JF and JN) scanned the same hand as the MRI using a Philips HDI 5000 (Best) machine with 12–5 and 15–7 MHz linear transducers and were unaware of the clinical examination findings. The interrater agreement between these assessors for this group of patients has been previously reported8. Power Doppler (PD) was assessed using a pulse repetition frequency of 750 Hz and medium wall filter and gain was adjusted until background signal was removed. Each joint was scanned in both longitudinal and transverse planes from the dorsal aspect. For the small joints of the hand, the second to fifth MCP joints and the second to fifth PIP joints were examined. Greyscale (GS) and PD were scored separately on a 0–3 semiquantitative scale for each joint imaged. A GS score of ≥ 2 and/or a PD score ≥ 1 were used to identify US active joints. The GS and PD scores were summated to give an overall score for “inflammation” (total possible score of 48)9. Erosions were defined as a definite cortical break seen in 2 planes and scored as present or absent at the joint level, so the maximum score for erosions per hand was 8.
Statistical analysis
The original TICOPA study was appropriately powered for its clinical outcome, but no formal power calculation was made for this substudy. Only matched (baseline and followup) MRI and US data, and combined MRI, US, and clinical data were used. There was no data imputation. The clinical composite outcome (PASDAS) was derived as previously described6. Significance was assumed at a level of 5%; no correction was made for multiple comparisons. Interrater reliability for aggregate MR scores was assessed using the intraclass correlation coefficient. The magnitude of MR variable response was compared using the standardized response mean (SRM), calculated as the mean difference between time-points divided by the SD of the difference10. Statistical testing was carried out using SPSS v21 (IBM Corp.).
RESULTS
For the TICOPA study, 206 patients were recruited and of these, 85 entered imaging substudies. Clinical characteristics of the patients in this study were male/female 40/45; mean age 45.1 years; mean tender joint count 11.7; mean swollen joint count 7.3; mean skin score (Psoriasis Area Severity Index) 2.7; mean CRP (mg/dl) 23.9. The majority of patients (n = 59, 69%) presented with polyarticular disease (≥ 5 joints involved). Baseline disease activity was high (mean PASDAS score 5.1) and significant within-group changes in clinical outcomes were seen (TC group, mean change in PASDAS score 2.2, p < 0.0001; StdC 1.1, p = 0.03) but between-group differences were not significant (F = 3.6, p = 0.06). In the imaging substudies, paired observations (baseline and 48 weeks) were available for 61 participants for the MRI and 78 participants for the US groups, with complete paired MRI, US, and clinical data for 50 participants. The demographics of each of these groups (MRI, US, and matched) were very similar (Appendix 1).
MRI results
Interobserver intraclass correlation coefficient for paired observations varied by feature: scores (95% CI) for synovitis 0.85 (0.74–0.91), flexor tenosynovitis 0.73 (0.54–0.85), periarticular inflammation 0.82 (0.69–0.89), bone edema 0.76 (0.59–0.86), bone erosion 0.86 (0.76–0.92), and bone proliferation 0.25 (0.30–0.57). The data for both readers was combined and expressed as the mean. The results for the MRI scores, for each joint level, at each time-point and each treatment group are given in Table 1. At the MCP joint, a significant difference between baseline and 48 weeks was seen in the TC arm for synovitis, flexor tenosynovitis, periarticular inflammation, bone edema, and global inflammation, but not for bone erosion and bone proliferation. Comparable changes were seen in the StdC arm of the study. At the PIP joint, the changes were similar with the exception of bone edema. At the DIP joint, the differences were less pronounced, with only flexor tenosynovitis and global inflammation for both arms of the study significantly different between baseline and followup. SRM varied from 0.70 (periarticular inflammation at the MCP joint in the TC arm) to −0.39 (erosions at the DIP joint in the TC arm) and were generally larger for the TC arm. ANCOVA for individual components of the score (synovitis, tenosynovitis, periarticular inflammation, bone edema, global inflammation, bone erosion, and bone proliferation) at each joint level did not show any difference between the 2 treatment groups at 48 weeks for any of the comparisons (statistics not shown).
MRI scores for each PsAMRIS feature at the metacarpophalangeal joint level, for each treatment group at each timepoint.
MRI scores for each PsAMRIS feature at the proximal interphalangeal joint level, for each treatment group at each timepoint.
MRI scores for each PsAMRIS feature at the distal interphalangeal joint level, for each treatment group at each timepoint.
US results
The results for the US examination at each time-point, and each treatment group, for MCP and PIP joints, are given in Table 2. For about two-thirds of cases, inflammation (synovitis) was represented by a GS score of ≥ 2. A significant difference was seen for the inflammation score between baseline and 48 weeks for both treatment groups. However, there was no difference in scores between treatment groups at 48 weeks (F = 0.38, p = 0.75). For erosions, scores were low (median of 0 for both groups at baseline and 48 weeks) and no significant differences within or between groups were seen at joint or aggregate level.
Ultrasound scores for each group at baseline and 48 weeks.
Relationship between MRI and US data and clinical outcomes
MRI and US scores at baseline, and their change scores, were highly significantly correlated (Table 3). A significant correlation was found between baseline PASDAS scores and MRI global inflammation scores from the MCP and PIP joint regions (Spearman ρ for MCP, PIP, and DIP joint inflammation and PASDAS were 0.46, 0.63, and 0.35, respectively). However, a non-significant positive correlation was found between baseline US inflammation and baseline PASDAS score. Non-significant positive correlations were found between the change in PASDAS score from baseline to Week 48 and the change in global inflammation MRI score over the same time period. A significant positive correlation was found between the change in PASDAS score from baseline to Week 48 and the change in US “inflammation” score (ρ = 0.37, p = 0.02).
Relationship between MRI and US scores at baseline and difference between scores at 48 weeks.
DISCUSSION
In this substudy of the TICOPA trial, the individual low-field MRI inflammation scores reflected a modest degree of inflammation but consistent with another report using the PsAMRIS scoring method in PsA11. Although a within-group improvement in the inflammation components of the PsAMRIS score was demonstrated for the TC group, the improvements were modest overall, as reflected by the SRM, but larger than those seen in the StdC group. However, the “whole body” clinical improvements were reflected in the single-hand MR improvement scores, thus indicating construct validity of the change scores. It is also worth noting that the MRI scans in this analysis were low-field scans where there are limitations to the images, such as low resolution and difficulty visualizing the DIP joints, and there was lack of contrast agent to help define inflammation. The relatively oligoarticular features of PsA, where individual joints may be affected in an asymmetrical distribution, compared to RA, which is more symmetrical and polyarticular, should also be recognized12. In this situation imaging may show large changes in individual joints but collectively, over the whole hand, the magnitude of change may be smaller when compared to polyarticular disease.
US inflammation scores improved in both treatment groups, and there was a significant association between baseline and change in US score and the equivalent clinical scores. In this study, therefore, both US and MRI were responsive, aligned with baseline clinical scores, and in the case of US, aligned with change in clinical scores. It should be noted that MRI and US assessed slightly different joint sets.
In all the imaging/clinical comparisons made in this study, it must be remembered that the imaging focused on the peripheral joints of a single hand, whereas the clinical score is more comprehensive, with both patient-reported measures, joint counts, measures of dactylitis and enthesitis, and an acute-phase reactant. Although the PASDAS response has been shown to correlate with radiographic progression scores13, in our study the use of treatments without proven disease-modifying abilities, such as methotrexate, could lower the effect size and interfere with attempts to demonstrate relationships between clinical course and imaging changes, and the TICOPA study was not powered to demonstrate this. It is also worth noting that the design of the TICOPA study does not allow direct comparison of drug efficacy between conventional synthetic disease-modifying antirheumatic drugs and biologic drugs.
The TICOPA study demonstrated improved clinical outcomes using a treat-to-target approach in early PsA, but there were no differences in radiographic progression between groups. In the current analysis, a substudy of TICOPA, there were similar within-group improvements in clinical outcomes but a significant change in most of the inflammatory components of an extremity MRI score in the TC group over the 48-week study, and a significant improvement in US inflammation scores in both groups. A significant difference between the treatment groups for the change in clinical scores was not found in this substudy, and the imaging modalities also did not demonstrate a between-group difference. It must be remembered that both groups received active treatment for 48 weeks, there being no placebo group in our study. Good correlation between baseline and change scores for MRI and US was found, and good correlation between baseline MRI imaging and clinical scores. Overall, few erosions were seen and there was little progression over 48 weeks in either group.
MRI assesses a greater range of pathologies compared to US yet more recent US machines can now give much better detail compared to those used in this paper. Future studies of this kind using US could include an assessment of enthesitis and tenosynovitis, which may improve responsiveness of a more “global” inflammation construct. In this context, dactylitis reflects many of the pathologies seen in PsA, including synovitis, enthesitis, and tenosynovitis, but reliable US assessments of dactylitis have not yet been developed14.
The limitations of our study relate to both the modalities and the clinical context. This substudy was not powered to show a significant difference in imaging outcomes between 2 active therapies. Second, as noted above, the clinical composite used relates to total disease burden, yet the imaging was confined to a single hand. It may be that more extensive joint assessment, such as that obtained with total body MR, is more closely related to clinical scores such as the PASDAS. Third, the MR technique, being a peripheral scanner without the use of contrast, will have limited ability to demonstrate improvement in inflammation in any tissue.
The imaging substudy of TICOPA reported in this paper provides further validation for the use of both imaging modalities as outcome measures in this disease. The somewhat sporadic joint involvement of PsA, where only a few individual joints may be affected, makes aggregate imaging scores less responsive to change and future imaging studies should perhaps focus on polyarticular disease inclusion, or one manifestation, such as dactylitis, to demonstrate within-and between-group changes in response to treatment.
Acknowledgment
We thank Brian Witham and Dr. Jill Halstead for technical assistance with the MR scans.
APPENDIX 1. Demographics of patient groups imaged in this study.
Footnotes
The TICOPA study was supported by Arthritis Research UK (grant no 18825) and Pfizer. This study presents independent research supported by the NIHR Leeds Biomedical Research Centre (BRC). The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the UK Department of Health and Social Care.
- Accepted for publication October 18, 2019.
REFERENCES
In this issue
