Points to consider for designing trials in systemic sclerosis patients with arthritic involvement

Abstract

Although musculoskeletal involvement is quite common in SSc (arthritic in particular), there have been few trials and even fewer controlled trials of therapeutic agents in arthritis in SSc. In addition, there have been only three outcome measures that have been validated for use in trials of SSc arthritis: the HAQ Disability Index, the Cochin Hand Function Scale and the Hand Mobility in SSc scale. The purpose of this article is to present evidence-based points to consider for the design of trials in SSc patients with musculoskeletal involvement (joints in particular). In addition, we make an argument for including outcome variables that can be validated within a given trial for use in future trials.

Rheumatology key messages

• There have been few trials of therapeutic agents in arthritis in SSc.

  Only three outcome measures have been validated for use in trials of SSc arthritis.

• Unvalidated outcome variables should be included for validation in future SSc arthritis trials.

Introduction

In RA, major progress has been made with regard to the treatment of joint swelling and pain. In part, progress in this area was dependent on the development of standardized outcome measures and efficient study designs. While joint pain and swelling as well as tendinopathy are common in SSc, no therapy has been approved for the arthritis of this disease. It is reasonable to assume that mild anti-inflammatory and moderate analgesic medications are effective in SSc, but more potent agents will need to be tested in a rigorous fashion for safety and efficacy in SSc arthritis, much as they were in RA and PsA. We propose evidence-based points to consider for the design of trials in SSc patients with musculoskeletal (joints in particular) involvement.

Study design

Double-blind randomized controlled trials (RCTs) are preferred for testing the efficacy and safety of a new treatment. For trials of therapeutic strategies or agents that are to be used in a given circumstance (e.g. analgesia, anti-inflammatory or disease modification in SSc arthritis) for which no active treatment has been found effective, a trial against placebo should be considered, usually on a background of standard care. For a trial in a circumstance for which there is a putative proven treatment, a double-blind RCT of a new agent against the putative active control could be considered (i.e. positively controlled). While such trials may not meet registration requirements initially, they may enable the development of valid measures and at the same time supply sufficient data for a sense of a drug’s efficacy and toxicity. These trials, using one or more validated outcomes, along with partially validated or unvalidated measures, may yet serve the best interests of SSc patients until additional validated measures become available [1].

Trial duration

Consideration of trial duration should be driven by the expected outcomes. For example: changes in the HAQ Disability Index (HAQ-DI) and tender and swollen joint counts (TJCs and SJCs, respectively) could conceivably occur over 12–24 weeks [2, 3]. Changes in hand function as measured by either the HAQ-DI, Cochin Hand Function Scale (CHFS) or both, or contractures, may also change over the same timeframe [4, 5]. And for regulatory purposes, trials are most often at least 12 months in duration, although specific considerations for shorter or longer trials may be driven by direct interactions with regulatory agencies [6–8].

Primary outcome

A validated measure for SSc should be used as the primary outcome [1]. Outcomes that are fully validated presently include the HAQ-DI, the CHFS or the Hand Mobility in SSc (Table 1) [1]. TJCs and SJCs have been used in SSc trials, but have usually been unresponsive when tested in RCTs. Unfortunately they are not yet fully validated in SSC [1]. It might be appropriate to consider using these measures or composite measures such as the ACR response criteria as exploratory measures and validating them in an early phase 2a study while using a validated measure as a primary outcome, thus allowing TJC and SLC to be used in phase 3 studies. Imaging techniques could also be considered. If not already validated, an attempt to validate them could be made during the trial [1]. Any appropriately validated measure should be used for the specific outcome of interest.

Table 1.

Outcome measures and their validity characteristics by the OMERACT filter in relation to SSc arthritis

Measure	Truth				Discrimination		Feasibility	Ready for use in clinical trials as primary outcome?	Primary or secondary outcome measures
	Face validity	Content validity	Construct validity	Criterion validity	Reliability	Sensitive to change
HAQ-DI	Y	Y	Y	Y	Y	Y	Y	Y	Prefer 2°
TJC	Y	Y	PV	NT	NT	PV	Y	N	2°, prefer 1°
SJC	Y	Y	PV	NT	NT	PV	Y	N	2°, prefer 1°
Synovitisa	Y	Y	NT	NT	NT	NT	Y	N	2°
Tendon friction rub	Y	Y	PV	NT	NY	PV	Y	N	2°
Contractures, small jointsb	Y	PV	Y	NT	NT	NT	?	N
Contractures, large joints	Y	PV	NT	NT	PV	N	Y	N
Finger to palm	Y	Y	Y	Y	Y	Y	Y	N, with reservations
ESR, CRP	Y	NT	NT	NT	PV	PV	Y	N	2°
HAMIS scale	Y	Y	PV	RT	NT	Y	Yc	UN	1°, prefer 2°
Cochin Hand Function Scale	Y	Y	Y	Y	Y	Y	Yc	UY	1°
SFAQ	Y	Y		N		?	Y	N
SHAQ-VAS	Y	Y	NT	NT	NT	Y	Y	N
SF-36	Y	Y	Y	Y	Y	Y	Y	Y	Prefer 2°
WHO-QoL	Y	Y	NT	NT	NT	NT	Y	N
Radiographic	Y	Y	NT	NT	NT	NT	Yc	N
US	Y	Y	NT	NT	NT	NT	Ya,c	N
MRI	Y	Y	NT	NT	NT	NT	Y	N
Physician’s global VASd	Y	Y	NT	NT	NT	NT	Y	N	2°
Patient global VAS	Y	Y	NT	NT	NT	NT	Y	N	2°
Patient pain VAS	Y	Y	NT	NT	NT	NT	Y	N	2°
SSc activity index	Y	Y	N	?	N	N	Y	N
(FTP) SSc disease severity	Y	?	Y	?	Y	Y	Y	N

Adapted from Clements PJ, Allanore Y, Khanna D, Singh M, Furst DE. Arthritis in systemic sclerosis: systematic review of the literature and suggestions for the performance of future clinical trials in systemic sclerosis arthritis. Seminars in Arthritis and Rheumatism 2012;41(6):801–14 [1]. Copyright 2012, with permission from Elsevier. ^aEither tender or swollen joints (usually a dichotomous value). ^bPossibly feasible in a 1 year multicentre randomized clinical trial. ^cHand joints only. ^dPhysician’s global is the global assessment when a physician asked about the change in the patient’s course (over 1–2 years) from baseline. 1°, primary; 2°, secondary; ?, unknown; FTP, finger-to-palm; N, no; NT, Not tested; PV, Partially validated; SFAQ, Scleroderma Functional Assessment Questionnaire; SF-36, 36-item Short Form Health Survey; SHAQ, Stanford HAQ; N, No; Y, Yes; VAS, visual analogue scale; WHO-QoL, World Health Organization quality of life; Y, yes.

Since baseline musculoskeletal disease may affect the degree of response, it would be useful to consider the limits of such disease activity at baseline guided by the principle that patients should have potentially reversible baseline disease activity. For example, patients might only be included if their HAQ-DI is ⩾1.0, CHFS is ⩾10 or 15 and/or their TJC is ⩾6 to reduce the chances of a significant floor effect.

Secondary outcomes

To support the primary outcome, secondary outcomes that have not been fully validated can be considered (Table 1) [1, 9]. These include TJC, SJC, ESR and CRP and the number of tendon friction rubs (although the latter three measures occur in only 20–30% of patients, resulting in statistical underpowering if another measure is primary). It would be useful to consider including measurements within the trial with the dual purpose of being supportive measures and to further validate these measures. When these outcomes are validated, they can be used as primary measures in future studies.

Exploratory outcomes

Other outcomes of interest such as X-rays, US, unvalidated patient response measures or serological measures can be considered as exploratory outcomes (Table 1) [1]. They can be included without consideration of statistical issues and would help to understand their variability and responsiveness. Storing serum, plasma or cells for future testing should be considered.

Patient inclusion criteria

Patients should have SSc as defined by the 2013 classification criteria of the ACR/EULAR for the diagnosis of SSc [10]. One might wish to include both limited and diffuse cutaneous subsets in a study. Since most outcomes have been developed in the diffuse cutaneous subset, this group is often used. However, if outcomes are developed in both limited and diffuse patients, recruitment may be easier and a more general population of SSc patients will be included.

Since most trials are in adults, patients ⩾18 years of age should be considered, although children could be included if the therapeutic risk/benefit and clinical and preclinical safety data allow it [6, 11]. A caveat with children: a confounder in children is growth.

Disease duration needs to be defined as needed to accomplish the objectives of the trial (see above). Gender should be considered, particularly if there is any reason to consider that response or adverse events may be different among men or women. Usually, however, gender is not limiting.

Patient exclusion criteria

Although exclusion of non-SSc rheumatic diseases is already part of the 2013 ACR/EULAR classification criteria, it bears repeating that careful consideration should be given to exclude other symptomatic rheumatic diseases or complications of rheumatic disease that might confound the ability to see a response. Thus one might wish to exclude OA, SLE, RA and FM if these diseases are likely to confound the outcome of tender or swollen joints if one wishes to validate these measures. If one wishes to include some potential confounding illnesses, it would be advisable to stratify these patients and/or consider their effect on response to therapy and statistical power. With potential confounding, the possibility of a false-negative conclusion may increase, requiring larger patient groups. On the other hand, if included, this can increase the generalizability of the results and may improve recruitment [6–8].

Other SSc-like diseases such as eosinophilic fasciitis, nephrogenic systemic fibrosis, scleromyxedema or environmental exposures associated with SSc-like disease should be excluded to improve the accuracy of diagnosis and sensitivity to change [12].

SSc-associated organ system involvement that might compromise the likelihood of survival should be considered and might be excluded (e.g. uncontrolled pulmonary arterial hypertension; the need for total parental nutrition; severe lung, heart or renal disease). These patients may confound the ability to show differences in treatment because they may be too sick to attend regularly or may not survive a longer trial, thus increasing mortality that might be ascribed to the test treatment [6–8].

Consideration should be given to the presence of other diseases that might compromise the survival of the patient to the end of the study or could compromise the ability to measure the outcomes of the trial (e.g. active cancer and diabetes with neuropathy) [6–8]. Consideration should also be given to whether anti-rheumatic medications that might compromise the ability to measure response should be excluded. These might include prednisone (or equivalent) >10 mg/day and immunosuppressive or anti-inflammatory medications. On the other hand, if one wishes to include patients who are using those medications that might make the results more generalizable, then it would be appropriate to consider if the doses of the medications are stable for some time prior to baseline, such as at least 1 month of stable medication prior to baseline [6–8].

Since renal, hepatic or cardiac function may affect the disposition of a test medication, the study design should consider the limits of acceptable renal, hepatic and cardiac function. For example, many trials require serum creatinine to be < 2.0 mg/dl and stable for a number of months and hepatic function tests to be normal [6–8].

States that might endanger the patient when using the test or control medication should be excluded (e.g. patients should not be pregnant or breastfeeding at the randomization visit or during the trial) [6–8].

Analysis

Analyses considered for trials in general need to be considered in SSc arthritis trials. It is important to be able to fully describe the patient population, to consider the number of patients needed for the trial, to describe the analysis of response and to describe how to handle inevitable missing data [6–8].

At a minimum, the following analyses should be considered: patient disposition (table or figure) and descriptive information by patient groups, including demographics, ancillary disease characteristics, concomitant medications and concomitant diseases, as well as baseline values for outcome measurements. Comparisons among the groups to establish baseline uniformity are strongly recommended; a predefined analysis plan for assessing the effect of the test medication or therapy is important and could include Student’s t-test, correlations (Pearson or Spearman), analysis of variance, analysis of covariance, Wilcoxon or other non-parametric tests, including logistic or linear modelling or generalized estimating equations. Because missing data are inevitable, some plan should be outlined for dealing with missing data (e.g. dropouts); in addition, an alpha value should be defined (a priori) at the traditional 5% error rate or a different error rate (such as 10–15% in proof-of-concept studies). Analysis plans should account for significantly different covariates at baseline. Appropriate tests can include analysis of variance, analysis of covariance, linear or logistic regressions, generalized estimating equations and survival analyses, among others. Adverse events are going to occur and need to be presented. Analysis of frequency in the group of interest compared with the control group can be analysed by chi-square (often using Fisher’s exact test for low sample size).

Conclusion

Double-blind RCTs of placebo on background therapy are often useful and should be considered. Active control trials can be done when a proven therapy is available. Both limited and diffuse cutaneous disease should be considered. Patients should have a minimum degree of active, potentially reversible, inflammatory arthritis. Trial durations are dependent on the chosen outcome and can vary from 12 weeks to 12 months. Only the HAQ-DI, Cochin Hand Function and Hand Mobility in SSc scales are validated for SSc arthritis trials, although TJC and SJC have been used. Other measures such as US and inflammatory markers can be included and are encouraged for validation.

Supplement

This paper is part of the supplement titled Points to consider: systemic sclerosis and was funded by an unrestricted educational grant from EULAR.

Funding: No specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

Disclosure statement: Y.A. received research grants or honorarium from Actelion, Bayer, Bristol-Myers Squibb, Boehringer, ChemoMab, Genentech/Roche, Galapagos, Inventiva, Medac, Pfizer, Sanofi/Genzyme and Servier. D.K. is supported by the National Institutes of Health/National Institute of Arthritis and Musculoskeletal and Skin Diseases K24 AR063120 and has received investigator-initiated grants and acts as a consultant to Actelion, Bristol-Myers Squibb, Bayer, Corbus, Cytori, ChemoMab, GlaxoSmithKline, Genentech/Roche and Sanofi-Aventis. D.E.F. has received grant/research support from AbbVie, Actelion, Amgen, Bristol-Myers Squibb, Corbus, National Institutes of Health, Novartis, Pfizer and Roche/Genetic and consulting fees from AbbVie, Actelion, Amgen, Bristol-Myers Squibb, Cytori, Novartis, Pfizer and Roche/Genentech. The other author has declared no conflicts of interest.