Evaluating Change in Disease Activity Needed to Reflect Meaningful Improvement in Quality of Life for Clinical Trials in Cutaneous Lupus Erythematosus

Abstract

Background:

Outcome measures of clinical trials in cutaneous lupus erythematosus (CLE) should reflect clinically meaningful improvement in disease activity, as measured by the Cutaneous Lupus Disease Area and Severity Index activity score (CLASI-A).

Objective:

We aim to define the degree of improvement in disease activity meaningful to a patient’s quality of life (QoL).

Methods:

The change in the CLASI-A in 126 patients needed to predict meaningful change in QoL, as defined by the Emotions and Symptoms subscales of Skindex-29, was evaluated by linear regression models.

Results:

In patients with an initial CLASI-A ≥8, a 42.1% or ≥7-point and a 31.0% or ≥5-point decrease in CLASI-A predicts meaningful improvement in the Emotions and the Symptoms subscales, respectively.

Limitations:

This is a retrospective study of prospectively collected data at a single site.

Conclusions:

A CLASI-A score ≥8 for trial entry allows for inclusion of patients with milder disease where CLASI-A improvement by ≥50% is clinically significant and meaningful.

Introduction

Cutaneous lupus erythematosus (CLE) can occur with or without other features of systemic lupus erythematosus (SLE).¹ The disease can impose a significant burden on patients’ lives, compromising mental and psychological health and impairing quality of life (QoL).² Furthermore, it has been established that an improvement in disease activity, as defined by the Cutaneous Lupus Disease Area and Severity Index (CLASI) score for activity (CLASI-A), is correlated to an improvement in quality of life, as measured by Skindex-29.^{3, 4}

The impact of CLE on QoL is severe when compared to the impact of several other skin conditions, and particularly affects the Emotions and Symptoms subscales of Skindex-29.³ In the past 50 years, there have been no new medications approved for the treatment of CLE.⁵ SLE trials face many obstacles, including complex trial end-points, efficacy measures, heterogeneity of the disease and prevalence of background medications.^{5, 6} The US Food and Drug Administration (FDA) has recommended endpoints for lupus that are organ-specific,⁷ and a goal is that implementation of the skin-specific endpoint may lead to FDA-approved treatments for CLE.^{5, 7} From a regulatory standpoint, it is important determine a meaningful change in the CLASI-A score that reflects the patients’ perspective.

While minimal clinically significant improvement in disease activity has been determined,^{3, 4, 8} there has been no quantification of the amount of change in the CLASI-A needed to predict a meaningful change in QoL, as determined by the change in the Emotions and Symptoms subscales. Clinical trials employing the CLASI currently use endpoints that have been discriminatory between effective treatments and placebo (e.g. percentage of patients with ≥ 50% improvement in CLASI-A in patients with a CLASI-A ≥ 10).^9–11 The goal of our project is to define the degree of change in the CLASI activity score correlating with a meaningful improvement in quality of life. This meaningful improvement in CLASI activity score is an important variable in the design and interpretation of future clinical trials.^{12, 13}

Materials and Methods

Patients

This study included 126 patients seen at the Autoimmune Skin Disease Clinic at the Hospital of the University of Pennsylvania who had elected to participate in a research database from 2006 to 2019. The diagnosis of CLE was based on clinical, laboratory and histopathological evidence and determined by an expert physician trained in immunodermatology. This prospective CLE database, established to monitor disease progression and changes in quality of life during routine clinical visits, is approved by University of Pennsylvania’s Institutional Review Board. Patients are not required to have a research visit after their clinical evaluation. We used physician-collected CLASI activity scores and patient-reported Skindex-29 scores for this study. Subjects were excluded if questionnaires were incomplete or if there was only one visit. Subjects with initial CLASI-A ≤ 3 were also excluded as a previous study has shown that there is no additional improvement in the Emotions and Symptoms subscales of Skindex-29 with further improvement in the CLASI scores in this low range.¹⁴ Patients were included if all the responses to Skindex-29 and CLASI scores were completely recorded and if they had at least two research visits.

Cutaneous Lupus Disease Area and Severity Index (CLASI)

The CLASI is a validated scoring system developed to provide an accurate way to measure clinical outcomes for therapeutic trials.¹⁵ The scoring system takes into consideration the activity of the disease and damage caused by the disease, with each scored separately. The scores are based on anatomical locations and the most severe lesion located in that location. Activity is measured by erythema, scale/hypertrophy, mucous membrane involvement, and non-scarring alopecia or hair loss in the past 30 days, with a maximum of 70 points, and is evaluated by a CLASI active score (CLASI-A). Damage is scored by dyspigmentation or scarring, and includes the extent of scarring alopecia in the scalp, with a maximum of 80 points. Higher activity scores indicate more severe disease, with a score of 0–9 indicating mild disease, 10–20 indicating moderate disease, and 21–70 indicating severe disease.¹⁶ CLASI activity scores have been shown to have higher correlations with Skindex-29 subscales than did CLASI damage scores.¹⁷

Skindex-29

The Skindex-29 was developed as a quality of life measurement tool specific to dermatology. It includes 29 questions that are categorized into three subscales: Emotions (Skindex-E), Symptoms (Skindex-S), and Functioning (Skindex-F). Each question is scored 0–4 and then normalized to a 100-point scale, with higher scores indicating worse quality of life. Subscale scores are calculated as the mean score of the questions specific for the individual subscale. Patients with CLE have particularly high scores in the Emotions and Symptoms subscales of Skindex-29, much more so than in the Functioning subscale.³ While the functioning subscale responds to changes in CLASI activity scores, it was found to have poor correlation with disease activity at all CLASI scores.^{4, 14} Due to poor correlation between the Skindex-F and CLASI activity scores, this study only considered Skindex-E and Skindex-S.¹⁴

Prior to this study the values of meaningful change in Skindex-29 subscales had not been established and to do so we used a prospective database of patients with dermatomyositis (DM) which collected the Dermatology Life Quality Index (DLQI) in addition to the Skindex-29 to determine meaningful change in the Emotions and Symptoms subscales of Skindex-29. Using the known value of meaningful improvement in the DLQI, which is a score decrease of 5 points,^{18, 19} we have found that a 9.38-point change in Skindex-E and a 7.37-point change in Skindex-S is meaningful change in quality of life in patients with a CDASI score ≥ 10, a point of disease activity that correlates well with DLQI, Skindex-E and Skindex-S. Both CLE and DM are inflammatory skin conditions that can have a similar magnitude and symptomology of the clinical presentation with the potential for multisystem manifestations.²⁰ The shared symptomology between the two conditions include photosensitivity, pruritus, inflammatory rash, and alopecia, amongst others, all of which have the potential to have a significant, negative impact on a patient’s QoL.^{3, 21–24} Given these similarities, we have applied results from previous studies of meaningful change in the Skindex in dermatomyositis to CLE.

Analysis

Analysis was performed with GraphPad Prism ver. 5 using a significance level of 0.05. Our patient population was characterized using descriptive statistics. All patients were evaluated together to confirm correlations between CLASI-A scores and Skindex-29 subscales using a Pearson correlation. Subsequently, patients with mild, moderate or severe initial CLASI-A scores were analyzed separately, with severity levels categorized as scores of 4–9, 10–20, and > 20, respectively. Patients with mild initial activity were stratified further by analyzing all patients with initial CLASI score >4, CLASI score >5, and at further increments of one-point increases. Using parametric correlations, change in CLASI-A scores was also correlated to changes in the Skindex-E and Skindex-S scores for patients with mild initial disease activity (CLASI-A 4–9) in two different subdivisions: CLASI-A 4–7 and CLASI-A 8–9. The percent change and difference between the CLASI-A scores of the first two visits were estimated and compared with the difference between each of the Skindex-29 subscale scores between the two visits using a linear regression analysis.

Meaningful change in each Skindex-29 subscale, which is the independent variable in the linear regression models and defined as a 9.38-point change in Skindex-E and a 7.37-point change in Skindex-S, was divided by each of the slopes of the linear equation to estimate the percent change and difference in CLASI that was associated with a meaningful change in the Emotions and Symptoms subscales of Skindex-29.

Results

Patient characteristics, including gender, ethnicity, SLE prevalence, and CLE subtype are summarized in Table 1. The median time between the initial visit and the first follow-up visit was 4.0 months (IQR=2.0–9.0) for all patients. In this study, forty-nine patients had an initial CLASI activity score in the mild category (CLASI-A 4–9), 46 patients were in the moderate category (CLASI-A 10–20) and 31 patients were in the severe category (CLASI-A >20). The median change in CLASI-A score was a decrease of 3.0 points (IQR=-8.0–1.0) and the median change in Skindex-29 subscales was a decrease of 10.0 points (IQR=-22.5–2.5) in Emotions and a decrease of 7.1 points (-17.9–3.6) in Symptoms. For all patients, change in the CLASI-A score was correlated with changes in both Skindex-29 subscales using a Pearson correlation: r = 0.498 (p < 0.0001) for Emotions and r = 0.475 (p < 0.0001) for Symptoms. Pearson correlations were also established between change in disease activity and change in Skindex-E (r² = 0.143, p = 0.0229) and change in Skindex-S (r² = 0.027, p = 0.3400) for patients in the lower range of mild initial disease activity (CLASI-A 4–7), and between the change in disease activity and change in Skindex-E (r² = 0.449, p = 0.0122) and Skindex-S (r² = 0.336, p = 0.0380) for patients in the upper range of mild initial disease activity (CLASI-A 8–9).

Table 1.

Patient characteristics

	n	%
Gender
Female	106	84.1
Male	20	15.9
Race
Caucasian	81	64.2
African American	31	24.6
Asian	7	5.6
Other	7	5.6
CLE/SLE (n=46)
Generalized DLE	15	32.6
Localized DLE	9	19.6
Tumid	1	2.17
Panniculitis	0	0.00
SCLE	10	21.7
ACLE	6	13.0
Multiple subtypes	5	10.9
CLE (n=80)
Generalized DLE	15	18.8
Localized DLE	19	23.8
Hypertrophic	2	2.50
Chilblains	1	1.25
Tumid	8	10.0
Panniculitis	1	1.25
SCLE	26	32.5
ACLE	0	0.00
Multiple subtypes	8	10.0

For both subscales, patients with severe initial disease required a smaller percent change in CLASI-A to achieve a meaningful change in QoL (Skindex-E, 25.42%; Skindex-S, 18.90%) than patients with moderate initial disease (Skindex-E, 49.11%; Skindex-S, 33.96%) (Table 2). When looking at the differences in CLASI-A, a smaller reduction in the CLASI-A score was needed to predict meaningful change in the Emotions and Symptoms subscales in patients with initial CLASI activity score of 8–9 (Table 2). In patients with an initial CLASI-A ≥ 8, an improvement of 42.1% in disease activity was associated with a meaningful improvement in the Emotions subscale and an improvement of 31.0% for the Symptoms subscale (Table 3). When looking for the difference in CLASI-A associated with meaningful impact on quality of life in patients with initial CLASI-A ≥ 8, a decrease in CLASI activity by ≥ 7 and ≥ 5 points predicts meaningful change for the Emotions and Symptoms subscales, respectively (Tables 3).

Table 2.

Percent change and difference needed in CLASI-A^a scores to predict meaningful improvement in Skindex-29 subscales (Emotions, Symptoms)^b in patients with a range of initial CLASI-A scores

		Percent Change in CLASI-A	Slope (95% CI)	p-value
Mild Disease Activity (CLASI-A 8–9)
n=13	Skindex-29 Subscale
	Emotions	48.85	0.192 (0.053–0.332)	0.0114
	Symptoms	38.19	0.193 (0.010–0.376)	0.0403
Moderate Disease Activity (CLASI-A 10–20)
n=46	Skindex-29 Subscale
	Emotions	49.11	0.191 (0.062–0.321)	0.0050
	Symptoms	33.96	0.217 (0.130–0.303)	<0.0001
Severe Disease Activity (CLASI-A > 20)
n=31	Skindex-29 Subscale
	Emotions	25.42	0.369 (0.165–0.573)	0.0010
	Symptoms	18.90	0.390 (0.111–0.669)	0.0080
		Δ CLASI-A1	Slope (95% CI)	p-value
Mild Disease Activity (CLASI-A 8–9)
n=13	Skindex-29 Subscale
	Emotions	4.265	2.199 (0.583–3.816)	0.0122
	Symptoms	3.277	2.249 (0.149–4.349)	0.0380
Moderate Disease Activity (CLASI-A 10–20)
n=46	Skindex-29 Subscale
	Emotions	7.415	1.265 (0.369–2.161)	0.0067
	Symptoms	5.051	1.459 (0.861–2.058)	<0.0001
Severe Disease Activity (CLASI-A > 20)
n=31	Skindex-29 Subscale
	Emotions	6.541	1.434 (0.734–2.135)	0.0002
	Symptoms	4.426	1.665 (0.728–2.601)	0.0011

^aHigher scores in CLASI-A indicate more severe disease, with a score of 0–9 indicating mild disease, 10–20 indicating moderate disease, and 21–70 indicating severe disease.

^bQuestions in the survey are normalized to 100 points, with higher scores indicating worse quality of life.

^cΔ CLASI-A is defined as the difference between two CLASI-A scores needed to result in a meaningful improvement in a patient’s quality of life, measured by Skindex-29

Table 3.

Percent change and difference needed in CLASI-A scores to predict meaningful improvement in Skindex-29 subscales (Emotions and Symptoms) in patients with CLASI-A ≥ 8

Initial CLASI-A ≥ 8		Percent Change in CLASI-A	Slope (95% CI)	p-value
n=90	Skindex-29 Subscale
	Emotions	42.06	0.223 (0.136–0.310)	<0.0001
	Symptoms	30.97	0.238 (0.153–0.322)	<0.0001
		Δ CLASI-A	Slope (95% CI)	p-value
Initial CLASI-A ≥ 8
n=90	Skindex-29 Subscale
	Emotions	6.871	1.365 (0.872–1.858)	<0.0001
	Symptoms	5.107	1.443 (0.965–1.921)	<0.0001

Discussion

Currently, clinical trials for lupus erythematosus use an efficacy measure of ≥ 50% improvement in the CLASI activity score in patients with initial CLASI-A ≥ 10.^9–11 However, many trials enroll only a subset of patients with that degree of skin severity. Our study confirms that, in patients with an initial CLASI-A ≥ 8, an improvement in CLASI-A by at least 42.1% and 31.0% is associated with a meaningful change in the Emotions and the Symptoms subscales of the Skindex-29, respectively (Table 3), which would allow for the inclusion of patients with milder disease to be included in trials that use an efficacy measure of ≥ 50% improvement in the CLASI activity score. A minimum CLASI activity score of eight used at trial entry would allow for the inclusion of patients with milder disease that might have been excluded from skin-specific endpoints. Whether this would discriminate optimally between effective treatments and placebo remains to be determined, but the clinical significance of cutaneous improvement defined here is supported by patient data.

We found that correlation between change in disease activity and change in the two Skindex subscales was much lower for patients on the low end of mild initial disease activity (CLASI-A 4–7) than with patients on the higher end of the spectrum (CLASI-A 8–9). This suggests that in patients with CLASI-A scores ≤ 7 it would be difficult to show meaningful improvement in QoL because of floor effects, while those on the more severe range of mild initial disease activity (CLASI-A scores of 8–9) can have meaningful improvement in their QoL without complete clearance of their disease.

Klein et al. identified a four-point or 20% decrease in CLASI activity as minimal clinically significant improvement in the CLASI activity score.¹⁶ For patients with an initial CLASI-A ≥ 8, a decrease in activity by at least seven and five points is not only a clinically significant improvement but also is associated with a meaningful impact on the Emotions and Symptoms subscales, respectively (Table 3). A lower magnitude of improvement in disease activity is needed to predict meaningful change in Symptoms compared to Emotions; if CLASI score decreases enough to meaningfully impact Emotions, then there is also an associated meaningful improvement in Symptoms.

For patients with initial disease activity in the severe range of mild disease (patients with CLASI-A score of 8–9), accomplishing an actual score improvement of seven-points, or near total clearance of disease activity would require complete resolution of the skin lesions. For these patients, using a percent improvement of 50% as a marker for meaningful improvement in Skindex-E correlates with a meaningful improvement in quality of life without disease clearance. Ultimately, the choice of using percent improvement or absolute decrease in the CLASI-A score for trials would come down to preference and feasibility.

CLE has a widely variable clinical presentation across multiple subtypes, some with a higher potential to be a scarring process. In this manuscript, a variety of subtypes are represented by our patient population but future research will evaluate meaningful change in disease activity between subtypes. In addition, a database of patients with dermatomyositis was used to determine meaningful change in the Skindex-29 subscales. While both CLE and DM are inflammatory skin conditions that can have a similar magnitude and symptomology of the clinical presentation with the potential for multisystem manifestations, there is potential for differences in meaningful change in the Skindex-29 subscales. ^{3, 20–24} For the purposes of this study, we believe that using patients with DM to evaluate meaningful change in Skindex-29 for patients with CLE is reasonable; however, future studies should evaluate meaningful changes in Skindex-29 using patients with CLE.

We find that currently established efficacy measures of at least a 50% improvement in disease activity is predictive of meaningful change in patients with initial CLASI activity score ≥ 8. Using a CLASI activity score ≥ 8 for trial entry allows for the inclusion of patients with milder disease for whom improvement of disease activity by ≥ 50% can result in a meaningful impact on quality of life, as determined by the Emotions and Symptoms subscales of Skindex-29. Our findings begin to establish appropriate trial endpoints by determining clinically significant change in disease activity associated with meaningful changes in patients’ quality of life.

Capsule Summary.

• We aim to define the degree of change in CLASI activity scores that correlate with meaningful improvement in quality of life in patients with cutaneous lupus.

• Analysis of clinical trials can include patients with milder disease and use endpoints that are meaningful to patients to advance new treatments.

Acronyms and Abbreviations

CLE

Cutaneous Lupus Erythematosus

SLE

Systemic Lupus Erythematosus

QoL

Quality of life

HRQoL

Health-related quality of life

CLASI

Cutaneous Lupus Disease Area and Severity Index

CLASI-A

Cutaneous Lupus Disease Area and Severity Index activity score

FDA

US Food and Drug Administration

Skindex-E

Emotions subscale of Skindex-29

Skindex-S

Symptoms subscale of Skindex-29

Skindex-F

Functioning subscale of Skindex-29

DM

Dermatomyositis

DLQI

Dermatology Life Quality Index

CDASI

Cutaneous Dermatomyositis Disease Area and Severity Index