Search
Close
Search
Advanced Search
Search Menu
AI Discovery Assistant

Abstract

Objectives

Heterogeneity of SLE patients in clinical trials remains a challenge for developing new therapies. This study used a combinatorial analysis of four molecular biomarkers to define key sources of heterogeneity.

Methods

Combinations of IFN (high/low), anti-dsDNA (+/−) and C3 and C4 (low/normal) were used to subset n = 1747 patients from two randomized phase III trials. A dichotomous classification scheme defined SLE (+) as: IFN (high), anti-dsDNA (+), C3 (low) and/or C4 (low). SLE (−) required all of the following: IFN (low), anti-dsDNA (−), C3 (normal) and C4 (normal). Additional analyses subset the data further by IFN, anti-dsDNA and complement.

Results

The trials enrolled n = 2262 patients of which n = 1747 patients had data for IFN, anti-dsDNA, C3 and C4 at baseline. There were n = 247 patients in the SLE (−) population and n = 1500 patients in the SLE (+) population. The SLE (−) population had more mucocutaneous and musculoskeletal disease at baseline, while SLE (+) had more haematological, renal and vascular involvement. There was lower concomitant medication use in the SLE (−) population for corticosteroids and immunosuppressants, except for MTX. Time to severe flare was significantly longer in SLE (−) vs SLE (+) (P < 0.0001) and SRI-4 response rate was significantly lower in SLE (−) vs SLE (+) (P = 0.00016). The USA had more SLE (−) patients (22%) than Mexico/Central America/South America (10%), Europe (7%) and the rest of the world (5%).

Conclusion

Combinatorial analysis of four molecular biomarkers revealed subsets of SLE patients that discriminated by disease manifestations, concomitant medication use, geography, time to severe flare and SRI-4 response. These data may be useful for designing clinical trials and identifying subsets of patients for analysis.

systemic lupus erythematosus, biomarker, autoantibody, anti-dsDNA, complement, interferon
Rheumatology key messages

  • SLE patients from a P3 trial were categorized by IFN, anti-dsDNA, C3 and C4 status.

  • Patients lacking molecular markers of SLE distinguished from biomarker positive patients on multiple clinical parameters.

  • Biomarker negative patients have distinct disease characteristics that may impact clinical trial outcomes.

Introduction

The paucity of success in late stage clinical trials in SLE has been a persistent and challenging problem [1–4]. While some agents likely failed due to lack of efficacy, there are confounding factors that can contribute, including disease heterogeneity, background polypharmacy, complex composite endpoints and geographical heterogeneity associated with large multinational trials [5].

If heterogeneity in SLE trials could be accurately analysed and categorized, future trials might benefit from better control and/or reduced heterogeneity. Toward this end, we recently analysed the composite endpoint, SLE Responder Index (SRI), in two large phase III trials [6]. Using a signal-to-noise analysis approach, it was possible to isolate the inputs that impact the efficacy signal from those that contribute noise and increase heterogeneity [7].

The current study employs a molecular approach to add objective laboratory measurements that further define sources of signal and noise in SLE trials. Significant progress has been made in defining SLE phenotypes at the molecular level using transcriptomic profiling technology [8–13]. Several studies have confirmed that a major phenotype in SLE is the type 1 interferon signature [14–17]. In a large phase III trial, 75% of patients had elevated IFN signature at baseline, and multivariate analysis revealed that elevated IFN signature, anti-dsDNA antibody and low complement at baseline were predictors of time to severe disease flare [18]. However, these analyses revealed only a modest correlation between elevated IFN signature and clinical disease manifestations at baseline.

At present, no single biomarker in SLE has been sufficiently informative to enable application of precision medicine. However, a combination of molecular markers might provide improved discrimination to better subset patients. This analytical approach requires candidate molecular markers that meet the following criteria: (i) adequate sensitivity and specificity for SLE, (ii) capability to distinguish SLE subsets, (iii) adequate prevalence in a defined patient population, and (iv) amenable to rational categorization.

We incorporated four molecular markers that met the above criteria to subset SLE patients using data derived from two large, well-characterized global phase III trials. IFN status was derived from transcript profiling and anti-dsDNA and complement (C3 and C4) from central laboratory determinations. These parameters met criteria of being relatively specific to SLE, intermediate prevalence in the trial set, and amenable to rational categorization as they can be readily dichotomized.

Combinatorial analyses rapidly create large numbers of subsets—four dichotomous parameters, for example, yield 16 combinations—so we initially condensed the subsets into two logical populations, a SLE (−) population that was IFN (low), anti-dsDNA (−) and had C3 and C4 in the normal range and a SLE (+) population that included patients with any combination of IFN (high) or anti-dsDNA (+) or low C3 or low C4. We then used the SLE (−) and SLE (+) strategy to subset patients by baseline disease characteristics and post-randomization outcomes to evaluate the multivariate molecular phenotype in a well-defined population of clinical trial patients. A second analysis strategy further subset the data by eight combinations that included IFN (high/low), anti-dsDNA (+/−) and complement (low/normal) at baseline.

The results emphasize the SLE (−) population, as this subpopulation with normal IFN and normal SLE serology readily distinguished from SLE (+) at baseline and post-randomization. At baseline, SLE (−) had different disease profile, less corticosteroid use and geographic bias. Post-randomization, SLE (−) had fewer severe flares and lower SRI-4 response levels at week 52.

Methods

Molecular and biochemical baseline parameters were obtained in two multinational phase III SLE trials (n = 2262 patients) that evaluated the impact of an anti-B-cell activating factor (anti-BAFF) antibody on SLE disease activity [1, 19]. The samples studied were obtained from a subset of patients who participated in ILLUMINATE-1 (876 of 1164; NCT01205438) and ILLUMINATE-2 (880 of 1124; NCT01196091), including only those patients who had given written informed consent for genetic testing and for whom samples were available. Patients were randomly assigned 1:1:1 to one of two treatment arms or placebo. The studies complied with the Declaration of Helsinki. The locally appointed ethics committee approved the research protocol and informed consent was obtained from the subjects (or their legally authorized representative).

ANA (≥1:80) and SLE Disease Activity Index (SLEDAI) ≥6 were required at entry. Patients with serious active renal or central nervous system disease were excluded. Four dichotomous baseline parameters were used to categorize patients: IFN gene signature (high/normal), anti-dsDNA (+/−), C3 (low/normal) and C4 (low/normal).

Two categorization schemes were employed. The first categorization was limited to two groups to minimize erroneous conclusions based on multiple comparisons. This categorization included an SLE (−) population, where all four molecular markers had to be negative, and an SLE (+) population, where any one of four molecular markers could be positive. Thus, SLE (−) required all of the following: IFN (low), anti-dsDNA (−), C3 (normal) and C4 (normal), while SLE (+) was defined by any of the following: IFN (high), anti-dsDNA (+), C3 (low) and/or C4 (low).

Additional analyses used the initial classifier of IFN (high) and IFN (low) to dichotomize the population, followed by addition of anti-dsDNA and complement status. To limit the number of groups to n = 8, C3 (low) and/or C4 (low) were combined into a single category as these reflect similar pathophysiologic processes. Patients were further defined as double seropositive [anti-dsDNA (+) and C3/C4 (low)], double seronegative [anti-dsDNA (−) and C3/C4 (normal)], triple positive [IFN (high), anti-dsDNA (+) and C3/C4 (low)] and triple negative [IFN (low), anti-dsDNA (−) and C3/C4 (normal)].

Geography was initially divided into the following regions: USA/Canada, Mexico and Central/South America, Europe, and rest of the world, which included sites in Western Europe, Eastern Europe, the Middle East, Russia, Southeast Asia, India, Australia, New Zealand and Taiwan. Within the USA/Canada region, Canada only contributed 12/769 (1.6%) of patients in the combined intention-to-treat population from both studies. Thus, for purposes of clarity, this region is defined as USA in the analyses.

The IFN gene signature was measured using a 34-gene signature as previously described [18]. In brief, total RNA was extracted from whole blood and RNA quality was assessed post-extraction. An Affymetrix Kit (Thermo Fisher Scientific, Waltham, MA, USA) was used to prepare complementary DNA that was hybridized to the GeneChip Human Transcriptome Array 2.0 (HTA 2.0) (Thermo Fisher Scientific). Initial analysis of 164 pre-selected IFN-inducible genes was reduced to a 34-gene signature to define high and low expression subgroups. Anti-dsDNA, C3 and C4 were measured at a central laboratory (Covance, Princeton, NJ, USA). Anti-dsDNA was measured using the Bio-Rad Laboratories (Hercules, CA, USA) Enzyme Immunoassay for IgG anti-dsDNA.

Patient populations defined by baseline molecular markers were analysed relative to other baseline variables for concomitant medications, geography and disease activity by SLEDAI organ system. They were also analysed for two post-baseline outcomes, time to severe flare and SRI-4 at week 24 and week 52.

All P-values are considered nominal as all analyses were retrospective and were not prespecified prior to database lock.

Results

IFN, anti-dsDNA and complement at baseline

In the evaluable study population of n = 1747 patients, 1318 (75.4%) were IFN (high), 429 (24.6%) were IFN (low), 1002 (57.4%) were anti-dsDNA (+) and 729 (41.7%) had low complement levels at baseline. This resulted in n = 247 (14.1%) patients in the SLE (−) population and n = 1500 patients (86%) in the SLE (+) population (Table 1).

Table 1
Open in new tab

Distribution of SLE (+) and SLE (−) patients by IFN, anti-dsDNA and complement status

Criterian (%)
SLE (+)IFN (high), anti-dsDNA (+) and C3 (low) or C4 (low)536 (30.7)
IFN (high) only351 (20.1)
IFN (high) and anti-dsDNA (+)316 (18.1)
IFN (high) and C3 (low) or C4 (low)115 (6.6)
Anti-dsDNA (+) only104 (6)
Anti-dsDNA (+) and C3 (low) or C4 (low)46 (2.6)
C3 (low) or C4 (low) only32 (1.8)
SLE (−)IFN (low), anti-dsDNA (−), C3 (normal) and C4 (normal)247 (14.1)
Criterian (%)
SLE (+)IFN (high), anti-dsDNA (+) and C3 (low) or C4 (low)536 (30.7)
IFN (high) only351 (20.1)
IFN (high) and anti-dsDNA (+)316 (18.1)
IFN (high) and C3 (low) or C4 (low)115 (6.6)
Anti-dsDNA (+) only104 (6)
Anti-dsDNA (+) and C3 (low) or C4 (low)46 (2.6)
C3 (low) or C4 (low) only32 (1.8)
SLE (−)IFN (low), anti-dsDNA (−), C3 (normal) and C4 (normal)247 (14.1)
Table 1
Open in new tab

Distribution of SLE (+) and SLE (−) patients by IFN, anti-dsDNA and complement status

Criterian (%)
SLE (+)IFN (high), anti-dsDNA (+) and C3 (low) or C4 (low)536 (30.7)
IFN (high) only351 (20.1)
IFN (high) and anti-dsDNA (+)316 (18.1)
IFN (high) and C3 (low) or C4 (low)115 (6.6)
Anti-dsDNA (+) only104 (6)
Anti-dsDNA (+) and C3 (low) or C4 (low)46 (2.6)
C3 (low) or C4 (low) only32 (1.8)
SLE (−)IFN (low), anti-dsDNA (−), C3 (normal) and C4 (normal)247 (14.1)
Criterian (%)
SLE (+)IFN (high), anti-dsDNA (+) and C3 (low) or C4 (low)536 (30.7)
IFN (high) only351 (20.1)
IFN (high) and anti-dsDNA (+)316 (18.1)
IFN (high) and C3 (low) or C4 (low)115 (6.6)
Anti-dsDNA (+) only104 (6)
Anti-dsDNA (+) and C3 (low) or C4 (low)46 (2.6)
C3 (low) or C4 (low) only32 (1.8)
SLE (−)IFN (low), anti-dsDNA (−), C3 (normal) and C4 (normal)247 (14.1)

In the SLE (+) population, the most prevalent phenotype was IFN (high) and anti-dsDNA (+) [n = 852 (56%)]. The least prevalent population in the SLE (+) population qualified only by low complement. Of the n = 32 patients that qualified for SLE (+) based exclusively on low complement, n = 19 had only C3 (low) and n = 6 had only C4 (low).

The SLE (−) population had less haematological, renal and vascular involvement, and more mucocutaneous and musculoskeletal disease as defined by SLEDAI organ system. The SLE (−) population also had lower SLEDAI and fewer patients with SLEDAI ≥10 at baseline (Table 2). This difference in mean SLEDAI was 2.4 points, which primarily reflects the 2 or 4 points awarded for a positive test in the immunological organ system that was excluded from SLE (−).

Table 2
Open in new tab

Baseline SLEDAI score and SLEDAI organ system prevalence in SLE (+) and SLE (−) patients

SLE (+) (n = 1500)SLE (−) (n = 247)P-value
Mean SLEDAI10.78.3<0.001
SLEDAI ≥10934 (62.3)69 (27.9)<0.001
 SLEDAI organ system
 Immunological1186 (79.1)13 (5.3)<0.0001
 Mucocutaneous1357 (90.5)237 (96)0.0047
 Musculoskeletal1296 (86.4)235 (95.1)0.0001
 Haematological147 (9.8)4 (1.6)<0.0001
 Renal135 (9)11 (4.5)0.0167
 Vascular121 (8.1)6 (2.4)0.0016
 Cardiovascular/respiratory116 (7.7)23 (9.3)0.3957
 Constitutional29 (1.9)2 (0.8)0.2152
 CNS25 (1.7)8 (3.2)0.0926
SLE (+) (n = 1500)SLE (−) (n = 247)P-value
Mean SLEDAI10.78.3<0.001
SLEDAI ≥10934 (62.3)69 (27.9)<0.001
 SLEDAI organ system
 Immunological1186 (79.1)13 (5.3)<0.0001
 Mucocutaneous1357 (90.5)237 (96)0.0047
 Musculoskeletal1296 (86.4)235 (95.1)0.0001
 Haematological147 (9.8)4 (1.6)<0.0001
 Renal135 (9)11 (4.5)0.0167
 Vascular121 (8.1)6 (2.4)0.0016
 Cardiovascular/respiratory116 (7.7)23 (9.3)0.3957
 Constitutional29 (1.9)2 (0.8)0.2152
 CNS25 (1.7)8 (3.2)0.0926
Table 2
Open in new tab

Baseline SLEDAI score and SLEDAI organ system prevalence in SLE (+) and SLE (−) patients

SLE (+) (n = 1500)SLE (−) (n = 247)P-value
Mean SLEDAI10.78.3<0.001
SLEDAI ≥10934 (62.3)69 (27.9)<0.001
 SLEDAI organ system
 Immunological1186 (79.1)13 (5.3)<0.0001
 Mucocutaneous1357 (90.5)237 (96)0.0047
 Musculoskeletal1296 (86.4)235 (95.1)0.0001
 Haematological147 (9.8)4 (1.6)<0.0001
 Renal135 (9)11 (4.5)0.0167
 Vascular121 (8.1)6 (2.4)0.0016
 Cardiovascular/respiratory116 (7.7)23 (9.3)0.3957
 Constitutional29 (1.9)2 (0.8)0.2152
 CNS25 (1.7)8 (3.2)0.0926
SLE (+) (n = 1500)SLE (−) (n = 247)P-value
Mean SLEDAI10.78.3<0.001
SLEDAI ≥10934 (62.3)69 (27.9)<0.001
 SLEDAI organ system
 Immunological1186 (79.1)13 (5.3)<0.0001
 Mucocutaneous1357 (90.5)237 (96)0.0047
 Musculoskeletal1296 (86.4)235 (95.1)0.0001
 Haematological147 (9.8)4 (1.6)<0.0001
 Renal135 (9)11 (4.5)0.0167
 Vascular121 (8.1)6 (2.4)0.0016
 Cardiovascular/respiratory116 (7.7)23 (9.3)0.3957
 Constitutional29 (1.9)2 (0.8)0.2152
 CNS25 (1.7)8 (3.2)0.0926

There was significantly lower concomitant medication use in the SLE (−) population for corticosteroids and immunosuppressants, with the exception of MTX, which was similar in the two groups (Table 3).

Table 3
Open in new tab

Concomitant medication utilization at baseline in the SLE (+) and SLE (−) groups

Baseline therapySLE (+)
(n = 1500)		SLE (−)
(n = 247)		P-value
Corticosteroids1166 (77.7)121 (49)<0.0001
Immunosuppressants654 (43.6)77 (31.2)0.0002
AZA305 (20.3)26 (10.5)0.0003
MTX181 (12.1)38 (15.4)0.1445
Mycophenolate mofetil145 (9.7)9 (3.6)0.0020
Antimalarials997 (66.5)178 (72.1)0.0823
Baseline therapySLE (+)
(n = 1500)		SLE (−)
(n = 247)		P-value
Corticosteroids1166 (77.7)121 (49)<0.0001
Immunosuppressants654 (43.6)77 (31.2)0.0002
AZA305 (20.3)26 (10.5)0.0003
MTX181 (12.1)38 (15.4)0.1445
Mycophenolate mofetil145 (9.7)9 (3.6)0.0020
Antimalarials997 (66.5)178 (72.1)0.0823
Table 3
Open in new tab

Concomitant medication utilization at baseline in the SLE (+) and SLE (−) groups

Baseline therapySLE (+)
(n = 1500)		SLE (−)
(n = 247)		P-value
Corticosteroids1166 (77.7)121 (49)<0.0001
Immunosuppressants654 (43.6)77 (31.2)0.0002
AZA305 (20.3)26 (10.5)0.0003
MTX181 (12.1)38 (15.4)0.1445
Mycophenolate mofetil145 (9.7)9 (3.6)0.0020
Antimalarials997 (66.5)178 (72.1)0.0823
Baseline therapySLE (+)
(n = 1500)		SLE (−)
(n = 247)		P-value
Corticosteroids1166 (77.7)121 (49)<0.0001
Immunosuppressants654 (43.6)77 (31.2)0.0002
AZA305 (20.3)26 (10.5)0.0003
MTX181 (12.1)38 (15.4)0.1445
Mycophenolate mofetil145 (9.7)9 (3.6)0.0020
Antimalarials997 (66.5)178 (72.1)0.0823
Table 5
Open in new tab

Baseline disease characteristics by IFN, anti-dsDNA and complement status

Subgroup criteriaIFN (high) groups (1–4)
IFN (low) groups (5–8)
Group 1 (n = 536)Group 2 (n = 351)Group 3 (n = 316)Group 4 (n = 115)Group 5 (n = 104)Group 6 (n = 46)Group 7 (n = 32)Group 8 (n = 247)IFN (−) (n = 429)
IFN(+)(+)(+)(+)(−)(−)(−)(−)(−)
Anti-dsDNA(+)(−)(+)(−)(+)(+)(−)(−)(+ or −)
Low C3 or C4(+)(−)(−)(+)(−)(+)(+)(−)(+ or −)
SLEDAI
 SLEDAI, mean12.48.910.110.710.111.19.88.39.2
 SLEDAI >10, %80.237.354.467.863.580.465.627.945.0
Concomitant medications
 Corticosteroids, %89.467.275.375.768.382.653.249.057.6
 Antimalarials, %64.665.068.472.266.467.475.072.170.4
 Immunosuppressant, %47.240.740.252.242.334.834.431.234.5
  AZA, %22.216.020.925.219.217.421.910.514.2
  MTX, %10.115.111.720.08.76.56.315.412.1
  MMF, %12.77.47.010.411.58.73.13.66.1
SLEDAI organ system
 CNS, %0.82.62.91.71.00.00.03.22.1
 Vascular, %10.37.46.010.45.84.43.12.43.5
 Musculoskeletal, %79.794.387.087.090.487.090.695.292.8
 Renal, %15.13.76.06.110.66.53.14.56.1
 Mucocutaneous, %87.996.989.294.884.682.690.696.091.4
 Cardiovascular/respiratory, %10.36.35.77.06.78.76.39.38.4
 Immunological, %99.812.099.499.1100.0100.096.95.345.2
 Constitutional, %1.72.02.23.51.02.20.00.80.9
 Haematological, %13.46.09.813.02.98.73.11.62.8
Geographical Region
 USA, %21.223.017.65.76.02.12.222.232.5
 Latin America, %39.614.919.28.54.02.51.510.017.9
 Europe, %40.418.115.97.07.03.41.66.67.0
Rest of world, %26.725.225.24.48.23.71.55.218.5
Subgroup criteriaIFN (high) groups (1–4)
IFN (low) groups (5–8)
Group 1 (n = 536)Group 2 (n = 351)Group 3 (n = 316)Group 4 (n = 115)Group 5 (n = 104)Group 6 (n = 46)Group 7 (n = 32)Group 8 (n = 247)IFN (−) (n = 429)
IFN(+)(+)(+)(+)(−)(−)(−)(−)(−)
Anti-dsDNA(+)(−)(+)(−)(+)(+)(−)(−)(+ or −)
Low C3 or C4(+)(−)(−)(+)(−)(+)(+)(−)(+ or −)
SLEDAI
 SLEDAI, mean12.48.910.110.710.111.19.88.39.2
 SLEDAI >10, %80.237.354.467.863.580.465.627.945.0
Concomitant medications
 Corticosteroids, %89.467.275.375.768.382.653.249.057.6
 Antimalarials, %64.665.068.472.266.467.475.072.170.4
 Immunosuppressant, %47.240.740.252.242.334.834.431.234.5
  AZA, %22.216.020.925.219.217.421.910.514.2
  MTX, %10.115.111.720.08.76.56.315.412.1
  MMF, %12.77.47.010.411.58.73.13.66.1
SLEDAI organ system
 CNS, %0.82.62.91.71.00.00.03.22.1
 Vascular, %10.37.46.010.45.84.43.12.43.5
 Musculoskeletal, %79.794.387.087.090.487.090.695.292.8
 Renal, %15.13.76.06.110.66.53.14.56.1
 Mucocutaneous, %87.996.989.294.884.682.690.696.091.4
 Cardiovascular/respiratory, %10.36.35.77.06.78.76.39.38.4
 Immunological, %99.812.099.499.1100.0100.096.95.345.2
 Constitutional, %1.72.02.23.51.02.20.00.80.9
 Haematological, %13.46.09.813.02.98.73.11.62.8
Geographical Region
 USA, %21.223.017.65.76.02.12.222.232.5
 Latin America, %39.614.919.28.54.02.51.510.017.9
 Europe, %40.418.115.97.07.03.41.66.67.0
Rest of world, %26.725.225.24.48.23.71.55.218.5
Table 5
Open in new tab

Baseline disease characteristics by IFN, anti-dsDNA and complement status

Subgroup criteriaIFN (high) groups (1–4)
IFN (low) groups (5–8)
Group 1 (n = 536)Group 2 (n = 351)Group 3 (n = 316)Group 4 (n = 115)Group 5 (n = 104)Group 6 (n = 46)Group 7 (n = 32)Group 8 (n = 247)IFN (−) (n = 429)
IFN(+)(+)(+)(+)(−)(−)(−)(−)(−)
Anti-dsDNA(+)(−)(+)(−)(+)(+)(−)(−)(+ or −)
Low C3 or C4(+)(−)(−)(+)(−)(+)(+)(−)(+ or −)
SLEDAI
 SLEDAI, mean12.48.910.110.710.111.19.88.39.2
 SLEDAI >10, %80.237.354.467.863.580.465.627.945.0
Concomitant medications
 Corticosteroids, %89.467.275.375.768.382.653.249.057.6
 Antimalarials, %64.665.068.472.266.467.475.072.170.4
 Immunosuppressant, %47.240.740.252.242.334.834.431.234.5
  AZA, %22.216.020.925.219.217.421.910.514.2
  MTX, %10.115.111.720.08.76.56.315.412.1
  MMF, %12.77.47.010.411.58.73.13.66.1
SLEDAI organ system
 CNS, %0.82.62.91.71.00.00.03.22.1
 Vascular, %10.37.46.010.45.84.43.12.43.5
 Musculoskeletal, %79.794.387.087.090.487.090.695.292.8
 Renal, %15.13.76.06.110.66.53.14.56.1
 Mucocutaneous, %87.996.989.294.884.682.690.696.091.4
 Cardiovascular/respiratory, %10.36.35.77.06.78.76.39.38.4
 Immunological, %99.812.099.499.1100.0100.096.95.345.2
 Constitutional, %1.72.02.23.51.02.20.00.80.9
 Haematological, %13.46.09.813.02.98.73.11.62.8
Geographical Region
 USA, %21.223.017.65.76.02.12.222.232.5
 Latin America, %39.614.919.28.54.02.51.510.017.9
 Europe, %40.418.115.97.07.03.41.66.67.0
Rest of world, %26.725.225.24.48.23.71.55.218.5
Subgroup criteriaIFN (high) groups (1–4)
IFN (low) groups (5–8)
Group 1 (n = 536)Group 2 (n = 351)Group 3 (n = 316)Group 4 (n = 115)Group 5 (n = 104)Group 6 (n = 46)Group 7 (n = 32)Group 8 (n = 247)IFN (−) (n = 429)
IFN(+)(+)(+)(+)(−)(−)(−)(−)(−)
Anti-dsDNA(+)(−)(+)(−)(+)(+)(−)(−)(+ or −)
Low C3 or C4(+)(−)(−)(+)(−)(+)(+)(−)(+ or −)
SLEDAI
 SLEDAI, mean12.48.910.110.710.111.19.88.39.2
 SLEDAI >10, %80.237.354.467.863.580.465.627.945.0
Concomitant medications
 Corticosteroids, %89.467.275.375.768.382.653.249.057.6
 Antimalarials, %64.665.068.472.266.467.475.072.170.4
 Immunosuppressant, %47.240.740.252.242.334.834.431.234.5
  AZA, %22.216.020.925.219.217.421.910.514.2
  MTX, %10.115.111.720.08.76.56.315.412.1
  MMF, %12.77.47.010.411.58.73.13.66.1
SLEDAI organ system
 CNS, %0.82.62.91.71.00.00.03.22.1
 Vascular, %10.37.46.010.45.84.43.12.43.5
 Musculoskeletal, %79.794.387.087.090.487.090.695.292.8
 Renal, %15.13.76.06.110.66.53.14.56.1
 Mucocutaneous, %87.996.989.294.884.682.690.696.091.4
 Cardiovascular/respiratory, %10.36.35.77.06.78.76.39.38.4
 Immunological, %99.812.099.499.1100.0100.096.95.345.2
 Constitutional, %1.72.02.23.51.02.20.00.80.9
 Haematological, %13.46.09.813.02.98.73.11.62.8
Geographical Region
 USA, %21.223.017.65.76.02.12.222.232.5
 Latin America, %39.614.919.28.54.02.51.510.017.9
 Europe, %40.418.115.97.07.03.41.66.67.0
Rest of world, %26.725.225.24.48.23.71.55.218.5

The USA region had the largest number and proportion of SLE (−) patients (Table 4). The probability of being SLE (−) in USA was more than double that of Mexico/Central/South America, Europe and the rest of the world.

Post-randomization outcome data using the SLE (−) and SLE (+) classification revealed a significant prolongation of time to severe flare in the SLE (−) population compared with the SLE (+) population (Fig. 1). The survival curves between SLE (−) and SLE (+) separated by 3 months and continued to show a greater rate of severe flare in SLE (+) over the course of the 1-year trial.

Time to severe flare in SLE (+) and SLE (−)
Fig. 1

Time to severe flare in SLE (+) and SLE (−)

Open in new tabDownload slide

SRI-4 rates were examined at week 24 and week 52 in SLE (−) and SLE (+) patients. In the all patient population at week 24, SRI-4 response rate was 41.3% (102/247) in SLE (−) vs. 49.5% (743/1500) in SLE (+) (P = 0.02). At week 52, SRI-4 response rate was 33.2% (82/247) in SLE (−) vs 46.3% (695/1500) in SLE (+) (P = 0.00016). In the placebo population at week 24, SRI-4 response rate was 36.4% (32/88) in SLE (−) vs 43.5% (217/499) in SLE (+) (P = 0.26). At week 52, SRI-4 response rate was 31.8% (28/88) in SLE (−) vs 39.7% (198/499) in SLE (+) (P = 0.2). In the active treatment arm at week 24, SRI-4 response rate was 44.0% (70/159) in SLE (−) vs 52.6% (526/1001) in SLE (+). At week 52, SRI-4 response rate was 34.0% (54/159) in SLE (−) vs 49.7% (497/1001) in SLE (+) (P < 0.001).

8× analyses

The three dichotomous variables of IFN, anti-dsDNA and complement (C3 and C4 combined) were analysed as eight unique categories of patients to provide greater granularity between the populations (Table 5 and Supplementary Table S1, available at Rheumatology online). The proportion of seropositive patients was lower in the IFN (low) group compared with the IFN (high) group for the anti-dsDNA [35% in IFN (low)/dsDNA (+) vs 64.6% in IFN (high)/dsDNA (+)], low complement [18.2% in IFN (low)/C3/C4 (low) vs 49.4% in IFN (high)/C3/C4 (low)] and double seropositive group [10.7% in IFN (low)/dsDNA (+)/C3/C4 (low) vs 40.7% IFN (high)/anti-dsDNA (+)/C3/C4 (low)]. Thus, patients that are dsDNA (−) and have normal complement levels are more likely to be IFN (−) than patients that are dsDNA (+) and/or low complement.

Table 4
Open in new tab

Distribution of patients by geographical region in the SLE (+) and SLE (−) groupsa,b

RegionTotalSLE (+)SLE (−)
Total17471500 (86)247 (14)
USA769 (44)598 (78)171 (22)

Mexico, Central and South America

402 (23)362 (90)40 (10)
Europe441 (25)412 (93)29 (7)
Rest of World135 (8)128 (95)7 (5)
RegionTotalSLE (+)SLE (−)
Total17471500 (86)247 (14)
USA769 (44)598 (78)171 (22)

Mexico, Central and South America

402 (23)362 (90)40 (10)
Europe441 (25)412 (93)29 (7)
Rest of World135 (8)128 (95)7 (5)
a

Nominal P < 0.0001 for difference in distributions among regions via chi square test.

b

Percentages for SLE (+) and SLE (−) reflect distribution of patients within specified region.

Table 4
Open in new tab

Distribution of patients by geographical region in the SLE (+) and SLE (−) groupsa,b

RegionTotalSLE (+)SLE (−)
Total17471500 (86)247 (14)
USA769 (44)598 (78)171 (22)

Mexico, Central and South America

402 (23)362 (90)40 (10)
Europe441 (25)412 (93)29 (7)
Rest of World135 (8)128 (95)7 (5)
RegionTotalSLE (+)SLE (−)
Total17471500 (86)247 (14)
USA769 (44)598 (78)171 (22)

Mexico, Central and South America

402 (23)362 (90)40 (10)
Europe441 (25)412 (93)29 (7)
Rest of World135 (8)128 (95)7 (5)
a

Nominal P < 0.0001 for difference in distributions among regions via chi square test.

b

Percentages for SLE (+) and SLE (−) reflect distribution of patients within specified region.

Corticosteroid utilization was lower in IFN (low) compared with IFN (high) patients across all serology categories. The impact of serology on corticosteroids at baseline was evident when comparing the double seronegative with the double seropositive groups. Within the IFN (low) subgroup, corticosteroids were used by 49.0% in the double seronegative compared with 82.6% in the double seropositive group. Similarly, within the IFN (high) group, corticosteroids were used by 67.2% in the double seronegative compared with 89.4% in the double seropositive group.

The proportion of patients that were IFN (high) was similar in Latin America (82.2%) and Europe (81.4%), and lower in the USA (67.5%). In Latin America and Europe, there were approximately four times as many IFN (high) patients compared with IFN (low). In contrast, in the USA there were approximately twice as many IFN (high) patients compared with IFN (low). Latin America and Europe also had more triple positive patients [IFN (high)/anti-dsDNA (+)/C3/C4 (low)] compared with the USA (39.6% and 40.4%, respectively, compared with 21.2%), while the USA had the highest proportion of triple negative patients [22.2% vs 10% (Latin America) and 6.6% (Europe)].

Discussion

Objective molecular markers have the potential to explain some of the heterogeneity observed in late stage clinical trials in SLE. However, single analytes have been limited in their ability to adequately discriminate baseline disease activity and clinical outcomes. This analysis employed a combinatorial analysis of four objective laboratory analytes to discriminate different populations of patients. Patients that were IFN (low), dsDNA (−) and complement (normal) were readily distinguished from patients that were positive for any combination of these parameters on multiple baseline demographics and disease characteristics. These analyses apply a laboratory-based molecular lens to reveal key sources of heterogeneity in SLE clinical trials.

There is an evolving set of molecular markers that distinguishes SLE from related clinical syndromes, and it is likely that more than one marker will be needed to adequately discriminate subsets within SLE. This study shows that a combinatorial analysis of four molecular parameters can discriminate SLE patient subsets by concurrent clinical manifestations, concomitant medications, geography, risk of subsequent flare and probability of achieving the desired clinical response as defined by the SRI-4 endpoint. It also highlights that IFN (high) vs INF (low) or seropositive vs seronegative is an oversimplification of the clinical phenotype.

While retrospective, the analyses were based on large, well-controlled trials in seropositive SLE patients with moderate to severe disease activity at baseline, which enabled the ability to characterize disease characteristics and outcomes in multiple subsets. The analysis of concurrent disease characteristics built on prior observations that there are only six prevalent disease manifestations in the study: arthritis, rash, alopecia, mucocutaneous, anti-dsDNA and low complement [7]. All other disease manifestations measured by SLEDAI-2K were prevalent in ≤11% of the ILLUMINATE population. Arthritis and rash were the most prevalent disease manifestations, and over-representation of these disease manifestations in the SLE (−) vs SLE (+) population was likely driven by the need to meet the study required inclusion criteria of SLEDAI ≥6 in the absence of points for serology.

The large sample size provided the opportunity to examine the SLE (−) subset of patients that lacked molecular markers of SLE disease activity. The SLE (−) subset readily distinguished themselves from patients with one or more molecular markers of SLE disease activity at baseline.

The SLE (−) population had lower prevalence of renal disease, which appeared to be influenced by anti-dsDNA status [20]. Vascular and haematological disease manifestations were also less prevalent in the SLE (−) population, and this suggests that these manifestations may be less prevalent in patients that lack serological indicators of disease.

The analyses document an important relationship between the molecular parameters and concomitant medication utilization. Corticosteroid utilization at baseline was less prevalent in the SLE (−) compared with the SLE (+) population. Almost 90% of triple positive patients were taking corticosteroids at baseline, compared with 49% taking steroids in the triple negative SLE (−) population. Corticosteroid utilization was also correlated with both serology and IFN status. While it might be anticipated that corticosteroid utilization would be more prevalent in SLE (+) vs SLE (−) due to investigator knowledge of serology status, it was notable that corticosteroid utilization was also lower in IFN (low) compared with IFN (high) across all serology categories in the 8× analysis. Importantly, investigators did not know IFN status at baseline as these assays were performed after the trials were completed. These observations suggest that IFN status does impact disease severity, and that greater utilization of corticosteroids in the IFN (high) population may attenuate the differences in disease outcomes between IFN (high) and IFN (low). However, the SLE (−) population should not be ignored, as they had more mucocutaneous and musculoskeletal disease, and were more common in the USA than the rest of the world.

Immunosuppressant utilization at baseline was generally less prevalent in the SLE (−) population compared with the SLE (+) population. The exception was MTX, which was numerically higher in the SLE (−) population. This observation was consistent with the higher prevalence of musculoskeletal disease in this population.

The geographic analysis was striking, showing clear distinction between regions. Notably, SLE (−) was much more prevalent in the USA vs Mexico/Central/South America and Europe. There was very limited Canadian enrolment in the trial, so the SLE (−) population is largely driven by United States sites. Recent evidence indicates that ancestry contributes to diversity in molecular signatures in SLE [21, 22]. Thus, differences between regions may be in part driven by different ancestorial populations. Alternatively, geographical differences in clinical enrolment practices may be driving the differences between the USA and other regions.

An intriguing observation was that SRI-4 response was consistently lower in SLE (−) compared with SLE (+) in both placebo and active treatment arms. A key goal in SLE clinical trial design is to minimize the placebo response rate, but we are not convinced that biasing recruitment to SLE (−) in order to achieve a lower placebo response rate is an effective strategy for efficacy trials. IFN status and serology appear to contribute to the lower placebo rate observed in the USA compared with other regions, but the USA region has also been associated with lower response rates in treatment groups relative to placebo in some trials [23, 24]. The data are consistent with the possibility that the USA region has a different molecular phenotype, on average, that is potentially less capable of responding to targeted immune therapy approaches.

There are several limitations to this study. All analyses presented here were retrospective and there was no correction for multiplicity. The multiplicity concern is amplified in the 8× analysis where the increased number of categories increases the risk that apparent differences are due to chance. However, the increased granularity also provides the opportunity for additional insight and hypotheses that can be tested in other datasets. While the evaluable population exceeded 1700 patients, 23% of the intention to treat (ITT) population did not have gene array analysis and were excluded from the analysis. The geographical analyses condensed patients from 377 sites into four regions, which could obscure important differences within regions. Additionally, the study enrolled patients that principally had joint, skin and immunological manifestations. The remaining organ systems had relatively low prevalence and substantially less power to detect differences between groups.

In conclusion, this combinatorial analysis provided distinct subsets of patients that can be discriminated by disease manifestations, concomitant medication use, geography and time to severe flare. The analyses identify important sources of heterogeneity in the clinical trials and may be useful for designing future clinical trials in SLE patients.

Funding: This work was supported by Eli Lilly and Company.

Disclosure statement: M.P.: none. S.D.W., R.E.H., M.D.L.: employees of, and shareholders in, Eli Lilly and Company.

Data availability statement

The clinical data from the ILLUMINATE trials reported in this manuscript are available at ClinicalTrials.gov. The data have also been posted to the TransCelerate Shared Investigator Platform.

Supplementary data

Supplementary data are available at Rheumatology online.

References

1

Isenberg
DA
,
Petri
M
,
Kalunian
K
,
Tanaka
Y
,
Urowitz
MB
,
Hoffman
RW.
  
Efficacy and safety of subcutaneous tabalumab in patients with systemic lupus erythematosus: results from ILLUMINATE-1, a 52-week, phase III, multicentre, randomised, double-blind, placebo-controlled study
.
Ann Rheum Dis
 
2016
;
75
:
323
–
31
.

Google Scholar

Crossref
Search ADS
PubMed

 
2

Clowse
ME
,
Wallace
DJ
,
Furie
RA
  et al. ,
Efficacy and safety of Epratuzumab in moderately to severely active systemic lupus erythematosus: results from two phase III randomized, double-blind, placebo-controlled trials
.
Arthritis Rheumatol
 
2017
;
69
:
362
–
75
.

Google Scholar

Crossref
Search ADS
PubMed

 
3

Merrill
JT
,
Shanahan
WR,
,
Scheinberg
M
,
Kalunian
KC
,
Wofsy
D
,
Martin
RS.
  
Phase III trial results with blisibimod, a selective inhibitor of B-cell activating factor, in subjects with systemic lupus erythematosus (SLE): results from a randomised, double-blind, placebo-controlled trial
.
Ann Rheum Dis
 
2018
;
77
:
883
–
89
.

Google Scholar

Crossref
Search ADS
PubMed

 
4

Furie
RA
,
Morand
EF
,
Bruce
IN
  et al.  
Type I interferon inhibitor anifrolumab in active systemic lupus erythematosus (TULIP-1): a randomised, controlled, phase 3 trial
.
Lancet Rheum
 
2019
;
1
:
e208
–
19
.

Google Scholar

Crossref
Search ADS

 
5

Merrill
JT
,
Manzi
S
,
Aranow
C
  et al.  
Lupus community panel proposals for optimising clinical trials: 2018
.
Lupus Sci Med
 
2018
;
5
:
e000258
.

Google Scholar

Crossref
Search ADS
PubMed

 
6

Furie
RA
,
Petri
MA
,
Wallace
DJ
  et al.  
Novel evidence-based systemic lupus erythematosus responder index
.
Arthritis Rheum
 
2009
;
61
:
1143
–
51
.

Google Scholar

Crossref
Search ADS
PubMed

 
7

Kalunian
KC
,
Urowitz
MB
,
Isenberg
D
  et al.  
Clinical trial parameters that influence outcomes in lupus trials that use the systemic lupus erythematosus responder index
.
Rheumatology (Oxford
)
2018
;
57
:
125
–
3
.

Google Scholar

Crossref
Search ADS
PubMed

 
8

Arazi
A
,
Rao
DA
,
Berthier
CC
  et al.  
The immune cell landscape in kidneys of patients with lupus nephritis
.
Nat Immunol
 
2019
;
20
:
902
–
14
.

Google Scholar

Crossref
Search ADS
PubMed

 
9

Jourde-Chiche
N
,
Whalen
E
,
Gondouin
B
  et al.  
Modular transcriptional repertoire analyses identify a blood neutrophil signature as a candidate biomarker for lupus nephritis
.
Rheumatology (Oxford)
 
2017
;
56
:
477
–
87
.

Google Scholar

PubMed
OpenURL Placeholder Text

 
10

Kotliarov
Y
,
Sparks
R
,
Martins
AJ
  et al.  
Broad immune activation underlies shared set point signatures for vaccine responsiveness in healthy individuals and disease activity in patients with lupus
.
Nat Med
 
2020
;
26
:
618
–
29
.

Google Scholar

Crossref
Search ADS
PubMed

 
11

Chiche
L
,
Jourde-Chiche
N
,
Whalen
E
  et al.  
Modular transcriptional repertoire analyses of adults with systemic lupus erythematosus reveal distinct type I and type II interferon signatures
.
Arthritis Rheumatol
 
2014
;
66
:
1583
–
95
.

Google Scholar

Crossref
Search ADS
PubMed

 
12

Smith
MA
,
Henault
J
,
Karnell
JL
  et al.  
SLE plasma profiling identifies unique signatures of lupus nephritis and discoid lupus
.
Sci Rep
 
2019
;
9
:
14433
.

Google Scholar

Crossref
Search ADS
PubMed

 
13

Der
E
,
Suryawanshi
H
,
Morozov
P
  et al.  
Tubular cell and keratinocyte single-cell transcriptomics applied to lupus nephritis reveal type I IFN and fibrosis relevant pathways
.
Nat Immunol
 
2019
;
20
:
915
–
27
.

Google Scholar

Crossref
Search ADS
PubMed

 
14

Baechler
EC
,
Batliwalla
FM
,
Karypis
G
  et al.  
Interferon-inducible gene expression signature in peripheral blood cells of patients with severe systemic lupus
.
Proc Natl Acad Sci USA
 
2003
;
100
:
2610
–
5
.

Google Scholar

Crossref
Search ADS
PubMed

 
15

Bennett
L
,
Palucka
AK
,
Arce
E
  et al.  
Interferon and granulopoiesis signatures in systemic lupus erythematosus blood
.
J Exp Med
 
2003
;
197
:
711
–
23
.

Google Scholar

Crossref
Search ADS
PubMed

 
16

Kirou
KA
,
Lee
C
,
George
S
,
Louca
K
,
Peterson
MG
,
Crow
MK.
  
Activation of the interferon-α pathway identifies a subgroup of systemic lupus erythematosus patients with distinct serologic features and active disease
.
Arthritis Rheum
 
2005
;
52
:
1491
–
503
.

Google Scholar

Crossref
Search ADS
PubMed

 
17

Kennedy
WP
,
Maciuca
R
,
Wolslegel
K
  et al.  
Association of the interferon signature metric with serological disease manifestations but not global activity scores in multiple cohorts of patients with SLE
.
Lupus Sci Med
 
2015
;
2
:
e000080
.

Google Scholar

Crossref
Search ADS
PubMed

 
18

Hoffman
RW
,
Merrill
JT
,
Alarcón-Riquelme
MM
  et al.  
Gene expression and pharmacodynamic changes in 1,760 systemic lupus erythematosus patients from two phase III trials of BAFF blockade with tabalumab
.
Arthritis Rheumatol
 
2017
;
69
:
643
–
54
.

Google Scholar

Crossref
Search ADS
PubMed

 
19

Merrill
JT
,
van Vollenhoven
RF
,
Buyon
JP
  et al.  
Efficacy and safety of subcutaneous tabalumab, a monoclonal antibody to B-cell activating factor, in patients with systemic lupus erythematosus: results from ILLUMINATE-2, a 52-week, phase III, multicentre, randomised, double-blind, placebo-controlled study
.
Ann Rheum Dis
 
2016
;
75
:
332
–
40
.

Google Scholar

Crossref
Search ADS
PubMed

 
20

Linnik
MD
,
Hu
JZ
,
Heilbrunn
KR
,
Strand
V
,
Hurley
FL
,
Joh
T.
  
Relationship between anti-double-stranded DNA antibodies and exacerbation of renal disease in patients with systemic lupus erythematosus
.
Arthritis Rheum
 
2005
;
52
:
1129
–
37
.

Google Scholar

Crossref
Search ADS
PubMed

 
21

Catalina
MD
,
Bachali
P
,
Yeo
AE
  et al.  
Patient ancestry significantly contributes to molecular heterogeneity of systemic lupus erythematosus
.
JCI Insight
 
2020
;
5
:
e140380
.

Google Scholar

Crossref
Search ADS

 
22

Goulielmos
GN
,
Zervou
MI
,
Vazgiourakis
VM
,
Ghodke-Puranik
Y
,
Garyfallos
A
,
Niewold
TB.
  
The genetics and molecular pathogenesis of systemic lupus erythematosus (SLE) in populations of different ancestry
.
Gene
 
2018
;
668
:
59
–
72
.

Google Scholar

Crossref
Search ADS
PubMed

 
23

Furie
R
,
Petri
M
,
Zamani
O
  et al. ; BLISS-76 Study Group.
A phase III, randomized, placebo-controlled study of belimumab, a monoclonal antibody that inhibits B lymphocyte stimulator, in patients with systemic lupus erythematosus
.
Arthritis Rheum
 
2011
;
63
:
3918
–
30
.

Google Scholar

Crossref
Search ADS
PubMed

 
24

Navarra
SV
,
Guzmán
RM
,
Gallacher
AE
  et al. ; BLISS-52 Study Group.
Efficacy and safety of belimumab in patients with active systemic lupus erythematosus: a randomised, placebo-controlled, phase 3 trial
.
Lancet
 
2011
;
377
:
721
–
31

Google Scholar

Crossref
Search ADS
PubMed

 
© The Author(s) 2021. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For permissions, please email: [email protected]
This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://dbpia.nl.go.kr/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
Topic:
Issue Section:
Clinical Science
Download all slides
  • Supplementary data

    • Supplementary data

    keab144_Supplementary_Data - pdf file

    Comments

    0 Comments
    Comments (0)
    Submit a comment
    You have entered an invalid code
    Thank you for submitting a comment on this article. Your comment will be reviewed and published at the journal's discretion. Please check for further notifications by email.