See related commentary on pg 1329 ORIGINAL ARTICLE

Intrinsic Expression Subsets of Diffuse Cutaneous Systemic Sclerosis Are Stable in Serial Skin Biopsies Sarah A. Pendergrass1, Raphael Lemaire2, Ian P. Francis2, J. Matthew Mahoney1, Robert Lafyatis2 and Michael L. Whitfield1

Skin biopsy was analyzed by DNA microarray from 13 diffuse cutaneous systemic sclerosis (dSSc) patients enrolled in an open-label study of rituximab, 9 dSSc patients not treated with rituximab, and 9 healthy controls. These data recapitulate the patient ‘‘intrinsic’’ gene expression subsets described previously, including fibroproliferative, inflammatory, and normal-like groups. Serial skin biopsies showed consistent and non-progressing gene expression over time, and importantly, the patients in the inflammatory subset do not move to the fibroproliferative subset, and vice versa. We were unable to detect significant differences in gene expression before and after rituximab treatment, consistent with an apparent lack of clinical response. Serial biopsies from each patient stayed within the same gene expression subset, regardless of treatment regimen or the time point at which they were taken. Collectively, these data emphasize the heterogeneous nature of SSc and demonstrate that the intrinsic subsets are an inherent, reproducible, and stable feature of the disease that is independent of disease duration. Moreover, these data have fundamental importance for the future development of personalized therapy for SSc; drugs targeting inflammation are likely to benefit those patients with an inflammatory signature, whereas drugs targeting fibrosis are likely to benefit those with a fibro- proliferative signature. Journal of Investigative Dermatology (2012) 132, 1363–1373; doi:10.1038/jid.2011.472; published online 9 February 2012

INTRODUCTION and lSSc, there is a heterogeneous range of skin and internal Systemic sclerosis (SSc) is a multisystem autoimmune organ involvement. Approaches that objectively quantify disorder with a hallmark of skin fibrosis and thickening along disease heterogeneity and predict internal organ involvement with significant internal organ involvement (Mayes et al., are critically needed. 2003). SSc has historically been divided into limited and Previous genome-wide gene expression studies in SSc skin diffuse disease based on the extent of skin involvement, with identified disease-specific gene expression signatures in both limited cutaneous SSc (lSSc) involving skin restricted to the lesional and non-lesional skin biopsies that are distinct from regions below the elbows, knees, and face, and diffuse those found in healthy controls (Whitfield et al., 2003; cutaneous SSc (dSSc), including more proximal skin. The Gardner et al., 2006; Milano et al., 2008). In addition, we degree of skin involvement has a direct correlation with SSc have shown that distinct gene expression signatures divide prognosis and internal organ complications (Barnett et al., SSc patients into ‘‘intrinsic subsets’’, capturing the clinical 1988; Scussel-Lonzetti et al., 2002). However, within dSSc heterogeneity of limited versus diffuse SSc, but extending this heterogeneity by revealing that patients with dSSc fall into several different subsets based on gene expression in the skin 1Department of Genetics, Dartmouth Medical School, Hanover, (Milano et al., 2008). These results suggested that distinct 2 New Hampshire, USA and Boston University School of Medicine, Arthritis pathogenic mechanisms may drive disease in different Center, Boston, Massachusetts, USA patients or at different stages of the disease. We previously Correspondence: Michael L. Whitfield, Department of Genetics, Dartmouth Medical School, 7400 Remsen, Hanover, New Hampshire 03755, USA. identified four intrinsic gene expression subsets: a ‘‘diffuse- E-mail: [email protected]; or Robert Lafyatis, Boston proliferation’’ group comprised completely of patients with University School of Medicine, Medical Campus, Evans 501, 72 East Concord dSSc (here referred to as fibroproliferative), showing increased Street, Boston, Massachusetts 02118-2526, USA. E-mail: [email protected] expression of associated with cell proliferation that Abbreviations: dSSc, diffuse cutaneous SSc; GO, ; lSSc, limited could be further subdivided into two groups: ‘‘diffuse 1’’ and cutaneous SSc; MRSS, modified Rodnan skin score; PPAR-g, peroxisome proliferation–activated receptor-g; SSc, systemic sclerosis ‘‘diffuse 2’’; an ‘‘inflammatory’’ group comprised of dSSc, Received 10 February 2011; revised 22 November 2011; accepted 27 lSSc, and morphea samples, showing increased expression November 2011; published online 9 February 2012 of genes associated with inflammation; a ‘‘limited’’ group

& 2012 The Society for Investigative Dermatology www.jidonline.org 1363 SA Pendergrass et al. Stable Subsets in Systemic Sclerosis

comprised primarily of patients with lSSc; and the ‘‘normal- Patient biopsies taken at different time points show similar like’’ group of dSSc and lSSc patients, showing gene patterns of gene expression expression similar to healthy controls. A weak relationship We carried out a second analysis to specifically explore whether was found between disease duration and these intrinsic patients showed significant alterations in gene expression subset subsets, suggesting that they might reflect evolution of the over time (intrinsic-by-time point analysis). Genes were disease process rather than biologically distinct pathogenic selected that showed the most consistent expression at a processes. single time point for each patient, but had the most diverse Here we recapitulate the intrinsic subsets, show these expression between time points (1,888 genes, false discovery subsets are stable over time, and that treatment with ritu- rate of 1.58%). Organizing the samples by hierarchical ximab fails to alter skin gene expression. These data illustrate clustering shows that serial biopsies from 13 of 14 patients that patients with an inflammatory signature do not go on to group together, even though this analysis emphasizes develop a fibroproliferative signature, suggesting a possible differences between time points (Figure 1b and Supplemen- explanation as to why, in the past, broad-spectrum anti- tary Figure S1 online), indicating that serial biopsies are inflammatory agents may not have worked in SSc. It also more similar to each other than to any other samples over indicates that different pathogenic mechanisms drive disease the 6 months to 2 years analyzed. The dendrograms for the pathogenesis within phenotypically similar patients with intrinsic-by-time point and intrinsic-by-patient analyses are dSSc and that this heterogeneity can be consistently and remarkably similar, confirming that gene expression varies reproducibly detected by analyzing skin gene expression, little across time (Figure 1). (Additional analyses are available having broad implications for the future development of in the Supplementary Material online.) therapies for SSc. Distinct pathways are associated with each intrinsic gene RESULTS expression group dSSc skin biopsies can reproducibly be divided into ‘‘intrinsic’’ Distinct sets of genes were associated with each subset that gene expression subsets corresponded to specific biological processes in the skin, We analyzed skin biopsies from dSSc patients for gene represented by gene ontology (GO) biological processes expression changes indicative of patient-specific heterogene- (Milano et al., 2008; Figure 2; and Supplementary Data file ity. Gene expression was measured in 60 skin biopsies from S3 online). Genes associated with the inflammatory group are 22 patients with dSSc and 9 healthy controls (Supplementary enriched for the GO biological processes of immune system Table 1 online). A total of 89 microarrays were hybridized, response and inflammatory response (Pp0.001, The Data- which included 29 technical replicates. All patients were base for Annotation, Visualization, and Integrated Discovery biopsied at a lesional forearm site; a subset was also biopsied analysis (DAVID); Figure 2d and e) and include IFN-induced at a non-lesional back site. Clinical data can be found in genes, such IFIT1, IFIT2, and OAS3. This group of genes is Supplementary Table 2 online. also enriched for the GO biological processes of vasculature Skin biopsies from dSSc patients were analyzed before and development (Pp0.01), including the genes vascular en- after treatment with rituximab for gene expression changes. dothelial growth factor C (VEGFC) and endoglin (ENG), as Consistent with the lack of clinical response (Lafyatis et al., well as genes associated with fibrosis (COL6A3, COL6A1, 2009), we did not find a significant change in gene expression COL5A2, COL5A1, COL1A1,andCOL1A2), collagen oligo- associated with rituximab treatment. Instead, gene expression meric matrix protein (COMP), and matrix metalloproteinase was nearly identical between serial biopsies of patients before 9(MMP9) (Varga and Jimenez, 1995; Jimenez et al., 1996; and after treatment (see Supplementary Material online; Ramirez et al., 2006; Liu and Zhang, 2008). Supplementary Data File S1 online; and Supplementary Two groups of dSSc patients showed increased expression Figure S2 online). of the proliferation signature indicative of dividing cells We previously selected a set of ‘‘intrinsic’’ genes that showed (Figure 2g) and low expression of the inflammatory signature consistent non-changing gene expression between the lesional (Figure 2d), labeled diffuse 1 (blue) and diffuse 2 (red, forearm and non-lesional back biopsies, but showed the largest showing a more prominent proliferation signature) (Whitfield changes between different patients (Milano et al., 2008), et al., 2002). Genes associated with this subset are enriched allowing us to compare differences between patients rather for the GO biological processes of mitosis, m-phase of the than between lesional/non-lesional biopsies. This resulted in mitotic cell cycle, segregation (Pp0.001), and the identification of patient subsets based on gene expression. DNA metabolic process (Pp0.05). These include the cell These groups were labeled fibroproliferative, inflammatory, cycle regulators CDCA8, CDC2, the kinesins KIF2C, KIF11, limited, and normal-like based on the biological gene expres- and cyclins CCNB2 and CCNB1. sion programs that predominated in each subset. We used Pathways more prominent in this study than seen pre- the same strategy to classify patients in this independent viously include metabolism (Milano et al., 2008), cohort of patients. Hierarchical clustering using 2,377 with increased expression in the normal-like and diffuse 1 intrinsic genes (false discovery rate of 0.4%) recapitulated subsets (Figure 2c). Enriched GO biological processes included the major intrinsic subsets, including the fibroproliferative metabolism and (Pp0.001), (diffuse 1 and diffuse 2), inflammatory, and normal-like which contained the peroxisome proliferation–activated groups (Figures 1a and 2). receptor-g (PPAR-g) coactivator a 1 gene. A group of genes

1364 Journal of Investigative Dermatology (2012), Volume 132 SA Pendergrass et al. Stable Subsets in Systemic Sclerosis

*P - 0.05 *P - 0.01 * * * * * A A_r dSSc_7_FA dSSc_7_FA_r RIT3_6MOS_F dSSc_3_FA dSSc_3_FA_r dSSc_3_FA_r dSSc_3_FA_r dSSc_3_B_r dSSc_3_B_r dSSc_3_B_r dSSc_6_FA dSSc_6_FA_r dSSc_6_FA_r dSSc_2_FA dSSc_2_6MOS_FA dSSc_1_FA dSSc_1_10MOS_FA dSSc_8_FA RIT15_6MOS_FA RIT15_BASE_FA RIT15_BASE_FA_r RIT15_BASE_FA_r RIT8_6MOS_FA RIT8_6MOS_B RIT8_BASE_B RIT10_36MOS_FA RIT10_6MOS_FA RIT13_6MOS_FA RIT13_BASE_FA RIT5_6MOS_B RIT5_6MOS_FA RIT5_BASE_FA RIT5_BASE_B RIT6_6MOS_FA RIT6_BASE_FA RIT6_6MOS_B RIT6_BASE_B RIT6_BASE_B_r RIT7_6MOS_FA RIT7_6MOS_B RIT7_BASE_FA RIT7_PREBASE_FA RIT7_PREBASE_B Normal_1 Normal_1_r Normal_1_r Normal_2 Normal_2_r Normal_3 Normal_3_r Normal_9 Normal_9_r Normal_8 Normal_8_r Normal_4 Normal_6 Normal_6_r Normal_7 Normal_5 Normal_5_r dSSc_5_FA dSSc_5_FA_r dSSc_4_FA dSSc_4_B dSSc_4_B_r dSSc_9_FA RIT11_6MOS_FA RIT11_BASE_FA RIT11_BASE_FA_r RIT11_6MOS_B RIT11_6MOS_B_r RIT11_BASE_B RIT17_20MOS_FA RIT17_6MOS_FA RIT17_BASE_FA RIT12_6MOS_FA RIT12_6MOS_FA_r RIT12_BASE_FA_r RIT2_6MOS_B RIT14_6MOS_FA RIT14_6MOS_FA_r RIT14_6MOS_FA_r RIT4_BASE_FA RIT4_BASE_FA_r RIT4_BASE_FA_r RIT3_6MOS_B RIT3_BASE_FA_r RIT3_BASE_B RIT3_6MOS_F

* * * * * * * RIT17_20MOS_FA RIT17_BASE_FA RIT11_6MOS_B RIT11_6MOS_B_r RIT11_6MOS_FA RIT11_BASE_FA RIT11_BASE_FA_r RIT11_BASE_B RIT12_6MOS_FA RIT12_6MOS_FA_r RIT12_BASE_FA RIT14_6MOS_FA RIT14_6MOS_FA_r RIT14_6MOS_FA_r RIT14_BASE_FA RIT14_BASE_FA_r RIT14_BASE_FA_r dSSc_6_FA dSSc_6_FA_r dSSc_6_FA_r RIT3_6MOS_B RIT2_6MOS_B dSSc_3_FA_r dSSc_3_FA_r dSSc_3_FA_r dSSc_3_FA_r dSSc_3_B_r dSSc_3_B_r dSSc_3_B_r RIT3_BASE_FA RIT3_BASE_B dSSc_1_FA dSSc_1_10MOS_FA dSSc_8_FA dSSc_2_FA dSSc_2_6MOS_FA RIT15_6MOS_FA RIT15_BASE_FA RIT15_BASE_FA_r RIT15_BASE_FA_r RIT10_36MOS_FA RIT10_6MOS_FA RIT13_6MOS_FA RIT13_BASE_FA RIT8_6MOS_FA RIT8_6MOS_B RIT8_BASE_B RIT5_6MOS_B RIT5_6MOS_FA RIT5_BASE_FA RIT5_BASE_B RIT6_6MOS_A RIT6_6MOS_B RIT6_BASE_FA RIT6_BASE_B RIT6_BASE_B_r RIT7_6MOS_FA RIT7_6MOS_B RIT7_BASE_FA RIT7_PREBASE_FA RIT7_PREBASE_B Normal_1 Normal_1_r Normal_1_r Normal_2 Normal_2_r Normal_3 Normal_3_r dSSc_5_FA dSSc_5_FA_r dSSc_7_FA dSSc_7_FA_r RIT3_6MOS_FA_r RIT3_6MOS_FA dSSc_4_FA dSSc_4_B dSSc_4_B_r dSSc_9_FA Normal_9 Normal_9_r Normal_5 Normal_5_r Normal_4 Normal_8 Normal_8_r RIT11_6MOS_FA Normal_6_r Normal_6 Normal_7

Figure 1. Gene expression over time in systemic sclerosis (SSc) skin. Shown are the hierarchical clustering dendrograms of the (a) ‘‘intrinsic-by-patient’’ and (b) ‘‘intrinsic-by-time point’’ analyses. Dendrogram branches are colored by subtype: normal-like (green), inflammatory (purple), diffuse 1 (blue), and diffuse 2 (red), which represent the fibroproliferative group. Statistically significant branches are indicated by an asterisk. Black bars below the sample identifiers indicate arrays from skin biopsies from the same patient that clustered together; yellow bars below the identifiers identify arrays for a single patient that split between groups. Black arrows connect longitudinal samples. Overlaid between the two dendrograms are shaded bars indicating arrays that changed intrinsic subset between the two analyses. dSSc, diffuse cutaneous systemic sclerosis; RIT, samples in the rituximab study. with related function, but a different expression pattern, Intrinsic gene expression groups are stable over time centers on PPAR-g gene expression (Figure 2f) (Wei et al., Two results from this study suggest that the intrinsic subsets 2010). Genes within this PPAR-g pathway–related group are not dependent on disease duration. The four major intrinsic include several PPAR-g target genes: CD36 (Huang et al., subsets identified in our previous study, where individuals 2004), (LPL)(Schoonjanset al., 1996), had variable and longer disease duration (Milano et al., stearoyl-CoA desaturase (SCD) (Vondrichova et al., 2007), 2008), were also found in this study cohort where all patients and catalase (CAT) (Girnun et al., 2002). have early-stage disease (Supplementary Figure S5a online). If We summarized the differentially regulated biological path- the subsets were dependent on disease duration, then we ways by averaging the genes associated with each (Supplemen- would expect a skewing toward the early subset in these data. tary Material online and Supplementary Figures S3–S4 online). In addition, there is no significant difference in disease We observe an increase in pathways associated with immune duration between subsets measured here (Supplementary system activation (Supplementary Figure S3a online) and an Figure S5b online). Collectively, these findings indicate that increase in gene expression associated with B cells, CD8 þ T gene expression subsets are an inherent feature of the clinical cells, leukocytes, macrophages, and IFN-treated keratino- dSSc phenotype and that this feature is independent of cytes and fibroblasts (Supplementary Figure S3b online). The disease duration. diffuse 2 subset showed an increase in cell cycle–related Despite the consistent gene expression, modified Rodnan processes, decreases in , steroid and fatty acid skin score (MRSS) did change in patients who provide long- metabolism, as well as enrichment of genes associated with itudinal biopsies (Supplementary Figure S6 online). In all, 7 of activated peripheral blood mononuclear cells, TNF-a-treated the 15 patients showed increases in skin score, 2 patients keratinocytes, T cells, and dendritic cells (Supplementary (RIT7 and RIT14) showed a decrease and the remaining 6 Figure S3a and b online). The normal-like and diffuse 1 group showed little change. In all cases, patients maintained a showed a decrease in immune activation and increases in stable pattern of gene expression. The range of MRSS for . patients in this data set has a slightly broader distribution than

www.jidonline.org 1365 1366 tbeSbesi ytmcSclerosis Systemic in Subsets Stable Pendergrass SA ora fIvsiaieDraooy(02,Vlm 132 Volume (2012), Dermatology Investigative of Journal

bv einBelow Median Above 1.00

0.6 Normal_1 Normal_1_r 0 Normal_1_r Normal_2 Normal_2_r

–0.6 Normal_3 al. et Normal_3_r Normal_9 –1.00 Normal_9_r Normal_8 Normal_8_r Normal_4 Normal_6 Normal_6_r Normal_7 Normal_5 Normal_5_r dSSc_5_FA

Normal-like dSSc_5_FA_r dSSc_7_FA dSSc_7_FAA_r RIT3_6MOS_FA_r RIT3_6MOS_FA * dSSc_4_FA dSSc_4_B dSSc_4_B_r dSSc_9_FA RIT11_6MOS_FA Diffuse 1 RIT11_BASE_FA RIT11_BASE_FA_r RIT11_6MOS_B RIT11_6MOS_B_r RIT11_BASE_B RIT17_20MOS_FA RIT17_6MOS_FA RIT17_BASE_FA RIT12_6MOS_FA RIT12_6MOS_FA_r RIT12_BASE_FA RIT2_6MOS_B RIT14_6MOS_FA

Inflammatory RIT14_6MOS_FA_r RIT14_6MOS_FA_r RIT14_BASE_FA RIT14_BASE_FA_r

RIT14_BASE_FA_r * RIT3_6MOS_B RIT3_BASE_FA_r RIT3_BASE_B dSSc_3_FA

Diffuse 2 dSSc_3_FA_r dSSc_3_FA_r dSSc_3_FA_r dSSc_3_B_r * dSSc_3_B_r dSSc_3_B_r WEE1 homolog WEE1 CDC42SE2 kinase21) CIT Citron(rho-interacting, serine/threonine CCNA2 CyclinA2 FOXM1 Forkhead box M1 CENPE CentromereproteinE,312kDa CDC20 celldivisioncycle20homolog CENPF CentromereproteinF CDC2 Celldivisioncycle2,G1toSandG2M CDCA5 Celldivisioncycleassociated5 IFI27 factor elementbindingtranscription 1 SREBF1 Sterolregulatory ACAA2 Acetyl-coenzymeAacyltransferase 2 ELOVL5 ELOVL family member5,elongationoflongchainfatty acids SC5DL Sterol-C5-desaturase-like FADS1 FDFT1 DHCR7 7-Dehydrocholesterol reductase MVK Mevalonate kinase SLC27A2 ITGB8 PPARA Peroxisome proliferative activated receptor, alpha PLA2G3 PhospholipaseA2,group III PEX7 Peroxisomal biogenesisfactor 7 IFITM3 IFITM1 IFIT2 MX1 Myxovirus resistance1,interferon-inducible proteinp78 G1P2 Interferon, alpha-inducible protein HSD11B2 ALDH5A1 PPARGC1A CCNB2 KIF2C Kinesinf CCNB1 CyclinB1 CDCA8 Celldivisioncycleassociated8 CDC45L CCNE1 CyclinE1 CDC7 celldivisioncycle7 PDK4 Pyruvate dehydrogenase kinase, isoenzyme 4 SCD Stearoyl-CoA desaturase CD36 antigen(collagentypeI,thrombospondinreceptor) PPARG Peroxisome proliferative activated receptor, gamma LPL Lipoproteinlipase HRASLS3 RAS-like suppressor3 RARRES2 CAT Catalase TNFRSF1A Tumor necrosisfactor receptorsuperfamily, 1A TNFSF10 Tumor necrosisfactor (ligand)superfamily, member10 PDGFRB Platelet-derived growth factor receptor, beta COL6A1 COL1A1 2 THBS2 Thrombospondin receptor,IL13RA1 Interleukin-13 alpha1 ENG Endoglin CCL2 Chemokine(C-Cmotif)ligand2 COL8A1 1 THBS1 Thrombaspondin receptor IL4R Interleukin-4 LNK Lymphocyte adaptorprotein IFI16 CTHRC1 CTGF Connective tissuegrowth factor FAP COL6A3 COL5A1 NRP1 Neuropilin1 C3 Complementcomponent3 COL4A1 metalloproteinase9 MMP9 Matrix protein matrix oligomeric COMP Cartilage TNFRSF12A Tumor necrosisfactor receptorfamily, member12A VEGFC Vascular endothelialgrowth factor C PTX3 Pentaxin-related gene, rapidly inducedby IL-1beta LTB IL27RA HLA-DPB1 PRKCA ProteinkinaseC, alpha TRAF5 TNF receptor-associatedfactor 5 CXCL1 Chemokineligand1 TAP1 Transporter1 STAT1 Signaltransducer andacivator 1 oftranscription factorIRF7 Interferon-regulatory 7 LGALS9 Lectin,galactoside-binding,soluble, 9 TIMP1 Tissue inhibitorofmetalloproteinase1 IFITM2 dSSc_6_BA dSSc_6_FA_r Lymphotoxin beta(TNFsuperfamily, member3) Fibroblast activation protein,alpha Interferon, alpha-inducible protein27 Interferon, gamma-inducible protein16 Interferon-induced repeats2 proteinwithtetratricopeptide dSSc_6_FA_r Integrin, beta8 Integrin, Farnesyl-diphosphate farnesyltransferase 1 Fatty aciddesaturase 1 Interf Interferon inducedtransmembrane protein1 Interferon inducedtransmembrane protein2 Interleukin-27 receptor, Interleukin-27 alpha CyclinB2 dSSc_2_FA CDC45celldivisioncycle45-like Collagen,type VI, alpha1 Collagen,typeI,alpha1 Collagen,type VIII, alpha1 Collagen,type VI, alpha3 Collagen,type V, alpha1 Collagen,typeIV, alpha1 Collagen triple helixrepeatcontaining1 Collagentriple Solute carrier family Solutecarrier member2 27(fatty acidtransport), Hydroxysteroid (11-beta)dehydrogenase 2 Aldehyde dehydrogenase 5family, memberA1 dSSc_2_6MOS_FA Majorhistocompatibilitycomplex, classII,DPbeta1 Retinoic acidreceptorresponder2 CDC42smalleffector 2 Peroxisome proliferative activated receptor, gamma,coact1,alpha

eron inducedtr dSSc_1_FA

amily member2C dSSc_1_10MOS_FA dSSc_8_FA RIT15_6MOS_FA RIT15_BASE_FA RIT15_BASE_FA_r ansmembrane protein3 RIT15_BASE_FA_r RIT8_6MOS_FA RIT8_6MOS_B

RIT8_BASE_B * RIT10_36MOS_FA * P-0.05 * P-0.01 RIT10_6MOS_FA RIT13_6MOS_FA RIT13_BASE_FA RIT5_6MOS_B RIT5_6MOS_FA RIT5_BASE_FA RIT5_BASE_B RIT6_6MOS_FA RIT6_BASE_FA RIT6_6MOS_B RIT6_BASE_B RIT6_BASE_B_r RIT7_6MOS_FA * RIT7_6MOS_B RIT7_BASE_FA NM_013261 NM_003390 NM_020240 NM_007174 NM_001237 NM_202002 NM_001813 NM_001255 NM_016343 NM_001786 NM_080668 NM_004701 NM_006845 NM_031966 NM_018101 NM_003504 NM_001238 NM_003503 NM_002612 NM_005063 NM_001001547 NM_138711 NM_000237 NM_007069 NM_002889 NM_001752 NM_001065 NM_003810 NM_002609 NM_001848 Z74615 L12350 NM_001560 NM_000118 NM_002982 NM_003246 NM_000418 NM_005475 NM_005531 NM_138455 NM_001901 NM_004460 NM_004369 NM_000093 NM_003873 NM_000064 NM_001845 NM_004994 NM_000095 NM_016639 NM_005429 NM_002852 NM_002341 NM_004843 NM_002121 NM_002737 NM_004619 NM_001511 NM_000593 NM_139266 NM_004031 NM_009587 NM_006435 NM_021034 NM_003641 NM_001547 NM_002462 NM_005101 NM_005532 NM_001005291 NM_006111 NM_006918 NM_013402 NM_004462 NM_001360 NM_003645 NM_001001930 NM_015715 NM_000288 NM_000196 NM_170740 RIT7_PREBASE_FA RIT7_PREBASE_B SA Pendergrass et al. Stable Subsets in Systemic Sclerosis

the skin scores in Milano et al. (2008) (Supplementary Figure similar to those found here using the same set of genes S5c online). When lSSc patients are excluded, the MRSS of (Milano et al., 2008; Figure 3c). The fibroproliferative and proliferative and inflammatory groups though broader in inflammatory groups in the two data sets are indicated with distributions are otherwise similar to our past study (Milano the fibroproliferative groups in red and the inflammatory et al., 2008) (Supplementary Figure S5c and d online). The group in purple. Subsets of overlapping genes between the inflammatory group has the widest range of MRSS scores, groups are indicated. The inflammatory signature is more whereas the normal-like group consistently shows lower MRSS prominent in the data set presented here, whereas the scores in both data sets. Autoantibodies did not show a proliferation signature is more prominent in our previous significant association with intrinsic subset (Supplementary report (Milano et al., 2008). Therefore, the original 995 genes Tables S2 online). One diffuse 1 patient was anti-scl-70 can stratify an independent cohort into the intrinsic patient positive (1/4; P ¼ 0.52, Fisher’s exact test), two inflammatory subsets. patients were anti-RNA polymerase III positive (2/7; P ¼ 0.12), and three unclassified patients were anti-scl-70 Surrogate gene biomarkers of MRSS positive (2/3; P ¼ 0.051) and anti-centromere positive (1/3; We previously reported a 177-gene expression signature that P ¼ 0.16). Diffuse 2 patients were negative for all three could serve as surrogate biomarker of MRSS (Milano et al., measured autoantibodies (P ¼ 0.52). 2008). We refined this by identifying 44 genes that were present in the 177-gene signature, and also found in the Independent validation by immunohistochemistry intrinsic-by-patient analysis of this study (Figure 2). Organiz- To validate the mRNA expression, we analyzed proteins for ing the samples using these 44 genes revealed two major COMP and IFN-induced transmembrane protein 3 (IFITM3) subdivisions of samples. One included both dSSc patients in representative biopsies spanning the intrinsic subsets. and healthy controls (Figure 4a, group 1), whereas the other Both showed highest expression in the inflammatory subset included only dSSc patients (Figure 4a, group 2). Subjects consistent with gene expression data (Figure 2d and in these groups showed a significant difference in mean Supplementary Figure S7 online). Immunohistochemical MRSS (group 1: mean 17.92, standard deviation 11.61; staining results paralleled and confirmed the gene expression group 2: mean 25.90, standard deviation 8.52; t-test, findings. COMP showed highest expression in the inflamma- P ¼ 0.005; Figure 4b). tory subset and lowest in the diffuse 2 subset (Po0.05), with slightly higher expression in the diffuse 1 subset (Supple- Gene expression gradients are evident within a gene mentary Figure S7a online); protein staining was most expression subset prevalent in dermal fibroblasts of SSc patients of the To further power the analysis of SSc skin gene expression inflammatory subset (Supplementary Figure S7c and g across time scales that cannot be easily captured with online), while absent in controls (Supplementary Figure S7e longitudinal biopsies, we combined the gene expression online). IFITM3 showed highest expression in both the analyses from both this and our previous study (Milano et al., inflammatory and diffuse 2 subsets (Supplementary Figure 2008). The combined data set includes skin biopsies from 39 S7b online) and lowest expression in the diffuse 1 subset patients with dSSc, 7 with lSSc, 3 with morphea, 1 patient (Po0.05), with staining around the microvasculature in the with eosinophilic fasciitis, and skin biopsies from 15 healthy skin (Supplementary Figure S7d and h online). These data control subjects, totaling 121 biopsies. After distance-weighted confirm and extend the gene expression findings at the discrimination adjustment to remove systematic differences protein level. (Benito et al., 2004), intrinsic analysis was performed, 3,551 probes selected (false discovery rate of 0.07%; Figure 5a), Validation of the 995-gene intrinsic subset gene set and the combined data sets clustered hierarchically. The two We next determined whether the 995 genes selected in our data sets recapitulated the major intrinsic subsets (Figure 5a). previous study could stratify the cohort of patients described Gene expression in the ‘‘fibroproliferative’’ and ‘‘inflamma- here into the intrinsic subsets (Milano et al., 2008), and thus tory’’ groups both showed gradients of gene expression could be developed into a classifier for subset stratification. In within their respective groups (Figure 5a). In the proliferation total, 808 genes that passed basic quality filters were used to group, the dendrogram split samples with high expression of organize the samples by hierarchical clustering (Figure 3b), the signature (‘‘Prolif 1’’) and low expression (‘‘Prolif 2’’). showing that 26 out of 31 dSSc skin biopsies from different Similarly, the inflammatory group contained samples with anatomical sites or time points were grouped together by high (‘‘Inflam 1’’) and low (‘‘Inflam 2’’) expression of the patient (Figure 3a). The subsets identified previously are signature. To determine if the intensity of these signatures

Figure 2. Recapitulation of the intrinsic subsets. In all, 2,377 genes were selected from 89 arrays (31 individuals) by ‘‘intrinsic-by-patient’’ analysis. (a) Hierarchical clustering dendrogram shows the normal-like (green branches), inflammatory (purple), and fibroproliferative groups (diffuse 1 (blue), and diffuse 2 (red)). Significance of clustering was determined by Statistical Significance of Clustering (SigClust). Healthy control identifiers are green and diffuse cutaneous systemic sclerosis (dSSc) are black. RIT indicates samples in the rituximab study. Black bars indicate patient samples that clustered together; the yellow bar indicates arrays from patient RIT3 that did not cluster together. (b) Heat map of genes and samples clustered hierarchically. (c) Fatty acid synthesis genes. (d and e) Inflammatory and collagen genes. (f) Peroxisome proliferation–activated receptor-g (PPAR-g) genes. (g) Proliferation cluster genes.

www.jidonline.org 1367 1368 tbeSbesi ytmcSclerosis Systemic in Subsets Stable Pendergrass SA ora fIvsiaieDraooy(02,Vlm 132 Volume (2012), Dermatology Investigative of Journal

bv einBelow Median Above 1.00 dSSc_3_B_r dSSc_3_B_r dSSc_3_B_r 0.6 RIT3_BASE_FA_r Normal_1_r Normal_1 Normal_1_r Normal_2

0 al. et Normal_2_r Normal_6 Normal_6_r Normal_7 –0.6 Normal_3 Normal_3_r dSSc_5_FA –1.00 dSSc_5_FA_r dSSc_7_FA dSSc_7_FA_r RIT3_6MOS_FA_r * RIT3_6MOS_FA SCGB2A1 Secretoglobin,family2A,member1 MPEG1 Macrophageexpressedgene1 CPA3 CarboxypeptidaseA3(mastcell) VSIG4 V-setandimmunoglobulindomaincontaining4 GIMAP6 GTpase,IMAPfamilymember6 LILRB2 Leukocyteimmunoglobulin-likereceptor PPIC PeptidylprolylisomeraseC(cyclophilinC) DDX58 DEAD(Asp–Glu–Ala–Asp)boxpolypeptide58 CD8A CD8antigen,alphapolypeptide(p32) TAP1 Transporter1,ATP-bindingcassette IFITI MX1 Myxovirus(influenzavirus)resistance1,interfron-inducibleprotein MRCL3 Myosinregulatorylightchain MSN Moesin LUM Lumican COL6A3 Collagen,typeVI,alpha3 MARCKS Myristoylatedalanine-richproteinkinaseCsubstrate CDW52 antigen(CAMPATH-1antigen) PTPRC Proteintyrosinephosphatase,receptortype,C APOL3 ApoliporoteinL,3 ICAM2 Intercellularadhesionmolecule2 CD33 antigen(gp67) GBP3 Guanylatebindingprotein3 FXYD2 FXYDdomaincontainingiontransportregulator2 GSG2 Haspin CENPE CentromereproteinE,312kDa LOC123876 Xenobiotic/medium-chainfattyacid:CoAligase ATAD2 ATPasefamily,AAAdomaincontaining2 APOH ApoliporoteinH(beta-2-glycoproteinI) TTR Transthyretin(prealbumin,amyloidosistypeI) IL23A Interleukin-23,alphasubunitp19 RAD51AP1 RAD51associatedprotein1 PPFIA4 Proteintyrosinephosphataseinteractingprotein CBLL1 Cas_Br_M(murine)ecotropicretroviraltransformingsequence-like1 FXN Frataxin ZSCAN2 ZincfingerandSCANdomaincontaining2 HPS3 Hermansky-Pudlaksyndrome3 CACNG6 Calciumchannel,voltage-dependent,gammasubunit6 OGDHL Oxoglutaratedehydrogenase-like CDT1 DNAreplicationfactor SYT6 SynaptotagminVI RIT3_6MOS_B RIT3_BASE_B Interferon-inducedproteinwithtetratricopeptiderepeats1 RIT14_6MOS_FA RIT14_6MOS_FA_r RIT14_6MOS_FA_r RIT14_BASE_FA RIT14_BASE_FA_r RIT14_BASE_FA_r dSSc_2_FA dSSc_6_FA dSSc_6_FA_r dSSc_6_FA_r dSSc_2_FA dSSc_3_FA_r dSSc_3_FA_r dSSc_3_FA_r dSSc_3_FA_r dSSc_1_FA dSSc_1_10MOS_FA dSSc_8_FA RIT5_6MOS_FA RIT5_BASE_FA RIT5_6MOS_FA RIT5_BASE_FA * RIT5_BASE_FA_r RIT5_BASE_FA_r RIT8_6MOS_FA RIT8_6MOS_B RIT10_36MOS_FA RIT10_36MOS_FA RIT13_36MOS_FA RIT13_BASE_FA dSSc_4_FA dSSc_4_B dSSc_4_B_r dSSc_9_FA Normal_9 Normal_9_r Normal_4 Normal_5 Normal_5_r Normal_8 Normal_8_r RIT11_6MOS_B RIT11_6MOS_B_r RIT11_6MOS_FA RIT11_BASE_FA RIT11_BASE_B RIT11_BASE_FA_r RIT17_20MOS_FA RIT17_6MOS_FA RIT17_BASE_FA RIT2_6MOS_B RIT12_6MOS_FA RIT12_6MOS_FA_r

RIT12_BASE_FA * RIT5_6MOS_B RIT5_BASE_B RIT6_6MOS_FA RIT6_BASE_FA RIT6_6MOS_B RIT6_BASE_B RIT6_BASE_B_r

RIT7_6MOS_FA *P RIT7_6MOS_B

RIT7_BASE_FA -0.05 RIT7_preBASE_FA RIT7_preBASE_B RIT8_BASE_B SA Pendergrass et al. Stable Subsets in Systemic Sclerosis

were associated with disease duration or MRSS, we compared with ‘‘Prolif 2’’ and a trend towards a higher compared the distributions between these two parameters MRSS (Po0.05). In contrast, the ‘‘Inflam 2’’ group, which had within each group (Figure 6). The ‘‘Prolif 1’’ group had a a higher inflammatory gene expression signature, showed a significantly longer disease duration (t-test, P ¼ 0.0027) significantly higher average MRSS (t-test, P ¼ 0.016), with

Group 1 Group 2

RECK Reversion-inducing-cysteine-rich protein with kazal motifs FBLN1 Fibulin 1 PDGFRL Platelet-derived growth factor receptor-like KAZALD1 Kazal-type serine protease inhibitor domain 1 OSR2 Odd-skipped related 2 IGFBP5 -like growth factor binding protein 5 PTGIS l2 (prostacyclin) synthase GHR Growth hormone receptor ECM2 protein 2, female organ and specific PCOLCE2 Procollagen C-endopeptidase enhancer 2 ROBO3 Roundabout, axon guidance receptor, homolog 3 NR3C1 Nuclear receptor subfamily 3, group C, member 1 CFHL1 Complement factor H-related 1 CRTAP Cartilage associated protein TMOD3 Tropomodulin 3 (ubiquitous) TNFRSF12A Tumor necrosis factor receptor superfamily, member 12A LGALS8 Lectin, galactoside-binding, soluble, 8 (galectin 8) CENPE Centromere protein E, 312 kDa CDC7 CDC7 cell division cycle 7 1.00 0.6 0 –0.6 –1.00

Above Median Below

40

30 MRSS 20

10

Group 1 Group 2

Figure 4. Surrogate gene expression biomarkers of modified Rodnan skin score (MRSS). (a) Probes that matched the 177 genes with correlations above |0.5| from Milano et al. (2008) were extracted from the ‘‘by-patient’’ intrinsic analysis, resulting in 44 genes. Hierarchical clustering results in two groups. Group 1 (red branches) includes dSSc and healthy control skin biopsies, whereas group 2 (black branches) includes primarily dSSc skin biopsies. The first row of bars below the dendrogram indicates the intrinsic subset assignment in the ‘‘by-patient’’ analysis (normal-like, green; diffuse 1, blue; diffuse 2, red; inflammatory, purple; unclassified, black). The second row of bars indicates sample diagnosis, dSSc (red) or healthy control (black). (b) Box plot comparison of MRSS between the two groups shows a statistically significant difference in MRSS (two-sample t-test, P ¼ 0.005). The MRSS at time of biopsy is plotted with open circles.

Figure 3. Concordance between data sets. Hierarchical clustering of 808 genes that matched the 995 intrinsic genes from Milano et al. (2008) (187 did not pass the basic filtering criteria). (a) Hierarchical clustering dendrogram shows normal-like (green), inflammatory (purple), diffuse-proliferation (diffuse 1 (blue) and diffuse 2 (red)), and limited groups (yellow). Significant branches are indicated by asterisk. Black bars indicate subject samples that clustered together. (b) Heat map for the 808 genes. (c) Heat map of the original 995 ‘‘intrinsic’’ genes in Milano et al. (2008). The fibroproliferative groups that are between the two data sets are connected in red and the inflammatory groups connected in purple; genes found in the respective clusters of both data sets are indicated. dSSc, diffuse cutaneous systemic sclerosis; RIT, samples in the rituximab study.

www.jidonline.org 1369 SA Pendergrass et al. Stable Subsets in Systemic Sclerosis

Prolif 1 Prolif 2 Inflam 1 Inflam 2

Proliferative Inflammatory Lim Normal - like

e 1.00 0.6 0 –0.6 –1.00

Above Median Below

Figure 5. Consistent classification and expression gradients within subsets. Data from this study and from Milano et al. (2008) were merged to create a single data set of 164 arrays. In all, 3,551 intrinsic genes were selected (false discovery rate of 0.07%). (a) Heat map of 2D hierarchical clustering. (b) Clustering dendrogram with branches indicating the intrinsic subset each sample was assigned in the independent data set analyses (proliferative (red), inflammatory (purple), limited (yellow), and normal-like (green)). The first row of bars below the dendrogram indicates patient diagnosis (dSSc, red; limited cutaneous SSc (lSSc), yellow; morphea and eosinophilic fasciitis, blue; healthy controls, green). The second row of bars indicates the data set the samples were obtained: Milano et al., red; this study, black. (c) Inflammatory gene cluster. (d) Proliferation gene cluster. (e) Fatty acid synthesis gene cluster.

little difference in disease duration across the subgrouping. patients having significantly less skin inflammation as part of Thus, this analysis suggests that the gene expression within the pathological process. The data presented here indicate a group changes intensity with increased disease severity that the gene expression subsets are stable over time. We and/or duration. show that measuring gene expression of skin biopsies from patients at different time points consistently classifies the DISCUSSION patients into the same intrinsic subsets. In addition, these data SSc is a progressive disease, with skin going through various taken from a cohort of patients with shorter disease duration phases that can begin with edema, then progressive fibrosis, compared with our previous studies of patients with longer and in some cases a skin softening late in the disease. Past disease duration show the same disease subsets. studies have suggested that SSc skin pathology evolves from Strikingly, although patients do not appear to move between inflammatory to fibrotic changes over time (Fleischmajer subsets over time, patients within inflammatory and particularly et al., 1977, 1978; Roumm et al., 1984; Kraling et al., 1995). fibroproliferative subsets show changes in the intensity of the Our molecular analyses suggest that inflammatory changes in signature associated with disease duration. Similar changes in the skin are not part of an evolving process, leading to intensity of gene expression was found in a study of limited fibrosis, but rather represent a subset of patients, with other SSc patients with and without PAH, suggesting that as disease

1370 Journal of Investigative Dermatology (2012), Volume 132 SA Pendergrass et al. Stable Subsets in Systemic Sclerosis

subsets of SSc patients. The disease duration of the cohort

40 analyzed here (o2 years) is more homogeneous and more 200 35 typical of when the disease is most actively progressing 150 compared with our previous report. Despite these differences, 30 the results show reproducibility of patient subsets in a com-

100 MRSS 25 pletely new patient population from a different clinical center. 50 20 The reproducibility and stability of these gene expression-based Disease duration (months) 0 15 subsets confirms that they reflect fundamental underlying Prolif 1 Prolif 2 Prolif 1 Prolif 2 pathogenic processes that differ between patients in the different subsets, rather than stages in the progressive disease. The addition of more serial biopsies covering a larger span of 70 60 40 time, coupled with longitudinal clinical data, should provide 50 prognostic information for patients in each group. 30 40 A subset of patients in this analysis was treated with

30 MRSS 20 rituximab, a mAb that depletes mature B cells. In that open- 20 label trial of rituximab, patients did not show a significant 10 change in skin score, pulmonary function, or other measures

Disease duration (months) 10 of organ involvement, although immunohistochemistry showed Inflam 1 Inflam 2 Inflam 1 Inflam 2 depletion of both peripheral and skin-resident B cells (Lafyatis Figure 6. Association between disease duration and modified Rodnan skin et al., 2009). Several other unblinded clinical trials have score (MRSS) within the proliferation and inflammatory subsets. Above the suggested some potential beneficial effect of rituximab on dendrogram in Figure 5, the ‘‘Prolif 1’’, ‘‘Prolif 2’’, ‘‘Inflam 1’’, and ‘‘Inflam 2’’ skin and/or lung disease in patients with SSc (Wesson et al., groups are indicated. These are groups showing differences in the intensity 2008; Bosello et al., 2010; Daoussis et al., 2010; Smith et al., of gene expression within the proliferative and inflammatory subsets. Box plots show differences in disease duration and MRSS between (a and b) 2010). Rituximab may prove to have value in SSc, and it is ‘‘Prolif 1’’ and ‘‘Prolif 2’’, as well as between (c and d) ‘‘Inflam 1’’ and particularly intriguing to consider its potential efficacy in SSc- ‘‘Inflam 2’’ are apparent. Disease duration (months) is plotted for each associated interstitial lung disease where B cell infiltration is blopsy at time collection (a and c). MRSS is plotted for each blopsy in often quite prominent (Lafyatis et al., 2007). However, our panels b and d. studies here are consistent with our clinical trial results, suggesting no or modest effect on skin disease. We therefore believe that rituximab is more likely to show efficacy for SSc- becomes more severe, the intensity of the changes increases associated interstitial lung disease and that lung disease (Pendergrass et al., 2010). Our data are consistent with would be a better target for a large-scale trial of B cell previous observations that the expression of pro-fibrotic depletion. genes in non-lesional SSc fibroblasts tends to be intermediate Comparing the results of Chung et al. (2009), who reported between that observed in healthy and lesional SSc fibroblasts two patients with dSSc that showed a response to treatment (Chen et al., 2005). Thus, our data suggest that different with imatinib mesylate (Chung et al., 2009), with the results biological processes may underpin dSSc pathogenesis in dif- reported here in rituximab-treated patients suggests that when ferent patient subsets, and that the intensity of these processes a clinical response is evident, one is also likely to find a gene change over time. Although we cannot exclude the possibility expression response. Thus, gene expression may be useful as that analysis of serial biopsies over a longer period of time a surrogate outcome measure and for stratifying likely patient would show patients changing subsets, the data here support responders in clinical trials. As the patients with an inflam- the notion that a patient could be assigned to a specific matory signature do not go on to display a fibroproliferative intrinsic subset early in their disease course. These observa- signature, and vice versa, it is essential to target patients with tions then have the potential of suggesting treatments that a therapy that is appropriate to their gene expression block the immune response if a patient displays an inflam- signature. Therefore, it will be imperative to identify patient matory signature, or treatments that block fibrosis, if they subsets in clinical trials as drugs may target only those in a posses a fibroproliferative signature. Thus, as the pathogenic single subset. mechanisms underlying intrinsic subsets are identified, patient subset identification might permit specific targeting of patho- genic processes. A TGFb-responsive signature appears to MATERIALS AND METHODS drive at least in part pathogenesis in the fibroproliferative Patient selection, biopsy processing, and microarray subset of patients (Sargent et al., 2009), emphasizing the hybridization contribution of this cytokine to the phenotype of SSc (Leask, All study participants gave written, informed consent under Boston 2009; Varga and Pasche, 2009), while IL-13 appears to play University Medical Center Institutional Review Board, an approved an important role in the inflammatory subset (Greenblatt protocol. The study conformed to Declaration of Helsinki principles. et al., 2012). Forearm lesional skin and, for a subset of patients, non-lesional back This data set recapitulates and validates our originally skin were collected by punch biopsies (3–6mm) from 13 patients defined inflammatory, diffuse-proliferation, and normal-like enrolled in the rituximab study (Lafyatis et al., 2009), as well as from

www.jidonline.org 1371 SA Pendergrass et al. Stable Subsets in Systemic Sclerosis

9 additional patients with early dSSc (o1.5 years from diagnosis). Daoussis D, Liossis SN, Tsamandas AC et al. (2010) Experience with A total of 60 skin biopsies were collected from 22 patients with dSSc rituximab in scleroderma: results from a 1-year, proof-of-principle study. Rheumatology (Oxford) 49:271–80 and 9 healthy controls (Supplementary Tables S1 and S2 online). RNA purification and microarray hybridization was carried out Fleischmajer R, Gay S, Meigel WN et al. (1978) Collagen in the cellular and fibrotic stages of scleroderma. Arthritis Rheum 21:418–28 essentially as described (Milano et al., 2008). Detailed methodology Fleischmajer R, Perlish JS, West WP (1977) Ultrastructure of cutaneous is provided in Supplementary Text online. cellular infiltrates in scleroderma. Arch Dermatol 113:1661–6 Gardner H, Shearstone JR, Bandaru R et al. (2006) Gene profiling of Bioinformatic and statistical analyses scleroderma skin reveals robust signatures of disease that are imperfectly Intrinsic gene analysis was carried out as described previously reflected in the transcript profiles of explanted fibroblasts. Arthritis (Supplementary Material online; Milano et al., 2008). Gene expression Rheum 54:1961–73 data were organized by average linkage hierarchically clustering Girnun GD, Domann FE, Moore SA et al. (2002) Identification of a functional peroxisome proliferator-activated receptor response element in the rat using Cluster 3.0 (http://bonsai.hgc.jp/~mdehoon/software/cluster/ catalase promoter. Mol Endocrinol 16:2793–801 software.htm) and visualized using Java TreeView (http://jtreeview. Greenblatt MB, Sargent JL, Farina G et al. (2012) Interspecies comparison of sourceforge.net/). Significance of clustering was determined by human and murine scleroderma reveals IL-13 and CCL2 as disease Statistical Significance of Clustering (Liu et al., 2008). Enriched GO subset-specific targets. Am J Pathol; e-pub ahead of print 11 January 2012 biological processes were determined using The Database for Huang H, Campbell SC, Bedford DF et al. (2004) Peroxisome proliferator- Annotation, Visualization, and Integrated Discovery (Huang da activated receptor gamma ligands improve the antitumor efficacy of thrombospondin peptide ABT510. Mol Cancer Res 2:541–50 et al., 2007) (Supplementary Data File S5 online). Module maps Huang da W, Sherman BT, Tan Q et al. (2007) The DAVID Gene Functional were created with Genomica. Correlation between gene expression Classification Tool: a novel biological module-centric algorithm to and clinical parameters were calculated using MATLAB (Mathworks, functionally analyze large gene lists. Genome Biol 8:R183 Natick, MA). Statistics were carried out using R (http://www. Jimenez SA, Hitraya E, Varga J (1996) Pathogenesis of scleroderma. Collagen. r-project.org/). Distance-weighted discrimination, utilizied the java Rheum Dis Clin North Am 22:647–74 implementation (Benito et al., 2004). Gene expression data from this Kraling BM, Maul GG, Jimenez SA (1995) Mononuclear cellular infiltrates in study is available from NCBI GEO (http://www.ncbi.nlm.nih.gov/geo/; clinically involved skin from patients with systemic sclerosis of recent accession number GSE32413). onset predominantly consist of monocytes/macrophages. Pathobiology 63:48–56 Lafyatis R, Kissin E, York M et al. (2009) B cell depletion with rituximab in CONFLICT OF INTEREST patients with diffuse cutaneous systemic sclerosis. Arthritis Rheum MLW, RL, and SAP have filed patent applications for gene expression 60:578–83 biomarkers in scleroderma. MLW is a scientific founder and holds an interest in Celdara Medical, LLC, which is aiming to translate these discoveries into Lafyatis R, O’Hara C, Feghali-Bostwick CA et al. (2007) B cell infiltration in clinical use. systemic sclerosis-associated interstitial lung disease. Arthritis Rheum 56:3167–8 Leask A (2009) Signaling in fibrosis: targeting the TGF beta, endothelin-1 and ACKNOWLEDGMENTS CCN2 axis in scleroderma. Front Biosci (Elite Ed) 1:115–22 This work was supported by NIH grant U01-AR055063 to RL and MLW, and by a grant from the Scleroderma Research Foundation to MLW. MLW is a Liu T, Zhang J (2008) Detection of V, III and I type collagens of dermal tissues Hulda Irene Duggan Arthritis Investigator. SAP was supported by the NIAMS in skin lesions of patients with systemic sclerosis and its implication. Autoimmunity and Connective Tissue Training grant T32 AR007575-11 and J Huazhong Univ Sci Technol Med Sci 28:599–603 by funds from the Arthritis Foundation. Liu Y, Hayes DN, Nobel A et al. (2008) Statistical significance of clustering for high-dimension, low-sample size data. J Am Statist Assoc 103:1281–93

SUPPLEMENTARY MATERIAL Mayes MD, Lacey Jr JV, Beebe-Dimmer J et al. (2003) Prevalence, incidence, survival, and disease characteristics of systemic sclerosis in a large Supplementary material is linked to the online version of the paper at http:// US population. Arthritis Rheum 48:2246–55 www.nature.com/jid Milano A, Pendergrass SA, Sargent JL et al. (2008) Molecular subsets in the gene expression signatures of scleroderma skin. PLoS ONE 3:e2696 REFERENCES Pendergrass SA, Hayes E, Farina G et al. (2010) Limited systemic sclerosis patients with pulmonary arterial hypertension show biomarkers of Barnett AJ, Miller MH, Littlejohn GO (1988) A survival study of patients with inflammation and vascular injury. PLoS ONE 5:e12106 scleroderma diagnosed over 30 years (1953–1983): the value of a simple cutaneous classification in the early stages of the disease. J Rheumatol Ramirez F, Tanaka S, Bou-Gharios G (2006) Transcriptional regulation of the 15:276–83 human alpha2(I) collagen gene (COL1A2), an informative model system Benito M, Parker J, Du Q et al. (2004) Adjustment of systematic microarray to study fibrotic diseases. Matrix Biol 25:365–72 data biases. Bioinformatics 20:105–14 Roumm AD, Whiteside TL, Medsger Jr TA et al. (1984) Lymphocytes in the Bosello S, De Santis M, Lama G et al. (2010) B cell depletion in diffuse skin of patients with progressive systemic sclerosis. Quantification, progressive systemic sclerosis: safety, skin score modification and IL-6 subtyping, and clinical correlations. Arthritis Rheum 27:645–53 modulation in an up to thirty-six months follow-up open-label trial. Sargent JL, Milano A, Bhattacharyya S et al. (2009) A TGFbeta-responsive Arthritis Res Ther 12:R54 gene signature is associated with a subset of diffuse scleroderma with Chen Y, Shi-Wen X, van Beek J et al. (2005) Matrix contraction by dermal increased disease severity. J Invest Dermatol 130:694–705 fibroblasts requires transforming growth factor-beta/activin-linked kinase Schoonjans K, Peinado-Onsurbe J, Lefebvre AM et al. (1996) PPARalpha and 5, heparan sulfate-containing proteoglycans, and MEK/ERK: insights PPARgamma activators direct a distinct tissue-specific transcriptional into pathological scarring in chronic fibrotic disease. Am J Pathol 167: response via a PPRE in the gene. EMBO J 15:5336–48 1699–1711 Scussel-Lonzetti L, Joyal F, Raynauld JP et al. (2002) Predicting mortality in Chung L, Fiorentino DF, Benbarak MJ et al. (2009) Molecular framework for systemic sclerosis: analysis of a cohort of 309 French Canadian patients response to imatinib mesylate in systemic sclerosis. Arthritis Rheum with emphasis on features at diagnosis as predictive factors for survival. 60:584–91 Medicine (Baltimore) 81:154–67

1372 Journal of Investigative Dermatology (2012), Volume 132 SA Pendergrass et al. Stable Subsets in Systemic Sclerosis

Smith V, Van Praet JT, Vandooren B et al. (2010) Rituximab in diffuse WeiJ,GhoshAK,SargentJLet al. (2010) PPARgamma downregulation by TGFss cutaneous systemic sclerosis: an open-label clinical and histopatho- in fibroblast and impaired expression and function in systemic sclerosis: logical study. Ann Rheum Dis 69:193–7 a novel mechanism for progressive fibrogenesis. PLoS ONE 5:e13778 Varga J, Jimenez SA (1995) Modulation of collagen gene expression: its Wesson RN, Sparaco A, Smith MD (2008) Chronic pancreatitis in a patient relation to fibrosis in systemic sclerosis and other disorders. Ann Intern with malnutrition due to anorexia nervosa. JOP 9:327–31 Med 122:60–2 Whitfield ML, Finlay DR, Murray JI et al. (2003) Systemic and cell type- Varga J, Pasche B (2009) Transforming growth factor beta as a therapeutic specific gene expression patterns in scleroderma skin. Proc Natl Acad Sci target in systemic sclerosis. Nat Rev Rheumatol 5:200–6 USA 100:12319–24 Vondrichova T, de Capretz A, Parikh H et al. (2007) COX-2 and SCD, markers Whitfield ML, Sherlock G, Saldanha AJ et al. (2002) Identification of genes of inflammation and , are related to disease activity in periodically expressed in the human cell cycle and their expression in Graves’ ophthalmopathy. Thyroid 17:511–7 tumors. Mol Biol Cell 13:1977–2000

www.jidonline.org 1373