Anti Interferon-Inducible Protein 16 Antibodies Associate With Digital Gangrene in Patients With Scleroderma

Transcription

1 ARTHRITIS & RHEUMATOLOGY Vol. 68, No. 5, May 2016, pp DOI /art VC 2016, American College of Rheumatology Anti Interferon-Inducible Protein 16 Antibodies Associate With Digital Gangrene in Patients With Scleroderma Zsuzsanna H. McMahan, Ami A. Shah, Dhananjay Vaidya, Fredrick M. Wigley, Antony Rosen, and Livia Casciola-Rosen Objective. To examine the association between anti interferon-inducible protein 16 (anti IFI-16) antibodies and clinical features of scleroderma. Methods. Sera from a discovery sample of 94 patients with scleroderma and 47 healthy controls were assayed for anti IFI-16 antibodies by enzyme-linked immunosorbent assay, and associations were examined using regression analyses. Since anti IFI-16 autoantibodies were found to be strongly associated with digital gangrene in the discovery sample, a subsequent case control study (with subjects matched 1:1 on disease duration) was designed for further exploration. Cases were patients with scleroderma and digital gangrene, while controls were patients with scleroderma and Raynaud s phenomenon alone (n 5 39 matched pairs). Nonparametric, unadjusted matched pairs analysis as well as univariate and multivariable conditional logistic regression analyses were performed. Results. In the discovery sample, anti IFI-16 antibodies were more prevalent in patients with scleroderma than in healthy controls (18% versus 2%; Supported by a Scientist Development Award from the Rheumatology Research Foundation, the Maryland Stem Cell Research Fund, and the Jerome L. Greene Foundation Scholar Award (to Dr. McMahan), NIH grants R56-AR and P30-AR (to Dr. Casciola-Rosen), NIH grant DE (to Drs. Rosen and Casciola- Rosen), NIH/National Institute of Arthritis and Musculoskeletal and Skin Diseases grant K23-AR (to Dr. Shah), the Scleroderma Research Foundation and Martha McCrory Professorship (to Dr. Wigley), and the Johns Hopkins Center for Child and Community Health Research (to Dr. Vaidya). Zsuzsanna H. McMahan, MD, Ami A. Shah, MD, Dhananjay Vaidya, MD, Fredrick M. Wigley, MD, Antony Rosen, MD, Livia Casciola-Rosen, PhD: Johns Hopkins University School of Medicine, Baltimore, Maryland. Dr. Vaidya has received consulting fees from Consumable Science, Inc. (more than $10,000). Address correspondence to Livia Casciola-Rosen, PhD, 5200 Eastern Avenue, Suite 5200, Mason F. Lord Building Center Tower, Baltimore, MD lcr@jhmi.edu. Submitted for publication May 28, 2015; accepted in revised form December 15, P ). Patients with anti IFI-16 antibodies, compared to anti IFI-16 antibody negative patients, were more likely to have limited scleroderma (77% versus 46%; P ), a longer disease duration (median 15.2 years [interquartile range ] versus 6.0 years [interquartile range ]; P < 0.01), digital gangrene (24% versus 4%; P ), and a low diffusing capacity for carbon monoxide (DLCO) (P < 0.01). In the case control study, 35 (45%) of 78 patients were anti IFI-16 antibody positive. Anti IFI-16 antibody levels were significantly higher in cases with digital gangrene than in matched controls (P ). In analyses adjusted for age, cutaneous scleroderma subtype, smoking, and DLCO, high anti IFI-16 antibody levels were associated with the presence of digital gangrene (adjusted odds ratio 2.3, 95% confidence interval , P ). The odds of having digital gangrene increased with higher anti IFI-16 antibody titers, in a dose-dependent manner. Conclusion. Anti IFI-16 antibodies are associated with digital gangrene in patients with scleroderma. Longitudinal prospective studies exploring anti IFI-16 antibodies as a disease biomarker, and biologic studies investigating the pathogenicity of these antibodies, are warranted. Scleroderma is an autoimmune disease that is often associated with the development of a systemic vasculopathy, dermal fibrosis, and visceral organ failure. Although the etiology of scleroderma is unknown, many have hypothesized that it is primarily a vascular disease mediated by autoimmunity. Evidence for this hypothesis arises from several clinical and biologic observations. First, the onset of Raynaud s phenomenon is typically one of the earliest signs of disease and may manifest years before the diagnosis of scleroderma is appreciated. Second, the presence of other vascular complications, including pulmonary hypertension, scleroderma renal crisis, 1262

2 ANTI IFI-16 ANTIBODIES AND DIGITAL GANGRENE IN SCLERODERMA 1263 cardiac disease, and striking microvascular changes in the capillaries, are prominent. Third, these clinical features suggest that vascular injury occurs early in scleroderma, preceding fibrosis, and progresses over time, ultimately resulting in both macrovascular and microvascular disease in multiple organs (1 3). Interferon-inducible protein 16 (IFI-16) is a member of the interferon-inducible p200 protein family (4). Accumulated experimental evidence has attributed a wide array of functions to the p200 family of proteins, including roles in activities such as cell differentiation, apoptosis, cell growth regulation, and autoimmunity (4). Antibodies to IFI-16 are present in 20 30% of patients with scleroderma, particularly in those with limited skin disease (5,6). Patients with limited skin disease are at greater risk, compared to patients with other scleroderma subtypes, of developing severe vascular manifestations of scleroderma, including pulmonary arterial hypertension (PAH) and severe digital ischemia (7,8). In this regard, it is interesting that the levels of IFI-16 are constitutively enriched in vascular endothelial cells (9). A prior study demonstrated a significant association between anti IFI-16 antibodies and cardiac disease, as measured by the presence of pericarditis, congestive heart failure, arrhythmias requiring treatment, conduction system abnormalities, or diastolic dysfunction. However, there is no indication that an association between anti IFI-16 antibody positivity and severity of Raynaud s phenomenon or digital loss has ever been evaluated (5,6). In this study, we examined whether patients with scleroderma who test positive for anti IFI-16 antibodies have any associated clinical features or complications of scleroderma. Findings made in our initial discovery sample showed that patients with anti IFI-16 antibodies are at greater risk of digital gangrene. These observations were confirmed in a subsequent case control study. We also showed that serum anti IFI-16 antibody levels are highest in the time period closest to the ischemic event. These findings highlight the potential utility of evaluating anti IFI-16 antibodies as a biomarker of peripheral vascular disease, manifesting as digital gangrene, in patients with scleroderma. PATIENTS AND METHODS Cohorts. The discovery sample comprised 94 patients randomly selected from the Johns Hopkins Scleroderma Center cohort database who were being evaluated as part of routine clinical care at the center and had an available serum sample. All patients in the discovery cohort either met the American College of Rheumatology (ACR) 1980 preliminary classification criteria for scleroderma (10) or had at least 3 of 5 features of the CREST (calcinosis, Raynaud s phenomenon, esophageal dysmotility, sclerodactyly, telangiectasias) syndrome in scleroderma (all meeting the ACR/European League Against Rheumatism [EULAR] 2013 classification criteria [11]). Patients were classified as having either the diffuse or limited cutaneous subtype of scleroderma, based on the extent of skin involvement (12). For a pilot longitudinal study, we subsequently selected 5 individuals with anti IFI-16 antibodies from the discovery sample for whom a minimum of 5 serum samples, obtained at various times, were banked and available for study. All serum samples were obtained during routine clinic visits at the Johns Hopkins Scleroderma Center and were stored at 2808C. As a comparator control group, 47 healthy subjects without a history of scleroderma or other autoimmune disease were selected. Since the analyses of our initial discovery sample suggested an association between anti IFI-16 autoantibodies and digital gangrene, we performed a matched case control study to enrich the analyses for the event of interest and to confirm the initial findings. All patients with scleroderma again met either the ACR 1980 classification criteria for scleroderma (10) or had at least 3 of 5 features of the CREST syndrome in scleroderma (all but 1 of these patients met the ACR/EULAR 2013 criteria [11]). All patients were evaluated between March 25, 1991 and August 26, 2014, and cases were selected on the basis of a history of digital gangrene and the availability of a blood sample that was obtained within 6 months preceding or following the ischemic event. Thirty-nine cases meeting these criteria were available in our cohort. The 39 cases were matched to 39 disease control subjects who were characterized as having scleroderma and Raynaud s phenomenon alone, without a history of any digital ischemic events (maximum score of 1 on the Medsger severity score for Raynaud s phenomenon [13]). Cases and controls were matched on the basis of disease duration, because this feature was a potential confounder in the initial cross-sectional study. For cases, disease duration was defined as the interval from the appearance of the first scleroderma symptom (Raynaud s phenomenon or non Raynaud s phenomenon) to the time of the serum sampling closest to the clinic visit for digital gangrene. Control serum samples were identified to represent comparable disease duration at the time of the case serum sampling. We sought to determine whether anti IFI-16 antibody levels were highest immediately around the time of the digital gangrene event. We therefore compared anti IFI-16 antibody levels between 2 groups of patients with digital gangrene: 1) those who were positive for anti IFI-16 antibodies and had an available blood sample obtained close to the ischemic event (within 6 months preceding or following the event) (n 5 20 of the 39 original cases), and 2) those who were positive for anti IFI-16 antibodies and had an available blood sample obtained farther from the ischemic event (6 24 months before or after the event) (n 5 10 of the 23 patients tested). Written informed consent was obtained from all subjects. The present study was approved by the Johns Hopkins Institutional Review Board. Clinical phenotyping. Demographic and clinical data, including age, sex, race, smoking status, disease duration, scleroderma subtype, specific organ involvement, and autoantibody status, were previously obtained for each patient at the time of the clinic visit. Digital gangrene was defined using the Medsger severity score for Raynaud s phenomenon (where 0 5 no Raynaud s phenomenon, 1 5 Raynaud s phenomenon with or without vasodilator required, 2 5 digital pitting scars, 3 5 digital tip ulcera-

3 1264 McMAHAN ET AL tions, and 4 5 digital gangrene) (13). This definition of digital gangrene requires demarcated ischemic territory on a digit (usually at the tip) that is dark, cool, and often intensely painful. Moreover, the gangrenous area must be clearly demarcated from viable painless flesh at the proximal border. This definition does not include nondigital vascular lesions, and also excludes palpable purpura, erythema nodosum, oral or nasal ulcers, rheumatoid nodules, genital ulcers, traumatic ulcers, and stasis ulcers due to venous insufficiency (13). The definition does include both wet and dry types of gangrene. The presence of non sclerodermarelated vascular disease (peripheral arterial disease) was evaluated by chart review of cases with gangrenous lesions on the toes. Additional measures of Raynaud s phenomenon severity were also determined using the Medsger severity score for Raynaud s phenomenon (13). Skin involvement was scored according to the modified Rodnan skin thickness score (scale 0 51) (14). Internal organ involvement was assessed using previously published criteria (13). Specific measures of pulmonary function were determined on the basis of findings on pulmonary function tests (forced vital capacity [FVC] and single-breath diffusing capacity for carbon monoxide [DLCO], each measured as the absolute value as well as the percent predicted value for sex and age). A low DLCO was defined as a value that was,70% predicted. For this study, patients were considered to have evidence of PAH when the mean pulmonary arterial pressure measured $25 mm Hg and the pulmonary capillary wedge pressure measured #15 mm Hg. Cardiac, pulmonary, gastrointestinal, or renal involvement was considered present when the relative Medsger severity score was $1 (13). Evidence of the sicca complex was determined on the basis of clinical symptoms of dry mouth and dry eyes. A positive test result for anticentromere, anti RNA polymerase III, anti-rnp, and anti topoisomerase I antibodies was ascertained from each subject s clinical findings. Enzyme-linked immunosorbent assays (ELISAs). IFI- 16 ELISA. Serum levels of anti IFI-16 antibodies were quantitated using an ELISA specific for human IFI-16. Full-length recombinant human IFI-16, expressed and purified as previously described (15), was used to coat 96-well ELISA plates (100 ng/ well, overnight at 48C). Washes were included between this and all subsequent steps, and all further incubations were carried out at room temperature for 1 hour. Wells were blocked with 5% bovine serum albumin (BSA) in phosphate buffered saline 0.05% Tween 20 (PBST), followed by incubation with human serum samples diluted 1:400 in 1% BSA PBST. A horseradish peroxidase labeled goat anti-human antibody (1:10,000 dilution; Jackson ImmunoResearch) was added to each well, and color development was subsequently performed with SureBlue peroxidase reagent (KPL) before the absorbance at 450 nm was read. A positive reference serum (1:400 dilution with an optical density [OD] value in the linear range) was included in every IFI-16 ELISA; all absorbances were calibrated relative to this reference absorbance. Sera were considered to be positive for anti IFI-16 antibodies if their log-transformed calibrated OD value was.2 SD above the mean log-transformed OD value in the 47 healthy controls. Centromere ELISA. Serum levels of anticentromere antibodies were quantitated using a commercially available ELISA kit (Inova Diagnostics), in accordance with the manufacturer s protocol. Statistical analysis. Variables were log transformed when a non-normal distribution was apparent. In the first crosssectional analysis (discovery cohort), the evaluation for associations between dichotomous variables was done using chi-square or Fisher s exact tests. For parametric data, t-tests were used to evaluate for significant differences in the mean values of continuous variables between 2 groups. In a subset of scleroderma patients with the highest anti IFI-16 antibody titers who had longitudinal serum samples available, we evaluated whether anti IFI-16 autoantibody levels over time were associated with the severity of Raynaud s phenomenon as assessed using the patient-reported Scleroderma Health Assessment Questionnaire (SHAQ) (16). Regression analyses were used to assess this association, and the correlation of repeated measurements in the same patient was taken into account. In the matched case control analysis (using a study design in which cases and controls were matched 1:1), we determined the association of anti IFI-16 antibodies with severity of scleroderma complications. For cases, we included all patients with scleroderma from our database who had a history of digital gangrene, based onascoreof 4ontheMedsgerscale for the severity of Raynaud s phenomenon (13), and for whom there was an available serum sample obtained within 6 months before or after the ischemic event. Anti IFI-16 antibody positivity and levels in the serum of cases and disease controls (patients with scleroderma and a maximum score of 1 on the Medsger scale for the severity of Raynaud s phenomenon) were compared. Controls were matched to cases based on disease duration, which was defined as the interval from the appearance of the earliest symptom, either Raynaud s phenomenon or non Raynaud s phenomenon, to the time of the serum sampling. Matching was done using Stata, version 11 (StataCorp). For every case sample, 1 control sample with the nearest disease duration was chosen. To examine whether anti IFI-16 antibody levels may peak at the time of the digital gangrene event, a second group of patients with scleroderma and digital gangrene was selected based on availability of serum samples obtained within 6 24 months of the ischemic event. The associations between 2 dichotomous variables in the matched case control study were analyzed using McNemar s test. Paired t-tests were used to evaluate the differences in mean values of parametric continuous variables between cases and controls. The nonparametric sign test was used to evaluate the difference in non-normally distributed continuous variables between 2 groups. The dependence of digital gangrene on the risk factors considered was evaluated using univariate conditional logistic regression analyses. Significant associations were tested using the regression coefficients, and the association between risk factors and outcome was expressed as the odds ratio (OR) and the corresponding 95% confidence interval (95% CI), or as a P value (considered significant at the level of P # 0.05). A multivariable conditional logistic regression model was then constructed based on the use of statistically significant covariates from the simple conditional logistic regression, as well as clinically relevant covariates that were fixed in the model. We evaluated the final model for collinearity, and no variables were removed. Using the Wilcoxon-Mann-Whitney test, we then compared antibody levels in anti IFI-16 positive patients with digital gangrene from the case control study whose blood samples were obtained within 6 months before or after the event (n 5 20 of 39) to antibody levels in anti IFI-16 positive patients whose blood samples were obtained at 6 24 months before or after the event (n 5 10 of 23). In our initial analysis, we did not detect an association between digital gangrene and anticentromere antibody status.

4 ANTI IFI-16 ANTIBODIES AND DIGITAL GANGRENE IN SCLERODERMA 1265 Because this may reflect variability in the assay methods used for clinically obtained test results, we systematically performed an anticentromere antibody ELISA (as detailed above). We then looked at the association between digital gangrene (dependent variable) and the presence of anticentromere autoantibodies using univariate and multivariate conditional logistic regression analyses. Furthermore, we examined whether the association between digital gangrene and anti IFI-16 antibody status varied according to anticentromere antibody status, by modeling the interaction between these 2 antibodies. RESULTS Association of anti IFI-16 antibodies with scleroderma disease characteristics. Sera from 94 patients with scleroderma (meeting the ACR 1980 criteria or the ACR/ EULAR criteria including CREST syndrome) and 47 healthy controls were assayed for anti IFI-16 antibodies by ELISA. Anti IFI-16 antibodies were detected in 17 (18%) of 94 patients with scleroderma compared to 1 (2%) of 47 healthy controls (P ) (Figure 1A). Table 1 summarizes the associations between anti IFI-16 antibody positivity and the demographic and disease variables studied in the scleroderma group. Of the 17 anti IFI-16 antibody positive patients, 14 (82%) were female, 11 (65%) were white, and 6 (35%) were African American. Moreover, 7 (41%) of the 17 anti IFI-16 antibody positive patients had smoked at some point in their lifetime. Compared to patients who were anti IFI-16 negative, patients with anti IFI-16 antibodies were more likely to have limited scleroderma (77% versus 46%; P ), a longer disease duration (median 15.2 years, interquartile range [IQR] versus 6.0 years, IQR ; P, 0.01), severe Raynaud s phenomenon as manifested by digital gangrene (24% versus 4%; P ), and a low DLCO (median 6 IQR % predicted versus % predicted; P, 0.01) (Figure 1B). The FVC was similar in both groups (75% predicted in antibody-positive patients versus 81% predicted in antibody-negative patients; P ), and pulmonary involvement, as measured by the Medsger severity scale, was also comparable between the 2 groups (94% in antibody-positive patients versus 80% in antibodynegative patients; P ). Patients with anti IFI-16 antibodies did not have a significantly increased risk of pulmonary hypertension, as defined by right-sided heart catheterization, although the number of patients with PAH in this group was small (Table 1). Of the anti IFI-16 positive scleroderma patients, 5 (31%) of 16 tested also had anti topoisomerase I antibodies, 3 (18%) of 17 tested were anticentromere antibody positive, and 4 (31%) of 13 tested were anti RNA polymerase III antibody positive, whereas none were positive for anti-rnp. The distribution of Figure 1. Higher levels of anti interferon-inducible protein 16 (anti IFI-16) antibodies (Ab) in scleroderma patients compared to healthy controls, and association of anti IFI-16 with a lower diffusing capacity for carbon monoxide (DLCO). A, Anti IFI-16 antibody levels (calibrated OD units) in patients with scleroderma (n 5 94) and healthy controls (n 5 47) were assayed by enzyme-linked immunosorbent assay. Symbols represent individual subjects. P between groups. B, The DLCO was compared between patients with scleroderma who were positive for anti IFI-16 antibodies (n 5 17) and those who were negative for anti IFI-16 antibodies (n 5 77). Values are presented as box plots, where the boxes represent the 25th to 75th percentiles, the lines within the boxes represent the median, and the lines outside the boxes represent the 10th and 90th percentiles. P, 0.01 between groups. Color figure can be viewed in the online issue, which is available at onlinelibrary.wiley.com/journal/doi/ /art.39558/abstract. coexisting antibodies between those with and those without anti IFI-16 antibodies was not significantly different. Because there was an association between anti IFI- 16 antibodies and severe digital ischemia, we performed a small pilot study to explore whether anti IFI-16 antibody levels changed over time and whether this might be associated with patient-reported severity of Raynaud s phenomenon as assessed by the SHAQ. Banked longitudinal serum samples were available from 5 scleroderma patients with

5 1266 McMAHAN ET AL Table 1. Demographic and disease characteristics of the 94 scleroderma patients in the discovery cohort according to the presence or absence of anti IFI-16 antibodies* Anti IFI-16 antibodies Variable Negative (n 5 77) Positive (n 5 17) P Age, mean 6 SD years Female, no. (%) 61 (79.2) 14 (82.4) 1.00 Race, no. (%) 0.13 White 64 (83.1) 11 (64.7) African American 11 (14.3) 6 (35.3) Other 2 (2.6) 0 (0) Ever smoker, no. (%) 31 (40.3) 7 (41.2) 0.94 Limited scleroderma subtype, no. (%) 35 (45.5) 13 (76.5) 0.03 Disease duration, median (IQR) years 6.0 ( ) 15.2 ( ),0.01 Modified Rodnan skin thickness score, 13.0 ( ) 7.0 ( ) 0.14 median (IQR) (range 2 45) Severe GI involvement (score $3), no. (%) 12 (15.6) 1 (5.9) 0.45 Severe Raynaud s phenomenon (score.3), no. (%) 3 (3.9) 4 (23.5) 0.02 Lung involvement (score $1), no. (%) 62 (80.5) 16 (94.1) 0.29 Cardiac involvement (score $1), no. (%) 27 (35.1) 9 (52.9) 0.17 Tendon friction rub, no. (%) 26 (33.8) 4 (23.5) 0.57 Kidney involvement, no. (%) 18 (23.4) 5 (29.4) 0.60 Sicca complex, no. (%) 54 (70.1) 15 (88.2) 0.22 Pulmonary function, median (IQR) % predicted FVC 80.7 (20.8) (n 5 76) 75.2 (21.6) (n 5 17) 0.33 DLCO (n 5 71) (n 5 16),0.01 RVSP, median (IQR) mm Hg 32.5 ( ) (n 5 58) 36 ( ) (n 5 12) 0.23 Pulmonary arterial hypertension, no./total (%) 5/19 (26.3) 5/8 (62.5) 0.10 Autoantibody positive, no./total (%) Anti topoisomerase I 20/76 (26.3) 5/16 (31.3) 0.69 Anticentromere 15 (19.5) 3 (17.7) 1.00 Anti RNP 8/74 (10.8) 0/16 (0) 0.34 Anti RNA polymerase III 21/70 (30) 4/13 (30.8) 1.00 * Associations were assessed using chi-square or Fisher s exact tests for categorical data, t-tests for parametric continuous data, and the Wilcoxon-Mann-Whitney test for nonparametric continuous data. Anti IFI-16 5 anti interferoninducible protein 16; IQR 5 interquartile range; GI 5 gastrointestinal; FVC 5 forced vital capacity; DLCO 5 diffusing capacity for carbon monoxide; RVSP 5 right ventricular systolic pressure. Duration was determined from the time of the first scleroderma symptom (Raynaud s phenomenon or non Raynaud s phenomenon) (range 0 51 years). Cutoff values for involvement were determined on the basis of the Medsger severity scale. high titers of anti IFI-16 antibodies (.5 samples per patient). In the longitudinal analysis of sera from these 5 patients, we found a significant association between anti IFI-16 antibody levels and the severity of Raynaud s phenomenon (b coefficient 0.16, 95% CI , P ). Association of anti IFI-16 antibodies with digital gangrene. Since the association between anti IFI-16 antibody positivity and digital gangrene in the initial cross-sectional analysis was observed in only a small number of patients with digital gangrene (n 5 7), we subsequently designed a matched case control study to enrich for the outcome of interest (digital gangrene) while controlling for scleroderma disease duration, in order to confirm this association. In addition, since the longitudinal analysis from the first cross-sectional study suggested that high anti IFI-16 antibody levels might be correlated with clinically worsening Raynaud s phenomenon, we considered this element in the study design. After matching, the disease duration was not significantly different between cases and controls (median 7.81 years [IQR ] versus 7.79 years [IQR ]; P ). The difference in disease duration among all matched pairs was a median of 0.09 years (IQR 24.0 to 1.10) (Table 2). All demographic characteristics, including age, sex, race, smoking status, disease subtype, and antibody status, were not different between those with and those without digital gangrene in the confirmatory study (Table 2). Presence of lung involvement, cardiac involvement, renal involvement, and sicca complex (dry eyes and dry mouth) also did not differ between cases and controls. The mortality was greater in patients with digital gangrene than in controls (46% versus 13%; P ). Although the numbers were small, there was no significant difference in the distribution of cardiac disease in cases (P ) and control subjects (P ) who were alive compared with those who had died. Among patients with digital gangrene for whom data on coexisting antibodies were available, 11 (33%) of 33 were anticentromere antibody positive, 11

6 ANTI IFI-16 ANTIBODIES AND DIGITAL GANGRENE IN SCLERODERMA 1267 Table 2. Clinical characteristics of the cases and controls in the confirmatory study* Variable Cases (n 5 39) Controls (n 5 39) P Age, mean 6 SD years Female, % White, % Ever smoker, % Limited scleroderma subtype, % Disease duration, median (IQR) years 9.88 ( ) 10.4 ( ) 0.87 From onset of Raynaud s phenomenon (range 0 66) From first scleroderma symptom (Raynaud s or 7.81 ( ) 7.79 ( ) 0.85 non Raynaud s phenomenon) (range 0 56) Difference in disease duration among pairs, Referent 0.09 (24.0 to 1.10) median (IQR) years Modified Rodnan skin thickness score, 9 (4 17) 5 (2 16) 0.18 median (IQR) (range 0 45) Severe GI involvement (score $3), % Lung involvement (score $1), % Cardiac involvement (score $1), % Kidney involvement, % Sicca complex, % Pulmonary function, median (IQR) % predicted FVC 63.6 (20.8) (n 5 34) 74.9 (19.4) (n 5 34) 0.05 DLCO (n 5 31) (n 5 31) 0.02 Autoantibody positive, no./total (%) Anti topoisomerase I 11/30 (37) 5/37 (14) 0.09 Anticentromere 11/33 (33) 12/37 (32) 1.00 Anti-RNP 4/29 (14) 4/32 (13) 1.00 Anti IFI-16 antibodies Median (IQR) titer 0.59 ( ) 0.47 ( ) 0.02 % positive * Cases were scleroderma patients with digital gangrene. Controls were scleroderma patients with Raynaud s phenomenon alone. Associations were assessed using McNemar s test for categorical data, paired t-test for parametric continuous data, and sign test for nonparametric continuous data. IQR 5 interquartile range; GI 5 gastrointestinal; FVC 5 forced vital capacity; DLCO 5 diffusing capacity for carbon monoxide; anti IFI anti interferon-inducible protein 16. Cases and controls were matched on this variable. Cutoff values for involvement were determined on the basis of the Medsger severity scale. (37%) of 30 were positive for anti topoisomerase I, 4 (14%)of29werepositiveforanti-RNPantibodies,and none had anti RNA polymerase III antibodies. Seventeen (44%) of 39 patients had documented digital ulcers preceding the gangrenous event. The DLCO was significantly lower in cases with digital gangrene compared to disease controls (52% predicted versus 67% predicted; P ). The prevalence of scleroderma patients with anti IFI-16 antibodies was significantly higher in the confirmatory case control study (45%) compared to the discovery sample (18%) (P, 0.01). This was consistent with an enrichment of patients with scleroderma having an outcome of digital gangrene in the confirmatory study (39 [50%] of 78, compared to 7 [7%] of 94 in the discovery sample). A comparison of anti IFI-16 antibody levels in cases and controls using the nonparametric sign test confirmed the findings of significantly higher anti IFI-16 antibody levels in cases compared to matched controls (P ) (Figure 2). To further evaluate the association of digital gangrene and other relevant scleroderma characteristics that were normally distributed, we used univariate conditional logistic regression. Every 10% decrease in the DLCO value was associated with a 32% higher odds of having digital gangrene (OR 1.32, 95% CI , P ). There were no significant associations between any other clinical or demographic variables and digital gangrene. We subsequently constructed a multivariable model to further examine the relationship between digital gangrene and anti IFI-16 antibody levels, with adjustments made for the DLCO and other clinically relevant covariates (age, smoking status, and limited versus diffuse cutaneous subtype). In this model, the odds of having digital gangrene were 2.3 times higher with each unit increase in anti IFI-16 antibody levels (adjusted OR 2.3, 95% CI , P ) (detailed results available upon request from the corresponding author). To further examine the relationship between anti IFI-16 antibody levels and digital gangrene, we studied 3 different cutoff levels for defining anti IFI-16 positivity: 2 SD, 3 SD, and 4 SD above the mean OD value in normal controls (Table 3). When a cutoff of 2 SD was used to define anti IFI-16 antibody positivity, the adjusted OR for

7 1268 McMAHAN ET AL Figure 2. Anti interferon-inducible protein 16 (anti IFI-16) antibody levels in cases compared to controls in the confirmatory study. Anti IFI-16 antibody levels were determined in serum samples from patients with scleroderma and digital gangrene (cases; n 5 39) and disease controls, comprising patients with scleroderma and Raynaud s phenomenon alone, with no history of digital ischemic events (maximum score of 1 on the Medsger severity scale for Raynaud s phenomenon [RP1]) (n 5 39). Values are presented as box plots, where the boxes represent the 25th to 75th percentiles, the lines within the boxes represent the median, and the lines outside the boxes represent the 10th and 90th percentiles. Symbols represent outliers. P between groups. Color figure can be viewed in the online issue, which is available at onlinelibrary.wiley.com/journal/doi/ /art.39558/abstract. nearest to the ischemic event, we selected additional patients with scleroderma and associated digital gangrene who had blood samples obtained at 6 24 months from the time of the ischemic event (n 5 23). Anti IFI-16 antibody levels in anti IFI-16 positive patients from this group (n 5 10 of 23) were compared to those in anti IFI-16 positive patients with digital gangrene from the case control study who had blood samples obtained within 6 months of the event (n 5 20 of 39). We found that anti IFI-16 antibody levels were significantly higher closer to the ischemic event (z 522.2, P ) (Figure 3). Findings of sensitivity analyses. Because it is well established in the literature that anticentromere antibodies are associated with digital loss in scleroderma (17,18), we were surprised to find that there was no association between the presence of anticentromere autoantibodies and digital gangrene in our data sets. Because the anticentromere antibody status in our database was obtained from patients clinical records, we attributed this discrepancy to variations in testing across different laboratories. We therefore reassessed anticentromere antibody status in all of the cases and controls by performing an anticentromere ELISA. When examining for anticentromere positivity, we found a significant association with the outcome of digital gangrene in the univariate model (OR 3.0, 95% CI , P ), and a significant associa- presence of digital gangrene was 4.2 (95% CI ; P ). Using a cutoff of 3 SD, the adjusted OR for presence of digital gangrene was 7.7 (95% CI ; P ). Finally, at a cutoff of 4 SD, the adjusted OR for presence of digital gangrene was 10.5 (95% CI ; P ). Thus, patients with higher levels of anti IFI-16 antibodies are more likely to have digital gangrene. Highest anti IFI-16 antibody levels evident within 6 months of the ischemic event. To determine whether anti IFI-16 antibody levels are highest in the time period Table 3. Higher anti IFI-16 antibody level cutoffs associated with higher odds of digital gangrene* Anti IFI-16 positive status cutoff Adjusted OR 95% CI P 2 SD SD SD * Anti interferon-inducible protein 16 (anti IFI-16) antibody positive status in patients with scleroderma was stratified according to cutoff levels of positivity, shown as 2, 3, or 4 SDs above the mean optical density values in normal controls (matched with cases on disease duration). The odds ratios (ORs) with 95% confidence intervals (95% CIs) were determined in multivariable models that were adjusted for age, cutaneous scleroderma subtype, diffusing capacity for carbon monoxide, and smoking status. Figure 3. Higher levels of anti interferon-inducible protein 16 (anti IFI-16) antibodies in serum samples obtained from patients with scleroderma within 6 months preceding or following the digital gangrene event. Anti IFI-16 antibody levels (calibrated OD units) were determined in scleroderma patients with digital gangrene who were positive for anti IFI-16 antibodies (n 5 30 of 62), stratified according to the time interval between serum sampling and clinic visit for digital gangrene (within 6 months before or after the event [n 5 20 of 39] versus within 6 24 months of the event [n 5 10 of 23]). Values are presented as box plots, where the boxes represent the 25th to 75th percentiles, the lines within the boxes represent the median, and the lines outside the boxes represent the 10th and 90th percentiles.

8 ANTI IFI-16 ANTIBODIES AND DIGITAL GANGRENE IN SCLERODERMA 1269 tion in the multivariable model (OR 4.8, 95% CI , P ). However, the association between digital gangrene and anti IFI-16 antibody status did not vary by anticentromere antibody status (data not shown). It was interesting, however, that the prevalence of patients positive for both anti IFI-16 antibodies and anticentromere antibodies was higher among cases with digital gangrene (n 5 20 of 62) than among disease controls without digital gangrene (n 5 4of62)(P, 0.01). We then performed a second sensitivity analysis to evaluate whether the association between anti IFI-16 antibodies and digital gangrene was influenced by the presence of coexisting risk factors in patients with macrovascular disease. In this analysis, we excluded 5 case control pairs with coexisting macrovascular disease who had gangrene in the toes. We found that excluding this patient subset had no effect on the statistically significant association between anti IFI-16 antibody levels and digital gangrene (P ). DISCUSSION This study evaluated the association of anti IFI-16 antibodies with clinical features of scleroderma. Our initial exploratory study assessed anti IFI-16 antibodies in 94 patients with scleroderma and revealed an association between anti IFI-16 antibodies and digital gangrene. In a small longitudinal analysis performed on sera from 5 anti IFI-16 positive patients identified in the exploratory group, we found that anti IFI-16 antibody levels correlated temporally with symptoms of worsening Raynaud s phenomenon. We therefore designed a confirmatory study using a case control design that included 78 patients, 39 of whom had digital gangrene and had a banked serum sample available that was obtained within 6 months of the ischemic event. In this case group, we confirmed the association between anti IFI-16 antibodies and digital gangrene, after adjusting for confounders. We also determined that anti IFI-16 antibody positive patients with scleroderma have a significantly lower DLCO (a potential measure of pulmonary vascular disease), but no significant difference in the FVC, compared to those without this specificity (Table 1). In aggregate, these findings suggest that anti IFI-16 antibodies may be markers of systemic vascular disease in scleroderma. The association of anti IFI-16 antibody levels with digital gangrene, combined with the observation that those levels are highest in patient samples obtained near the time of the ischemic event, suggests that the anti IFI- 16 immune response and digital ischemia might be related mechanistically, either proximally or distally. For example, IFI-16 is known to be prominently expressed in vascular endothelial cells (9,19), and is up-regulated by oxidative stress (a likely feature of severe vasospasm in Raynaud s phenomenon [20]). It is therefore possible that during vascular injury, IFI-16 becomes available to drive the immune response, and higher antibody levels in the serum reflect this injury. Alternatively, anti IFI-16 antibodies or other IFI-16 directed immune effector pathways (e.g., cytotoxic T lymphocytes) may themselves participate in injuring endothelial cells. A growing body of data suggests that some autoantibodies can penetrate the cell membrane of live cells, translocate into the subcellular compartments containing the associated autoantigen, and induce cellular dysfunction or death (21,22). It will be of interest to determine whether anti IFI-16 autoantibodies have this capacity, as well as whether IFI-16 reactive CD8 cells have cytotoxic effects against endothelial cells. The results of our study suggest that anti IFI-16 antibody levels may be more clinically relevant than the anti IFI-16 antibody positive versus negative status. Although the percentage of patients who tested positive for anti IFI-16 antibodies was higher among those with severe digital ischemia compared to controls, the significant difference between groups was observed only when autoantibody levels were compared. This difference was likely driven by the data for the patientswithparticularly high antibody titers in the group with digital gangrene, as represented by the higher IQR of antibody levels in cases with digital gangrene relative to controls (see Table 2). Indeed, patients with the highest antibody levels, which were 3 and 4 SDs above the mean values in normal controls, appear to have the highest odds of digital gangrene, even after adjusting for other clinically relevant covariates (e.g., age, limited versus diffuse disease subtype, smoking, and DLCO) (Table 3). It will be interesting to determine in future longitudinal studies whether absolute autoantibody levels or the change in antibody levels over time have utility in terms of predicting incipient digital ischemia. Importantly, only a subset of patients with scleroderma will develop severe digital ischemia in the form of digital gangrene during the course of their disease. A survey of the Johns Hopkins Scleroderma Center database showed that 337 (;10%) of 3,224 patients with scleroderma had a history of digital gangrene. Others have reported digital loss in patients with scleroderma at a prevalence of 16 20% (18,23). Digital gangrene in scleroderma has significant consequences in patients, including pain, tissue loss, risk of hospitalization, and long-term disability (3,24). It is widely accepted that a subset of patients with autoantibodies targeting centromere are at particularly high risk of digital loss (18). Our data, and the data from prior studies, demonstrate that a significant proportion of patients with anticentromere antibodies have coexisting anti IFI-16

9 1270 McMAHAN ET AL autoantibodies (31% in the discovery cohort) (6). Although we did not observe any change in the association between digital gangrene and anti IFI-16 autoantibodies in relation to anticentromere antibody status, the high prevalence of patients positive for both antibodies among the cases suggests that further exploration of this association may be warranted. Studies to date have not demonstrated an association between anticentromere autoantibody titers and measures of disease activity, greatly limiting their use in real-time to predict incipient digital vascular events(25).however, our data do suggest that anti IFI-16 antibodies may have a role as a novel longitudinal biomarker if the levels do indeed change over time and are correlated with disease activity. Additional studies exploring the clinical utility of assessing longitudinal anti IFI-16 antibody titers within individuals and across populations will be important. Anti IFI-16 autoantibodies have been previously described in scleroderma. Other investigators (5,6) have reported a 20 30% prevalence of this specificity among patients, and our findings in the discovery sample are consistent with this. Associations between anti IFI-16 antibodies and Raynaud s phenomenon associated complications of scleroderma have not been previously reported (5,6), perhaps because digital gangrene is a rare event, and a cohort of all-comer scleroderma patients might be underpowered to detect such an association. Additionally, our data suggest that anti IFI-16 antibody levels may change over time, peaking around the time of the digital gangrene event. By seeking to determine whether the relationship between anti IFI-16 antibodies and digital gangrene was more striking at the time nearest to the event, we were able to detect this association, whereas it would otherwise have been difficult. These findings are limited by the retrospective study design, which inevitably led to some missing data, including data related to other coexisting antibodies. As noted above, the cross-sectional nature of this study limits our ability to definitively determine longitudinal anti IFI-16 antibody trends in an individual or across the population. Confirming these findings in a longitudinal prospective study will be an important next step in understanding the utility of anti IFI-16 antibodies as a marker of systemic vascular disease in scleroderma. ACKNOWLEDGMENTS We thank Adrianne Woods and Margaret Sampedro for their involvement in generation of the data set and sample acquisition. AUTHOR CONTRIBUTIONS All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Casciola-Rosen had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study conception and design. McMahan, Shah, Vaidya, Rosen, Casciola-Rosen. Acquisition of data. McMahan, Wigley, Casciola-Rosen. Analysis and interpretation of data. McMahan, Shah, Vaidya, Wigley, Rosen, Casciola-Rosen. REFERENCES 1. Wigley FM. Clinical practice: Raynaud s phenomenon. N Engl J Med 2002;347: Gabrielli A, Avvedimento EV, Krieg T. Scleroderma. N Engl J Med 2009;360: Steen V, Denton CP, Pope JE, Matucci-Cerinic M. Digital ulcers: overt vascular disease in systemic sclerosis. Rheumatology (Oxford) 2009;48 Suppl 3:iii Veeranki S, Choubey D. Interferon-inducible p200-family protein IFI-16, an innate immune sensor for cytosolic and nuclear double-stranded DNA: regulation of subcellular localization. Mol Immunol 2012;49: Mondini M, Vidali M, De Andrea M, Azzimonti B, Airo P, D Ambrosio R, et al. A novel autoantigen to differentiate limited cutaneous systemic sclerosis from diffuse cutaneous systemic sclerosis: the interferon-inducible gene IFI-16. Arthritis Rheum 2006;54: Costa S, Mondini M, Caneparo V, Afeltra A, Airo P, Bellisai F, et al. Detection of anti-ifi-16 antibodies by ELISA: clinical and serological associations in systemic sclerosis. Rheumatology (Oxford) 2011;50: Wigley FM, Wise RA, Miller R, Needleman BW, Spence RJ. Anticentromere antibody as a predictor of digital ischemic loss in patients with systemic sclerosis. Arthritis Rheum 1992;35: Steen V, Medsger TA Jr. Predictors of isolated pulmonary hypertension in patients with systemic sclerosis and limited cutaneous involvement. Arthritis Rheum 2003;48: Gariglio M, Azzimonti B, Pagano M, Palestro G, De Andrea M, Valente G, et al. Immunohistochemical expression analysis of the human interferon-inducible gene IFI-16, a member of the HIN200 family, not restricted to hematopoietic cells. J Interferon Cytokine Res 2002;22: Preliminary criteria for the classification of systemic sclerosis (scleroderma). Subcommittee for scleroderma criteria of the American Rheumatism Association Diagnostic and Therapeutic Criteria Committee. Arthritis Rheum 1980;23: Van den Hoogen F, Khanna D, Fransen J, Johnson SR, Baron M, Tyndall A, et al classification criteria for systemic sclerosis: an American College of Rheumatology/European League Against Rheumatism collaborative initiative. Arthritis Rheum 2013;65: LeRoy EC, Black C, Fleischmajer R, Jablonska S, Krieg T, Medsger TA Jr, et al. Scleroderma (systemic sclerosis): classification, subsets, and pathogenesis. J Rheumatol 1988;15: Medsger TA Jr, Silman AJ, Steen VD, Black CM, Akesson A, Bacon PA, et al. A disease severity scale for systemic sclerosis: development and testing. J Rheumatol 1999;26: Clements P, Lachenbruch P, Seibold J, White B, Weiner S, Martin R, et al. Inter and intraobserver variability of total skin thickness score (modified Rodnan TSS) in systemic sclerosis. J Rheumatol 1995;22: Morrone SR, Wang T, Constantoulakis LM, Hooy RM, Delannoy MJ, Sohn J. Cooperative assembly of IFI-16 filaments on dsdna provides insights into host defense strategy. Proc Natl Acad Sci U S A 2014;111:E62 71.

10 ANTI IFI-16 ANTIBODIES AND DIGITAL GANGRENE IN SCLERODERMA Steen VD, Medsger TA Jr. The value of the Health Assessment Questionnaire and special patient-generated scales to demonstrate change in systemic sclerosis patients over time. Arthritis Rheum 1997;40: Herrick AL, Heaney M, Hollis S, Jayson MI. Anticardiolipin, anticentromere and anti-scl-70 antibodies in patients with systemic sclerosis and severe digital ischaemia. Ann Rheum Dis 1994;53: Wigley FM, Wise RA, Miller R, Needleman BW, Spence RJ. Anticentromere antibody as a predictor of digital ischemic loss in patients with systemic sclerosis. Arthritis Rheum 1992;35: Wei W, Clarke CJ, Somers GR, Cresswell KS, Loveland KA, Trapani JA, et al. Expression of IFI 16 in epithelial cells and lymphoid tissues. Histochem Cell Biol 2003;119: Gugliesi F, Mondini M, Ravera R, Robotti A, de Andrea M, Gribaudo G, et al. Up-regulation of the interferon-inducible IFI- 16 gene by oxidative stress triggers p53 transcriptional activity in endothelial cells. J Leukoc Biol 2005;77: Deng SX, Hanson E, Sanz I. In vivo cell penetration and intracellular transport of anti-sm and anti-la autoantibodies. Int Immunol 2000;12: Douglas J, Gardner L, Levin M. Antibodies to an intracellular antigen penetrate neuronal cells and cause deleterious effects. J Clin Cell Immunol 2013;4: Nihtyanova SI, Brough GM, Black CM, Denton CP. Clinical burden of digital vasculopathy in limited and diffuse cutaneous systemic sclerosis. Ann Rheum Dis 2008;67: Silva I, Almeida J, Vasconcelos C. A PRISMA-driven systematic review for predictive risk factors of digital ulcers in systemic sclerosis patients. Autoimmun Rev 2015;14: Tramposch HD, Smith CD, Senecal JL, Rothfield N. A long-term longitudinal study of anticentromere antibodies. Arthritis Rheum 1984;27: DOI: /art Clinical Images: Chrysiasis Chrysiasis is a distinctive and permanent skin pigmentation induced by prolonged treatment with parenteral gold salts. It is a photoinduced phenomenon in which gold deposition in light-exposed skin results in visible pigmentation. Severity of the pigmentation is dose dependent. Changes begin to occur at an approximate threshold of 19 mg/kg elemental gold, equivalent to the amount present after a cumulative dose of 49 mg/kg gold sodium aurothiomalate (GST). Light microscopy reveals aggregates of gold in the reticular and papillary dermis, in a predominantly perivascular distribution. As chrysiasis is permanent, sunlight avoidance is mandatory; treatment with Q-switched lasers is strongly contraindicated in patients who have received gold therapy, as this will result in permanent worsening/discoloration. The patient shown here (left) was treated with GST for rheumatoid arthritis for a period of 30 years, from 1985 to 2015.This photograph, obtained in 2015, depicts grayish skin color in sun-exposed areas, with normal skin color in creases and upper eyelids. Light microscopy (right) reveals the presence of fine dark pigment granules in interstitial dermis and macrophages. Chrysiasis is becoming increasingly less frequent and will probably become obsolete as the use of gold salts decreases and development of new biologic drugs continues. However, facial appearance should be checked in patients who have received gold therapy, and evidence of chrysiasis should be recorded prior to institution of laser therapy. Alejandro Fueyo-Casado, MD Javier Pedraz-Mu~noz, MD, PhD Lucia Campos-Mu~noz, MD Hospital Clınico San Carlos de Madrid and Universidad Complutense de Madrid Dolores Velez-Velazquez, MD Labco Diagnostics Eduardo Lopez-Bran, MD, PhD Hospital Clınico San Carlos de Madrid and Universidad Complutense de Madrid Madrid, Spain