Faculty Bibliography

Background/Purpose: Patients with SLE may be at increased risk for developing a venous thromboembolism (VTE), particularly in the postpartum period. The Royal College of Obstetricians and Gynaecologists (RCOG) guideline for postpartum VTE prophylaxis is unique in its inclusion of "active" SLE as an actionable risk factor. In this guideline, a score >= 3 drives a formal recommendation for a 6-week prophylactic treatment course with enoxaparin. Although not defined, "active" SLE alone scores 3 points. The inclusion of SLE raises concerns regarding appropriate attribution and subsequent management decisions. The current study applied the RCOG model to a cohort of postpartum SLE patients to determine whether these patients a) qualify as having "active" SLE b) have other risk factors for VTE c) received the recommended prophylaxis and d) had a postpartum VTE.
Method(s): The retrospective study comprised 55 pregnancies in 49 patients fulfilling criteria for classification of SLE based on ACR, SLICC or EULAR/ACR definitions consecutively seen over the last 5 years. Disease activity at delivery was assessed by the SLEPDAI using SELENA and Hybrid SELENA definitions for scoring proteinuria. Remission was assigned by applying the DORIS (Definitions of Remission in SLE) criteria. Patients not in remission were considered to have "active" SLE, even if a low level with only one clinical domain scored. RCOG scoring was calculated for each patient prior to and after delivery.
Result(s): The median age was 32 years (IQR 29-36 years) and the median BMI was 26.6 kg/m2 (IQR 23.0-30.9 kg/m2), with 49.1% African-American, 16.4% Asian, 29.1% White, 5.5% Other and 32.7% of Hispanic ethnicity. The median SELENA and Hybrid SELENA SLEPDAI scores were 2.0 (IQR 0-6) and 2.0 (IQR 0-5) respectively. The components of the RCOG model with each of its elements scored for the cohort (Table 1). 34 pregnancies (61.8%) were in DORIS remission throughout pregnancy. 21 (38.2%) were not in DORIS remission at delivery and received 3 points on the RCOG model, since by not achieving remission their SLE could be considered at least mildly active. Of these pregnancies, only 19% were recommended for VTE prophylaxis despite RCOG score >= 3. Only 35.7% of pregnancies in DORIS remission, but with 3 points for non-SLE related VTE risk factors, were recommended for VTE prophylaxis (Table 2). Of the 20 pregnancies in remission with an RCOG score < 3 after assessing all risk factors, 15% were nevertheless recommended for VTE prophylaxis. In contrast, of the 14 inactive pregnancies with RCOG score >= 3 for non-SLE activity factors, only 35.7% were recommended for VTE prophylaxis. No patients had a postpartum VTE regardless of therapy.
Conclusion(s): These data reveal that even for SLE patients in remission at the time of delivery, points for SLE alone should not automatically be assigned on the RCOG model. However, those who are in remission may still warrant VTE prophylaxis if other non-SLE related risk factors are present. Although no patient had a postpartum VTE, prophylactic anticoagulation should be instituted only when clinically appropriate. The healthcare team should carefully consider disease activity before applying 3 points for the diagnosis of SLE

Background/Purpose: SLE is characterized by autoantibody production. The most common lupus-specific serology is the anti-dsDNA antibody (anti-DNA). Traditionally, anti-DNA is measured by an enzyme immunoassay or equivalent such as multiplex flow immunoassay (EIA), which is considered sensitive. In contrast, the crithidia luciliae immunofluorescence test (CLIFT) is considered more specific. Serial measurement of anti-DNA is often used to monitor lupus disease activity. With NYU's extensive multi-race/ethnic lupus registry, we studied the relationship between these two methods, their association with lupus nephritis (LN), and their ability to predict subsequent flares.
Method(s): Using the NYU Lupus Registry of patients who meet ACR, SLICC, or EULAR criteria, we identified patients who had one or more simultaneous anti-DNA results by multiplex EIA and CLIFT. We report on their concordance (e.g., always, never, or fluctuating), association with LN, and ability to predict flare within 90 days using the SELENA-SLEDAI Flare Index. To account for degree of positivity, we defined tertiles for EIA and CLIFT as low positive [11-50 and 1:10-1:40], mild positive [51-200 and 1:80-1:320], and high positive [> 200 and > 1:640]. Table 1. 586 total visits with paired EIA and CLIFT. H = high positive M = mild positive L = low positive defined in Methods section Table 2. Relationship between EIA, CLIFT and hypocomplementemia with lupus nephritis. Chi-square test comparing significance between 60/100 v 72/100 (EIA/CLIFT) p = 0.07 Table 3. Relationship between EIA, CLIFT, hypocomplementemia and flare within 90 days Results: 207 patients had one or more paired anti-DNA results generating 586 paired results (Table 1). Cohort demographics: 92% Female, 22% Hispanic ethnicity, 24% Black, 16% Asian, 49% White, 10% Other. Overall, 377 pairs were always concordant, and 209 were never concordant. 236 pairs demonstrated titer concordance and 350 with titer discordance. Of the 207 patients, 64 patients had only one and 143 patients had two or more paired tests. Of the 64 patients, 46 were always concordant: 18 had positive EIA/CLIFT and 28 had negative EIA/CLIFT. The remaining 18 patients were never concordant: 8 had +EIA/-CLIFT and 10 had-EIA/+CLIFT. The concordance of the 143 patients with multiple paired results: 73 always, 23 never, and 47 fluctuating. Whether by one or multiple paired tests, 41/207 patients were never concordant. 100 of the 207 patients had LN associated with +EIA in 60 and +CLIFT in 72 (Table 2). Hypocomplementemia was present in 88% of +EIA and 89% of +CLIFT patients with LN. 51 visits in 42 patients had paired anti-DNA results and a SELENA-SLEDAI Flare Index assessment within 90 days. 7 patients had mild flares with +EIA in 4 and +CLIFT in 3. 4 patients had severe flares with +EIA and +CLIFT in 3 (Table 3). Low C3 and or C4 occurred in 1 of 7 (14%) mild flares and in 4 of 4 (100%) severe flares.
Conclusion(s): Our data demonstrate that discordance of positivity between two assays for anti-DNA occurred in 41/207 (20%) patients, in 207/586 (36%) visits and in 350/586 (60%) visits magnitude of positivity nonconcordant. EIA positivity is associated with LN less often than CLIFT positivity. Flares were infrequent and associated with either EIA or CLIFT positivity, with severe flares more likely if accompanied by hypocomplementemia. We recommend the utility of more than one anti-DNA assay in routine monitoring for lupus disease activity

Background Lupus nephritis can occur as an isolated component of disease activity or be accompanied by diverse extrarenal symptoms that can adversely affect a patient's quality of life (QOL). Whether renal disease absent other activity is sufficient to decrease QOL is unknown. A lack of reported QOL impairment may place patients at risk for delayed diagnosis of nephritis or medication noncompliance yet nephritis trials have largely neglected QOL. As such, this study leveraged the multi-center multi-racial Accelerating Medicines Partnership (AMP) lupus nephritis cohort to assess QOL measured by PROMIS-29. Methods Patients (n=182) fulfilling ACR or SLICC criteria for SLE with a uPCR >= .5 and biopsy Class III, IV, V, or mixed were consecutively enrolled in AMP at the time of renal biopsy and clinical history, PROMIS-29, and disease activity as assessed by the hybrid SELENA-SLEDAI were recorded. Patients were determined to have extrarenal clinical activity if, after excluding all laboratory parameters from the SLEDAI, the score remained >= 1. Raw PROMIS-29 scores were transformed to t-scores with the mean of 50 +/- 10 representing the US population and a difference of 5 points considered clinically meaningful. PROMIS-29 physical and mental health summary scores were calculated according to published formulas. Results Forty-three percent of patients (n=78) had extrarenal clinical manifestations including vasculitis (4%), arthritis (39%), rash (45%), alopecia (42%), mucosal ulcers (13%), pleurisy (12%), pericarditis (8%), and fever (4%). Patients with isolated renal disease (n=104, 57%) did not have PROMIS-29 scores that differed clinically from the US population whereas patients with extrarenal disease reported deficits in physical functioning, fatigue, social functioning, and pain (table 1). Patients with extrarenal disease had significantly lower physical health summary scores compared to patients with isolated disease (median [IQR]: 40.31 [35.79, 47.02] p<0.001 vs. 48.6 [40.14, 57.08]) and significantly lower mental health summary scores (44.12 [38.63, 51.39], p=0.024 vs. 48.67 [40.51, 55.07]). Female and African American patients and those with nephrotic range proteinuria or undergoing first biopsy had significantly lower physical health summary scores, but mental health summary scores did not differ by these variables. Patients on greater than 20 mg of prednisone had both significantly lower physical and mental health summary scores compared to those on lower doses. PROMIS-29 scores did not differ by low complements, anti-dsDNA, or anti-Ro antibodies. Stepwise multivariable linear regression analysis demonstrated that the association between extrarenal disease and lower PROMIS-29 summary scores was primarily driven by arthritis and independent of potential confounders (tables 2 and 3). Conclusion The majority of patients had isolated renal disease and report a QOL similar to that of the general population. In contrast, those with extrarenal manifestations report significantly worse QOL outcomes. These results reinforce the critical importance of routine laboratory surveillance and medication compliance for nephritis even in patients with seemingly quiescent clinical disease since lupus nephritis is often asymptomatic

Background/Purpose: A decline of urine protein-to-creatinine ratio (UPCR) to < 0.5 is associated with better long-term preservation of kidney function in lupus nephritis (LN). UPCR < 0.5 defines complete response in guidelines and clinical trials when achieved after 1 or 2 years. Biomarkers of early response are needed to guide early treatment changes. We studied longitudinal urine proteomic profiles in LN to identify early predictors of proteinuric response.
Method(s): We quantified 1200 biomarkers (Kiloplex, RayBiotech) in urine samples collected on the day of (73%) or within 3 weeks (27%) of kidney biopsy and week 12, 24, or 52 in LN patients (ISN class III, IV, V, or mixed) with proteinuria > 1 g/d. Response was defined at one year from renal biopsy: Complete = UPCR < 0.5, serum creatinine (sCr) < 125% of baseline, prednisone <= 10mg/d; Partial = UPCR < 50% from baseline but >0.5, sCr < 125% of baseline, but prednisone allowed to 15 mg/d; Non responder = not meeting previous definitions.
Result(s): A total of 127 patients were included: 48 (38%) with pure proliferative LN (class III or IV), 41 (32%) with mixed LN (III or IV +/-V), and 38% (30%) with pure membranous LN. Response was complete in 34 (27%), partial in 29 (23%), and none in 64 (50%). There were no urinary biomarkers at baseline that predicted response. We then analyzed the changes in urinary proteins at 3 months compared to baseline. Patients who responded at 1 year showed an early decline in 51 urinary proteins led by CD163, IL-16, and macrophage mannose receptor (MMR, CD206) (Figure 1A) which matched the proteomic signature associated with histological activity (Figure 1B). No changes were observed in nonresponders. The decline of several urinary biomarkers at 3 months outperformed a decline in UPCR (clinical standard) in predicting the 1 year response (-Figure 2A). In particular, a decline of CD163 predicted 1 year response in ROC analysis with an area under the curve (AUC) of 83% compared to an AUC of 75% for UPCR decline (Figure 2B). In proliferative LN, urinary biomarkers displayed superior performance with an AUC of 91%, 86%, and 78% for the decline of MMR, CD163, and UPCR, respectively (-Figure 2C-D). Pathway enrichment analysis identified leukocyte activation, neutrophil degranulation, and matrix degradation as the main pathways reduced at 3 months in responders.
Conclusion(s): An early decline in urinary biomarkers of histological activity is associated with proteinuric response at 1 year. These findings indicate that effective immunosuppression induces by three months an immunological response in the kidney that can be noninvasively monitored in the urine. Biomarkers of immunological response outperformed early decline of UPCR, the standard of care, in predicting 1-year proteinuric response, especially in proliferative LN. Because biomarkers of immunological response parallel intrarenal activity, they could detect early treatment response/failure and allow early treatment changes. They could serve as surrogate endpoints in clinical trials. Longitudinal studies are needed to confirm that this immunological response is a better predictor of long-term kidney function preservation than proteinuric responses. Figure 1. Proteins linked to intrarenal activity (NIH activity index) decrease in treatment responders. Volcano plot of the changes of the urinary proteomic profile of treatment responders at 3 months (A) after kidney biopsy/treatment compared to baseline at time of biopsy. (B) Volcano plot displaying the Spearman's correlation coefficients of urinary biomarkers at time of biopsy with histological activity defined as the NIH activity index. Blue arrows indicate 4 examples of shared urinary biomarkers that indicate both intrarenal activity and response to treatment. FDR, false discovery rate; q, adjusted p value

Background Mitochondria are intracellular organelles derived from the endosymbiosis between an a-proteobacterium and a primitive eukaryotic cell. Mitochondria thus display proinflammatory and antigenic properties, when released into the extracellular milieu. Several cross-sectional studies reported increased levels of anti-mitochondrial antibodies (AMAs) in patients with systemic lupus erythematosus (SLE) and the antiphospholipid syndrome (APS). These autoantibodies also displayed correlations with the SLE disease activity index 2000 (SLEDAI-2K) and associations with various clinical manifestations (e.g. lupus nephritis, thromboses, carotid plaque). In the present study, we aim to detect AMAs against either whole organelles (AwMA), mitochondrial DNA (mtDNA) or RNA (mtRNA) through time in samples from patients included in the SLICC cohort. Methods Clinically relevant variables (e.g., sociodemographic variables, disease-specific outcomes including death and arterial vascular events (AVE)) were documented and biosamples were harvested upon patient enrolment in the SLICC cohort, as well as at each follow-up visit. AMA levels were measured by in-house direct ELISAs whereas SLE autoantibodies were detected by clinical laboratories. Healthy individuals, defined as having no known illnesses and infectious symptoms at the time of the blood draw, were recruited. 90% confidence intervals were calculated for both limits of the 95% nonparametric two-sided reference intervals for values measured in healthy donors. AMA values were segregated into 3 categories: Normal values were determined as within the inner limits of the range while values outside this range were characterized as abnormal, either lower or higher than the reference interval. (figure 1). Marginal Cox models with AMAs in 3 categories were adjusted for covariables and are presented as [hazard ratio (95% CI)]. Interactions with sex were tested in models with the AMAs as continuous variables. Results Sera from healthy individuals (n=126) or SLE patients included in the SLICC cohort, from their inclusion, up to 7 years of follow-up (n=1114 patients at baseline, 3577 samples in total). AwMA displayed lower correlations with antibodies to mitochondrial nucleic acids (versus AmtDNA: rs=0.37, and vs AmtRNA: rs=0.38), while antibodies to mitochondrial DNA or RNA shown higher correlations (rs=0.59). During our preliminary analyses on the distribution of the variables, We made intriguing observations regarding patients with AMA levels that were either lower or higher than those of healthy individuals. This information led us to categorize SLE patients as described in the methods and in figure 1. For each of the three antibodies assessed, SLE patients displayed more abnormal AMA levels at baseline than controls. The percentage of patients with higher levels of AwMA and AmtRNA increased at subsequent follow-up visits, while a slight decrease was observed for AmtDNA (figure 2). SLE patients with higher levels of AwMA showed higher risks of death [2.12 (1.18-3.83)]. It is of interest that an inverse relationship was found between AmtRNA and AVEs, with a small subset of patients with low levels of AmtRNA (n = 4), this autoantibody was associated with increased risks of this manifestation [4.46 (1.71-11.66)]. Additionally, patients with higher levels of AmtDNA and AmtRNA displayed increased risks of lupus nephritis [respectively: 3.05(2.05-4.54), and 1.56(1.12-2.18)]. Interestingly, there was an interaction with sex for AmtRNA levels effect on AVEs [males: 0.32 (0.11-0.99). Females: 1.56 (1.11-2.19)], and AmtDNA association with nephritis was only significant in female patients [4.00 (2.51-6.36)] (table 1). Conclusion These results show that AMAs display different associations with disease manifestations in various clusters of patients. These results prompt for further analyses by machine-learning in order to delineate clusters of clinical interests by adding AMAs in the routine serological assessment of SLE autoantibodies. Acknowledgements We acknowledge the contribution of the study participants, individual center support staff as well as investigators of the Systemic Lupus International Collaborating Clinics (SLICC) Inception Cohort project who for the past 20 years have made this study possible. LAY ABSTRACT The mitochondrion is a part of the cell that controls various biological mechanisms (e.g., energy supply, whether the cell should live or die, control, or produce various cellular components). They are derived, through evolution, from a microbe. Mitochondria may sometimes be jettisoned out of their host cell and subsequently elicit immune reactions - including the production of antibodies. Previous studies indicated that patients with autoimmune conditions such as systemic lupus erythematosus (SLE) and the antiphospholipid syndrome (APS) have antibodies against mitochondria in their blood stream. Presence of these antibodies was associated with increased disease activity and clinical manifestations of these diseases (e.g. kidney disease, arterial vessel disease). In this study, we studied blood samples harvested by an international group dedicated to the study of SLE [i.e., the SLE International Collaborating Clinics (SLICC) cohort] and observed that patients may be clustered into groups, upon their levels of antibodies and/or sex, allowing to have a better appreciation of their risks of death, vascular events, and kidney disease. These results might lead to improved diagnosis and/or prognosis in SLE and thus, in improved care and quality of life for the people living with lupus

Background Lupus nephritis (LN) is characterized by considerable variability in its clinical manifestations and histopathological findings. Understanding the cellular and molecular mechanisms underlying this heterogeneity is key for the development of personalized treatments for LN. Methods Droplet-based single-cell RNA-sequencing was applied to the analysis of dissociated kidney samples, collected from 155 LN patients with active kidney disease and 30 living donor controls as part of the Accelerating Medicines Partnership (AMP) in SLE consortium -a large-scale, multi-center study. 73,440 immune cells passing quality control were identified, spanning 134 cell subsets, representing various populations of tissue-resident and infiltrating leukocytes, as well as the activation states these cells assume as part of their diseaserelated activation and differentiation (figure 1). Principal component analysis (PCA) was used to characterize the variability in cell subset frequencies across the LN patients. Relationships between the resulting principal components (PCs) and the demographic, clinical and histopathological features of the patients were then assessed. Results The main source of variability in immune cell subset frequencies, as represented by the first PC (PC1), reflected the balance between lymphocytes and monocytes/macrophages. Subsequent PCs represented the balance between B cells and T cells (PC2); the levels of cytotoxic T lymphocytes and NK cells, as compared to plasma cells (PC3); and the degree of macrophage differentiation to an alternatively activated phagocytic profile (PC4). PC1 was significantly correlated with the Chronicity index, such that patients with a higher percentage of lymphocytes compared to monocytes/macrophages had a higher Chronicity score (rho = -0.422, p-value < 0.001; figure 2A). A high degree of macrophage differentiation, as represented by PC4, was associated with a high Activity score (rho = 0.387, p-value < 0.001; figure 2B), and, in addition, with proliferative or mixed histology class, compared to pure membranous nephritis (p-value = 0.001, Kruskal-Wallis test). The ratio of B cells to T cells, as represented by PC2, demonstrated a positive correlation with the Activity index (rho = 0.311, p-value < 0.001). We further identified a significant correlation of PC1 with age; specifically, older patients had a higher relative frequency of lymphocytes compared to monocytes/macrophages (rho = -0.239, p-value = 0.003). Our analysis indicated that these relations are not driven by demographic, clinical and technical sources of variation in our data, including race, ethnicity, the mixture of different nephritic classes, and the inclusion of both first and later biopsies. Conclusion Our work identifies distinct leukocyte populations active in different LN patients and, possibly, different stages of disease, and points to potential therapeutic targets, that must be validated in mechanistic studies. This approach may pave the way to personalized treatment of LN

Background/Purpose: Lupus nephritis (LN) is characterized by considerable variability in its clinical manifestations and histopathological findings. Understanding the cellular and molecular mechanisms underlying this heterogeneity is key for the development of personalized treatments for LN.
Method(s): Droplet-based single-cell RNA-sequencing was applied to the analysis of dissociated kidney samples, collected from 155 LN patients with active kidney disease and 30 living donor controls as part of a large-scale, multi-center study. 73,440 immune cells passing quality control were identified, spanning 134 cell subsets, representing various populations of tissue-resident and infiltrating leukocytes, as well as the activation states these cells assume as part of their diseaserelated activation and differentiation (Fig. 1). Principal component analysis (PCA) was used to characterize the variability in cell subset frequencies across the LN patients. Relationships between the resulting principal components (PCs) and the demographic, clinical and histopathological features of the patients were then assessed. Figure 1. Single-cell RNA-sequencing was used to profile immune cells isolated from the kidneys of LN patients and healthy controls. Five main lineages of cells were identified, as shown in a Uniform Manifold Approximation and Projection (UMAP) plot: myeloid cells, T/NK cells, B cells, plasma cells and dividing cells. The cells of each lineage were further split into finer subsets of cells (color-coded).
Result(s): The first PC (PC1), explaining 33% of the total variability in cell subset frequencies, reflected the balance between lymphocytes and monocytes/macrophages. The second PC (PC2), explaining an additional 21% of the total variability, represented the degree of macrophage differentiation to an alternatively activated phagocytic profile. The third and fourth PCs, bringing the total explained variability to 74%, were related to the balance between cell-mediated and humoral immune responses. PC1 was significantly correlated with the Chronicity index, such that patients with a higher percentage of lymphocytes compared to monocytes/macrophages had a higher Chronicity score (rho =-0.439, p-value < 0.001; Fig. 2A). A high degree of macrophage differentiation, as represented by PC2, was associated with a high Activity score (rho =-0.495, p-value < 0.001; Fig. 2B), and, in addition, with proliferative or mixed histology class, compared to pure membranous nephritis (p-value = 0.001, Kruskal-Wallis test). We further identified a significant correlation of these PCs with age; specifically, older patients had a higher relative frequency of lymphocytes compared to monocytes/macrophages, a lesser degree of macrophage differentiation, and a higher representation of cells putatively involved in a humoral immune response compared to a cell-mediated one.
Conclusion(s): These results identify distinct leukocyte populations active in different LN patients and, possibly, different stages of disease, and suggest potential therapeutic targets, that must be validated in mechanistic studies. This approach may pave the way to personalized treatment of LN

BACKGROUND:Complement activation has been associated with adverse pregnancy outcomes (APO) in SLE. Pregnant women with SLE were studied to evaluate whether complement dysregulation within the first two pregnancy trimesters predicts APO. METHODS:Pregnant women fulfilled classification criteria for SLE. APO included neonatal death, preterm delivery before 36 weeks and small for gestational age newborn. Pre-eclampsia was also evaluated. Erythrocyte complement receptor 1 (ECR1) and erythrocyte-bound C4d (EC4d) were measured by flow cytometry. Complement proteins C3 and C4 were measured by immunoturbidimetry and anti-double-stranded DNA by ELISA in serum. Statistical analysis consisted of t-test, confusion matrix-derived diagnostic analysis, and multivariate logistic regression. RESULTS:Fifty-one women had 57 pregnancies and 169 visits during the study. Baseline visits occurred mainly in the first (n=32) and second trimester (n=21). Fourteen (24.6%) pregnancies resulted in 21 APO with preterm delivery being the most common (n=10). ECR1 <5.5 net mean fluorescence intensity in the first trimester predicted APO with a diagnostic OR (DOR) of 18.33 (95% CI: 2.39 to 140.4; t-test p=0.04). Other individual biomarkers did not reach statistical significance. To estimate the likelihood of APO, we developed an algorithm that included the week of pregnancy, ECR1 and EC4d. From this algorithm, a Pregnancy Adversity Index (PAI) was calculated, and a PAI >0 indicated an elevated likelihood of pregnancy complications (DOR: 20.0 (95% CI: 3.64 to 109.97)). CONCLUSIONS:Low levels of ECR1 in early or mid-pregnancy are predictive of an APO. Incorporating the weeks of gestation and both ECR1 and EC4d generated a PAI, which further predicted serious pregnancy complications.