Session Information
Session Type: ACR Poster Session B
Session Time: 9:00AM-11:00AM
Background/Purpose: ANCA-associated vasculitis (AAV) has a propensity for heterogeneous organ involvement. ANCA specificity has increasingly been favored over clinical diagnosis (e.g., microscopic polyangiitis) for subgrouping AAV patients but also has its limitations. Latent class analysis (LCA) is an unbiased method to identify patient subgroups who share similar manifestations and might therefore share risk factors and/or outcomes.
Methods: The Partners AAV (PAAV) Cohort is a retrospective cohort established at Partners HealthCare, a large US hospital system. All PAAV patients are PR3- or MPO-ANCA+. We performed LCA to identify clusters according to baseline manifestations. We fitted models with 2-5 clusters and analyzed the one with superior fit statistics. Cases were assigned to a cluster based on posterior probability. We compared the distribution of baseline features between clusters using Chi square tests or analysis of variance, as appropriate. We used multivariable Cox regression to compare the rates of cardiovascular disease (CVD), deep vein thrombosis/pulmonary embolism (DVT/PE), end-stage renal disease (ESRD), and all-cause and cause-specific death between clusters. We adjusted for the competing risk of death and cause-specific death.
Results: A 3 cluster model fit best. Cluster 1 (N=143) was characterized by GN whereas cluster 2 (N=236) was characterized by upper respiratory tract disease and lung nodules more often than renal disease (Table 1). Cluster 3 (N=35) was distinguished by interstitial lung disease (ILD) which was present in <1% of clusters 1 and 2. In contrast to clusters 1 and 2, cluster 3 patients tended to be male (51%) and older (63.9 ±13.4yrs vs 58.9±18.4yrs and 54.1±17.6yrs, respectively). PR3- and MPO-ANCA status were distributed evenly in cluster 2 (PR3-ANCA=53%) whereas MPO-ANCA+ was more common in clusters 1 (71%) and 3 (91%). Compared to cluster 2, patients in cluster 1 had a higher risk of ESRD (HR 6.48, 3.39-12.36), DVT (HR 2.98, 1.65-5.39), and death (HR 1.62, 1.12-2.34). Patients in cluster 3 had a higher risk of death from respiratory causes compared to cluster 2 (HR 4.43, 1.19-16.58); a similar trend was noted when comparing clusters 3 and 1 (HR 3.89, 0.95-16.04). The risk of CVD was not different across the 3 clusters.
Conclusion: In contrast to traditional AAV phenotyping by ANCA type or clinical diagnosis (e.g., granulomatosis with polyangiitis), LCA identified three unique AAV clusters. Besides organ involvement, clusters were distinguishable by baseline features and risk of key outcomes. AAV patients with ILD represent a unique subgroup, characterized by MPO-ANCA positivity, older age, and death from respiratory causes. In contrast, patients with primarily GN (i.e., Cluster 1) are at higher risk of DVT, ESRD, and all-cause death. Future studies might evaluate the cluster-specific response to treatment.
|
Table 1: AAV Clusters: Baseline Features and Long-Term Outcomes |
Glomerulonephritis Cluster (N=143) |
Upper Respiratory Tract and Nodule Cluster (N=236) |
ILD Cluster (N=35) |
|
Proportion of Cohort (%)* |
34.6% |
56.4% |
9.0% |
|
Cluster Input Variables (%)* |
|
|
|
|
Arthralgia/Arthritis |
13.9% |
26.7% |
18.3% |
|
Purpura |
2.2% |
5.5% |
<1% |
|
Conjunctivitis/Episcleritis |
0.7% |
4.7% |
<1% |
|
Scleritis |
2.2% |
8.5% |
0.2% |
|
Nasal |
11.4% |
27.2% |
6.0% |
|
Sinus |
12.7% |
30.6% |
19.7% |
|
Subglottic |
<1% |
3.9% |
<1% |
|
Conductive Hearing Loss |
1.4% |
5.6% |
2.7% |
|
Nodules/Cavitary Leisons |
13.3% |
19.9% |
9.2% |
|
Other Infiltrate |
9.0% |
11.6% |
99.2% |
|
ILD |
<1% |
<1% |
85.6% |
|
DAH |
24.7% |
7.4% |
12.1% |
|
Hematuria |
<1% |
12.4% |
10.6% |
|
RBC Cast |
97.3% |
10.3% |
42.0% |
|
Acute Kidney Injury |
95.1% |
12.0% |
36.5% |
|
Sensory Neuropathy |
1.4% |
4.3% |
2.7% |
|
Mononeuritis Multiplex |
1.6% |
8.4% |
5.5% |
|
Probability of Cluster Membership (mean, ±SD)** |
87.2% (±12.6%) |
96.2% (±9.4%) |
88.5% (±13.9%) |
|
Case Demographics |
|||
|
Male (N, %) |
58 (41%) |
90 (38%) |
18 (51%) |
|
Age (mean, ±SD)† |
58.9 (±18.4) |
54.1 (±17.6) |
63.9 (±13.4) |
|
BVAS/WG† |
5.4 (±1.5) |
3.6 (±2.0) |
4.1 (±1.8) |
|
PR3-ANCA+† |
41 (29%) |
126 (53%) |
3 (9%) |
|
Former or Current Smoker |
65 (46%) |
117 (50%) |
21 (60%) |
|
Outcomes (Adjusted for Age, Sex, and ANCA-type) |
|||
|
Risk of Death (HR, 95% CI) |
1.62 (1.12-2.34) |
1 (ref) |
0.94 (0.50-1.77) |
|
Risk of Respiratory Death¥ |
1.14 (0.35-3.74) |
1 (ref) |
4.43 (1.19-16.58) |
|
Risk of ESRD†† |
6.48 (3.39-12.36) |
1 (ref) |
2.50 (0.80-7.83) |
|
Risk of CVD†† |
1.13 (0.69-1.88) |
1 (ref) |
0.98 (0.51-1.90) |
|
Risk of DVT†† |
2.98 (1.65-5.39) |
1 (ref) |
1.34 (0.45-4.02) |
|
*All reported proportions are probabilities conditional on latent class membership **Generally, a probability of cluster membership > 70% is considered strong †P<0.01 by Chi Square or ANOVA ¥Adjusted for competing risk of other causes of death ††Adjusted for competing risk of death |
|||
To cite this abstract in AMA style:
Wallace Z, Zhang Y, Stone JH, Choi HK. Interstitial Lung Disease in ANCA-Associated Vasculitis Defines a Unique Subgroup of Patients at High Risk for Respiratory Death: A Cluster Analysis [abstract]. Arthritis Rheumatol. 2018; 70 (suppl 9). https://acrabstracts.org/abstract/interstitial-lung-disease-in-anca-associated-vasculitis-defines-a-unique-subgroup-of-patients-at-high-risk-for-respiratory-death-a-cluster-analysis/. Accessed .« Back to 2018 ACR/ARHP Annual Meeting
ACR Meeting Abstracts - https://acrabstracts.org/abstract/interstitial-lung-disease-in-anca-associated-vasculitis-defines-a-unique-subgroup-of-patients-at-high-risk-for-respiratory-death-a-cluster-analysis/