Clinical Characteristics and Outcomes of Polyarteritis Nodosa – an International Study

Abstract

Objective:

To describe the demographics, clinical features, disease course, and survival of polyarteritis nodosa (PAN) through an international collaboration (GLOBAL-PAN).

Methods:

Patients with PAN recruited between 1990 and 2020 from observational cohorts of 9 countries across Europe, Japan, and North America. Eligibility was retrospectively defined using the European Medicines Agency (EMA) classification algorithm. Patients with PAN related to hepatitis B virus (HBV) (n=12) and two monogenic diseases mimicking PAN, deficiency of adenosine deaminase 2 enzyme (n=16) or familial Mediterranean fever (n=11), were excluded. Data regarding organ involvement, relapse, disease-related damage, and survival were analyzed.

Results:

358 patients (female/male: 174/184), including systemic PAN (sPAN, n=282) and cutaneous PAN (cPAN, n=76), were included. Twenty-five were pediatric-onset. Mean (SD) age at diagnosis was 44.3 (18.1) years. Constitutional symptoms (71.5%), cutaneous involvement (70.5%), musculoskeletal findings (69.1%), and neurologic features (48.0%) were common manifestations. Among patients with sPAN, gastrointestinal involvement, and proteinuria over 400 mg/day were reported in 52.2% and 11.2%, respectively. During a median (inter-quartile range) 59.6 (99.5) months of follow-up, relapse occurred in 48.5% of patients. One, 5- and 10-year survival rates for sPAN were 97.1%, 94.0%, and 89.0%, respectively. Predictors of mortality for sPAN included age ≥ 65 years at diagnosis, serum creatinine at diagnosis >140 μmol/L, gastrointestinal manifestations, and central nervous system (CNS) involvement.

Conclusion:

The spectrum of PAN remains a complex, multi-faceted disease. Relapse is common. Age ≥ 65 years and serum creatinine >140 μmol/L at diagnosis, gastrointestinal and CNS involvement are independent predictors of mortality in sPAN.

INTRODUCTION

Polyarteritis nodosa (PAN) is a necrotizing vasculitis predominantly affecting medium-sized arteries that presents in pediatric and adult populations (1). The estimated prevalence in Europe is 31 cases per million, with an annual incidence of 0.9–8.0 per million (2, 3). The 1990 American College of Rheumatology (ACR) criteria for the classification of PAN did not distinguish classical PAN from microscopic polyangiitis (MPA) and preceded the introduction of antineutrophil cytoplasmic antibodies (ANCA) for the classification of vasculitis (4). Bin 1993 the Chapel Hill Consensus Conference (CHCC) Nomenclature of Vasculitis, differentiated MPA from PAN and application of this approach to population data led to a reduction in the frequency of PAN diagnoses (1, 5).

Understanding of PAN has evolved in recent decades (6). Although cutaneous PAN (cPAN) has been defined as a form of PAN limited to skin with no systemic involvement, the revised CHCC-2012 classified cPAN as a single-organ vasculitis (1). Additionally, hepatitis B virus (HBV)-related PAN was classified as vasculitis associated with probable etiology (1). The relationship of PAN with familial Mediterranean fever (FMF) has been known for over 60 years and PAN is generally accepted as a clinical comorbidity of FMF (7). In the last decade, deficiency of adenosine deaminase 2 (DADA2), was recognized with clinical, histopathologic, and radiologic features indistinguishable from idiopathic PAN (6, 8, 9). Among 25 patients with newly defined, monogenic disease of adulthood caused by somatic mutations (VEXAS syndrome: Vacuoles, E1 enzyme, X-linked, Autoinflammatory, Somatic), three had a diagnosis of PAN (10).

The low prevalence of PAN necessitates a global effort to clarify the characteristics and outcome of this disease spectrum. This report describes the analysis of GLOBAL-PAN, a collaboration of the European Vasculitis Society (EUVAS), the Vasculitis Clinical Research Consortium (VCRC), and other networks in many countries, that agreed to contribute their cohort data to this study.

This initial project of GLOBAL-PAN aimed to i) identify the demographic and clinical features of idiopathic PAN; ii) assess similarities and differences between systemic and cutaneous PAN; iii), investigate whether pediatric-onset PAN is similar to adult-onset PAN; and iv) analyze the outcomes and relapse risk of PAN.

METHODS

Patients

Retrospective chart review [Italy, Japan, Mexico, Slovenia, Sweden, the United Kingdom (UK), the Turkey] and prospective, but not as an inception cohort [Canada, the United States of America )] from study centers of the GLOBAL-PAN consortium in 9 countries were included in this study. Patients diagnosed with PAN between 1990 and 2020 and fulfilling the European Medicines Agency (EMA) algorithm (this algorithm differentiates ANCA-associated vasculitis from PAN) were included in this study (11). An objective and characteristic radiologic or histopathologic finding of PAN is essential in the EMA algorithm. To analyze idiopathic PAN, patients with vasculitis with probable associated etiologies (e.g. hepatitis B virus; HBV) and monogenic diseases resembling PAN (FMF and DADA2) were excluded (Figure 1).

Figure 1:

Flow-chart of the patient selection and classification

All enrolled patients were categorized as:

Systemic PAN (sPAN): when there was evidence for extra-cutaneous organ or systemic involvement including neurologic findings [neuropathy, mononeuritis multiplex and central nervous system (CNS); cranial angiographic findings, transient ischemic attack], gastrointestinal (GI), kidney, or testicular involvement.

Cutaneous PAN (cPAN): when there was only skin involvement with or without musculoskeletal or constitutional symptoms.

Adult-onset PAN: patients aged 18 years or older at the date of diagnosis of PAN.

Pediatric-onset PAN (12): patients aged < 18 years at the date of diagnosis of PAN.

Demographic and clinical phenotypic differences were assessed between patient groups diagnosed before or after 2000.

Data collection

The following data elements were recorded in all patients: demographic and clinical characteristics of PAN at baseline and during follow-up observation, including constitutional, cutaneous, musculoskeletal, neurologic, gastrointestinal, kidney, cardiac, and testicular manifestations.

Organ involvement was assessed using the definitions for the modified disease extent index (DEI) (13). The level of proteinuria was determined according to the cut-off (>400 mg/day) proposed by the French Vasculitis Study Group (FVSG) (14). Histopathology was rated as diagnostic by investigators at each center. Findings on vascular imaging (conventional angiography, computed tomography angiography (CTA), or magnetic resonance imaging angiography (MRA)), such as, arterial microaneurysm and occlusions/infarcts attributed to PAN, were recorded (11).

Disease activity at diagnosis was measured using the Birmingham Vasculitis Activity Score (BVAS) (15). Relapse was defined as occurrence of new or worsening manifestations of vasculitis requiring treatment change. Clinical definitions, assessment parameters, and outcome measures collected are outlined in the Supplementary Material. Organ damage was estimated using the Vasculitis Damage Index (VDI) at last available visit (16). The Five-Factor Score (FFS) was calculated according to findings at diagnosis (17). The cause of death was recorded as PAN-related or not, according to the treating physician’s assessment

Statistical analysis

For continuous and normally distributed variables, data are presented as means and standard deviation (±SD), and Student’s t-test was used for comparison between groups. For continuous and non-normally distributed variables, data were presented as median and interquartile ranges (IQR), with the Mann-Whitney U-test used for comparison. Prevalences and counts of clinical characteristics were given as calculated based on the total number of patients for whom those data were available. The differences in the frequency of categorical variables among groups were quantified using the χ2 or Fisher exact test. Patients’ survival and relapse-free survival were assessed by Kaplan-Meier curves and log-rank test was used for comparison between groups. To determine independent predictors of mortality, possible predictors at diagnosis were identified with univariate analyses (p < 0.20, Kaplan-Meier analysis and log-rank test; Supplementary Figure 1) and were further entered into the Cox proportional hazards models with backward LR model selection (Supplementary Table 1). Results were expressed as hazard ratios (HRs) and 95% confidence intervals. P-values of <0.05 were considered as significant in all statistical tests. The SPSS 24.0 and GraphPad Prism 9.0 software packages were used for the analysis.

Ethics

Within the VCRC institutional review board/ethics board approval was received at all study sites and all patients signed informed consent forms. For the Mexican cohort, the hospital Institutional Review Board approved the study (Reference 2662). The study was approved by the Ethical Board in Sweden (2010_517 and 2020–00697). For the United Kingdom (UK) cohort, in accordance with individual National Health Service Research Ethics Committee guidelines, ethical approval was not required for this work because it is composed of retrospective data. The medical ethics committee of Hacettepe University approved the study for the rest of the cohort (23.02.2021, GO 21/213).

RESULTS

Demographic and clinical characteristics

Of 460 patients originally identified as having PAN, 63 were excluded due to missing data (n=4) or not fulfilling the EMA algorithm (due to unavailable biopsy and angiography, n=48; normal biopsy unavailable angiography, n=6; unavailable biopsy but negative angiography, n=5). Twelve patients with HBV-PAN, 16 with DADA2, and 11 with FMF were excluded even though they fulfilled the EMA algorithm (Figure 1). 358 patients were included in the final analysis (USA & Canada: 112, Turkey: 73, Italy: 50, UK: 39, Japan: 34, Mexico: 29, Slovenia: 11, and Sweden: 10). Baseline demographic and clinical features of the patients are shown in Table 1. 184 (51.4%) were male and mean (SD) age at diagnosis was 44.3 (18.1) years. The most common manifestations were constitutional symptoms (n=253, 71.5%), cutaneous (n=251, 70.5%), musculoskeletal (n=244, 69.1%), and neurologic (n=170, 48.0%) involvement (Table 1).

Table 1.

The Demographic and Clinical Characteristics of Patients with Systemic and Cutaneous Polyarteritis Nodosa

	All Patients (n=358)	Systemic Polyarteritis Nodosa (sPAN; n=282)	Cutaneous Polyarteritis Nodosa (cPAN; n=76)	P-Value (sPAN vs cPAN)
Age at onset, years	44.3 (18.1)	45.1 (17.6)	41.5 (19.4)	0.074
Sex, male	184 (51.4)	157 (55.7)	27 (35.5)	0.002
Pediatric-onset disease	25 (7)	20 (7.1)	5 (6.6)	1
	253 (71.5)	219 (78.8)	34 (44.7)
Constitutional symptoms				<0.001
Fever	146 (42.6)	122 (45.2)	24 (32.9)	0.063
Weight loss	151 (44.2)	136 (50.6)	15 (20.5)	<0.001
Cutaneous Manifestations	251 (70.5)	175 (62.5)	76 (100)	<0.001
Nodule	90 (25.5)	53 (19.1)	37 (48.7)	<0.001
Livedo reticularis	94 (26.4)	70 (25)	24 (31.6)	0.245
Necrosis/Ulceration	57 (16)	39 (13.9)	18 (23.7)	0.051
Panniculitis	32 (13.2)	19 (9.7)	13 (27.1)	0.003
Raynaud phenomenon	24 (10.8)	21 (12.9)	3 (5.1)	0.141
Urticarial rash	32 (29.1)	22 (27.8)	10 (32.3)	0.648
Limb edema	31 (28.2)	23 (29.1)	8 (25.80	0.817
	244 (69.1)	198 (71)
Musculoskeletal findings			46 (62.2)	0.158
Myalgias	162 (48.1)	138 (51.7)	24 (34.3)	0.011
Arthralgias	171 (50.6)	142 (53.2)	29 (40.8)	0.082
	76 (22.3)	59 (22)
Arthritis			17 (23.3)	0.874
Leg tenderness			13 (36.1)	0.695
	58 (33.3)	45 (32.6)
Neurologic manifestations		170 (61.2)
Peripheral neuropathy		147 (52.7)	*	N/A*
Central Nervous System Involvement		47 (16.8)
Gastrointestinal manifestations		145 (52.2)
Abdominal pain		60 (21.8)	*	N/A*
Cholecystitis		3 (1.1)
Pancreatitis		5 (1.8)
Bleeding		33 (11.7)
Perforation		1 (0.4)
Renal involvement		133 (48.2)
Disease related Hypertension		69 (36.3)
			*	N/A*
Proteinuria >400 mg/day		29 (11.2)
S. creatinine >140 μmol/L		24 (10.5)
Cardiac involvement		24 (8.5)	*	N/A*
Pleural effusion		19 (9.7)	*	N/A*
Testicular pain/tenderness (for males)		35 (12.6)	*	N/A*
Positive angiography findings		149/168 (88.6)
Gastrointestinal
			*	N/A*
Kidney		78 (46.1)
Others		22 (13.1)
BVAS at diagnosis		10.2 (9.1)	*	N/A*
Distribution according to FFS
FFS=0	115 (51.6)	74 (41.3)	43(97.7)
FFS=1	71 (31.8)	68 (38)	1 (2.3)	<0.001
FFS≥2	37 (16.5)	37 (20.7)	0
Disease Extent Index	5.1 (2.8)	5.7 (2.8)	3 (1.4)	<0.001
VDI, median (IQR)	1 (0–3)	2 (1–3)	0 (0–1)	<0.001

^*Parameters valid only for systemic polyarteritis nodosa

Percentages are given as valid percent.

Variables are labeled as mean (SD) or n(%) unless otherwise stated.

Abbreviations: BVAS: Birmingham Vasculitis Activity Score; FFS: Five Factor Score; IQR: Interquartile Range; N/A:Not applicable; PAN: polyarteritis nodosa; SD: standard deviation; VDI; Vasculitis Damage Index

Histopathologic evaluation was documented for 266 (74.3%) patients. Biopsy sites were skin (n=115), kidney (n=14), gastrointestinal (n=10), muscle (n=6), nerve (n=6), vessel (n=2), testis (n=2), and not reported (n=111). The biopsy was reported as diagnostic in 90.5% of these cases. Among 179 with vascular imaging (CTA, MRA, or catheter-based angiography), 149 (83.2%) had diagnostic findings for PAN (Table 1).

Two-hundred and eighty-two (78.8%) patients had sPAN and 76 (21.2%) had cPAN (Table1). Mean age at diagnosis was 45.1 (17.6) years for sPAN and 41.5 (19.4) years for cPAN (p=0.074). Demographic and clinical characteristics of sPAN were compared to cPAN. Male gender frequency was 55.7% in sPAN whereas 35.5% in cPAN (p=0.002). Constitutional symptoms were more common in sPAN (219/278, 78.8%) compared to cPAN (34/76, 44.7%), p<0.001. Cutaneous manifestations were seen in 175/280 (62.5%) of the patients with sPAN with skin nodules less frequent compared to cPAN (53 (19.1) vs. 37 (48.7), p<0.001).

The most frequent organ/system involvement in patients with sPAN was neurologic involvement (170/279, 60.9%), including peripheral neuropathy (147/279, 52.7%) and CNS involvement (47/279, 16.8%). GI involvement was reported in 145/278 (52.2%) of those with sPAN, including GI bleeding in 33/282 (11.7%). Data regarding abdominal surgery was available in 183 patients, 30 of whom required surgical procedures for PAN-related GI complications. Kidney involvement was observed in 133/276 /48.2%) with sPAN. Disease-related hypertension was seen in 69/190 (36.3%) and proteinuria >400 mg/day recorded in 29/259 (11.2%). Median (IQR) serum creatinine level at diagnosis was 70.7 (35.4) μmol/L and 24/204 (10.5%) patients had a serum creatinine level >140 μmol/L. Cardiac involvement was present in 24/281 (8.5%) of patients. BVAS, was reported in 266 patients with mean (SD) 10.2 (9.1).

Fifty-four patients were diagnosed before 2000 and 304 after. The pre-2000 had a higher rate of proteinuria and higher BVAS at diagnosis, respectively, (51.9% vs 36.7%, p=0.04;11.4 (7.9) vs 8.5 (8.7), p=0.03).

Twenty-five patients had pediatric-onset PAN with a mean (SD) age at diagnosis of 11.9 (2.8) years (Table 2). There was no difference between those with pediatric and adult-onset PAN in the proportion of cPAN (20% vs. 21.3%, p=0.88). Female predominance was observed in pediatric-onset PAN (68% in pediatric-onset vs 47.1% in adult onset (p=0.06) (Table 2). Compared to adult-onset PAN, pediatric-onset PAN was associated with more frequent fever, and CNS, GI, and kidney involvement. No patient with pediatric-onset PAN had cardiac or testicular involvement.

Table 2.

The Demographic and Clinical Characteristics of Patients with Polyarteritis Nodosa According to the Age at Diagnosis of Disease

	Adult-Onset Polyarteritis Nodosa (n=333)	Pediatric-Onset Polyarteritis Nodosa (n=25)	P-Value
Age at diagnosis of disease, years	46.8 (16.3)	11.9 (2.8)	<0.001
Sex, male	176 (52.9)	8 (32)	0.061
Constitutional symptoms	232 (70.5)	21 (84)	0.174
Fever	130 (40.8)	16 (66.7)	0.018
Weight loss	142 (44.5)	9 (39.1)	0.669
Cutaneous Manifestations	233 (70.4)	18 (72)	1
Nodule	83 (25.3)	7 (28)	0.812
Livedo reticularis	83 (25.1)	11 (44)	0.057
Necrosis/Ulceration	55 (16.6)	2 (8)	0.396
Panniculitis	28 (12.7)	4 (18.2)	0.505
Raynaud phenomenon	22 (10.5)	2 (16.7)	0.624
Urticarial rash	30 (29.7)	2 (22.2)	1
Limb edema	28 (27.7)	3 (33.3)	0.710
Musculoskeletal manifestations	226 (68.5)	18 (78.3)	0.484
Myalgias	149 (47.5)	13 (56.5)	0.518
Arthralgias	156 (49.4)	15 (68.2)	0.122
Arthritis	70 (21.9)	6 (27.3)	0.597
Leg tenderness	55 (34)	3 (25)	0.753
Neurologic manifestations	158 (48)	12 (48)	1.000
Peripheral neuropathy	141 (42.7)	6 (24)	0.091
Central Nervous System Involvement	40 (12.1)	7 (28)	0.034
Gastrointestinal manifestations	130 (39.4)	15 (62.5)	0.032
Abdominal pain	50 (15.8)	10 (40)	0.005
Cholecystitis	3 (1)	-	N/A
Pancreatitis	5 (1.5)	-	N/A
Bleeding	29 (8.7)	4 (16)	0.270
Perforation	1 (0.3)	-	N/A
Renal involvement	118 (37.2)	15 (60)	0.032
Disease related hypertension	60 (27.8)	9 (42.9)	0.206
Proteinuria	26 (8.6)	3 (13)	0.445
Serum creatinine >140 μmol/L	24 (8.9)	-	N/A
BVAS at diagnosis	8.9 (8.7)	10 (8.7)	0.538
Distribution according to FFS*
FFS=0	103 (51.2)	12 (54.5)
			0.887
FFS=1	65 (32.3)	6 (27.3)
FFS>=2	33 (16.4)	4 (18.2)
Disease Extent Index	5.1 (2.8)	5.9 (2.3)	0.066
VDI, median (IQR)	2 (0–3)	0 (0–1)	0.001
Relapse	136 (46.6)	14 (82.4)	0.005

Percentages are given as valid percent.

Variables are labeled as mean (SD) or n(%) unless otherwise stated

Abbreviations: BVAS: Birmingham Vasculitis Activity Score; DEI: Disease Extent Index; FFS: Five Factor Score; IQR: Interquartile Range; N/A: Not applicable; PAN: polyarteritis nodosa; SD: standard deviation; VDI; Vasculitis Damage Index

Outcome

Disease duration was noted in 329 patients. During a median (IQR) disease duration of 59.6 (99.5) months, relapse occurred in 150/309 (48.5%) patients. Of the 150 who relapsed, 73 had data on time to relapse indicating 57 (78%) relapses occurred in the first two years (Figure 2A). The rate of relapse was slightly higher in cPAN compared to sPAN: 19/49=38.8% vs 54/168=32.1%, respectively, p=0.38. Time to first relapse was comparable in sPAN and cPAN (Figure 2B). Of 17/25 (68%) patients with pediatric-onset PAN with relapse data available, 14 had relapses. From the available data on 13 of these relapsed patients, 61.5% had relapsed in the first two years (Figure 2C).

Figure 2:

Evaluation of the first relapse in patients with polyarteritis nodosa.

Survival curves for relapse for all patients (Fig2A), comparison of systemic vs cutaneous polyarteritis nodosa (Fig2B) and pediatric-onset polyarteritis nodosa vs adult-onset polyarteritis nodosa (Fig2C)

During follow up, 34/358 (9.5%) patients died (one with cPAN). The relationship of cause of death with PAN was recorded in 26. Eight were PAN-related. Four patients deceased due to GI disease [perforation (n=2), bleeding (n=1), intestinal hemorrhage and necrosis(n=1)]. Other causes were cerebrovascular accident (n=2) and kidney failure (n=2). Of 18 whose deaths were not related to PAN, 16 has causes of death available. Eight had malignancy (colon cancer:3, cancer, not defined:2, one for each, pancreas cancer, malign melanoma, myelodysplastic syndrome/acute myeloid leukemia). Five had infectious (sepsis:2, one for each, pulmonary nocardia infection, cytomegalovirus pneumonitis, necrotizing pneumonia). Three had other causes as heart failure, pulmonary thromboembolism and pulmonary oedema. The 1-, 5-, and 10- year survival rates of patients with sPAN were 97.1%, 94.0%, and 89.0%, respectively (Figure 3). No difference was found in the survival of patients with relapsing vs. non-relapsing disease. FFS was recorded in 179 of patients with sPAN. One hundred and five (58.6%) patients had FFS ≥1 and FFS was related to mortality (p=0.006)(Figure 3E). Independent predictors of mortality for sPAN included age at diagnosis ≥ 65 years (hazard ratio (HR) 3.85, 95%CI 1.23–12.06, p=0.02); serum creatinine at diagnosis >140 μmol/L (HR 4.93, 95%CI 1.61–15.12, p=0.005); GI involvement (HR 3.51, 95%CI 1.13–10.88, p=0.03); and CNS involvement (HR 3.56, 95%CI 1.17–10.80, p=0.02) (Table 3).

Figure 3:

Evaluation of the mortality and survival analysis in patients with polyarteritis nodosa

Survival curves for all (Fig3A), comparison of systemic vs cutaneous polyarteritis nodosa (Fig3B), relapsing and non-relapsing course polyarteritis nodosa (Fig3C) and pediatric-onset vs adult-onset polyarteritis nodosa (Fig3D). according to the Five Factor Score at diagnosis (Figure 3E)

Table 3.

Predictors of Mortality in Patients with Systemic Polyarteritis Nodosa

	Univariable analysis* (Log-Rank Test)		Multivariable analysis Initial Model† Cox Proportional Hazards		Multivariable analysis Final Model† Cox Proportional Hazards
Predictors	Hazard Ratio (95% CI)	p	Hazard Ratio 95% CI	p	Hazard Ratio 95% CI	p-Value
Age ≥ 65	4.01 (1.11–14.53)	<0.001	3.87 (1.23–12.12)	0.02	3.85 (1.23–12.06)	0.02
Male Gender	1.93 (0.91–4.10)	0.09	3.48 (0.74–16.33)	0.12	3.43 (0.75–15.61)	0.11
Constitutional Symptoms	1.61 (0.58–4.50)	0.43			-	-
Central Nervous System Involvement	2.46 (0.79–7.72)	0.03	3.65 (1.18–11.30)	0.03	3.56 (1.17–10.80)	0.02
Peripheral Neuropathy	0.64 (0.30–1.39)	0.26	-	-	-	-
Pleural Effusion	0.72 (0.25–2.10)	0.58	-	-	-	-
Cardiac Involvement	0.73 (0.20–2.58)	0.66	-	-	-	-
Musculoskeletal Manifestations	0.53 (0.21–1.34)	0.11	0.87 (0.28–2.70)	0.81	-	-
Disease-Related Hypertension	0.79 (0.35–1.78)	0.58	-	-	-	-
Proteinuria (≥400mg/day)	2.54 (0.80–8.07)	0.03	1.11 (0.23–5.44)	0.23	-	-
Serum Creatinine > 140 μmol/L	4.78 (1.01–22.68)	0.001	5.04 (1.07–23.79)	0.04	4.93 (1.61–15.12)	0.005
Gastrointestinal Manifestations	2.19 (1.02–4.69)	0.04	3.39 (1.06–10.87)	0.04	3.51 (1.13–10.88)	0.03
Cutaneous Involvement	0.95 (0.43–2.09)	0.90	-	-	-	-

^*Survival curves for each of the univariable analysis are shown in Supplementary Figure 1.

^†Three steps of Cox proportional hazards model with Backward Elimination (Likelihood Ratio: LR) Method selection is shown in Supplementary Table 1

DISCUSSION

PAN remains a poorly understood form of systemic vasculitis with heterogenous presentations, secondary, and monogenic mimickers, a situation that complicates and delays diagnosis and development of therapies. This report describes the clinical characteristics and outcome of 358 patients with PAN from 9 countries in three continents and permitted comparisons of pediatric-onset to adult-onset PAN, and sPAN to cPAN, and the comprehensive assessment of vasculitis-related activity and damage.

PAN is a challenging disease for the physician because of the lack of specific laboratory markers, the multisystem nature of the disease, and the broad list of differential diagnoses. The current analysis aimed to develop and describe a cohort of patients in whom the diagnosis of PAN was firm, by using the EMA algorithm to discriminate PAN from ANCA-associated vasculitis and having at least one objective finding of PAN (with either angiography or biopsy).

Most patients in the current study had sPAN with a mean age at diagnosis in the 5^th decade with constitutional features, and neurologic and GI involvement as the most common manifestations. Ten percent of patients experienced GI bleeding or abdominal surgery for PAN-related GI complications. In kidney disease in PAN, the capillaries and glomeruli are spared, and predominantly medium-sized arteries are affected. Renovascular involvement mainly causes hypertension and presents with mild proteinuria and, rarely, with impaired kidney function. In this study, half of the patients with sPAN had kidney involvement, particularly disease-related hypertension (36.3%) with only 10% of had proteinuria (>400 mg/day) and 10% significantly decreased kidney function. Cardiac involvement was rare. Compared to sPAN, patients with cPAN had a female predominance, more skin nodules, and fewer constitutional symptoms, consistent with a prior report of cPAN (18).

In the comparison of patients diagnosed before 2000 and after 2000, the earlier group had higher disease activity and higher rates of proteinuria. These differences may be due to several confounding factors; such as lack of clear data on treatment. Additionally, the Chapel Hill Consensus definitions essentially classify-out MPA from PAN (1).Two thirds of pediatric-onset PAN patients were female in contrast with prior report of pediatric-onset PAN from Europe and the Americas finding no gender difference (19). Excluding patients with HBV-PAN and monogenic diseases allowed for a better comparison of pediatric-onset to adult-onset PAN. which found that arteries of the kidney and GI tract were frequently affected in pediatric-onset patients. Even after exclusion of patients with monogenic diseases, patients with pediatric PAN more often signal characteristics of PAN-mimickers such as higher rate of fever, abdominal pain and CNS involvement than do patients with adult-onset sPAN, Therefore, perhaps there are additional genetic drivers to be discovered in children with PAN.

The relapse rate of 50% in 5 years is higher than previously reported. However, the rate of relapse in PAN differs depending on the presence of a viral etiology to the vasculitis. One-third of the patients in both a French cohort and US cohort had infection with HBV or HCV (14, 18). In the French cohort the rate of relapse among patients with HBV-PAN was 10.6% and non-HBV-PAN was 28% which is comparable to the current cohort’s relapse rate in sPAN (32.1%) (14). In the current study there was a trend towards more relapses among patients with cPAN, similar to previous reports (14, 18). The pediatric-onset patients who were followed into adult life in this study experienced more relapses compared to patients with adult-onset PAN. A previously published study of PAN revealed longer courses of induction treatment in adult-onset patients compared to childhood-onset patients, possibly contributing to the higher relapses seen in childhood-onset cases (20). Furthermore, different immunosuppressive drugs may be prescribed for adults and children with PAN.

The 5-year survival for sPAN in the current study was 94.0%, higher than reported in previous studies (14, 18). It is not clear to what extent improved mortality rates may be due to more effective treatment, earlier recognition of the disease, or changes in diagnosis and classification.

It is important to understand prognostic factors in any organ- and life-threatening disease such as PAN. In 1996, the FFS was introduced as a prognostic tool and included high serum creatinine, proteinuria, cardiomyopathy, gastrointestinal and CNS involvement. However, this score was derived from a heterogeneous cohort of 119 patients with non-HBV PAN, 89 with HBV-PAN, 52 with MPA, and 82 with EGPA (17). In the FVSG PAN cohort, age >65 years, hypertension, and gastrointestinal manifestations requiring surgery or at least consultation with a surgeon were identified as independent predictors of death (14). The five-factor score was revisited in 2011, the items proposed in the second paper did not work well in the PAN subgroup (21). The current project investigated the prognostic value of items in the FFS and some previously identified for ANCA-associated vasculitis and other primary vasculitides. In a multivariate analysis older age (>65 years), CNS involvement, having GI disease, and a baseline serum creatinine >140 mml/L were each associated with higher mortality in sPAN. On the contrary to the first FFS, proteinuria >1g and cardiac involvement, were not found as independent predictors for mortality in this study, but there were only a small number of events. What is not clear is how changes in the approach to treatment may have changed in the past 20 years and how such changes may have influenced the course of disease in PAN.

This study has several strengths, including a large sample size for a rare condition that allowed for subgroup analyses defined by age, and cutaneous vs. systemic disease. Excluding the 39 patients with DADA2/FMF/HBV-PAN who fulfilled the EMA algorithm for PAN provided a homogenous PAN cohort. Additionally, the clinical manifestations of PAN were cataloged in a similar manner at all study sites. The large numbers of participating centers across 9 countries in Asia, Europe, and North America increases the generalizability of the results.

Limitations of this study were the mix of prospectively and retrospectively collected data, with different approaches to patient assessment, but diagnostic consensus was optimized by use of the standardized EMA algorithm There was a moderate amount of missing data, although all calculations took missingness into account. Additionally, there was no collection of data on the effects of treatment. The evolution of cPAN to sPAN was also not evaluated. The recently-described VEXAS syndrome is known to be associated with PAN; however, screening of patients with PAN for VEXAS syndrome was not conducted among the patients in this cohort, nor was systematic testing for other genetic variants causing a polyarteritis phenotype.

There is a need for better understanding of the causes of PAN and the reasons for the disease heterogeneity. The authors favor defining VEXAS syndrome and DADA2 as unique disease entities rather than secondary forms of PAN. In the case of FMF, the authors decided to use FMF-associated vasculitis rather than FMF-associated PAN, a nomenclature similar to that used for FMF-associated IgA vasculitis. Thus, patients with HBV-PAN, FMF, and DADA2 were excluded to better identify a cohort of patients with idiopathic PAN.

In conclusion, this study helps better define the demographic and clinical characteristics of patients with PAN, differentiates sPAN from cPAN. These data confirm that relapse in PAN is common, and that several clinical features are associated with higher mortality. These findings are valuable to both clinicians treating patients with PAN and investigators studying this complex disease.