Reappraisal of Renal Arteritis in ANCA-associated Vasculitis: Clinical Characteristics, Pathology, and Outcome

Abstract

Background 

Renal involvement in ANCA–associated vasculitis (AAV) is associated with poor outcomes. The clinical significance of arteritis of the small kidney arteries has not been evaluated in detail.

Methods 

In a multicenter cohort of patients with AAV and renal involvement, we sought to describe the clinicopathologic characteristics of patients with AAV who had renal arteritis at diagnosis, and to retrospectively analyze their prognostic value.

Results 

We included 251 patients diagnosed with AAV and renal involvement between 2000 and 2019, including 34 patients (13.5%) with arteritis. Patients with AAV-associated arteritis were older, and had a more pronounced inflammatory syndrome compared with patients without arteritis; they also had significantly lower renal survival (P=0.01). In multivariable analysis, the ANCA renal risk score, age at diagnosis, history of diabetes mellitus, and arteritis on index kidney biopsy were independently associated with ESKD. The addition of the arteritis status significantly improved the discrimination of the ANCA renal risk score, with a concordance index (C-index) of 0.77 for the ANCA renal risk score alone, versus a C-index of 0.80 for the ANCA renal risk score plus arteritis status (P=0.008); ESKD-free survival was significantly worse for patients with an arteritis involving small arteries who were classified as having low or moderate risk, according to the ANCA renal risk score. In two external validation cohorts, we confirmed the incidence and phenotype of this AAV subtype.

Conclusions 

Our findings suggest AAV with renal arteritis represents a different subtype of AAV with specific clinical and histologic characteristics. The prognostic contribution of the arteritis status remains to be prospectively confirmed.

ANCA-associated vasculitis (AAV) describes a group of diseases characterized by the inflammatory injury of small vessels with a pathogenesis primarily related to ANCA directed against proteinase 3 (PR3) or myeloperoxidase (MPO), according to the 2012 revised Chapel Hill classification.^1,2 Renal involvement, present in 60%–80% of patients with AAV at diagnosis,^3,4 is associated with significant morbidity and higher mortality rates in AAV.^5–7 The main histologic feature of kidney injury in AAV is an extracapillary crescentic pauci-immune (negative immunofluorescence) GN. This glomerular pattern, together with arteriolitis and medullary angiitis, characterizes the AAV injury of the renal microcirculation. Hence, the first histopathologic classification proposed by Berden et al. in 2010, was based exclusively on glomerular injury and has been shown to predict renal outcome.⁸ Recently, Brix et al. developed the ANCA renal risk score (ARRS) to predict ESKD. Unlike Berden's classification, ARRS combines histopathological findings—percentage of normal glomeruli together with tubular atrophy and interstitial fibrosis—with baseline GFR.⁹ In addition to lesions of the microcirculation, AAV may also affect small arteries of the kidney parenchyma, namely, interlobular and arcuate arteries with a prevalence varying from 10% to 35%, according to the series.^10–15 Although well recognized by pathologists, this pattern of arterial injury is still largely unknown in terms of clinical presentation and outcome.¹⁴ Indeed, none of the above classifications and scores included arteritis of the small renal arteries as a prognostic factor. However, isolated case reports and small series have suggested patients with AAV and arteritis (AAV_A+) on kidney biopsy (KB) may display more severe and active disease than patients with AAV but without arteritis (AAV_A−).^11,14

The aim of the study was to describe the clinicopathological features and outcomes of patients who were AAV_A+.

Materials and Methods

Study Design

The Paris Descartes Nephropathology unit at Necker Hospital routinely receives KBs for first line histopathological diagnosis from Necker, Cochin, and Pompidou hospitals in Paris, Foch Hospital in Suresnes, and Félix Guyon Hospital in Saint-Denis La Réunion. Patients referred from these centers between January 1, 2000 and March 31, 2019 and listing diagnoses of AAV on KB were identified from the Paris Descartes Nephropathology unit database for inclusion in this study. On the basis of the presence or absence of arteritis in interlobular and/or arcuate arteries, the patients were divided into AAV_A+ and AAV_A− groups, respectively. The median follow-up after diagnosis was 34 (interquartile range [IQR], 5–55) months for patients who were AAV_A+ and 42 (IQR, 16–95) months for patients who were AAV_A−.

Participants

This study was limited to patients aged >18 years with AAV and biopsy-proven kidney involvement, defined by pauci-immune renal vasculitis involving renal glomeruli, with or without positive ANCA serology. Thus, patients with polyarteritis nodosa according to the Chapel Hill 2012 criteria, namely, vasculitis involving medium or small arteries without GN or vasculitis in arterioles, capillaries, or venules, and not associated with ANCAs, were excluded.² Patients with concomitant Goodpasture’s disease, defined by the presence of glomerular basement membrane antibodies and/or positive linear IgG staining on immunofluorescence study, were also excluded. Only the first KB was considered the index biopsy for each patient. Repeated KBs from the same patient were not included. Patients with inadequate KB, defined as <8 glomeruli and/or <2 arterial sections (interlobular and/or arcuate arteries) were not included in the study. Patients with incomplete medical files at the time of KB and/or without follow-up were also excluded. All patients were followed from the day of the KB (the index date of the analyses) until hemodialysis, death, or date of final data extraction before March 31, 2020. Clinical and biologic data were retrospectively collected from the patients’ charts. Clinical data included: sex, age at the time of KB, history of hypertension, smoking, diabetes mellitus, neoplasia, and history of AAV without renal involvement before KB, treatment regimen for induction and maintenance therapy, dialysis, kidney transplantation, or death at last follow-up. Extrarenal organ involvement was included if it was directly attributable to active vasculitis. Among such attributable involvements, heart manifestations included pericarditis and congestive heart failure, and severe gastrointestinal manifestations included bowel perforation, bleeding, and/or pancreatitis. Peripheral neurologic manifestations were included if confirmed by electromyoneurography. Pulmonary manifestations included alveolar hemorrhage, interstitial pneumoniae, and nodules attributed to AAV. Biologic data, at the time of KB, included ANCA testing by immunofluorescence and/or ELISA, serum creatinine, eGFR using the Modification of Diet in Renal Disease equation,^9,16 urinary protein-creatinine ratio, urinary sediment, serum albumin, C-reactive protein (CRP), and hemoglobin level. Biologic data, at last follow-up, included serum creatinine and eGFR. Codes were used to ensure strict donor and recipient anonymity. Our study complies with the 2000 Declaration of Helsinki. The protocol was approved by the local ethics committee (Comité d’Éthique pour la Recherche Assistance Publique-Hôpitaux de Paris 2020–07–12; Institutional Review Board 00011928).

Kidney Biopsies

Light microscopic specimens were fixed in formalin acetic acid and paraffin embedded. The 3 µm thick sections were stained with hematoxylin eosin, by periodic acid–Schiff, by Masson trichrome, and Jones silver stains. immunofluorescence specimens were studied with antiheavy chain antibodies to IgG, IgA, and IgM, light chain antibodies (kappa and lambda), and antibodies to complement factors (C3 and C1q) and to fibrinogen (Dako, Denmark). Histopathological review of the KB glass slides was performed by two observers altogether (I.B. and J.P.D.V.H.).

Glomerular injury was assessed according to Berden et al. with evaluation of the number and percentage of normal, crescentic cellular and fibrous, and globally sclerotic glomeruli.⁸ ARRS was calculated as described by Brix et al. using the percentage of normal glomeruli, the percentage of interstitial fibrosis and tubular atrophy (IF/TA) and eGFR at diagnosis, estimated by the Modification of Diet in Renal Disease equation.⁹ Interstitial inflammation was evaluated semiquantitatively and considered significant if >25% of the total cortical parenchyma was inflamed. Periglomerular and interstitial granuloma were also assessed.

Patients who were AAV_A+ were defined by the presence of arteritis involving the small kidney arteries (interlobular and/or arcuate arteries). Arteritis was defined by fibrinoid necrosis of the arterial wall and/or inflammatory infiltrate affecting the media of the artery.² The following vascular lesions were evaluated: fibrinoid necrosis, thrombosis occlusive or not, type of inflammation, that is, leukocytoclastic, mixed lymphocytic, and neutrophilic, and granulomatous inflammation, characterized by predominantly macrophagic infiltrate with occational giant cells, and chronic scarred lesions.

Outcome Measures

The primary outcome was ESKD defined by an eGFR <15 ml/min per 1.73m², the need for dialysis, or kidney transplantation at last follow-up. The secondary outcome was defined by death at last follow-up.

External Validation

Two independent cohorts were recruited to compare the incidence, the phenotype, and the prognostic value of the arteritis status with the training cohort: (1) an international retrospective cohort of 145 patients with AAV recently published by van Daalen et al. (the international cohort)¹⁷ and (2) a French retrospective multicenter cohort of 214 patients with AAV (the RENVAS [Renal Vasculitis]) cohort.¹⁸ Drs Bajema, Wester Treje, and van Daalen reviewed all of the patients in the international cohort, whereas Dr. Gnemmi and J.P.D.V.H. reviewed all of the patients with AAV_A+ from the RENVAS cohort.

Statistical Analysis

Qualitative variables were described as frequencies and compared by chi-squared tests or Fisher’s exact test, as necessary. Quantitative variables were described as mean or median values with the standard deviation or interquartile range, when appropriate. Cumulative survival curves for the time-to-event analyses were constructed according to the Kaplan–Meier method and compared with the log-rank test. Berden histopathological classification and ARRS were calculated according to the initial publication, and subgroups of patients were delineated according to validated thresholds.^8,9 Cox univariate regression was used to evaluate association between validated classifications, vascular injury status, or baseline characteristics and survival outcomes. Two different multivariable models were developed, one for each major risk classifications (first model, ARRS forced; second model, Berden classification forced) to analyze the independent association between the vascular injury status and outcomes. Candidate factors were selected when the univariate likelihood ratio test P value was <10%. Stepwise backward elimination was performed to obtain the final multivariable model (P value for removal was 5%). Two sensitivity analyses were performed to analyze the independent association between arteritis status and renal prognosis: (1) analysis of quantitative variables as continuous variables and (2) after forcing in the multivariable model, all variables significantly imbalanced between ANCA_A+ and ANCA_A− (potential confounders). Discrimination of the models on the basis of validated classifications with or without the vascular injury status was evaluated by calculating Harrel’s concordance statistic and compared using the STATA’s “somersd” package as previously recommended.^17,18,19 We performed internal validation procedures using bootstrap resampling. We randomly generated 1000 bootstrap samples. In each sample, we applied the same methodological approach than the one applied in the initial cohort, including (1) automated variable selection process (two models developed were ARRS and Berden classification forced), (2) comparison of discrimination of models with or without arteritis status, and (3) comparison of renal prognosis of patients with or without arteritis. In the external validation cohorts, two analyses were performed: (1) comparison of the clinical phenotype of patients who were ANCA_A+ and ANCA_A− and (2) ESKD survival-free according to the arteritis status in patients who were low and moderate risk (ARRS) using Kaplan–Meier curves, log-rank test, and Cox models. Statistical significance was set at P≤0.05. All tests were two sided. Statistical analyses were performed using STATA 15.1 (StataCorp LP, College Station, TX, USA).

Results

Patient Characteristics

Between January 2000 and March 2019, 372 KB from five different French centers were considered renal AAV after histopathological analysis. Iterative biopsies from the same patient (n=80), inadequate biopsies (n=10), biopsies with incomplete medical records (n=14), no follow-up data (n=14) or incorrect diagnosis (n=3) were excluded. Thus, 251 patients with AAV and biopsy-proven renal involvement were enrolled in this study, including 34 (13.5%) with histologic arteritis (AAV_A+) and 217 (86.5%) without (AAV_A−) (Supplemental Figure 1).

Baseline clinical and biologic variables are listed in Table 1. Of note, 241 (96%) patients had positive ANCA serology, but all patients had vasculitis affecting the glomeruli. Patients who were AAV_A+ were older (72 versus 62 years, respectively, P=0.003), with a less frequent history of arterial hypertension (23.5% versus 40.1%, P=0.06). MPO ANCA was the most frequent ANCA serotype in both groups (76.5% of AAV_A+ and 63.6% of AAV_A−, P=0.21). There was no statistically significant difference between AAV_A+ and AAV_A− regarding eGFR (20 ml/min per 1.73m² versus 24 ml/min per 1.73m², respectively, P=0.43), urinary protein-creatinine ratio (1.1 g/g versus 1.6 g/g, P=0.12) and dialysis requirement (17.6% versus 16.6%, P=0.80) at diagnosis. However, patients who were AAV_A+ had a more pronounced inflammatory syndrome (CRP 177 versus 31 mg/L, P<0.001; serum albumin 2.3 versus 3.0 g/dl, for AAV_A+ and AAV_A−, respectively, P<0.001) and more frequent peripheral neuropathy (32.4% versus 13.8%, P=0.007) and digestive symptoms (11.8% versus 1.8%, P=0.01). All patients received induction immunosuppressive treatment with methylprednisolone pulses associated with rituximab or cyclophosphamide. The type of induction therapy did not statistically significantly differ between the two groups.

Table 1. - Clinical and biologic variables at diagnosis

Variables	Total	ANCA-associated Vasculitis A-	ANCA-associated Vasculitis A+	P
	n=251	n=217	n=34
Age, yrs	63 (52–73)	62 (51–71)	72 (58–79)	0.003
Male/female ratio	124:127	111:106	13:21	0.16
Hypertension	95 (37.8)	87 (40.1)	8 (23.5)	0.06
Smoking	47 (18.7)	40 (18.4)	7 (20.6)	0.7
Neoplasia	29 (11.6)	22 (9.52)	7 (20.6)	0.09
Diabetes	24 (9.6)	22 (10.1)	2 (5.9)	0.75
History of extrarenal vasculitis	22 (8.3)	19 (8.2)	3 (8.8)	0.91
ANCA specificity a				0.21
Proteinase 3	77 (30.7)	71 (32.7)	6 (17.6)
MPO	164 (65.3)	138 (63.6)	26 (76.5)
None	11 (4.4)	9 (4.1)	2 (5.9)
Renal function at the time of diagnosis
eGFR, ml/min per 1.73m2	24(11–46)	24(12–47)	20(10–40)	0.43
Serum creatinine, μmol/L	230 (128–404)	230 (128–400)	271 (160–501)	0.51
Dialysis dependance	42 (16.7)	36 (16.6)	6 (17.6)	0.8
UPCR, g/g	1.5 (0.9–2.54)	1.6 (1–3)	1.1 (0.8–2)	0.12
Positive hematuria	242 (96.4)	211 (97.2)	31 (91.2)	0.11
CRP at diagnosis, mg/L	46 (11–133)	31 (9–92)	177 (111–220)	<0.001
Albuminemia at diagnosis, g/dl	2.9 (2.5–3.3)	3.0 (2.6–3.4)	2.3 (1.9–2.9)	<0.001
Hemoglobin level at diagnosis, g/dl	9.7 (8.7–11)	9.9 (8.6–11)	9.4 (8.9–9.9)	0.07
Organ involvement
Renal-limited vasculitis	53 (21.1)	49 (22.6)	4 (11.8)	0.18
Ears, nose, and throat	77 (30.7)	68 (31.3)	9 (26.5)	0.57
Pulmonary involvement	130 (51.8)	112 (51.6)	18 (52.9)	0.89
Gastrointestinal involvement	8 (3.2)	4 (1.8)	4 (11.8)	0.01
Peripheral nervous system	41 (16.3)	30 (13.8)	11 (32.4)	0.007
Cardiac involvement	11 (4.4)	9 (4.1)	2 (5.9)	0.65
Cutaneous involvement	22 (8.8)	20 (9.2)	2 (5.9)	0.75
Induction therapy
Cyclophosphamide	191 (76.1)	163 (75.1)	28 (82.4)	0.27
Rituximab	56 (22.3)	50 (23.0)	6 (17.6)	0.48
Plasma exchange	46 (18.3)	41 (18.9)	5 (14.7)	0.56
Maintenance therapy
Azathioprine	101 (40.2)	81 (37.3)	20 (58.8)	0.02
Rituximab	124 (49.4)	110 (50.7)	14 (41.2)	0.30
Other	13 (5.2)	13 (6.0)	0 (0.0)	0.14

Quantitative data are presented as median (interquartile range) or mean±SD and qualitative data as n (%), as appropriate. UPCR, urine protein-creatinine ratio.

^aOne patient was positive for anti-MPO and PR3 ANCA in the AAV_A− group.

Renal Pathology

KB were comparable between the two groups regarding total number of glomeruli, the percentages of normal glomeruli, and IF/TA (Table 2).

Table 2. - Renal pathology

Variable	Total	ANCA-associated Vasculitis A−	ANCA-associated Vasculitis A+	P
	(n=251)	(n=217)	(n=34)
Total number of glomeruli	20 (14–25)	20 (14.5–25)	20 (13.75–25)	0.9
Normal glomeruli	5 (1–11)	5 (1–10)	5 (0.75–13)	0.8
Sclerotic glomeruli	3 (1–8)	4 (1–9)	2 (0–5)	0.03
IF/TA ≥25%	98 (39.0)	87 (40.1)	11 (32.4)	0.45
Granuloma a	66 (26.3)	49 (22.6)	17 (50)	0.001
Interstitial infiltrate b	84 (33.5)	76 (35)	8 (23.5)	0.24
Arteriolitis	33 (13.1)	27 (12.4)	6 (17.6)	0.41
Berden classification				0.04
Focal	85 (33.9)	75 (34.6)	10 (29.4)
Crescentic	49 (19.5)	42 (19.3)	7 (20.6)
Mixed	60 (23.9)	46 (21.2)	14 (41.2)
Sclerotic	57 (22.7)	54 (24.9)	3 (8.8)
ARRS				0.03
Low	101 (40.2)	91 (41.9)	10 (29.4)
Moderate	76 (30.3)	59 (27.2)	17 (50.0)
High	74 (29.5)	67 (30.9)	7 (20.6)
Arterial lesions	34 (13.5)	0 (0)	34 (100)	—
Fibrinoid necrosis	33 (13.1)	—	33 (97.1)	—
Thrombosis	12 (4.8)	—	12 (35.3)	—
Endothelitis	3 (1.2)	—	3 (8.8)	—
Inflammation type c	26 (10.4)	—	26 (76.5)	—
Mixed
Granulomatous	15 (6)	—	15 (44.1)	—
Leukocytoclastic	1 (0.4)	—	1 (2.9)	—
Chronic scarred lesions	6 (2.4)	—	6 (17.6)	—

Quantitative data are presented as median (interquartile range) or mean (±SD) and qualitative data as n (%), as appropriate.

^aGranulomatous inflammation including interstitial and vascular compartments.

^bInterstitial infiltrate was considered significant if >25% of total cortical parenchyma was inflamed.

^cIn a single patient, different types of arterial inflammation can be present.

In patients who were AAV_A+, arteritis affected interlobular and arcuate arteries in 34 (100%) and 10 (29.4%) patients, respectively. The median number of arterial sections with arteritis was two, ranging from one (in 11 KB) to eight arteries. Segmental or circumferential fibrinoid necrosis of the arterial wall was observed in 33 (97.1%) KB, among patients who were AAV_A+ (Figure 1, A, D, and F), associated with mural or occlusive thrombosis in 12 patients (35.3%) (Figure 1G). Arterial wall inflammation was characterized by mixed lymphocytic, monocytic and neutrophilic mild-to-moderate palisading infiltrates in 26 (76.5%) patients (Figure 1, A and D), with epithelioid cell and multinucleated giant cell granulomas in 15 (44.1%) patients (Figure 1, B and E). Only one patient showed prominent leukocytoclastic inflammation in an arcuate artery, associated with ischemic necrosis of the renal cortex (Figure 1F). In addition to active arteritis, chronic scarred lesions were observed in six (17.6%) patients (Figure 1H).

Figure 1.: Pathologic features of ANCA-associated vasculitis with arteritis. (A) Fibrinoid necrosis and palisading mixed inflammatory infiltrate in an interlobular artery in an active ANCA-related GN with glomerular segmental fibrinoid necrosis, in renal AAV with anti-PR3 antibodies. (B) Granulomatous arteritis in an interlobular artery in a renal AAV with anti-MPO antibodies. Note the presence of periglomerular and interstitial (*) granulomas. (C) Purely necrotic arteritis in an interlobular artery with scarce vascular inflammatory infiltrate in a renal AVV with anti-MPO antibodies. (D) Palisading mixed infiltrate composed of macrophages, lymphocytes, and neutrophils in an ANCA-associated arteritis with anti-MPO antibodies. (E) Granulomatous arteritis with infiltration of the arterial wall by macrophagic infiltrate and periarterial granuloma with multinucleate giant cells, in a renal AAV with anti-PR3 antibodies. (F) Leukocytoclastic (insert) and fibrinoid necrosis in an arcuated artery with ischemic necrosis of the cortex (*) in a patient with anti-MPO antibodies. (G) Occlusive thrombosis complicating an interlobular arteritis in a patient of anti-MPO ANCA-related GN. (H) Scarred arteritis with fibrous luminal occlusion and macrophagic inflammation of the arterial wall with disruption of the elastica, in a renal AAV with anti-PR3 antibodies.

There was no statistically significant difference regarding the presence of a significant interstitial infiltrate in KB (23.5% versus 35% in AAV_A+ and AAV_A− KB, respectively (P=0.24). However, AAV_A+ KB presented more frequent granuloma, including interstitial and vascular epithelioid/giant cell granulomas (50% versus 22.6%, P=0.001). Arteriolitis was observed in 17.6% versus 12.4% of AAV_A+ and AAV_A− KB, respectively, P=0.41. According to the glomerular histopathological classification proposed by Berden et al.,⁸ patients who were AAV_A+ belonged to the sclerotic GN class in 8.8% versus 24.9% of patients who were AAV_A−, mixed GN in 41.2% versus 21.2%, focal GN in 29.4% versus 34.6%, and crescentic GN in 20.6% versus 19.3%, respectively (P=0.04).

Likewise, according to the ARRS,⁹ AAV_A+ were classified as low risk for 29.4% of the patients versus 41.9% of patients who were AAV_A−, moderate risk for 50.0% versus 27.2%, and high risk for 20.6% versus 30.9%, respectively (P=0.03) (Table 2). These results showed that arterial injury in AAV had a heterogeneous histologic presentation on KB, affecting one or multiple arteries and showing multiple patterns of inflammation with frequent granulomatous inflammation and occasional scarred arterial lesions from previous flares of the disease. AAV_A+ patients were distributed in all categories of the current prognosis’ classification.

Arteritis Was Independently Associated with Renal Function Decline

The median follow-up after diagnosis was 42 (IQR, 12–89) months, representing 1178 patient-years. Patients who were AAV_A+ were followed for 34 (IQR, 5–55) months after diagnosis and those who were AAV_A− for 42 (IQR, 16–95) months. The primary endpoint occurred in 75 patients (29.9%). ESKD-free survival was significantly poorer in patients who were AAV_A+ compared with AAV_A− (hazard ratio [HR], 2.20, 95% confidence interval [95% CI], 1.21 to 4.00, P=0.01, Figure 2). After a median follow-up of 42 months, 12 patients who were AAV_A− and eight patients who were AAV_A+ had died, with a decreased overall survival of AAV_A+ in Kaplan–Meier analysis (log-rank test, P<0.001) (Supplemental Figure 2). In univariable Cox regression analysis, 12 predictive variables were associated with the risk of ESKD occurrence, including four clinical characteristics (age at diagnosis, history of diabetes mellitus, history of arterial hypertension, and a renal limited vasculitis), three biologic variables (ANCA directed against PR3, eGFR, and hemoglobin at diagnosis), two histologic variables (number of normal glomeruli and IF/TA), two prognostic classifications (Berden and ARRS), and the presence of arteritis on the index KB (Table 3).

Figure 2.: ESKD-free survival of patients with ANCA-associated vasculitis with or without arteritis on index KB. ¹Follow-up time: 42 (IQR, 16–95) months for patients with AAV without arteritis; ²follow-up time 34 (IQR, 5–55) months for patients with AAV and arteritis. Kaplan–Meier curves, univariable Cox model.

Table 3. - Univariable analysis (Cox proportional HR): ESKD survival free

Characteristics	Variable	Label	Number of Patients	Number of Events	HR	95% CI	P
Clinical variables	Age	≤63 years old	130	31	1	—	0.001
		>63 years old	121	44	2.32	1.44 to 3.75
	Sex	Female	127	41	1	—	0.41
		Male	124	34	0.82	0.52 to 1.30
	Relapsing AAV	No	229	69	1	—	0.96
		Yes	22	6	0.98	0.42 to 2.26
	History of hypertension	No	156	37	1	—	0.01
		Yes	95	38	1.81	1.15 to 2.85
	History of diabetes mellitus	No	227	63	1	—	0.001
		Yes	24	12	2.84	1.51 to 5.34
	Smoking history	No	204	64	1	—	0.60
		Yes	47	11	0.84	0.44 to 1.61
	Renal limited vasculitis	No	198	51	1	—	0.007
		Yes	53	24	1.95	1.2 to 3.18
	Cyclophosphamide induction therapy	No	60	17	1	—	0.20
		Yes	191	58	0.70	0.40 to 1.22
	Rituximab induction therapy	No	195	62	1	—	0.84
		Yes	56	13	1.06	0.58 to 1.96
	Azathioprine maintenance therapy	No	150	44	1	—	0.12
		Yes	101	31	0.67	0.41 to 1.10
	Rituximab maintenance therapy	No	127	46	1	—	0.59
		Yes	124	29	0.88	0.54 to 1.42
Biologic variables	ANCA PR3 a	No	174	59	1	—	0.03
		Yes	77	16	0.53	0.31 to 0.93
	ANCA MPO a	No	87	21	1	—	0.16
		Yes	164	54	1.43	0.86 to 2.38
	ANCA negative	No	240	70	1	—	0.07
		Yes	11	5	2.36	0.95 to 5.89
	eGFR, ml/min per 1.73m2b,c	≥60	39	1	1	—	<0.001
		≥30 and <60	58	6	3.54	0.43 to 29.5
		<30 or dialysis	154	68	20.0	2.8 to 80.2
	UPCR, g/g b	<1.5	120	28	1	—	0.17
		≥1.5	131	47	1.39	0.87 to 2.23
	C-reactive protein, mg/dl b	<46	129	38	1	—	0.85
		≥46	122	37	1.05	0.66 to 1.65
	Serum hemoglobin, g/dl b	<9.7	139	50	1	—	0.02
		≥9.7	112	25	0.56	0.35 to 0.91
Histologic variables	Normal glomerulic,d	>25%	144	20	1	—	<0.001
		10%–25%	34	12	2.23	1.09 to 4.58
		<10%	73	43	5.09	2.99 to 8.68
	IF/TA c , d	≤25%	153	22	1	—	<0.001
		>25%	98	53	4.44	2.70 to 7.31
	Granuloma c	No	185	60	1	—	0.30
		Yes	66	15	0.74	0.42 to 1.31
	Significant interstitial infiltrate d , e	No	167	46	1	—	0.56
		Yes	84	29	1.15	0.72 to 1.83
	Berden classification	Focal	85	7	1	—	<0.001
		Mixed	60	21	4.64	1.97 to 10.94
		Crescentic	49	14	2.91	1.17 to 7.24
		Sclerotic	57	33	8.50	3.75 to 19.30
	ARRS	Low	101	6	1	—	<0.001
		Moderate	76	20	3.74	1.49 to 9.39
		High	74	49	14.79	6.33 to 34.56
	AAV_A+ c	No	217	61	1	—	0.02
		Yes	34	14	1.99	1.11 to 3.59

UPCR, urine protein-creatinine ratio.

^aOne patient was positive for both anti-MPO and PR3 ANCA.

^bAt diagnosis.

^cItem included in the ARRS.

^dOn index KB.

^eInterstitial infiltrate was considered significant if more than 25% of total cortical parenchyma was inflamed.

We built two different multivariable Cox-regression models, one for each validated prognostic classification. In both patients, the presence of an arteritis on KB at diagnosis was an independent risk factor for ESKD (Table 4). In model 1, ARRS, age at diagnosis, history of diabetes mellitus, and arteritis on index KB (HR, 3.10; 95% CI, 1.65 to 5.81, P=0.001) were independently associated with ESKD. In model 2, Berden classification, age at diagnosis, kidney function at diagnosis, and arteritis on index KB (HR, 2.01; 95% CI, 1.09 to 3.72, P=0.03) were independently associated with ESKD. Two sensitivity analyses were performed and confirmed the independent association of arteritis status with renal prognosis (continuous variables, Supplemental Table 1; confounding factors, Supplemental Table 2).

Table 4. - Multivariable analysis (Cox proportional HR): ESKD survival-free

Variables	Label	Number of Patients	Number of Events	HR	95% CI	P
Model 1
Age	≤63 years old	130	31	1	—	0.001
	>63 years old	121	44	2.42	1.42 to 4.13
ARRS	Low	101	6	1	—	<0.001
	Moderate	76	20	2.98	1.06 to 8.37
	High	74	49	16.7	6.51 to 42.70
AAV_A+	No	217	61	1	—	0.001
	Yes	34	14	3.10	1.65 to 5.81
History of diabetes	No	227	63	1	—	0.02
	Yes	24	12	2.10	1.11 to 3.99
Model 2
Age	≤63 years old	130	31	1	—	0.001
	>63 years old	121	44	2.17	1.33 to 3.55
Berden classification	Focal	85	7	1	—	<0.001
	Mixed	60	21	2.44	1.00 to 5.93
	Crescentic	57	33	1.53	0.59 to 3.96
	Sclerotic	49	14	4.73	1.97 to 11.37
AAV_A+	No	217	61	1	—	0.03
	Yes	34	14	2.01	1.09 to 3.72
eGFR at diagnosis, ml/min per 1.73 m2	≥60	39	1	1	—	0.003
	≥30 and <60	58	6	2.17	0.25 to 18.56
	<30 or dialysis	154	68	7.94	1.03 to 61.21

Arteritis Status on Index KB Significantly Improved The Discrimination of ARRS

In the whole cohort, the Berden classification and ARRS were significantly associated with renal survival (Figure 3, A and B). The discrimination of the ARRS (concordance index, 0.77; 95% CI, 0.73 to 0.82) was significantly higher compared with the discrimination of the histopathological classification (concordance index, 0.69; 95% CI, 0.63 to 0.75, P=0.007).

Figure 3.: ESKD-free survival of the whole cohort, according to (A) the Berden histopathological classification or to (B) the ARRS, at diagnosis. Kaplan–Meier curves, log-rank test. P<0.001 for both analyses.

The addition of the arteritis status on the index KB significantly improved the discrimination of the ARRS (ARRS alone: concordance index, 0.77; 95% CI, 0.73 to 0.82; ARRS + arteritis status: concordance index, 0.80; 95% CI, 0.75 to 0.84, P=0.008).

The ESKD-free survival was significantly worse in patients who were AAV_A+ compared with patients who were AAV_A− in both patients who are low and moderate risk according to the ARRS, but did not statistically significantly differ in patients who are high risk (Figure 4).

Figure 4.: ESKD-free survival of patients with ANCA-associated vasculitis with or without arteritis on index KB, according to the ARRS. Kaplan–Meier curves, log-rank test.

Internal Validation

The arteritis status was independently associated with renal prognosis in 83.3% and 70.6% of randomly generated samples, for the ARRS-forced and the Berden classification-forced models, respectively. The addition of the arteritis status on the index KB significantly improved the discrimination of the ARRS in 75.4% of the samples (ARRS alone: concordance index, 0.78; 95% CI, 0.72 to 0.82; ARRS + arteritis status: concordance index, 0.80; 95% CI, 0.76 to 0.84). In patients classified as low or moderate risk according to the ARRS, the ESKD-free survival was significantly worse in patients who were AAV_A+ compared with AAV_A− in 78.9% of samples.

External Validation

Two independent cohorts were recruited: (1) 145 patients with AAV from an international cohort recently described by van Daalen et al.¹⁷ and (2) 214 patients with AAV from the RENVAS cohort.¹⁸ The clinical and biologic data available for the three cohorts are described in Supplemental Table 3.

The prevalence of arteritis of small renal arteries was similar in the three cohorts, representing 11% of the international cohort, 9.9% of the RENVAS cohort, and 13.5% in the training cohort (P=0.45).

Supplemental Table 4 describes the characteristics of patients who were AAV_A+ and those who were AAV_A− in the three cohorts. As in our cohort, patients who were AAV_A+ in the RENVAS cohort were significantly older (mean 75 versus 64 years, P=0.007) and displayed a more pronounced inflammatory syndrome (mean CRP, 134 versus 68 mg/l, P=0.006) than patients who were AAV_A−.

In the international cohort, in patients classified as low and moderate risk according to the ARRS, ESKD-free survival was significantly worst in patients who were AAV_A+ compared with patients who were AAV_A− (HR, 5.40, 95% CI, 2.04 to 8.26, P=0.05, Supplemental Figure 3A). In the RENVAS cohort, renal survival of AAV_A− and patients who were AAV_A+ in the same risk categories did not statistically significantly differ (HR, 0.87; 95% CI, 0.45 to 3.06, P=0.66 Supplemental Figure 3B).

Discussion

In this study, we described the phenotype of AAV with arteritis of small renal arteries (AAV_A+) in a large series of AAV with renal involvement. In the AAV_A+ group, patients were older at diagnosis, displayed a more intense inflammatory syndrome, and presented with more systemic disease. AAV_A+ KB presented with more granulomatous inflammation and less sclerotic glomeruli than AAV_A− KB. Patients who were AAV_A+ had a poorer renal prognosis and patient survival as compared with the AAV_A− group of patients. The presence of an arteritis involving small arteries was independently associated with renal prognosis. The arteritis status significantly improved the statistical performance of the ARRS for low and moderate risk categories. In the international cohort, the prognostic value of the arteritis status was validated in patients categorized as low and intermediate risk according to the ARRS.

In three independent cohorts gathering 610 patients with AAV, the prevalence of renal arteritis of small arteries varied from 9.9% in the RENVAS cohort, 11% in the international cohort, and 13.5% in our cohort. This is in agreement with previous studies^12,15 and slightly less than others (18.6% of “interstitial vasculitis” by Bajema et al. and 22.2% by Vizjak et al.), possibly because these previous papers have included arteriolitis in their definition of vascular involvement.^10,11 Regarding clinical presentation, AAV_A+ were significantly older than AAV_A− at diagnosis and displayed a very intense inflammatory syndrome. Those results were confirmed in the RENVAS cohort. Moreover, they displayed a severe systemic extrarenal involvement, with more neurologic and gastrointestinal diseases. Such a higher prevalence of extrarenal damage in AAV_A+ was also reported by Endo et al. in a small case-control series.¹⁴ From a pathologic point of view, AAV_A+ also present certain specificities. Although most of the patients were associated with MPO-ANCA, granulomatous inflammation was observed in 50% of the AAV_A+ KB. Moreover, sclerotic class according to Berden classification was underrepresented in AAV_A+ KB, suggesting a more acute form of the disease.

Our findings raise the question of the biologic meaning of parenchymal arteritis in AAV. As suggested by higher levels of CRP in patients who were AAV_A+, the involvement of parenchymal arteries could be associated with disease activity. The formation of neutrophil extracellular traps is linked to the pathogenesis of AAV by contributing to endothelial damage, the stimulation of lymphocytes, the production of cytokines, and activation of the alternative complement cascade.^20–23 This leads to the formation of a vicious circle that could amplify the vascular damage throughout renal vascular bed in AAV. High levels of DNA-MPO complexes have been associated with disease activity in AAV.²¹ A correlation between CRP and the production of neutrophils extra-cellular traps has been demonstrated in other disease.^24,25 Experimental models²⁶ and multi-omics studies²⁷ could be valuable approaches for deciphering the pathogenesis of AAV_A+.

For almost two decades, constant effort has been made to stratify ESKD risk in AAV.²⁸ The Berden classification,⁸ on the basis of four purely histopathologic categories according to the prominent type of glomerular lesion, has been extensively validated.^17,29–31 More recently, Brix et al. developed a composite predictive model on the basis of both histopathological features (percentage of normal glomeruli, severity of interstitial fibrosis) and eGFR.⁹ Recent studies, although on the basis of a limited number of patients, have demonstrated the good discrimination of this model.^17,32,33 Here, in a large cohort of 251 consecutive patients with AAV, we validated the Berden classification⁸ and the prognostic score proposed by Brix et al.⁹ We showed that ARRS had a better discrimination than the histopathological classification proposed by Berden et al. These results are in agreement with several recent studies.^9,32–34

Importantly, we found that arteritis affected the renal outcome in low and moderate risk categories of the ARRS. Although a worse renal outcome was confirmed in patients who were AAV_A+ categorized as low/intermediate risk according to the ARRS in the international cohort, we failed to confirm the prognostic value of arteritis in the RENVAS cohort.

Our study has several limitations. Although multicenter and international, this study remains retrospective with a modest AAV_A+ sample size. As a result, we were not able to conclude on the prognostic effect of the arteritis status with certainty. Likewise, we could not directly link vascular injury with mortality. Nevertheless, it emphasizes the frailty of these patients and the severity of this AAV form. Finally, because parenchymal arteritis is a well-known and well-recognized feature of AAV by renal pathologists, this study was not designed to address the question of the reproducibility of its diagnosis. However, the histopathological review of the patients who were AAV_A+ from the RENVAS cohort showed a 100% agreement between J.P.D.V.H. and V.G. The same range of incidence of AAV_A+ in the three cohorts also suggest a similar recognition of these lesions by pathologists from different centers.

In conclusion, we reassessed renal parenchymal arteritis in AAV and identified a specific subtype of renal AAV with different clinical, biologic, and histologic presentations. Addressing the biologic differences between AAV_A+ and patients who were AAV_A− would improve our understanding of AAV pathophysiology. Eventually, prospective studies are warranted to address the prognosis value of the arteritis status.

Disclosures

A. Karras reports receiving honoraria from AbbVie, Amgen, Gilead, and Roche Pharmaceuticals. B. Terrier reports receiving honoraria from AstraZeneca, GlaxoSmithKline, Laboratoire Français du Fractionnement et des Biotechnologies, Roche Chugai, Grifols, Terumo Blood and Cell Technologies, and Vifor. I. Bajema reports having consultancy agreements with Aurinia, Boehringer Ingelheim, GlaxoSmithKline, and Novartis; reports being a scientific advisor or member of board of directors of Renal Pathology Society, the Editorial Board of Glomerular Diseases (online journal); and reports other interests/relationships as Director of Bajema Institute of Pathology, and Vice-President of European Vasculitis Society. All remaining authors have nothing to disclose.

Funding

None.