Alveolar Hemorrhage in Antineutrophil Cytoplasmic Antibody–Associated Vasculitis: Results of an International Randomized Controlled Trial (PEXIVAS)

Abstract

Rationale

Diffuse alveolar hemorrhage (DAH) is a life-threatening manifestation of antineutrophil cytoplasmic antibody–associated vasculitis (AAV). The PEXIVAS (Plasma Exchange and Glucocorticoids in Severe Antineutrophil Cytoplasmic Antibody–Associated Vasculitis) (NCT00987389) trial was the largest in AAV and the first to enroll participants with DAH requiring mechanical ventilation.

Objectives

Evaluate characteristics, treatment effects, and outcomes for patients with AAV with and without DAH.

Methods

PEXIVAS randomized 704 participants to plasma exchange (PLEX) or no-PLEX and reduced or standard-dose glucocorticoids (GC). DAH status was defined at enrollment as no-DAH, nonsevere, or severe (room air oxygen saturation of ⩽ 85% as measured by pulse oximetry, or use of mechanical ventilation).

Measurements and Main Results

At enrollment, 191 (27.1%) participants had DAH (61 severe, including 29 ventilated) and were younger, more frequently relapsing, PR3 (proteinase 3)-ANCA positive, and had lower serum creatinine but were more frequently dialyzed than participants without DAH (n = 513; 72.9%). Among those with DAH, 8/95 (8.4%) receiving PLEX died within 1 year versus 15/96 (15.6%) with no-PLEX (hazard ratio, 0.52; confidence interval [CI], 0.21–1.24), whereas 13/96 (13.5%) receiving reduced GC died versus 10/95 (10.5%) with standard GC (hazard ratio, 1.33; CI, 0.57–3.13). When ventilated, ventilator-free days were similar with PLEX versus no-PLEX (medians, 25; interquartile range [IQR], 22–26 vs. 22–27) and fewer with reduced GC (median, 23; IQR, 20–25) versus standard GC (median, 26; IQR, 25–28). Treatment effects on mortality did not vary by presence or severity of DAH. Overall, 23/191 (12.0%) with DAH died within 1 year versus 34/513 (6.6%) without DAH. End-stage kidney disease and serious infections did not differ by DAH status or treatments.

Conclusions

Patients with AAV and DAH differ from those without DAH in multiple ways. Further data are required to confirm or refute a benefit of PLEX or GC dosing on mortality.

Original clinical trial registered with www.clinicaltrials.gov (NCT00987389).

At a Glance Commentary

Scientific Knowledge on the Subject

Diffuse alveolar hemorrhage (DAH) is a potentially life-threatening manifestation of antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV), but there are limited data on the effects of specific treatment approaches for DAH due to AAV. Historically, patients with DAH requiring mechanical ventilation have been excluded from clinical trials.

What This Study Adds to the Field

The PEXIVAS (Plasma Exchange and Glucocorticoids in Severe Antineutrophil Cytoplasmic Antibody–Associated Vasculitis) trial was the largest ever conducted in AAV and the first to allow enrollment of participants with DAH requiring mechanical ventilation. Thus, this trial presents a unique opportunity to examine the characteristics and outcomes of patients with ANCA-associated vasculitis with and without DAH, including severe presentations, and to evaluate the effects of randomized treatments—plasma exchange and two glucocorticoid dosing regimens. Our analysis includes baseline characteristics, mortality, and treatment effects. We found that participants with DAH differed from those without DAH in multiple ways, including baseline characteristics and a significantly increased risk of death at 30 days and 1 year. The randomized treatments did not have clear effects on mortality when looking at presence or severity of DAH.

Diffuse alveolar hemorrhage (DAH) is a severe manifestation of antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV) characterized by neutrophilic capillaritis with bleeding into the distal airspaces of the lungs, which can cause life-threatening respiratory failure (1–3). The treatment of DAH involves supportive care and treating the underlying AAV with glucocorticoids (GC) and other forms of immunosuppression (e.g., rituximab and/or cyclophosphamide) to reduce inflammation, halt tissue injury, and allow restoration of endothelial integrity (4, 5). Despite aggressive treatment, DAH in AAV can be fatal, with first-year mortality estimated at 15–50% (2, 3, 6–9). Furthermore, current treatment also increases the risk of serious infections, the most common cause of death in the first year in AAV, including in patients with DAH (10–17).

The optimal treatment for DAH in patients with AAV is uncertain. Plasma exchange (PLEX) has been used in organ- or life-threatening AAV, most frequently in patients with DAH and/or glomerulonephritis, with the rationale that swift removal of pathogenic antibodies and other mediators may expedite resolution of active inflammation (18–25). Similarly, high-dose GC are frequently used to control severe manifestations of AAV, but there are limited data to define the relative benefits and risks of dose regimens (4, 5).

The PEXIVAS (Plasma Exchange and Glucocorticoids in Severe Antineutrophil Cytoplasmic Antibody–Associated Vasculitis) trial was a randomized trial comparing the use of PLEX to no-PLEX, and two GC dosing regimens, in addition to cyclophosphamide or rituximab, in participants with severe AAV (NCT00987389) (26). This was the largest trial ever conducted in AAV, the first to evaluate GC dosing in a randomized fashion, and the first to allow enrollment of participants with DAH requiring mechanical ventilation. Importantly, PEXIVAS provides an opportunity to describe the characteristics and outcomes of patients with AAV with and without DAH and to estimate the effects of PLEX and GC dosing.

Study Design and Methods

PEXIVAS Trial

The protocol of the PEXIVAS trial was previously described (26, 27). Briefly, participants with severe AAV (granulomatosis with polyangiitis [formerly called Wegener's granulomatosis (WG)] or microscopic polyangiitis), including glomerulonephritis and/or DAH, were randomized into four treatment groups to evaluate the effects of PLEX and two GC dosing regimens on a composite primary endpoint of end-stage kidney disease or death. The initial immunosuppressive regimen, with either cyclophosphamide or rituximab, was determined by the local investigator, as was the dose of intravenous methylprednisolone (1–3 g over 3 d). Ethics approval was obtained at each site of the PEXIVAS trial. Written informed consent was provided by patients or surrogate decision-makers for trial inclusion and subsequent analyses.

Definitions

DAH due to AAV was defined at trial entry by chest imaging with bilateral pulmonary infiltrates and absence of an alternative explanation, such as volume overload or infection. In addition, participants needed at least one of four other findings: 1) evidence of DAH by bronchoscopy, such as increasingly bloody return on bronchoalveolar lavage; 2) hemoptysis; 3) anemia with hemoglobin <10 g/dl or drop in hemoglobin >1 g/dl that is otherwise unexplained; or 4) an elevated Dl_CO by pulmonary function testing.

Participants were categorized as having severe DAH if they had an oxygen saturation of ⩽85% on room air or were on mechanical ventilation.

End-stage kidney disease was defined as the need for dialysis for at least 12 consecutive weeks, or kidney transplantation.

Serious infection was defined as an infectious syndrome requiring hospitalization or intravenous antimicrobials (26).

Patient Selection

All participants randomized in PEXIVAS were included in this analysis. Participants who developed DAH after randomization were not considered to have DAH for these analyses. Randomization to treatment groups in PEXIVAS included stratification by severity of DAH (26, 27).

Statistical Analysis

This subgroup analysis was prespecified in the PEXIVAS trial protocol (27). Descriptive statistics were used for baseline characteristics and outcomes. Categorical variables were summarized by frequencies and proportions, then compared using chi-square test. Continuous variables were summarized using mean ± SD or median and interquartile range (IQR) and compared using a two-sample t test or Wilcoxon rank-sum test, respectively. Follow-up time was censored at 1 year. Outcomes were compared using Cox proportional hazards models, adjusted for the trial’s stratification characteristics (chosen for their likely association with mortality and kidney failure) of age (<60 vs. ⩾60 yr old), ANCA subtype (anti–proteinase 3 [anti-PR3] vs. anti-myeloperoxidase [anti-MPO]), serum creatinine (<500 μmol/L [5.6 g/dl] or ⩾500 μmol/L [5.6 g/dl]), and initial treatments (i.v. cyclophosphamide vs. oral cyclophosphamide vs. rituximab), as well as sex (male vs. female).

Among participants on mechanical ventilation, ventilator-free days were defined as days alive and free of ventilation within the first 30 days after randomization (rather than 28 d, to align with results for deaths at 30 d). Duration (in days) that the participant required ventilation were recorded during the trial and subtracted from 30 if they were successfully liberated from ventilation; ventilator-free days = 0 if the participant died within 30 days of mechanical ventilation, or if the participant was ventilated for >30 days. Results were summarized as median and IQR and compared by Kruskal-Wallis test. Sensitivity analysis was performed using the duration of ventilation.

All statistical analyses were performed using Stata version 15. A P value < 0.05 was considered statistically significant for all comparisons.

Results

Clinical Characteristics

Of 704 participants enrolled in PEXIVAS, 191 (27.1%) had DAH, with 61 (31.9% of DAH group) severe, including 29 (47.5% of the severe subgroup) on mechanical ventilation (Figure 1). No participants without DAH were on mechanical ventilation.

Figure 1.

Presence and severity of DAH among the 704 participants in PEXIVAS (Plasma Exchange and Glucocorticoids in Severe Antineutrophil Cytoplasmic Antibody–Associated Vasculitis) and distribution among treatment groups. DAH = diffuse alveolar hemorrhage; GC = glucocorticoids; PLEX = plasma exchange.

The characteristics of participants with and without DAH at enrollment are outlined in Table 1. Participants with DAH were younger, more frequently positive for anti-PR3 antibodies rather than anti-myeloperoxidase antibodies, more often had relapsing disease at enrollment, and had higher baseline scores on the Birmingham Vasculitis Activity Score/Wegener’s Granulomatosis (28). Participants with DAH had lower baseline creatinine concentrations but were more frequently started on hemodialysis. There were no differences in the proportions of participants randomized to PLEX or the reduced-dose GC regimen, but those with DAH more commonly received intravenous versus oral cyclophosphamide.

Table 1.

Clinical Characteristics of Participants in the PLEX and GCs in Severe Antineutrophil Cytoplasmic Antibody–Associated Vasculitis Trial by Presence of DAH

	Any DAH (n = 191)	No DAH (n = 513)	P Value
Age, yr, mean (SD)	61.1 (15.4)	63.9 (13.4)	0.018
Female	78 (40.8)	229 (44.6)	0.37
ANCA			<0.001
Anti-PR3	99 (51.8)	187 (36.5)
Anti-MPO	92 (48.2)	326 (63.6)
New diagnosis of AAV	164 (85.8)	477 (93.0)	0.004
BVAS/WG	11 (9–13)	7 (6–9)	<0.001
Creatinine, μmol/L	184 (115–320)	304 (212–442)	<0.001
Dialysis at baseline	48 (25.3)	92 (18.0)	0.033
Randomized to PLEX	95 (49.7)	257 (50.1)	0.93
Randomized to reduced GC	96 (50.3)	257 (50.1)	0.97
Immunosuppression			<0.001
Oral cyclophosphamide	44 (23.0)	197 (38.4)
Intravenous cyclophosphamide	107 (56.0)	247 (48.2)
Rituximab	40 (20.9)	69 (13.4)

Definition of abbreviations: AAV = ANCA-associated vasculitis; ANCA = antineutrophil cytoplasmic antibody; BVAS/WG = Birmingham Vasculitis Activity Score for Wegener’s Granulomatosis; DAH = diffuse alveolar hemorrhage; GC = glucocorticoid; MPO = myeloperoxidase; PLEX = plasma exchange; PR3 = proteinase 3.

Data are presented as n (%) or median (25th–75th interquartile range) unless otherwise noted.

Looking specifically at participants with DAH at enrollment, baseline characteristics of participants with severe DAH compared with those with nonsevere DAH are presented in Table 2. A high proportion of participants in both groups had hemoptysis at presentation. The median oxygen saturation on room air for those with severe DAH was 81%, versus 93% in those with nonsevere DAH. Among the 29 participants with DAH on mechanical ventilation, 15 were randomized to receive PLEX (95 total with any DAH) and 14 to the no-PLEX group (96 total with any DAH); 13 were randomized to standard GC regimen (95 total with any DAH) versus 16 to reduced GC regimen (96 total with any DAH).

Table 2.

Clinical Characteristics of Participants in the PLEX and GCs in Severe Antineutrophil Cytoplasmic Antibody–Associated Vasculitis Trial by Severity of DAH

	Nonsevere DAH (n = 130)	Severe DAH (n = 61)	P Value
Lowest SpO2	93 (91–95)	81 (78–85)	NA
Ventilation	0 (0)	29 (47.5)	NA
Hemoptysis	92 (70.7)	50 (82.0)	0.10
Creatinine, μmol/L	275 (141–456)	222 (137–391)	0.43
Dialysis at baseline	31 (23.9)	17 (28.3)	0.51

Definition of abbreviations: DAH = diffuse alveolar hemorrhage; NA = not applicable because these variables were used to define the nonsevere DAH and severe DAH groups; Sp_O2 = oxygen saturation on room air as measured by pulse oximetry.

Data are presented as n (%) or median (25th–75th interquartile range) unless otherwise noted.

Treatment Effects

The effect of PLEX on survival in participants with versus without DAH, adjusted for age, sex, ANCA subtype, kidney function, and initial treatments, is presented in Figure 2. For those with DAH, death occurred in 6 in the PLEX group and 12 in the no-PLEX group at 3 months and 8 versus 15 at 1 year (hazard ratio [HR], 0.52; 95% confidence interval [CI], 0.21–1.24) (P-interaction = 0.37 for any DAH vs. no DAH and P-interaction = 0.42 for nonsevere DAH vs. no DAH and 0.44 for severe DAH vs. no DAH) (Table 3).

Figure 2.

One-year survival by plasma exchange and severity of DAH. Diamond represents effect estimate, and horizontal line represents the 95% CI based on a Cox proportional hazards model adjusted for age, sex, antineutrophil cytoplasmic antibody subtype, kidney function, and initial treatments. CI = confidence interval; DAH = diffuse alveolar hemorrhage; HR = hazard ratio; PLEX = plasma exchange.

Table 3.

Risk of Death at 3 Months and 1 Year in PLEX and GCs in Severe Antineutrophil Cytoplasmic Antibody–Associated Vasculitis Trial Participants with or without DAH, and by Severity of DAH, by PLEX Treatment Allocation

	Died 3 Mo		Died 1 Yr		Effect of PLEX
	PLEX	No PLEX	PLEX	No PLEX	HR (95% CI)	P-interaction
Overall	18 (5.1)	21 (6.0)	25 (7.1)	32 (9.1)	0.74 (0.44–1.26)
No DAH	12 (4.7)	9 (3.5)	17 (6.6)	17 (6.6)	0.86 (0.43–1.71)
Any DAH	6 (6.3)	12 (12.5)	8 (8.4)	15 (15.6)	0.52 (0.21–1.24)	0.37
Nonsevere DAH	1 (1.6)	3 (4.6)	2 (3.1)	5 (7.6)	0.43 (0.08–2.31)	0.42
Severe DAH	5 (16.1)	9 (30.0)	6 (19.4)	10 (33.3)	0.45 (0.14–1.40)	0.44

Definition of abbreviations: CI = confidence interval; DAH = diffuse alveolar hemorrhage; HR = hazard ratio; PLEX = plasma exchange.

Effects of PLEX are expressed as HR over the first year and are adjusted for age, sex, antineutrophil cytoplasmic antibody subtype, baseline kidney function, and initial treatments.

The effect of GC dosing regimen on survival in participants with versus without DAH, adjusted for age, sex, ANCA subtype, kidney function, and initial treatments, is presented in Figure 3. For those with DAH, death occurred in 9 in the reduced GC group and 9 in the standard GC group at 3 months and in 13 and 10 at 1 year (HR, 1.33; 95% CI, 0.57–3.13) (P-interaction = 0.10 for any DAH vs. no DAH, 0.65 for nonsevere DAH vs. no DAH, and 0.08 for severe DAH vs. no DAH) (Table 4).

Figure 3.

One-year survival by GC regimen and severity of DAH. Diamond represents effect estimate, and horizontal line represents the 95% CI based on a Cox proportional hazards model adjusted for age, sex, antineutrophil cytoplasmic antibody subtype, kidney function, and initial treatments. CI = confidence interval; DAH = diffuse alveolar hemorrhage; GC = glucocorticoids; HR = hazard ratio.

Table 4.

Risk of Death at 3 Months and 1 Year in Patients with or without DAH, and by Severity of DAH, by GC Regimen Treatment Allocation

	Died 3 Mo		Died 1 Yr		Effect of Reduced GC
	Reduced GC	Standard GC	Reduced GC	Standard GC	HR (95% CI)	P-interaction
Overall	16 (4.5)	23 (6.6)	24 (6.8)	33 (9.4)	0.69 (0.41–1.17)
No DAH	7 (2.7)	14 (5.5)	11 (4.3)	23 (9.0)	0.46 (0.22–0.94)
Any DAH	9 (9.4)	9 (9.5)	13 (13.5)	10 (10.5)	1.33 (0.57–3.13)	0.10
Nonsevere DAH	1 (1.5)	3 (4.6)	3 (4.6)	4 (6.2)	0.62 (0.12–3.21)	0.65
Severe DAH	8 (25.8)	6 (20.0)	10 (32.3)	6 (20.0)	2.02 (0.64–6.39)	0.08

Definition of abbreviations: CI = confidence interval; DAH = diffuse alveolar hemorrhage; GC = glucocorticoid; HR = hazard ratio.

Effects of GC are expressed as HR over the first year and are adjusted for age, sex, antineutrophil cytoplasmic antibody subtype, baseline kidney function, and initial treatments.

Initial immunosuppression was specified by the treating clinicians at the time of randomization, before learning the allocated PLEX or GC regimen, rather than assigned by the trial. Given the differences in agent selected between those with versus without DAH, as presented in Table 1, the effects of these on death at 1 year among those with DAH were evaluated and no differences were found. Using intravenous cyclophosphamide as the referent, oral cyclophosphamide had a HR of 1.00 (95% CI, 0.57–1.78) and rituximab an HR of 0.79 (95% CI, 0.32–1.94).

Among participants with DAH requiring mechanical ventilation, there was no difference in ventilator-free days between the PLEX (median, 25; IQR, 22–26) and no-PLEX groups (median, 25; IQR, 22–27; P = 0.83). Participants receiving the reduced GC had fewer ventilator-free days (median, 23; IQR, 20–25) than those receiving the standard GC regimen (median, 26; IQR, 25–28; P = 0.009) (Figure 4). Sensitivity analysis comparing duration of mechanical ventilation yielded similar results (data not shown).

Figure 4.

Ventilator-free days within 30 days of randomization among participants with diffuse alveolar hemorrhage requiring mechanical ventilation. Box plots representing median ventilator-free days and 25th–75th interquartile range, represented by (A) PLEX treatment allocation and (B) GC regimen. GC = glucocorticoids; PLEX = plasma exchange.

The effects of PLEX or GC regimen on serious infections in the first year did not vary based on the presence or absence of DAH. Serious infections in the DAH group occurred in 40 versus 39 (PLEX vs. no-PLEX; odds ratio, 1.13; 95% CI, 0.61–2.10), and 38 versus 41 (reduced versus standard GC; odds ratio, 0.89; 95% CI, 0.48–1.65) (P-interaction = 0.85 for PLEX and 0.91 for GC regimen). This was also not impacted by the severity of DAH (P-interaction = 0.28 to 0.33 for PLEX and P-interaction = 0.92 for GC regimen) (Table 5).

Table 5.

Effects of PLEX and GC Dosing on Number of Patients with One or More Serious Infections within 1 Year

	PLEX [n (%)]	No PLEX [n (%)]	OR (95% CI)	P-interaction	Standard GC [n (%)]	Reduced GC [n (%)]	OR (95% CI)	P-interaction
Overall	149 (53)	131 (37)	1.25 (0.92–1.70)		146 (42)	134 (38)	0.85 (0.63–1.16)
No DAH	109 (42)	92 (36)	1.26 (0.88–1.81)		105 (41)	96 (37)	0.87 (0.61–1.25)
Any DAH	40 (42)	39 (41)	1.13 (0.61–2.10)	0.85	41 (43)	38 (40)	0.89 (0.48–1.65)	0.91
Nonsevere DAH	22 (34)	26 (39)	0.84 (0.39–1.84)	0.33	41 (43)	38 (40)	0.88 (0.40–1.91)	0.92
Severe DAH	18 (58)	13 (43)	1.68 (0.50–5.60)	0.28	16 (53)	15 (48)	0.92 (0.30–2.81)	0.92

Definition of abbreviations: CI = confidence interval; DAH = diffuse alveolar hemorrhage; GC = glucocorticoid; OR = odds ratio; PLEX = plasma exchange.

Overall Outcomes at 1 Year: Survival

Survival analysis by presence or absence of DAH, and by severity of DAH, is presented in Figure 5. Participants with DAH at enrollment had reduced survival compared with those without DAH (adjusted HR, 2.22; 95% CI, 1.29–3.81; P = 0.004). Among the 191 participants with DAH, 23 (12.0%) died within the first year, versus 34 of 513 (6.6%) among those without DAH (P = 0.019). The majority of deaths among participants with DAH occurred within the first 30 days after randomization (14/23; 60.9%) versus 11/34 (32.4%) for those without DAH (P = 0.033). In both groups, most deaths occurred within the first 3 months after randomization (18/23 [78.3%] with DAH and 21/34 [61.8%] without DAH). The group with severe DAH predominantly accounted for the reduced survival among those with DAH (adjusted HR, 5.17; 95% CI, 2.80–9.56; P < 0.001), with 16 of the 23 (69.6%) deaths in the DAH group occurring in those with severe DAH.

Figure 5.

One-year survival based on presence or absence of DAH at enrollment. Cox proportional hazards model adjusted for age, sex, antineutrophil cytoplasmic antibody subtype, kidney function, and initial treatments. DAH = diffuse alveolar hemorrhage.

Overall Outcomes at 1 Year: End-Stage Kidney Disease, Serious Infections

Twenty-five participants with DAH (13.1%) developed end-stage kidney disease within the first year, as did 22 of 130 (16.9%) of those with nonsevere DAH and 3 of 61 (4.9%) of those with severe DAH. This was similar to those without DAH (82/513; 16.0%; P = 0.34).

Serious infections occurred in 79/191 (41.4%) participants with DAH and 201/513 (39.2%) without DAH (P = 0.60). Among those with DAH, serious infections were less common in participants with nonsevere DAH (48/130; 36.9%) than those with severe DAH (31/61; 50.8%).

Discussion

There are limited published data on the outcomes of patients with DAH due to AAV. The PEXIVAS trial, the largest trial in AAV and the first to include participants with DAH requiring mechanical ventilation, presented a unique opportunity to examine characteristics and outcomes of patients with DAH due to AAV.

There are also few studies informing the characteristics of patients who develop DAH in AAV. Participants in PEXIVAS with DAH were younger, more commonly anti-PR3 positive, and more frequently had relapsing AAV rather than a new diagnosis at enrollment. Although participants with DAH had lower median creatinine at enrollment, they more commonly received hemodialysis. This may reflect the acuity of their presentation with active multiorgan disease, the diagnostic challenge of differentiating pulmonary edema from DAH, an increased propensity to pulmonary edema in patients with DAH and/or kidney injury, or other factors. Notably, approximately three-quarters of PEXIVAS participants with DAH had hemoptysis. This frequency, similar to that in other reports, highlights the severity of this patient cohort, while also serving as a reminder that lack of hemoptysis should not diminish clinical suspicion for DAH (2, 3, 6, 8, 9). Importantly, participants with DAH, especially those with severe DAH, were at increased risk of death, particularly within the first 30 days and 3 months.

The role of plasma exchange in DAH due to AAV remains an area of interest and debate. The numerical differences in mortality at 3 and 12 months for those with DAH between the PLEX and no-PLEX groups are of unclear clinical significance because they are based on small numbers of events and not statistically significant. Given that DAH can be acutely fatal both through flooding of alveoli with blood and concomitant ventilator-acquired lung injury, high inspired oxygen concentrations, and acute respiratory distress syndrome, ventilator-free days may be a more sensitive measure of treatment effects in the most severe DAH. There was no change in ventilator-free days with PLEX compared with no-PLEX, but there was a difference between GC dosing regimens, with fewer ventilator-free days among participants receiving the reduced-dose regimen. Although there was also a numerical difference in deaths between GC dosing regimens, this was again based on a small number of events and not statistically significant. Thus, the clinical significance of the difference in ventilator-free days is unclear. Whether longer duration of mechanical ventilation or higher dose and/or slower GC taper carries more risk of infection and other downstream sequelae should be considered on an individual basis. More research is necessary to understand optimal dosing of GCs in this population.

The strengths of this study include use of data from a large, well-defined group of patients with DAH due to AAV who received randomized treatments and that all data were collected prospectively. However, there are also several limitations to consider for this report. Although this is one of the largest cohorts of patients with DAH in AAV, it is still small to study relevant treatment effects, particularly when considering the severe DAH group, resulting in imprecise estimates. Bronchoscopy was not required to define DAH. This pragmatic decision avoided delays in enrollment and treatment of patients with this severe manifestation of AAV, recognizing that procedural availability may differ across the 95 participating clinical sites. The remaining three findings that could meet the definition of DAH (hemoptysis, otherwise unexplained anemia, or elevated diffusing capacity), in the context of appropriate radiographic appearance and clinical picture, are consistent with the presence of erythrocytes in the lower respiratory tract, through visualization or functionally. Nevertheless, it remains possible that some participants were misclassified as having DAH who in fact did not, which would bias the association between DAH and mortality to the null, particularly in the nonsevere category. The definition of severe DAH was set at an oxygen saturation of 85% on room air or the use of mechanical ventilation. The threshold to use invasive mechanical ventilation may have differed across the 95 centers. Similarly, the duration of ventilation before randomization may have differed across participants and sites, a limitation of starting the calculation of ventilator-free days at randomization. However, randomization attenuated differences between treatment groups, and sensitivity analysis using duration of ventilation yielded similar results to those for ventilator-free days. Finally, the initial immunosuppressive regimens, including the doses of pulse methylprednisolone, were left to the local investigators and could have changed after randomization, mitigating differences in the effects of the randomized therapies. For future trials to detect treatment effects in severe DAH, a rare, severe manifestation of a rare disease, a larger sample size driven by engagement of an even broader network would be necessary, together with consideration of alternative statistical approaches. Protocol restrictions intended to address the limitations outlined above may ultimately result in more difficulty recruiting and enrolling participants with severe DAH.

In conclusion, severe DAH represents an acute manifestation of AAV that is potentially life-threatening, particularly early during the disease course. These findings highlight the need for development of more effective, safer therapies for patients with severe DAH in AAV.