ComPLEX controversy: management of alveolar hemorrhage in antineutrophil cytoplasmic antibody-associated vasculitis

This is the authors' abstract. We don't add key points for this paper.

Despite PEXIVAS, the largest global randomized trial in antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV), there is ongoing debate about plasma exchange (PLEX) in diffuse alveolar hemorrhage (DAH), an uncommon but clinically challenging scenario.1,2 The Plasma Exchange and Glucocorticoids in Severe AAV (PEXIVAS) trial assessed the effect of PLEX (and glucocorticoid dosing) on the composite of death or end-stage kidney disease at 1 year among 704 participants with glomerulonephritis and/or DAH. Although it was the first trial in AAV to enroll participants with DAH requiring mechanical ventilation, it was not designed to answer the specific question of the role of PLEX in life-threatening DAH.1,3 The prespecified DAH subgroup analysis found no effect of PLEX on 1-year mortality or key shorter-term outcomes, including ventilator liberation and early mortality, but was limited by the small number, only 61 participants with severe DAH (room air oxygen saturation <85% or requirement of mechanical ventilation).2 Therefore, controversy continues. Given the rarity of AAV and acuity of DAH, it is unrealistic to expect another global randomized trial to definitively answer this question. Accordingly, other approaches are needed. In this issue of the Journal, Sanna and colleagues present a novel approach to this question in “Plasma exchange for severe alveolar hemorrhage in ANCA-associated vasculitis: emulation of a target trial.”4 Data from 184 patients admitted to ICUs across 42 French sites over 11 years comprise this well-characterized cohort. The authors provide detailed clinical and respiratory parameters and clearly define severe DAH using PEXIVAS criteria while additionally incorporating receipt of ≥6 L/min supplemental oxygen. Most patients underwent bronchoscopy, increasing the diagnostic certainty of DAH. The authors also explicitly excluded alternative causes of DAH such as pulmonary edema, which is often underappreciated. These aspects already make this multicenter retrospective cohort a valuable addition to the literature on severe DAH in AAV. The authors used a clone-censor-weight target trial emulation design to analyze their cohort.5,6 This approach accounts for two key challenges with observational data: confounding due to lack of randomization and immortal time bias, which happens when treatment initiation occurs after follow-up starts. The latter inflates the benefit of treatment because patients in the treatment arm must have survived until treatment receipt. Cloning patients meeting the inclusion criteria inherently creates 2 balanced treatment arms at baseline (ICU admission). Clones are subsequently censored when the treatment they actually receive is no longer compatible with the treatment arm they entered. For patients who received PLEX within 7 days after ICU admission (the “grace period”), their clone in the control arm is censored at the time of treatment receipt. For patients who did not receive PLEX within the grace period, their clone in the PLEX arm is censored at day 7. This cloning and artificial censoring addresses immortal time bias, such that the survival time for patients who died during the grace period (5 here), survival time pretreatment, and time of patients who survived the full grace period contribute to both arms. The artificial censoring is informative since individual characteristics often influence treatment receipt, meaning censored and uncensored individuals differ in important ways. Inverse probability-of-censoring weights address this bias by up-weighting uncensored observations to represent censored ones with similar characteristics. Statistical tests require bootstrapping to account for the artificial increase in events. In this trial emulation, all 184 patients were included in each arm after cloning. After the grace period, 126 patients were included in the PLEX arm and 53 allocated to the control group. Ultimately, no differences between groups were found for any of the outcomes: 30-, 90-, or 180-day mortality, duration of invasive mechanical ventilation, ICU or hospital length of stay. This target trial emulation used a relatively large (3 times as many patients with severe DAH as PEXIVAS), well-characterized cohort and focused on relevant clinical outcomes for this acute, life-threatening manifestation of AAV. The prespecified weighting parameters were appropriate: age, ANCA type, Birmingham Vasculitis Activity Score, serum creatinine, hemoptysis, Simplified Acute Physiology Score II, invasive mechanical ventilation, and induction immunosuppression. Nonetheless, sample size and power remain key limitations. The authors acknowledge the effective sample size of the emulated trial fell short of the 109 patients per group needed to achieve 80% power to detect a 14% difference in 30-day mortality, meaning it is possible a true effect of PLEX was missed because of power rather than biology. The 7-day treatment allocation window is another potential source of type II error. While allowing “grace” for the difficulty of diagnosing DAH and AAV, and variable resources across centers, an early benefit of PLEX still cannot be fully excluded. Critical assumptions to obtain an unbiased treatment estimate using trial emulation are positivity, consistency, correct model specification, and no unmeasured confounders. Although the authors accounted for several key clinical factors, standardized mean differences remained high for multiple parameters after weighting, most notably estimated glomerular filtration rate, indicating residual imbalance between the groups. These differences may be meaningful given the lack of analysis on the impact of concurrent kidney involvement. Finally, the lack of comparison data regarding adverse effects of PLEX limits understanding of intervention risks. With limitations of current approaches, underpowered studies, biological plausibility of PLEX in AAV, and longstanding clinical experience paired with the life-threatening nature of DAH, debate over the role of PLEX will likely continue. This well-designed study including a large number of patients with severe DAH still does not show an effect of PLEX. The consistency of these findings with those from PEXIVAS and a well-defined Mayo Clinic cohort using propensity score analysis is notable.2,7 The alignment across all outcomes assessed—ventilation, ICU stay, and survival—further supports the validity of these findings. Based on available data to date, PLEX should not be used routinely in AAV. If considered for life- or organ-threatening disease, there is no definitive evidence that the presence of DAH should drive the decision (Figure 1). Role of plasma exchange (PLEX) in severe diffuse alveolar hemorrhage (DAH) due to antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis (AAV). Although biologically plausible, a prior global randomized controlled trial and a new emulation of a target trial in the current issue4 have not found an effect of PLEX on short- or longer-term outcomes in severe DAH resulting from AAV. Acknowledging that these studies have important limitations, there is no definitive evidence to date that DAH should drive the decision to use PLEX in AAV. Ongoing early mortality with DAH due to AAV highlights the urgent need for new treatment strategies. Avacopan, an oral C5a receptor antagonist, was approved by the U.S. Food and Drug Administration in October 2021 for use in severe AAV alongside rituximab or cyclophosphamide after demonstrating its glucocorticoid-sparing effects in the ADVOCATE trial.8 Importantly, patients with DAH requiring mechanical ventilation were excluded from ADVOCATE, although observational data from Cambridge and Mayo Clinic are promising.9,10 Future research should investigate similarities and differences between responses of glomerulonephritis and DAH in AAV, and the role of precision therapies targeting tissue-level injury alongside those affecting upstream autoantibodies and mediators in circulation. All three authors contributed equally. Supplementary material is available at American Journal of Respiratory and Critical Care Medicine online. Please see the ICMJE disclosure forms, which have been provided as supplementary material.