Predicting Kidney Response to Plasma Exchange in ANCA-Associated Vasculitis: Need for Plausible Models

• ANCA
• glomerulonephritis
• plasma exchange
• anti-neutrophil cytoplasmic antibody-associated vasculitis
• acute kidney injury

Nezam et al.¹ developed a prediction model on the basis of the average treatment effect of plasma exchange (PLEX) to estimate in which patients the use of PLEX could lower the probability of death or RRT by 12 months. The model defined two groups: the group not recommended PLEX comprising 248 patients (68% did not receive PLEX, and 32% were treated with PLEX), and the group recommended PLEX comprising 177 patients (38% did not receive PLEX, and 62% were treated with PLEX). The main conclusions were based on the absolute risk difference for the occurrence of the outcome when comparing the probability of the outcome in patients who did not receive PLEX with those who received PLEX. The authors found that microscopic polyangiitis, myeloperoxidase-ANCA, higher serum creatinine, crescentic and sclerotic classes, and higher Brix score were more frequent in the group recommended PLEX, and they proposed a prediction score to predict the benefit of PLEX on the basis of those variables.

We cannot agree with the authors’ interpretation of the results. Supplemental Table 4 in their article shows two important messages that cannot be ignored. First, the absolute risk difference for death or RRT at 12 months was –4.8% (95% confidence interval, –14.9% to –5.3%) in the group not recommended PLEX versus –15.9% (95% confidence interval, –29.4% to –2.5%) in the group recommended PLEX. However, this corresponds to the same relative risk reduction between groups (4.8%/13.9%=34.5% versus 15.9%/51.6%=30.8%). Therefore, there is nothing particular to the group recommended PLEX because the effect of PLEX was similar in both groups. Second, the probability of death and RRT at 12 months was higher on the group recommended PLEX when compared with the group not recommended PLEX when patients did not receive PLEX (51.6% versus 13.9%) and when patients received PLEX (35.6% versus 9.1%). The authors found different characteristics between groups and inferred a prediction model. However, these variables are not predictors of response to PLEX because this was not different between groups (34.5% versus 30.8%) but rather reflect the risk factors for the outcomes of death or RRT at 12 months.

Patients with microscopic polyangiitis, myeloperoxidase-ANCA, lower eGFR at baseline, and sclerotic or severe kidney involvement are at higher risk of poor renal outcomes secondary to the established and irreversible fibrotic injury at the time of diagnosis.² PLEX is unlikely to reverse their course. In addition, the use of complex matching strategies did not achieve their purpose, and these results still reflect the confounding per indication due to differences in disease severity.

• Copyright © 2022 by the American Society of Nephrology