Commentary on Complications of Immunosuppressive Treatments for Glomerulonephritis

Introduction

In this Evidence-Based Nephrology review, Jefferson (1) summarizes the potential complications and adverse effects of immunosuppressive therapies for GN. Organizing the results of major clinical trials by disease category and type of immunosuppressive therapy, he describes the frequency of major treatment complications—including serious infections, leukopenias, malignancies, bone disease, and hyperglycemia—associated with each category of therapy. Taken together, these data support the notion that studies of immunosuppressive therapies in GN, including clinical trials and observational studies, should carefully examine the adverse effects of such therapies in addition to treatment efficacy.

(1) Many early studies of immunosuppressive therapies in glomerular disorders did not systematically ascertain and/or report adverse events. This is especially true in studies of IgA nephropathy, in which more recent studies have documented a high rate of serious infections with long-term glucocorticoid therapy, which was not reported in earlier trials.

(2) Overall rates of major infections may be lower with mycophenolate mofetil and rituximab compared with older immunosuppressive therapies (e.g., cyclophosphamide). However, serious infections due to opportunistic organisms may occur with all immunosuppressive treatments, including polyomavirus and Pneumocystis jirovecii linked with rituximab and herpesvirus and cytomegalovirus linked with mycophenolate mofetil. Comorbidity, the immunosuppressed state due to nephrotic syndrome, and the severity of kidney functional impairment may contribute to the risk of infection.

(3) Leukopenias are a serious complication of many immunosuppressive therapies, particularly cyclophosphamide but also, rituximab. In patients with ANCA-associated vasculitis, neutropenia may be a late-onset complication of rituximab treatment.

(4) Bone disease—including osteoporosis and avascular necrosis—is a recognized complication of glucocorticoid therapy. The risk of osteoporosis-related fracture is related to both peak and cumulative glucocorticoid dosing. Bone mineral density testing along with calcium and vitamin D therapy are recommended for patients receiving long-term therapy. Avascular necrosis risk is increased with high peak doses (>30 mg/d) of glucocorticoids and in younger patients with SLE nephritis.

(5) Risk of malignancy—especially nonmelanomatous skin cancer but also, myeloid and bladder cancer—is increased in patients with GN treated with cyclophosphamide. The magnitude of excess risk is directly related to the cumulative exposure to cyclophosphamide. An increased risk of bladder cancer was observed primarily in older studies of very high cumulative doses; more recent studies suggest no increased risk with lower cumulative doses typically used in contemporary protocols.

(6) In lupus nephritis, combination therapy with calcineurin inhibitors added to mycophenolate resulted in more frequent serious infections and overall adverse events.

Teaching Statement

The reliable assessment of adverse drug effects, including complications of immunosuppressive therapies, optimally synthesizes data from multiple sources, including safety signals from preclinical studies and clinical development, adverse events from randomized trials supporting regulatory approval, and outcomes detected in postapproval studies and pharmacosurveillance programs. Randomized trial data are valuable for assessing drug safety, because these studies can separate the effect of a drug from the characteristics of the people who receive it and use standardized procedures to collect adverse events. Greater precision in risk assessment may be achieved by combining adverse event data across randomized studies in meta-analysis. Nonetheless, the exclusive use of randomized trial data to evaluate harm due to medications has some important limitations.

(1) Trials tend to enroll relatively healthy participants and exclude those with multiple comorbidities, who may be more susceptible to the adverse effects of a novel drug.

(2) Trials tend to include comprehensive monitoring procedures to maximize safety. Although these procedures are designed to protect the safety of study participants, they are typically not reflective of monitoring procedures in real world clinical practice.

(3) Trials may include an insufficient number of participants to reliably estimate the rate of uncommon events.

(4) Trials may be of insufficient duration to observe long-term adverse outcomes.

A classic example showing the discrepancy between assessing drug safety in randomized trials and nonrandomized studies is the occurrence of hyperkalemia due to spironolactone treatment (1,2). In a randomized trial of 1663 patients with systolic heart failure, the incidence of serious hyperkalemia in the spironolactone group was low and nearly identical to that in the placebo group. In contrast, a large observational study reported that spironolactone use was associated with an estimated 20-fold greater risk of serious hyperkalemia. This difference in harm most likely reflects the intensive monitoring procedures of the randomized trial, which frequently measured serum potassium levels and reduced or held spironolactone for even small increases in potassium. In contrast, the procedures for monitoring serum potassium levels in clinical practice are less stringent and more variable.

Observational studies can provide important drug safety data that may not be easily obtained from trials. These studies can evaluate large numbers of users and nonusers of particular medications under real world conditions to identify rare and unintended adverse effects that may have been missed in trials. Observational studies are also useful for assessing the harms of medications among patients who have serious comorbidities that would prevent inclusion in trials, including advanced kidney disease. The primary limitation of observational studies of medication use is that potential differences in the characteristics of users versus nonusers of the medication could affect the frequency of adverse outcomes (i.e., confounding). This limitation can be reduced but not eliminated by careful measurement of potential confounding characteristics and comprehensive adjustment for these characteristics in the analysis.

Disclosures

None.