Keeping Up with the Times

The ANCA-associated vasculitides are characterized by systemic necrotizing vasculitis involving small vessels and accompanied by the presence of circulating ANCAs. Left untreated, they are fatal, and therapeutic advances in the last 2 decades have transformed them to a disease of relapsing and remitting course with associated morbidity. Nowadays, mortality from active disease is exceeded by mortality due to therapy-related adverse events during the first year (1). Identification of clinical and histologic factors predictive of treatment response, relapse risk, and kidney and patient survival remains an unmet need in the management of ANCA-associated vasculitides. Identification of these predictors would assist in personalization of therapy.

Kidney involvement is common, occurring in over 80% of patients during the disease course, and it portends higher mortality rate. Despite advances in therapy, kidney failure occurs in 20%–30% of patients and negatively influences patient survival. Identifying patients at risk for kidney failure is imperative to balance the risk of progression to dialysis against the risk of adverse events from immunotherapy. The prognostication of kidney disease continues to pose a significant clinical challenge given the lack of validated clinical and pathologic correlates, but the kidney biopsy has diagnostic and prognostic value in ANCA-associated vasculitis.

Considering the prognostic value of the kidney biopsy, an international working group of kidney pathologists proposed a classification in 2010 on the basis of the changes in glomerular pathology. ANCA-associated GN was categorized into four classes addressing the extent and age of glomerular lesions: the focal, crescentic, mixed, and sclerotic classes (2). The validation study, performed in a cohort of 100 patients enrolled in European vasculitis trials, demonstrated that this classification predicted 1- and 5-year kidney survival. The focal class had the best kidney prognosis, followed by the crescentic, mixed, and sclerotic classes. Subsequent validation studies in other cohorts showed differential outcome in one of the four classes, with most inconsistency between the crescentic and mixed classes. Multivariable analyses adjusting for kidney function, percentage of normal glomeruli, or tubular atrophy and interstitial fibrosis demonstrated that using the four classes did not improve outcome prediction (3,4). A meta-analysis of 17 studies encompassing 1540 adult patients found no difference in kidney survival between crescentic and mixed class GN (5). Moreover, up to 50% of patients in the sclerotic class respond to immunotherapy and recover kidney function.

Bajema and colleagues recommended using the percentage of normal glomeruli to predict kidney function in ANCA-associated GN already in 1999 when investigating a multicenter cohort. Forming the European Vasculitis Society (EUVAS) working group, the investigators reviewed biopsies of the methyl-prednisolone versus plasma exchange trial for severe kidney vasculitis and stated that normal glomeruli were the predictor of dialysis independence and improvement in kidney function, indicating that the unaffected part of the kidney is vital in determining kidney outcome (6). The ANCA classification working group decided then to use a cutoff for glomerular lesions of 50% (2). Hillhorst detected in 2013 that patients’ kidney survival of the crescentic and mixed classes, which initially had not differed, was indeed different when applying a cutoff for normal glomeruli of 25%. In a multicenter, prospective cohort of over 100 patients, a regression tree detected two cutoffs for normal glomeruli predicting kidney survival: at 23.3% and 5% (7). Similar to the clinical picture of advanced kidney disease, where very little kidney function seems often enough to keep patients dialysis independent, a significantly smaller amount of normal glomeruli than 50% seems to decide kidney survival in ANCA-associated GN.

The severity of kidney dysfunction at diagnosis has been associated with reduced kidney survival. Validation in large interventional cohorts demonstrated that an eGFR<15 ml/min independently affected kidney survival (1). Nonetheless, in a retrospective study of 155 patients with severe ANCA-associated GN, >14% of patients with an average eGFR at entry of <10 ml/min and highest chronicity score responded to immunosuppressive therapy (8). The combination of baseline kidney function and histologic parameters has been shown to be a better predictor of kidney outcome than baseline kidney function alone. Different glomerular lesions have extensively been investigated for their prediction value. In a beautiful investigation, the EUVAS working pathologists reviewed biopsies in great detail in 2002. Here, Hauer and colleagues showed that the amount of normal glomeruli correlated with kidney function at diagnosis, achieving a mediocre strength of correlation (r=0.53). Other glomerular parameters, however, failed to demonstrate more than weak Pearson correlation coefficients to kidney function at diagnosis and after 18 months of follow-up. The kidney function at diagnosis was the single predictor of kidney function after 18 months of sufficient strength (r=0.67). The reversibility of different glomerular injuries is controversial. The point of no return when a glomerulus fails to cope with a glomerular injury is unknown. In a more recent prospective cohort, different glomerular damage scores were tested (7). The percentage of normal glomeruli and, inversely, the score counting all glomerular lesions were the most accurate predictors of kidney survival, superior to scores omitting or reducing the effect of fibrinoid necrosis and segmental lesions.

A novel tool called the ANCA kidney risk score was developed and validated by Brix et al. (7). This clinicopathologic score includes three parameters: baseline eGFR, percentage of normal glomeruli, and degree of tubular atrophy and interstitial fibrosis on the kidney biopsy. A prospective cohort was investigated for clinical and histopathologic glomerular and interstitial lesions. A tree analysis detected cutoffs for normal glomeruli, tubular atrophy and interstitial fibrosis, and kidney function separating patient cohorts as per development of kidney failure, and no arbitrary cutoffs were set. According to Cox regression, risk factors were assigned points, and a risk score was calculated ranging from zero to 11 points. This risk score was used to define three differing risk groups for kidney failure: low (zero points), medium (two to seven points), and high (more than eight points). The risk model was then validated and confirmed in an independent patient cohort.

There are histopathologic scoring and classification systems in clinical use for IgA and diabetic nephropathy. Ten years on from the initial proposal of the classification for ANCA-associated GN and many validation studies later, the classification has not been adjusted yet. In this CJASN issue, the original investigators validated classification and score in a further cohort and performed a meta-analysis of 21 publications. van Daalen et al. (9) analyzed a retrospective multicenter cohort of 145 patients diagnosed over a period of 20 years (1991–2011). During a mean follow-up of 8 years, only 17% of patients developed kidney failure. The smallest biopsy included in the study contained five glomeruli. Interobserver variability accounted for a 68% agreement of allocating biopsies into the four classes. Results of the new cohort as well as the meta-analysis confirmed the previous findings that the focal class has the best outcome, the sclerotic class has the worst outcome, and the crescentic and mixed classes show similar kidney survival.

Investigating the aspect of improvement in kidney function in this validation cohort, the crescentic class demonstrated a temporary improvement in ΔeGFR at 1 year follow-up, but the mixed class did not. This difference was not detectable in the ΔeGFR at 5 years of follow-up. The authors stated that they feared the clinical and histopathologic differences would be lost if the crescentic and mixed classes were merged now. Therefore, they decided to re-evaluate the classification addressing tubulointerstitial changes and reinvestigating glomerular lesions.

Validating the ANCA score, the authors confirmed the distinction between the risk groups, with significantly differing outcomes of the low-, medium-, and high-risk groups (0%, 10%, and 31%, respectively). Rates of kidney failure were lower compared with the German cohort, which comparatively had a lower baseline eGFR. Since its publication in 2018, the score has been investigated in first-validation cohorts. A recent cohort from Turkey demonstrated similar findings to the initial German cohort investigated by Brix et al. (7). The rates of kidney failure were 8%, 42%, and 67% in the low-, medium-, and high-risk groups, respectively, over a median follow-up of 39 months (10). A further cohort from Mexico enriched with mixed and sclerotic class GN, a group at high risk of progression to kidney failure, confirmed the risk groups’ distinction. Kidney survival rates were 100%, 85%, and 37% according to the low-, medium-, and high-risk groups, respectively, after 36 months follow-up (11).

ANCA-associated vasculitides present in very different clinical scenarios. A risk stratification assisting in management of these vulnerable patients will help to personalize medicine. Applying clinicopathologic risk stratification will represent a major advance in the management of ANCA-associated GN. Investigating large, prospective cohorts of interventional trials will enable us to improve our prognostic tools; adjust cutoffs; and include additional prognostic clinical parameters such as age, ANCA subtype, and type of induction therapy. Scoring systems and classifications using parameters that predict outcome will adapt over time to achieve our goal to provide patients with a realistic prognosis and tailor their therapy accordingly.

Disclosures

D. Geetha reports receiving consultant fees from Aurinia and ChemoCentryx outside the submitted work. The remaining author has nothing to disclose.

Funding

None.