Renal Biopsy and the Prognosis of Lupus Nephritis

Abstract

At one time the prognosis of lupus nephritis was reported to be dire, with a 2-year survival of under 10 per cent [1]. This outlook was transformed by the advent of renal replacement therapy and effective immunosuppressive treatment for systemic lupus erythematosus. As a result, concern has focussed on whether patients who are at greatest risk of developing progressive renal failure may be identifiable early in their course, enabling the administration of potentially toxic therapeutic regimens to be limited to those with greatest need. Early histological studies reported an association between the presence of diffuse proliferative nephritis and poor renal outcome [2], in contrast to a relatively good outlook if histological appearances of focal proliferative nephritis or membranous nephritis were present. An association between diffuse proliferative disease and a poor prognosis was confirmed in some [3], although not all subsequent series [4]. Nevertheless, the prognostic value of renal biopsy was questioned by Freis and colleagues [5], who analysed their own and previous experience and claimed that renal histological appearances provided no additional prognostic information over that given by severity of proteinuria and elevation of blood urea. Some degree of renal histological abnormality is present in virtually all patients with SLE, even those without any clinical evidence of nephritis. If the relationship between renal histology and clinical outcome is unpredictable, and if clinical and laboratory variables offer an equally accurate prognosis, should renal biopsy be performed at all in these patients? Studies in the past decade have attempted to answer this question in two ways. Histopathological analyses have tried to define more clearly markers which may be related to prognosis. Those emerging include presence of marked subendothelial deposits on electron microscopy [6], derivation of composite scores of the activity and chronicity of nephritis [7], and presence of tubular atrophy [8], all of which have been associated with poorer outcome. Other studies have examined whether addition of specific histological information may increase the accuracy of predictive models based on clinical variables such as patient age, degree of renal impairment, and proteinuria. In this way, Whiting-O'Keeffe et al. [9] employing step-wise linear regression analysis, found that percentage of sclerotic glomeruli and presence of subendothelia l deposits added to their clinical predictive model, whereas histological classification did not. In 1989 Esdaile and colleagues analysed the long-term outcome in 87 patients with various forms of lupus nephritis [10], using Cox's proportional hazards model. They emphasized the value of poorly-apprec iated clinical markers, such as duration of renal disease, and overall severity of SLE, in determining outcome, and they