In 145 patients with focal segmental glomerulosclerosis (43), membranous glomerulonephritis (72), and membranoproliferative glomerulonephritis (30), 71% with normal renal function (NRF) and 63% with nephrotic syndrome (NS), the proteinuria was evaluated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) and classified into four main patterns: physiological (termed 70 kd), pure glomerular (150 kd), mixed with low molecular weight (LMW) proteins as low as 23 kd (23 kd), and mixed with very LMW proteins (20 to 10 kd; termed 10 kd). The relative frequencies were 70 kd, 0.7%; 150 kd, 1.4%; 23 kd, 61%; and 10 kd, 37%. Therefore, only the two patterns characterized by LMW ("tubular") proteins were compared to determine whether they have different clinical and prognostic significance. The serum creatinine (sCr) values (P < 0.0001), the degrees of proteinuria (P = 0.007), and the tubulointerstitial damage (P = 0.015) were significantly different in the two subgroups of patients with 23-kd and 10-kd LMW proteinuria; the difference for tubulointerstitial damage was at the limit of statistical significance after Bonferroni correction. In 82 patients with NRF at entry (sCr, 1.00 +/- 0.22 mg/dL; range, 0.6 to 1.4 mg/dL) and a follow-up of 46 +/- 22 months (range, 12 to 84 months), the predictive value of the 23-kd and 10-kd SDS-PAGE patterns on functional outcome (chronic renal failure [CRF] or clinical remission) was evaluated. A total of 12.5% of 64 patients with mixed 23-kd proteinuria and 50% of 18 patients with mixed 10-kd proteinuria developed CRF. At this time, the difference between the survival curves was highly significant (P = 0.0001), as it also was after correction for NS (P = 0.0002). When the statistical analysis was limited to 69 patients with sCr < or = 1.2 mg/dL, the difference was still highly significant (P = 0.0016), as after correction for NS (P = 0.0064). Clinical remission developed in 30% of 64 patients with 23-kd proteinuria and in 33% of 18 patients with 10-kd proteinuria; this difference was not significant. In a retrospective analysis of 20 patients (13 focal segmental glomerulosclerosis and seven membranous glomerulonephritis; 10 with the 23-kd pattern and 10 with the 10-kd pattern) treated with steroids alone or with steroids and cyclophosphamide, 80% of the patients with the 23-kd pattern and 30% of the patients with the 10-kd pattern were responsive to treatment (P = 0.025). The SDS-PAGE patterns of 54 patients with NRF at entry were again evaluated after 48 +/- 22 months: 11 patients who developed clinical remission had changed from a prevalent (91%) 23-kd pattern to a prevalent physiological (55%) or glomerular (36%) pattern; eight patients who had developed CRF showed an increase from 37% to 100% of the 10-kd pattern. In 35 patients with normal and stable renal function (sCr from 1.08 +/- 0.20 mg/dL to 1.06 +/- 0.19 mg/dL) who had persistent proteinuria (20 patients) or NS (15 patients), the rate of the 10-kd pattern increased from 6% to 46% (72% in persistent NS), suggesting an impairment of tubular protein reabsorptive function even without a concomitant impairment of glomerular filtration rate, a phenomenon that can be hypothetically attributed to tubular toxicity of persistent proteinuria. The characterization of proteinuria by SDS-PAGE in primary progressive glomerulonephritis is a useful clinical tool: it can be used to identify the main pathophysiologic determinants of excretion of LMW proteins and it has a predictive value on CRF outcome in patients with NRF, reducing the unpredictability of clinical evolution. In focal segmental glomerulosclerosis and membranous glomerulonephritis, it seems to be of predictive value on responsiveness to therapy; monitoring the SDS-PAGE patterns over time may give some insights into the relationship between the persistent protein loss and the progression of the disease.