The fractional clearance of neutral dextrans (theta D) with Einstein-Stokes radii between 30 and 64 A was determined in normal subjects (controls, N = 15) and in diabetic patients with heavy proteinuria (advanced nephropathy, N = 16) or trace proteinuria (early nephropathy, N = 8). When plotted on log normal probability coordinates, the correlation between theta D and radius in controls and in early diabetic nephropathy was linear, suggesting that glomerular pores form one population with a normal distribution. In advanced diabetic nephropathy, however, theta D for large molecules (radius greater than 46 A) was elevated and departed from linearity suggesting a bimodal pore size distribution within the glomerular membrane. A mathematical model was devised, which revealed the mean fraction of glomerular filtrate permeating the upper pore mode to be 0.009 +/- 0.002, and the pores to be totally nondiscriminatory toward molecules with radii up to 64 A. We conclude that the development of large pores (or defects) within the glomerular membrane in advanced diabetic nephropathy permits the unrestricted passage of large plasma proteins into the urine.

Ten postcardiac surgical patients with acute renal failure (ARF) were infused with inulin and dextran 40. Plasma and urine were then submitted to gel-permeation chromatography to ascertain the apparent fractional clearance profile for the dextrans. Compared to normal volunteer controls, the fractional clearance profile was substantially elevated for dextran molecules in the Einstein-Stokes radius (r) range 20-40 A. For the smaller molecules (r = 20-28 A), fractional dextran clearance in ARF was frequently in excess of unity. A simple mass conservation model which assumes that the "true" fractional dextran clearance profile for the glomerulus (in Bowman's space) in ARF is the same as that for normal controls, when applied to the experimental observations, revealed that in ARF, on the average, 50% of filtered inulin is lost by tubular backleakage. Furthermore, the model permitted an estimate of the permeability properties of the damaged tubular wall. This indicated tubular permeability not unlike that of the normal glomerulus to dextran molecules with r less than 30 A, but relative impermeability to larger dextran molecules.