Comparative evaluation of glomerular morphometric techniques reveals differential technical artifacts between focal segmental glomerulosclerosis and normal glomeruli

Abstract

Morphometric estimates of mean or individual glomerular volume (MGV, IGV) have biological implications, over and above qualitative histologic data. However, morphometry is time-consuming and requires expertise limiting its utility in clinical cases. We evaluated MGV and IGV using plastic- and paraffin-embedded tissue from 10 control and 10 focal segmental glomerulosclerosis (FSGS) mice (aging and 5/6th nephrectomy models) using the gold standard Cavalieri (Cav) method versus the 2-profile and Weibel-Gomez (WG) methods and a novel 3-profile method. We compared accuracy, bias and precision, and quantified results obtained when sampling differing numbers of glomeruli. In both FSGS and controls, we identified an acceptable precision for MGV of 10-glomerular sampling versus 20-glomerular sampling using the Cav method, while 5-glomerular sampling was less precise. In plastic tissue, 2- or 3-profile MGVs showed greater concordance with MGV when using Cav, versus MGV with WG. IGV comparisons using the same glomeruli reported a consistent underestimation bias with both 2- or 3-profile methods versus the Cav method. FSGS glomeruli showed wider variations in bias estimation than controls. Our 3-profile method offered incremental benefit to the 2-profile method in both IGV and MGV estimation (improved correlation coefficient, Lin's concordance and reduced bias). In our control animals, we quantified a shrinkage artifact of 52% from tissue processed for paraffin-embedded versus plastic-embedded tissue. FSGS glomeruli showed overall reduced shrinkage albeit with variable artifact signifying periglomerular/glomerular fibrosis. A novel 3-profile method offers slightly improved concordance with reduced bias versus 2-profile. Our findings have implications for future studies using glomerular morphometry.

Keywords: 3-Profile method; Cavalieri method; Weibel-Gomez method; glomerular morphometry.

FIGURE 1

Flow diagram of advantages and disadvantages of the four morphometric techniques in plastic‐ and paraffin‐embedded tissues.

FIGURE 2

Evaluation of methods to estimate mean glomerular volume. (a) Plot shows mean ± 95% CI limits of estimates of accuracy of 5‐, 10‐, or 15‐glomerular samples versus 20‐glomeruli samples in control, FSGS, and overall samples using MGV‐Cav. (b) Schematic illustrates the 3‐profile method. R1, R2, and R3 represent three radii obtained from sections through the potential glomerular sphere. (c, d) Heat maps show correlation matrices of Spearman and Pearson correlation coefficients, respectively, of plastic‐based MGV by different methods used here versus MGV‐Cav (20 glomeruli; 100×). Square brackets show p‐values of correlation in each. (e–g) Bland–Altman plots show the bias observed in MGV in each animal sample when comparing MGV‐Cav versus (e) 2‐profile (f) 3‐profile (g) Weibel–Gomez methods, respectively. Inlay shows mean and SD of bias in each comparison for controls and FSGS mice.

FIGURE 3

Evaluation of methods to estimate individual glomerular volume. (a, b) Heat maps show correlation matrices of Spearman and Pearson correlation coefficients, respectively, of plastic‐based IGV by different methods used here versus IGV‐Cav (20 glomeruli; 100×). All these coefficients had p‐values <0.05. (IGV‐2P (1) & (2) annotate two separate IGV‐2P measurements obtained from the three profiles that are available in each glomerulus when generating IGV‐3P). (c–e) Bland–Altman plots show the bias observed in IGV in each glomerulus from Control and FSGS samples when comparing IGV‐Cav versus (c) IGV‐2P (1) (d) IGV‐2P (2) (e) 3‐profile methods, respectively. (f–h) Bland–Altman plots show the bias observed in IGV in each glomerulus from control and FSGS samples when comparing IGV‐Cav versus (c) IGV‐2P (1) (d) IGV‐2P (2) (e) 3‐profile methods, respectively. Inlay tabulates mean and SD of bias in each comparison for controls and FSGS mice. (i) Dot‐plot and error bars (mean ± SEM) plot coefficient of variation obtained from IGV measurements (9.0 ± 1.9 glomeruli/animal) within each of the 10 FSGS animals. *Kruskal–Wallis with post‐test Dunn's test p < 0.05. ***Kruskal‐Wallis with post hoc Dunn's test p < 0.001.

FIGURE 4

Differential shrinkage artifact and periglomerular/glomerular fibrosis in FSGS animals. (a) Bland–Altman plots show ratios between paraffin:plastic measurements of MGV obtained in each control animal from 10 or 20 glomeruli within respective techniques [11 comparisons from 6 control animals (green dots)]. (b) shows these same ratios of MGVs obtained from 10‐glomeruli in each FSGS animal [6 comparisons from 10 FSGS mice (red dots); X‐axis shows the corresponding MGVs]. (c) Violin plot (dashed line at median) summarizes the shrinkage artifact obtained from all measurements between FSGS and control animals [mean ± SEM; ****p < 0.0001]. (d) Photomicrographs (20×) from kidney sections of representative FSGS and control animals within our dataset showing variable glomerular fibrosis (area positive for Trichrome stain) in each.