Spatial Heterogeneity of Glomerular Phenotypes Affects Kidney Biopsy Findings

Abstract

Key Points:

1. Glomeruli with pathologic changes are not homogeneously distributed throughout the kidney cortex.

2. Biopsies that do not include the kidney capsule may underdetect glomeruli with pathologic changes.

3. Location of glomeruli with pathologic changes may be related to underlying clinical characteristics.

Background: Detection of rare glomerular phenotypes can affect diagnosis in indication kidney biopsies and in kidney tissue used for research studies. Nephropathologists are aware of potential sampling error when assessing needle biopsy cores, but quantitative data are lacking.

Methods: Kidney tissue from patients undergoing total nephrectomy enrolled in an observational, cross-sectional cohort study was used to characterize glomeruli as typical or atypical, which included globally sclerotic glomeruli (GSGs), segmentally sclerotic glomeruli, ischemic-like, and imploding. A 2D map of the glomerular annotations was generated. Spatial centrality of atypical glomeruli using the L2 metric and differences in pairwise distances between typical or atypical glomeruli were calculated. To determine how the yield of capturing atypical glomerular phenotype was affected by biopsy depth (i.e., not including the renal capsule), simulated kidney biopsies were generated from the 2D map.

Results: The mean number of glomeruli in a nephrectomy specimen was 209 (SD 143), and GSGs were the most common type of atypical glomeruli (median: 13% [interquartile range: 5,31]). Typical glomeruli were more likely to be surrounded by other glomeruli (i.e., centrally located in the kidney cortex) than GSGs, segmentally sclerosed glomeruli, ischemic-like glomeruli, and imploding glomeruli. Atypical glomeruli were 7.3% (95% confidence interval, 4.1 to 10.4) closer together than typical glomeruli and were more likely to be closer together in older patients or those with hypertension. In simulated kidney biopsies, failure to capture the capsule was associated with underdetection of GSGs, ischemic-like glomeruli, and imploding glomeruli.

Conclusions: Spatial analysis of large sections of kidney tissue provided quantitative evidence of spatial heterogeneity of glomerular phenotypes including clustering of atypical glomeruli in individuals with hypertension or older age. Most importantly, deep kidney biopsies that lack subcapsular area underdetect atypical glomerular phenotypes, suggesting that capturing the renal capsule is an important quality control measure for kidney biopsies.

Graphical abstract

Figure 1

Spatial analysis of 2D maps from kidney tissue. (A) 2D maps were generated by annotating glomerular phenotypes in scanned images of periodic acid-Schiff-stained kidney sections of each sample and (B) recording their spatial information. (C) Cartoon representation of statistical centrality analysis using the L2 metric is shown in panels i and ii. The average distance between a centrally located atypical glomerulus and typical glomeruli (panel i) is less than the distance between the peripherally located atypical glomerulus and typical glomeruli (panel ii). Atypical glomeruli are in orange, whereas typical glomeruli are in gray. (D) Cartoon representation of pairwise distance analysis in panels i and ii. In panel i, distances between atypical glomeruli (dotted lines) are overall smaller than typical glomeruli (black lines). In panel ii, distances between atypical glomeruli (light gray lines) follow a similar distribution to the typical glomeruli (black lines) (E) Representative image of simulated kidney biopsy that captures the renal capsule. (F) Representative image of a simulated “deep” kidney biopsy that does not capture the renal capsule with an offset of 2.5 mm.

Figure 2

Distribution of glomerular phenotypes in kidney tissue. (A) Representative images of typical, GSGs, SSGs, ischemic-like, and imploding glomeruli. (B) Distribution of absolute numbers of atypical glomerular phenotypes within nephrectomy specimens. (C) Distribution of percent atypical glomerular phenotypes within nephrectomy specimens. GSG, globally sclerotic glomeruli; SSG, segmentally sclerotic glomeruli.

Figure 3

Spatial heterogeneity of atypical glomerular phenotypes. (A) Difference in statistical centrality using the L2 depth between normal glomeruli and atypical glomerular phenotypes, where a negative value indicates that the glomerular phenotype is more peripherally located. GSGs (P < 0.001), ischemic (P < 0.001), imploding (P < 0.001), and SSGs (P < 0.05). (B) Histogram of log ratio of atypical to typical pairwise distance ratio at the fifth percentile (reflecting nearest neighbors), where a negative number indicates that atypical glomeruli are closer to other atypical glomeruli than to typical glomeruli. Shaded bars indicate individuals with atypical glomeruli more than 20% closer together than typical glomeruli.

Figure 4

Probabilities of capturing N or more abnormal glomeruli using local logistic regression, where N indicates the number of atypical glomeruli captured in the simulated kidney biopsy. (A) Probability of capturing N=1–5 GSGs as it relates to the proportion of GSGs in the nephrectomy specimen. (B) Probability of capturing N=1–5 SSGs as it relates to the proportion of SSGs in the nephrectomy specimen. (C) Probability of capturing N=1–5 imploding glomeruli as it relates to the proportion of imploding glomeruli in the nephrectomy specimen. (D) Probability of capturing N=1–5 ischemic glomeruli as it relates to the proportion of ischemic glomeruli in the nephrectomy specimen. Solid lines indicate biopsies including the renal capsule. Dotted lines indicate deep biopsies.