Number of Glomeruli

reviewed manuscript: Bertram, J.F. (2013) Estimating Glomerular Number: Why We Do it and How. Clinical and Experimental Pharmacology and Physiology Frontiers in Research, 40(11):785-8.

Note that our comments/suggestions are italicized.

In this review Dr. Bertram briefly considers techniques, including unbiased stereology, used to count glomeruli in kidney. “Although design-based stereological methods are generally considered the gold-standard method, all current methods have limitations” (Bertram, 2013, abstract).  Nephron number can tell us about kidney development, kidney disease, and kidney function such as glomerular filtration rate and renal blood flow (Bertram, Introduction).

Design-based stereological methods 

The disector method can be used for estimating the number of glomeruli. “Counting approaches based on the disector principle are considered design based and, if used correctly, the resulting estimates are unbiased” (Bertram, Design-based stereological methods, second sentence). Two methods are available, the fractionator method and the NvVref method (Bertram, Design-based stereological methods).  We recommend using the fractionator method if possible, since the extra step of knowing the reference volume is not required. The disector may be implemented as a series of thin optical sections in thick physical sections; if using the fractionator method this is called the optical fractionator. Alternatively two or more pairs of contiguous thin physical sections may be used, and when teamed up with the fractionator method this is called the physical fractionator. When estimating the number of smaller particles, such as cells, the optical fractionator is more efficient than the physical fractionator because it is easier to focus through a thick physical section using a thin optical plane to find the leading edge of particles that are in the disector, than it is to obtain, orient and compare two thin physical sections. But in the case of these larger particles, the optical fractionator is less likely to ‘find’ the top of a given glomerulus, as in this example with one section that is at the limit of how thick it can be due to histological and visualization restrictions (thirty microns):

 

OFwhereas the physical fractionator, that uses pairs of thin sections, can have the thin sections spaced more appropriately for the size of the glomerulus, as illustrated here with two ten micron  sections spaced at sixty microns:

 

 

PF

 

with the wider spacing possible using the physical fractionator, the top of the glomerulus has been located since a cross section appears in the bottom but not the top thin section. There will still be the need to align the two ten micron sections with each other, and this should be considered when choosing between the optical and physical fractionator, but tools such as those in our programs Stereo Investigator®  and Brain Maker® , simplify this task. As for all unbiased stereology, we recommend using systematic random sampling.

Bertram lists limitations of the disector method (Bertram, Design based stereological methods). The first is the need for histology using glycomethacrylate that will result in ‘dimensional stability’ but we are not aware of any need to use this type of histology for the optical or physical fractionator. The second is that systematic, what he calls exhaustive, sampling is needed, but we don’t see any way to get around this requirement if data that is representative of the whole kidney is wanted. The third problem listed is that “counting glomeruli with disectors is time consuming and tedious” (Bertram, Design based stereological methods, second paragraph, first sentence), but if software such as our Stereo Investigator is used to implement the physical fractionator, and the amount of work is adjusted to obtain the required precision, we think the time spent sampling one kidney will be less than the six to eight hours given in the manuscript.

 

Methods that are Not Design-based Stereology

Three methods that are currently in use for counting glomeruli in kidney and a new method involving MRI and image segmentation thresholding are considered. One method is acid maceration that involves homogenizing the kidney and counting the glomeruli in known aliquots. “However, estimates from laboratory to laboratory often vary widely, and the kidney has to be destroyed” (Bertram, Acid maceration). Techniques like this will work “If the entire region of interest can be dedicated” and ‘the estimates will be accurate if it can be ensured that all nuclei can be extracted from this region” (West, 2012, 3.17). Another method involves counting the number of glomerular profiles per unit area. It is pointed out that this method is flawed. The profiles do not represent whole glomeruli, instead pieces of glomeruli are being counted, and the number of pieces encountered will not be unbiased; it will depend on the size and shape of the glomeruli. Instead of trying to prove that the glomeruli size and shape does not correlate with the experimentally manipulated group, unbiased stereology should be used. Furthermore, reporting profiles per area is subject to the volume-reference trap; for instance, if the control group undergoes a different amount of artifact shrinkage than the experimental group, the profiles per area count will be different but not due to a different number of glomeruli (Bertram, Number of glomerular profiles per unit area of histological section). “Knowing the number of glomerular profiles per unit area of section tells us nothing about the total number of glomeruli in the kidney” (Bertram, Number of glomerular profiles per unit area of histological section, last sentence). The third method listed is model-based stereology. Model based stereology attempts to correct bias after the data is collected. It requires ‘knowledge of glomerular geometry, such as glomerular calliper diameter, the glomerular size distribution and/or glomerular shape” (Bertram, Model-based stereological methods, third sentence), as well as glomerular orientation. If these parameters are guessed at, or for any other reason incorrect, the estimate will be incorrect. (Bertram, Model-based stereological methods). Design based stereology was created as an alternative to model based stereology. Finally, a new method is proposed that uses magnetic resonance imaging. This method involves perfusing the kidneys intravenously with cationic ferritin, dissecting them, and imaging them whole. Advantages are touted. It is said that this is the first time a glomerular size distribution could be obtained, the kidney is imaged whole, and the method is six times faster than unbiased stereology (Bertram, A new approach for counting glomeruli). However, unbiased stereology could be easily used to obtain a volume distribution by using, for instance, the nucleator probe. This MRI method relies on “image processing … to segment the images to identify specific structures and to segment the kidney from the surrounding tissue or background” (Bennett, et al., 2013, Image analysis, second paragraph). This is potentially a serious drawback as the choice of a threshold will affect the data.  Other drawbacks are the expense of the magnetic resonance scanner and the efficacy and safety of cationic ferritin (Bertram, A new approach for counting glomeruli).  Instead of using image processing and segmentation-thresholding to count the glomeruli, we think the physical fractionator probe should be carried out on the MRI data resulting in a method to efficiently estimate glomerular number in an unbiased way without having to section the kidney.

REFERENCES

Bennett, K.M., Bertram, J.F., Beeman, S.C. and N. Gretz (2013) The Emerging Role of MRI in Quantitative Renal Glomerular Morphology, American J. Physiology Renal Physiology, 304, F1252 – F1257.

Bertram, J.F. (2013) Estimating Glomerular Number: Why We Do it and How. Clinical and Experimental Pharmacology and Physiology Frontiers in Research, 40(11):785-8.

West, M.J. (2012) Basic Stereology for Biologists and Neuroscientists, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY