Dr. Peter R. Mouton, Professor of Stereology (University of So. Fla, Tampa) and Director & Chief Scientific Officer at SRC Biosciences, will host the SRC’s April “Free For All” Stereology Webinar.
The date and time will be Wednesday, April 17, 2019, from 12:00 noon to 2:00 pm.
- 12-12:50 pm: Overview of practical methods for avoiding methodological bias in stereology studies.
- 1-1:50 pm Update on NSF-funded project to automate the stereology data collection(currently in beta testing).
- After each presentation, there will be a 10-min Q&A session/break.
Dear Dr. Mouton,
I have a stereology question. I’m comparing counts of TH immunostained (TH+) neurons in non-human primates (monkey) and mice. How do I account for variation in sampling area? For instance, there are obvious differences between mice and monkeys for the reference volume (substantia nigra, pars compacta, SNpc).
Unless I control for these volume differences, how will I account for the large differences in the estimate numbers of TH+ neurons?
I suggest quantifying two parameters for each SNpc from mice and monkey brains: mean total number TH+ neurons (Total N); and 2) mean SN volume (SNvol). From these parameters you can report three values for your comparison: Total N, SNvol, and neuron density (Nv).
SNvol: You can use either Cavalieri-point counting method or planimetry to estimate SNvol. Both are comparable in terms of accuracy. However, since the Cavalieri-point counting method is not based on outlining, it’s usually more precise and efficient than planimetry.
Total N of TH+ neurons in SN: The optical fractionator quantifies the true differences in total neuron number, Total N, independent of differences in region volume or surface area. If you quantify Total N of TH+ neurons at about 100 locations on at least 6 systematic-random sampled sections through each mouse and monkey SNpc, the resulting mean Total N TH+ neurons will reflect this true differences that exist for each species.
With regard to precision, you would sample all SN (monkey and mouse) to a CE of about 0.10 (mean CE = 10%). This will standardize the variation due to sampling (within-animal variation) across individuals and species. Analyzing more individuals from each group typically lowers variability in results by capturing more between-individual variation.
Neuron density (Nv, number TH+ neurons per unit SN volume): To estimate Nv, divide the Total N by the SNvol for each individual animal. If you average Nv across animals in each group, you will get the mean density (Nv, number per unit volume), e.g., mean Nv of TH+ neurons in mouse SN.
Notes for Total N vs. Nv:
1) The interpretation of Total N (TH+ neurons) is straightforward — we expect Total N monkey >> Total N mouse. In contrast, interpretation of mean Nv can be tricky. Nv is a ratio of two parameters: number of TH+ neurons (numerator in Nv) and SNvol (denominator in Nv). If the tissue volume for one group, e.g., treatment group, undergoes differential shrinkage (atrophy), the Nv increases without a change in neuron number.
Tracking changes in neuron number by relying on Nv, rather than using total neuron number, is a well-known bias called the Reference Trap. The Reference Trap introduces potential bias by assuming the reference volume (denominator in Nv) is constant across groups, e.g., treatment vs. control; young vs. aged. Since we know aging and a wide variety of treatments affect the shrinkage that occurs during tissue processing, this assumption can be faulty and lead to biased results. For more information on this and other sources of bias in morphometric studies of biological tissue, see refs 1-3 below.
2) For a cross-species comparison, the Total N and SN volume will be higher in monkey brains than in mouse brains; however, the mean density (Nv) of TH+ neurons will be lower in monkey brain. This is due to less white matter per unit SN in mouse SNpc, causing TH+ neurons to pack more tightly than in the monkey SN.
Because Nv can vary due fluctuations in the denominator (V) in Nv, without any changes in the number of neurons, total N is the more reliable estimator than Nv for counts of TH+ neurons in SN in either mouse and/or monkey brains. For applications of this concept to a variety of real-world neuroscience studies, see Ref. 4 below.
- Mouton, PR. Unbiased Stereology: A Concise Guide. The Johns Hopkins Press, Baltimore, August, 2011.
- Mouton, PR. Applications Of Unbiased Stereology To Neurodevelopmental Toxicology, In Developmental Neurotoxicology Research: Principles, Models, Techniques, Strategies And Mechanisms (C. Wang And W. Slikke, Eds), John Wiley & Sons, Hoboken, N.J. pg. 53-77, 2011.
- Mouton, PR. Quantitative Anatomy Using Unbiased Stereology, in CRC Handbook of Imaging in Biological Mechanics (CP Neu, GM Genin, Eds), CRC Press, London, 579 pp., October 23, 2014.
- Mouton, PR. Neurostereology, Wiley-Blackwell Press, Boston, 280 pages, November 2013.
During the past decade unbiased stereology approaches have become far more prevalent in the biosciences. Still, bioscientists often approach me and other stereologists with publications using biased stereology (assumption- and model-based methods) to quantify morphological endpoints.
The questioner typically has a statement and/or question such as, “I want to quantify the same endpoint, e.g., total cell number. If reviewers accepted this paper without stereology, why shouldn’t I use their approach?”
Though published in reputable, peer-reviewed journals, more often than not these papers were typically published a dozen or more years ago.
Today, reviewers for journals, funding orgs, and regulatory agencies increasingly prefer studies with morphological endpoints based on unbiased stereology, rather than unverifiable assumptions and questionable models, e.g., “assume a cell is a sphere.” This trend shows no evidence of abating in the near future.
The bottom line is that no one can guarantee that stereology or any other method will ensure acceptance by a particular journal. It’s important to remember that publication depends on a number of factors: how well the authors avoids introducing bias; how much weight the authors placed on the results; other data supporting the conclusions; and the specific journal, editors, reviewers, all of which may have different policies toward the methods underlying morphometric studies.