Design-based stereology: Planning, volumetry and sampling are crucial steps for a successful study
Introduction
Design-based stereology represents a set of tools which can be employed to quantify three-dimensional structures using two-dimensional (e.g. microscopic) sections, based on a sound statistical and stochastic background. It has been recommended as the method of choice for quantification of structure in kidney (Madsen, 1999), lung (Hsia et al., 2010) and brain research (Saper, 1996). However, performing a sophisticated design-based stereological assessment is demanding and prone to numerous pitfalls which have to be identified and avoided well in advance, ideally before starting to harvest the organs. This review will focus on the most common pitfalls an investigator might encounter. Some of these pitfalls will be analyzed in detail and appropriate action will be presented. Various ways of performing volumetry and sampling of tissue for multiple purposes (e.g. light microscopy, electron microscopy and immunohistochemistry) will be presented.
Experience shows that conceptual planning is often underestimated in microscopic quantification. Quantitative results obtained by an intuitive approach can be biased, may be misinterpreted, or are simply wrong (Mendis-Handagama and Ewing, 1990). Careful preliminary planning of the entire stereologic workflow is the most important prerequisite to obtaining unbiased quantitative data from microscopic images. Hence, the key to a successful stereological study is the deliberation made during the planning phase, before even touching any tissue or material. A best-case scenario would be a core unit within a scientific organization that can provide stereological know-how and support in step with actual practice.
As an example, our units do not offer a full stereology service, whereby scientists would supply animals and receive final numeric results. We expect interested scientists to be closely involved in the entire process of sampling and probing, since their specific knowledge of the biological problem being investigated is fundamental. Ideally a scientist intending to conduct a quantitative evaluation would contact a stereologist during the initial planning stage of the experiment. At this early time point the stereological study can be optimally designed (see details in Section 1.2 and Fig. 1). The stereologist would then accompany the investigator through all the steps of the experiment.
In stereology, all steps of the quantification workflow are interdependent as each step builds on the previous one. The stereological workflow always includes two major steps: (i) unbiased tissue sampling and (ii) unbiased probing of microscopic images. Fig. 1 shows a flowchart summarizing a study design. First, the expected structural alterations of the experimental setup have to be identified and discussed on the basis of data from previous experiments or the literature. The quantification approach and its workload may vary considerably depending on the experimental question and the given biological circumstances. Therefore, efficiency considerations must always to be part of the design (Gundersen and Osterby, 1981, Mathieu et al., 1981, Gundersen et al., 1999).
Planning a stereological study comprises numerous steps, including the characterization of parameters to assess the appropriate tissue preparation and sampling scheme, the technique of acquiring images and the modus of probing the structures (Weibel, 1979, Gundersen et al., 1988a, Gundersen et al., 1988b, Hsia et al., 2010). Performing a stereological study, however, is not demanding regarding the necessity of specific equipment. Tissue sampling and slicing is done by sharp knives or razor blades and more complex slicing tools can be borrowed from a stereology service unit. Image sampling from sections is done on standard microscopes (light and electron microscope) equipped with a digital camera and a specimen stage allowing defined x/y movements. However, a motorized stage with programmable x/y step movements facilitates the task considerably. For the analysis of images on a computer monitor, readily available software generates the graphical probing elements (“test system”) superposed on the images (e.g. STEPanizer (Tschanz et al., 2011)). The most sophisticated (but also expensive) approach in light microscopy is to use a computerized stereology system (e. g. VISIOPHARM newCAST, Hørsholm, Denmark/MBF Bioscience, StereoInvestigator, Williston, USA/Stereology Resource Center, The Stereologer system, Florida, USA). Such systems allow the scientist to program entire image sampling sequences and estimation routines.
When carrying out a stereological approach for the first time, there are many potential sources of error. It is highly recommended to perform an experimental test-run from tissue preparation to acquisition of a microscopic image and estimation of parameters. This allows the scientist to detect difficulties and to check for errors before processing the costly material. Table 1 summarizes common critical situations which regularly appear in planning a stereological study for the first time.
Quantification by microscopy means working under unique conditions where biological material is often considerably manipulated. The following challenges have to be recognized and taken into account:
- •
A substantial reduction of material: After the whole preparation, sampling and sectioning process, only a minute fraction of the whole tissue will be microscopically analyzed. The demands on the sampling strategy with respect to an unbiased, accurate and efficient sample are high. Errors at an earlier processing stage can cumulate with every further sampling step (Cruz-Orive and Weibel, 1981). Uneven distribution of a feature within the reference space may complicate this task (Gundersen and Jensen, 1987, Gundersen et al., 1999).
- •
Chemical and physical influences during preparation: may distort the spatial arrangement within the tissue, very often in an uneven manner, resulting in disproportional and anisotropic changes (Dorph-Petersen et al., 2001, Hsia et al., 2010).
- •
Analysis on microscopic sections reduces the sample dimension from 3D to 2D: One should be aware that only profiles of structures are represented on the level of sections and unbiased estimations of object numbers are not possible (Sterio, 1984).
- •
In an anisotropic structure section orientation influences the probability of detection of features (Mathieu et al., 1983).
- •
Quantification on microscopic images normally generates densities: Comparing densities measured in a specimen often leads to misinterpretation. This phenomenon is called “reference trap” (Braendgaard and Gundersen, 1986). Densities have to be referenced to the absolute volume resulting in absolute values (multiplication by the organ volume (Scherle, 1970, Weibel, 1979, Michel and Cruz-Orive, 1988) or fractionator principle (Gundersen, 1986, Gundersen, 2002)). Absolute parameters are much more robust because they prevent the “reference trap”.
During all steps of a stereological workflow (see Section 1.2) these problems have to be addressed rigorously. Good quantification is essentially based on appropriate tissue preparation and sampling. Together these make up a major part of the manual workload in a stereological study.
The following example is an illustration of how crucial errors can be avoided just by considering the most basic rules of stereology and accurately planning the experimental steps.
We were contacted by a scientific group working on a very interesting and complex knock-out experiment. They had observed parenchymal alteration in the lung due to genetic modifications and intended to quantify those phenotypic changes versus wild-type controls. When they first contacted us, they proudly brought a box with 600 stained microscopic lung sections. They had processed the material in the following way: Lungs of 10 knock-out and 10 wild-type animals had been removed, randomly cut into 30 blocks and fixed by immersion, i.e. the blocks had been dipped in paraformaldehyde. The blocks had then been embedded in paraffin and one section per block had been cut and stained for further analysis. Each animal was therefore “amply” represented by 30 lung sections. There was no lung material left. The group wanted to detect alterations in tissue and air volume, surface area, mean linear intercept of alveoli, as well as the number of cells and alveoli. They intended to count alveolar numbers by counting the profiles of alveoli, assuming them to be of round shape.
Unfortunately, we had to reject support for their quantification attempt due to the following crucial mistakes:
- •
Missing original lung volume data.
- •
Inadequate tissue fixation. Instillation fixation would have been the appropriate procedure for these particular stereological questions.
- •
Lack of proper tissue and section sampling.
- •
Paraffin embedding, leading to inhomogeneous tissue shrinkage (Dorph-Petersen et al., 2001).
- •
Single section sampling (2D), not allowing the assessment of cell or alveolar numbers.
- •
Assumptions on spatial properties of objects (alveolar and cell shape).
Section snippets
Accuracy and precision
In general, the goal of a design-based stereological assessment is to obtain accurate quantities of three-dimensional structures within an organ using two-dimensional microscopic sections in combination with geometric probes such as test points or test lines, based on a sound stochastic and statistical background (Mayhew and Gundersen, 1996). In order to obtain unbiased and representative stereological data, the tissue under study must be representative of the whole organ. This means that every
Summary
The aim of a quantitative assessment of structure is to derive unbiased (accurate), sufficiently precise and statistically valid data in an efficient way for comparisons in experimental research including morphology, pathology and physiology. To be accurate and efficient, a sophisticated planning of the study is necessary as the investigator might be otherwise faced by a number of pitfalls. This planning phase should involve a power analysis of how many study subjects are needed (pilot study!)
Acknowledgement
The authors thank Sheila Fryk for valuable linguistic input.
References (49)
- et al.
The impact of recent stereological advances on quantitative studies of the nervous system
J. Neurosci. Methods
(1986) - et al.
A review of recent methods for efficiently quantifying immunogold and other nanoparticles using TEM sections through cells, tissues and organs
Ann. Anat.
(2009) - et al.
Estimation of surface area from vertical sections
J. Microsc.
(1986) - et al.
Estimate of the total number of neurons and glial and endothelial cells in the rat spinal cord by means of the optical disector
J. Comp. Neurol.
(1993) - et al.
Design-based stereology: introduction to basic concepts and practical approaches for estimation of cell number
Toxicol. Pathol.
(2010) Sampling Techniques
(1977)- et al.
Sampling designs for stereology
J. Microsc.
(1981) - et al.
Recent stereological methods for cell biology: a brief survey
Am. J. Physiol.
(1990) - et al.
Tissue shrinkage and unbiased stereological estimation of particle number and size
J. Microsc.
(2001) - et al.
The effects of glycol methacrylate as a dehydrating agent on the dimensional changes of liver tissue
J. Microsc.
(1992)
Stereology of arbitrary particles. A review of unbiased number and size estimators and the presentation of some new ones, in memory of William R. Thompson
J. Microsc.
The smooth fractionator
J. Microsc.
The efficiency of systematic sampling in stereology and its prediction
J. Microsc.
Optimizing sampling efficiency of stereological studies in biology: or ‘do more less well!’
J. Microsc.
The new stereological tools: disector, fractionator, nucleator and point sampled intercepts and their use in pathological research and diagnosis
APMIS
Some new, simple and efficient stereological methods and their use in pathological research and diagnosis
APMIS
The efficiency of systematic sampling in stereology—reconsidered
J. Microsc.
An official research policy statement of the American Thoracic Society/European Respiratory Society: standards for quantitative assessment of lung structure
Am. J. Respir. Crit. Care Med.
Total number and mean size of alveoli in mammalian lung estimated using fractionator sampling and unbiased estimates of the Euler characteristic of alveolar openings
Anat. Rec. A: Discov. Mol. Cell Evol. Biol.
Alveoli increase in number but not size from birth to adulthood in rhesus monkeys
Am. J. Physiol. Lung Cell Mol. Physiol.
Microscopy-based quantitative analysis of lung structure: application in diagnosis
Expert Opin. Med. Diagn.
Ultrastructural changes of the intracellular surfactant pool in a rat model of lung transplantation-related events
Respir. Res.
Stereological estimates of alveolar number and size and capillary length and surface area in mice lungs
Anat. Rec. (Hoboken)
Sampling problems in an heterogeneous organ: quantitation of relative and total volume of pancreatic islets by light microscopy
J. Microsc.
Cited by (82)
A new method for physical disector analyses of numbers and mean volumes of immunohistochemically labeled cells in paraffin sections
2021, Journal of Neuroscience MethodsCitation Excerpt :Fractionator sampling designs are also not influenced by embedding/processing-related shrinkage of tissue samples and sections (Boyce et al., 2010; Howard and Reed, 2005; Slomianka, 2021; Gundersen, 2002). Nevertheless, an adequate determination of the volume of the examined tissue samples or organs is absolutely recommendable in any unbiased quantitative morphological study (Tschanz et al., 2014), since this information is essential for determination of the absolute volumes of the appropriate tissue reference compartment(s) and for estimation of dimensional quantitative morphological parameters, such as total and mean volumes, surfaces and lengths. For estimation of mean particle volumes, fractionator designs are often combined with additional quantitative stereological methods, such as the nucleator (Howard and Reed, 2005; Gundersen et al., 2013; Mayhew and Gundersen, 1996).
- 1
Biomedical Research in Endstage and Obstructive Lung Disease Hannover (BREATH), Member of the German Center for Lung Research.