Sample size

The most basic method of normalization ensures that an experiment compares similar sample sizes and this is achieved by measuring tissue weight, volume or cell number. If no attempt to use similar sample sizes is made, unnecessary error may be introduced at the first stage of the assay. Furthermore, different extraction protocols may add to variability, especially as the extraction of RNA from differing amounts of tissue is usually subject to differing efficiencies. This is a particular problem when the extraction method becomes saturated and results in inefficient extraction of RNA from large samples. Unfortunately sample sizing is often not as simple as might be assumed. The examples below demonstrate that use of a similar sample size may still mean extracting RNA from highly dissimilar samples.

Figure 4.3

Colorectal cancer biopsy. Just over one third of the section is made up of tumour, the rest is adjacent normal colon. At least three cell types (epithelial, stromal and T cells) are clearly distinguishable.

Figure 4.4

Laser microdissected colorectal crypt. The crypt highlighted by an arrow in the first picture has been ablated in the second picture. The software measures the dimensions of the dissected sample and displays its area in |im2.

Longitudinal section of an epiphysis growth plate. Sample sizing comparisons between the larger nonproliferating cells with the proliferating cells are extremely difficult with this tissue. Any result must also consider the difference in the size of the active small cells and the less active larger cells.

Longitudinal section of an epiphysis growth plate. Sample sizing comparisons between the larger nonproliferating cells with the proliferating cells are extremely difficult with this tissue. Any result must also consider the difference in the size of the active small cells and the less active larger cells.

1. Sampling blood from HIV patients (Huggett et al., 2005) allows their stratification into groups based on their CD4+ T-cell numbers. Patients with >200 cells/ml may be otherwise healthy whereas those with counts of <200/ml are considered to have acquired immune deficiency syndrome (AIDS). Clearly, since AIDS patients have fewer CD4+ cells per volume of blood, normalizing cell mRNA copy numbers against blood volume alone would result in inaccurate and biologically meaningless quantification.

2. In vivo biopsies contain numerous cell types from different lineages and may contain different types of tissue, e.g. adjacent normal and cancer (Figure 4.3, color plate, between pages 38 and 39). This is resulting in a more widespread use of laser microdisssection and makes it feasible to normalize against the dissected area, allowing the reporting of copy numbers per area excised (Figure 4.4, color plate, between pages 38 and 39). Of course, one has to assume that the extraction efficiency and integrity of the RNAs extracted from different areas are similar, which may not be the case. It is acceptable to do this when comparing expression patterns from the same slide, as the tissue will have been subject to identical treatment. It is probably also acceptable to use this method for normalization when comparing different sections from the same tissue block. It remains to be seen whether it is acceptable to use this method for comparison between samples obtained from different tissue blocks.

3. Another problem when using similar sample sizes may occur when comparing samples that differ histologically. In situations where fibro-sis has occurred or if young tissue is being compared with older tissue the efficiency by which nucleic acid is extracted from the sample may be very different. If different nucleic acids amounts are extracted from the same amount of histologically different samples then this will lead to a directional shift and any measurement will be influenced by the decrease in extraction efficiency from one of the groups. Another problem relates to samples containing cell types that differ significantly in their respective sizes; again, this can interfere with accurate quantification (Figure 4.5, color plate, between pages 38 and 39).

4. Many of the problems associated with in vivo samples might be expected to be overcome in in vitro systems. However, culture can provide its own problems when sampling (Figure 4.6). Non-adherent cells that do not aggregate represent the ideal starting material however this scenario is the exception rather than the norm. Furthermore the cells may become phenotypically very different as a result of the variable being tested by the experiment. Cells can be treated with buffers and/or enzymes to cause them to detach or segregate facilitating counting, however these treatments are almost certainly going to effect gene expression.

Cleary, whilst it is essential to compare samples of similar size and composition, ensuring that this is actually the case is often more complicated than it may at first appear. Consequently, we stress the importance of careful consideration of sample selection and emphasize that additional methods of normalization must be used.

Figure 4.6

Normalization against cell number. It may be impossible to normalize against total cell number even when using tissue culture cells, as demonstrated by this neuronal cell line.

Figure 4.6

Normalization against cell number. It may be impossible to normalize against total cell number even when using tissue culture cells, as demonstrated by this neuronal cell line.

0 0

Post a comment