The issue of single sampling in pilot cellular studies.
In published studies of effects on cells it is customary to present data with the average of three values +/- SDs. This is particularly important where the SD is large or the assay not frequently published. In cases where the assay is well known however, and the cell type of interest well characterised, it can be a more realistic presentation of the exposure to offer one sample (subject to certain caveats) than multiple data collected from one culture.
This is because the counting procedure inevitably takes time, during which period the cells are degrading or changing in some other way (e.g. under the influence of light cells can change size, not to mention the effects of other environmental EM fields over time, notoriously difficult to exclude from the experimental set-up), and therefore the results are being confounded. In the case of exposure to electric or magnetic fields there is now a wealth of literature and a large number of protocols available to guide study design. It is clear from this literature that multiple cultures are unlikely to add much precision to the results gained from a single sample. Obviously single sampling has the advantage of lower cost (e.g. one third of the cost of three values) and by limiting the risk of extraneous exposures an arguably more accurate portrayal of the effect under study, and is therefore most of use for pilot studies where a broad range of investigations are underway. Single sampling can also often assist in the design of larger studies, and avoid costly, redundant, or needless procedures.
As one example of this take Moisescu, Pologea-Mararu et al.,
Effects of low intensity pulsed electric fields on the endocytosis of cells in culture
Persented at EBEA by Carol Davila Lab. (Bucharest) jointly with CNRS, Institut Gustave-Roussy, Villejuif, Paris
This study investigated why the exposure of cells in culture to low intensity pulsed electric fields increases the fluid phase endocytitic rate. One of these factors could be the resting transmembrane difference of potential. With the aim of defining if the resting potential transmembrane voltage of the cells plays a role in the field induced acceleration of the endocytosis, the cells were exposed to pulsed electric fields in mediums with different pH and various ionic compositions.
In the resulting data every column presented the average of three values. However the SDs were extremely small compared with their absolute values, and with small variances (variance is in any case simply the square of the SD). In that study single sampling would have produced the same results, and served to establish that the effect of the electric pulses is not modified by inoculation medium pH range changes in the range of pH between 6 and 8.
Whereas studies designed for peer review publication are best likely to achieve publication with multiple values, studies carried out for internal pilot reasons and for impecunious clients canl provide initial answers to research questions and assist materially in the design of larger studies where the results are promising. In the case of the work on the Personal Harmoniser the cellular studies were only a small part of the client’s investigations, and for which funding was limited. It was clearly stated in the conclusions to this study that a) the in vitro results presented were no guarantee that the effect would be observed in vivo, and b) that a clinical trial would be necessary to establish efficacy. Though the client has other research results available to him it is not for us to disclose those, only some of which we are privy to.
Schorr and Karten (1966) documented all articles in three issues of eleven of the most read medical journals from an intense biostatistical viewpoint, and found that of 149 papers only 28 were judged acceptable on their criteria, and nearly all the remainder were in need of biostatistical revisions. Nevertheless the large majority of these papers still added usefully to knowledge of their scientific area of interest.
Method of Trypan blue exclusion test for viability is a well tried and tested method of evaluating cellular viability, and lends itself to simple descriptive statistical analysis, since the cells are categorised simply on a yes/no basis, in that the simplest data consists of unordered dichotomous observations (0-1 scale). Using a standard hemacytometer and standardised cell concentrations it is easy to replicate such studies. The ratio of target population to total population is also easily derivable. The three major sources of error in cell studies are biological, temporal and measurement error. To avoid biological variation only one subject must be in the exposure conditions used for such investigations. To avoid temporal variation the count must be completed in a reasonably short time. To avoid measurement error only one observer should carry out the count. Since different counters often qualitatively differ, and lack standardisation. The need to restrict the counter however itself imposes a fatigue restriction on the data collection.
With the trypan test the likelihood measurement error is low, since it is not difficult to distinguish blue from white in the squares of equal area and volume seen with low magnification in a hemacytometer. According to the concentration chosen, moreover, the number of cells within each square is also quite controllable, so that there are not a too large number present for counting. These are some of the reasons why trypan blue exclusion offers a robust method amenable to statistical analysis. It is an assay which our laboratory has used for a number of years, and hence are comfortable with a single count approach in cases where the client’s primary objective is at a clinical not an in vitro level. Because of this we are comfortable with the reproducibility of the observations, their normal distributions in exposed and unexposed conditions, and through our control of the ambient EM environment (e.g. use of triple skinned mu metal containers, sterile conditions such as vertical laminar flow hoods, and the ability to control temperature to within 0.1degree C), the low likelihood of unusual occurrences.