Statistical analysis of adaptive response in sister chromatid exchanges in human lymphocytes after treatment with very low and extremely low doses of N-methyl-N'-nitro-N-nitrosoguanidine using a study design to control variability.

Previous studies have been interpreted as suggesting that low concentrations of N-methyl-N'-nitro-N-nitrosoguanidine (MNNG) have an adaptive effect in the cultured lymphocytes of responsive donors (that is, the cells are protected against the mutagenic effects of a subsequent challenge with a higher concentration of MNNG). The objectives of the present study were to investigate, under stringent experimental conditions, whether a protective effect exists at very low and extremely low doses of MNNG (10(-8) and 10(-24) M, respectively). Peripheral blood lymphocytes from a donor considered responsive in a previous study were stimulated to divide and were cultured under standard conditions. Pre-adaptive treatments with dilutions of MNNG were added to the cultures repeatedly before a challenge treatment with MNNG. Bromodeoxyuridine was added at the same time as the challenge treatment and, following mitotic arrest, cells were differentially stained so that the number of sister chromatid exchanges (SCEs) could be counted. The study was designed to address potential criticisms of earlier studies which did not include replicate cultures. Samples of blood were divided into two identical batches for independent processing. Five replicate cultures were prepared for each combination of pre-adaptive and challenge treatments in each batch. The complete experiment was repeated to provide a further test of the consistency of results. Five replicates per treatment combination were chosen in an attempt to provide an experiment of adequate statistical power. Considerable precautions were taken to minimise the effect of factors outside experimental control on the results. Scoring was done by three scorers. In order to minimise inter-scorer variation, 240 cells were scored at each treatment observation (five cells per-scorer, three scorers per culture, four cultures per batch, two batches per experiment and two experiments). The study was designed in this way to take account of the sources of variability to ensure that any response obtained would exceed that obtainable by experimental variability alone. A high level of quality assurance monitoring was undertaken throughout the investigation. Two measures of SCE induction were used: (i) the mean frequency of SCEs; (iii) proportion of cells with at least 20 SCEs. In both experiments, the challenge concentration of MNNG significantly increased SCE frequency. There were, however, highly significant differences between the two experiments. The proportion of high frequency cells (HFCs) in Experiment 1 was increased significantly; the proportion of HFCs was also increased in Experiment 2, but the increase was not statistically significant. The pre-adaptive concentrations of MNNG included an extremely low dilution of 6.8 x 10(-24) M and a very low dilution of 6.8 x 10(-8) M in Experiment 1 and 1.4 x 10(-7) M in Experiment 2. The various pre-adaptive concentrations used had no consistent protective effect against the SCE-inducing capacity of the challenge concentration of MNNG of 6.8 x 10(-6) M. It is concluded that an adaptive response to the alkylating agent MNNG could not be demonstrated in cultured human lymphocytes. Neither a very low nor an extremely low dilution of MNNG elicited an adaptive response in terms of SCE induction (measured either as SCE frequency or as proportion of HFCs). This is in contradiction to previous reports published by us and other groups. This study was carefully designed with large numbers of replicates, a preliminary statistical power calculation, predefined comparisons and extensive quality assurance at each treatment administration. Despite these precautions the variability between scorers and between batches was much larger than anticipated. This resulted in some statistically significant differences, but these are likely to be false positives. Our findings indicate the need for such methodological refinement in human cell adaptive response studies.