**3. Methodological meaning of statistical analysis in modern biomedical investigations**

Usual desire of physician to find unique for a certain patient individual diagnosis and treatment become to be an advanced direction of the modern probative medicine. Some statistical algorithms appeared which substantiate individual conclusions made earlier

intuitively. The latter usually expelled these conclusions from the area of science, although this does not correspond to the truth. Misunderstanding between practical physicians and experts in statistics is based on the mistakes in applications of statistical methods in biomedical investigations.

Study of Succinate Dehydrogenase and α-Ketoglutarate Dehydrogenase in Mitochondria Inside Glass-Adhered Lymphocytes Under Physiological Conditions – The Two Dehydrogenases as Counterparts of … 241

described, convenient even in examination of people for which the selection of conditions is limited compared with animals. We can add to them examples from our experience. A pair comparison of results in one person in contrast to group comparison belongs to these protocols. This list is continued by: performance of measurements at maximal possible identical time. Comparison with healthy control, taken on the same day and at the same hour, or at least for four seasons separately, for which the difference of reactions is known. Performance of a comparison with controls, taken in the same laboratory simultaneously with examined group instead of using averaged data for norm, which increases variations.

**Figure 3.** Changes in the activity of SDH and KDH and ROS formation in glass-adhered blood lymphocytes in rats during short-term painless psychoemotional stress (PES) in a box.

Owing to a great amount of objects (200-800) the presented values are statistically significant at the level

Data are presented as a total area (S) of formazan granules in 30 cells.

of 0.001 (99.99%).

Very interesting and useful monograph by Thomas A. Lang and Mishelle Secic: "How to report Statistics in Medicine. Annotated Guidelines for Authors, Editors, and Reviewers" [46] has the purpose to fill the gap between mathematicians and physicians. The chapter: "The difference between clinical and statistical significance" substantiates the importance of a single observation, for example, life-saving of one person, in spite of its statistical unreliability.

In this respect the statistical validity of CBCh method for individual measurement is of importance and will be briefly considered below. We share also the idea of the authors that statistical analysis in biomedical investigations should be actively used for design of experiment but not only for evaluation of the results. We can illustrate this by our experience.

1. Statistical treatment

Besides its crucial advantage in preservation of native structure organization of mitochondria CBCh method possesses one more important property. This is high statistical validity of a single result because it is obtained by computer quantitation of a multitude of microscopic images, which measure enzyme activity. This kind of calculation finds an ever increasing application in biomedical studies. In the case of dehydrogenase activity determination by CBCh method, both in rats and in people, even in a small part of experimental samples with minimal enzyme activity, the number of objects measured was about a hundred. In most cases it was 300 - 500 images, while in a considerable part of samples with maximal activity, the number of images was more than 700. According to the standard determination of statistical significance, it is practically maximal, as judged by measurements even a hundred of images. Therefore all the data obtained in our work by computer quantitation are statistically significant at the level of 0.001 (99.99%).

We did statistical analysis for our data and illustrate this result in Fig. 3 by putting in the bars presenting ± S.D. As seen, S.D. is negligible in contrast to non-computer treatment which deals with a considerably smaller number of measurements. Therefore, we believe it possible do not overload all the figures by designation of minor bars after preliminary explanation in Methods for the presented data sampling. Such description may be convenient for presentation of other data obtained by computer quantitation.

In case of more than single individual measurement, such as data presented in Fig.8, we indicate ± S.D. in diagrams.

2. Design of experiment for the purpose to diminish "uncontrolled" statistical variations of data.

The preliminary design directed to diminishing of "uncontrolled" statistical variations of data was considered in the Guidance cited and in the valuable monograph by Glanz [47]. In these handbooks, protocols of comparison of results of the action of different substances are

biomedical investigations.

1. Statistical treatment

indicate ± S.D. in diagrams.

of data.

intuitively. The latter usually expelled these conclusions from the area of science, although this does not correspond to the truth. Misunderstanding between practical physicians and experts in statistics is based on the mistakes in applications of statistical methods in

Very interesting and useful monograph by Thomas A. Lang and Mishelle Secic: "How to report Statistics in Medicine. Annotated Guidelines for Authors, Editors, and Reviewers" [46] has the purpose to fill the gap between mathematicians and physicians. The chapter: "The difference between clinical and statistical significance" substantiates the importance of a single observation, for example, life-saving of one person, in spite of its statistical unreliability.

In this respect the statistical validity of CBCh method for individual measurement is of importance and will be briefly considered below. We share also the idea of the authors that statistical analysis in biomedical investigations should be actively used for design of experiment but not only for evaluation of the results. We can illustrate this by our experience.

Besides its crucial advantage in preservation of native structure organization of mitochondria CBCh method possesses one more important property. This is high statistical validity of a single result because it is obtained by computer quantitation of a multitude of microscopic images, which measure enzyme activity. This kind of calculation finds an ever increasing application in biomedical studies. In the case of dehydrogenase activity determination by CBCh method, both in rats and in people, even in a small part of experimental samples with minimal enzyme activity, the number of objects measured was about a hundred. In most cases it was 300 - 500 images, while in a considerable part of samples with maximal activity, the number of images was more than 700. According to the standard determination of statistical significance, it is practically maximal, as judged by measurements even a hundred of images. Therefore all the data obtained in our work by

We did statistical analysis for our data and illustrate this result in Fig. 3 by putting in the bars presenting ± S.D. As seen, S.D. is negligible in contrast to non-computer treatment which deals with a considerably smaller number of measurements. Therefore, we believe it possible do not overload all the figures by designation of minor bars after preliminary explanation in Methods for the presented data sampling. Such description may be

In case of more than single individual measurement, such as data presented in Fig.8, we

2. Design of experiment for the purpose to diminish "uncontrolled" statistical variations

The preliminary design directed to diminishing of "uncontrolled" statistical variations of data was considered in the Guidance cited and in the valuable monograph by Glanz [47]. In these handbooks, protocols of comparison of results of the action of different substances are

computer quantitation are statistically significant at the level of 0.001 (99.99%).

convenient for presentation of other data obtained by computer quantitation.

described, convenient even in examination of people for which the selection of conditions is limited compared with animals. We can add to them examples from our experience. A pair comparison of results in one person in contrast to group comparison belongs to these protocols. This list is continued by: performance of measurements at maximal possible identical time. Comparison with healthy control, taken on the same day and at the same hour, or at least for four seasons separately, for which the difference of reactions is known. Performance of a comparison with controls, taken in the same laboratory simultaneously with examined group instead of using averaged data for norm, which increases variations.

**Figure 3.** Changes in the activity of SDH and KDH and ROS formation in glass-adhered blood lymphocytes in rats during short-term painless psychoemotional stress (PES) in a box. Data are presented as a total area (S) of formazan granules in 30 cells. Owing to a great amount of objects (200-800) the presented values are statistically significant at the level

of 0.001 (99.99%).

In connection with data in animal models, described in this paper, we shall consider in more detail the protocol of the selection of really statistical homogenous groups of animals, which was elaborated in our laboratory. The necessity to form statistically homogenous groups of animals for application of statistical analysis for treatment of results is well known, although often it is practically ignored. We found that, in the selection of animals assumed to be similar many factors still exist, that cause uncontrolled variations of the state of animals and preparations obtained. We select animals bred in the vivarium of our Institute within 1-3 days after birth. We do not use for one group animals from only one brood because it is known from physiological data and found by the CBCh method that at least three subgroups can be differentiated in the one brood according to physiological activity [48]. These subgroups correspond to weight distribution. For standard investigations we take animals with greater weight, close to each other on the day of selection - 4 weeks, the time of transfer of young rats to dry feed and taking off from mother milk suckling (100-110 g). It was shown in the special experiments that the prolongation of suckling to 6 weeks considerably improves CBCh data and weight increase as compared with 160-180 g, when suckling was stopped at 4 weeks. The differences are found also in animals at the age of 6 week, before maturation and at the age of 8 week, just after maturation (190-210 g). According to physiological data the differences between these ages are due not only to maturation but also to domination of adrenergic regulation before maturation and cholinergic regulation after maturation. We found manifestation of these physiological phenomena by the ratio of SDH and KDH activities. For standard investigations we have chosen the age of 8 weeks, however other ages can be used but in separate groups. Usually these details are not considered in experiments, which lead to the increase in variations of the results.

Study of Succinate Dehydrogenase and α-Ketoglutarate Dehydrogenase in Mitochondria Inside Glass-Adhered Lymphocytes Under Physiological Conditions – The Two Dehydrogenases as Counterparts of … 243

animals (upper panel) completely agree with results in the other group (lower panel), selected in different time. In both cases we observe maximal activation of dehydrogenases in the initial stage of stress - alarm reaction at 30 and 60 min, their further decline approaching the initial level to 120 min, however in the both cases with a small excess, which is considered in the description. Prolongation of stress to 180 min, presented at lower panel, shows increase in inhibition in time. On the reasons considered for upper panel we did not show minor bars in the lower panel as well as in the further similar figures in the text. However the number of images measured and statistical significance are given in legend.

The attention to the creative search of causes of result variations and to the real possibilities of their diminishing will improve efficiency of studies. This affords making smaller number of experiments with more definite results. Such approach presents more deep penetration of

**4. Demonstration of high physiological sensitivity of the CBCh method** 

**distinct borderline between intensive functioning and early dysfunction** 

Over a long initial period, the biochemical mechanisms of stress were studied under action of strong chemical or physical stressors, which are beyond the range of physiological adaptation. In spite of this, the responses of isolated mitochondria studied under pain immobilization stress for a long time, 24 h, were not pronounced, accounting for no more than 20% inhibition of succinate oxidation measured by the polarographic method [11, 50, 51]. The considerably higher sensitivity of the CBCh assay to changes in the physiological state is evident, as it reveals a great-range difference of SDH activity, 2-3 times and more, under the influence of much milder stressors compatible with life, such as the administration of ADR in submaximal dose or painless short-time restraint of rat in a narrow box for 30-120 min, psychoemotional stress (PES). A comparison of mitochondrial responses observed in isolated mitochondria and in mitochondria inside lymphocytes is convincing, considering that the state of animals after strong or physiological stress differed greatly. After strong stress, animals were exhausted and could not move, while animals

The changes in the SDH and KDH activities after the administration of ADR are presented in the upper panel of Figure 4. As shown, these activities are moderate and nearly equal in the initial quiescent state. Under the action of ADR, the initial acute phase of stress, an alarm reaction arises. It is clearly manifested by a pronounced (more than twofold!) increase in the SDH activity, and by a fall (about twentyfold!) in the KDH activity. Under 24-h restraint in a box, a stress phase of exhaustion develops. It is also clearly pronounced by a decrease in the SDH activity approximately to the initial level. It is very important that this decline does not mean a restoration of the initial state because the KDH activity nearly disappears. Therefore, the dynamics of SDH activity develops according to a bell-shaped curve with a maximum at early stages of stress. This corresponds to the classic stress dynamics followed by

**in models of increased adrenergic regulation. SDH/KDH ratio as a** 

statistical analysis into biomedical investigations.

after short-term stress looked like intact.

physiological data.

Besides careful choice of animals, maintenance of conditions of treatment of isolated tissue preparation is of importance. Usual measurement of data in so called "parallel" samples for analysis taken from liquid preparation: suspension of mitochondria, tissue homogenate or liquid blood is indeed not simultaneous but subsequent in time. It is carried out subsequently, often during prolonged time of storage, while preparation is changed, which gives variations of results. These variations reflect not only changes in the state of the preparation itself but also its response to the pronounced changes in time of cosmophysical environment. This problem is fundamentally investigated in many year studies of S. Shnoll [49]. Many investigators confess that they meet such variations when make repeated measurements. However, observed deviations seem to be obscure and are neglected. Our method of preparation of fixed smears, which are more stable in time, allows one to carry out analysis much more close to simultaneous. Elimination of the considered sources of variations affords the possibility to observe practically completely identical results for different animals in the same physiological state within the selected group of 6-8 species. Such example is given at the upper panel of Figure 3. As shown under 120 min PES CBCh data for two animals are identical. Statistical probability of data for individual organism, considered in paragraph 1, allows one to compare the influence of investigated factors on individual animals in the same group. Comparison the results obtained in one group of

the results.

In connection with data in animal models, described in this paper, we shall consider in more detail the protocol of the selection of really statistical homogenous groups of animals, which was elaborated in our laboratory. The necessity to form statistically homogenous groups of animals for application of statistical analysis for treatment of results is well known, although often it is practically ignored. We found that, in the selection of animals assumed to be similar many factors still exist, that cause uncontrolled variations of the state of animals and preparations obtained. We select animals bred in the vivarium of our Institute within 1-3 days after birth. We do not use for one group animals from only one brood because it is known from physiological data and found by the CBCh method that at least three subgroups can be differentiated in the one brood according to physiological activity [48]. These subgroups correspond to weight distribution. For standard investigations we take animals with greater weight, close to each other on the day of selection - 4 weeks, the time of transfer of young rats to dry feed and taking off from mother milk suckling (100-110 g). It was shown in the special experiments that the prolongation of suckling to 6 weeks considerably improves CBCh data and weight increase as compared with 160-180 g, when suckling was stopped at 4 weeks. The differences are found also in animals at the age of 6 week, before maturation and at the age of 8 week, just after maturation (190-210 g). According to physiological data the differences between these ages are due not only to maturation but also to domination of adrenergic regulation before maturation and cholinergic regulation after maturation. We found manifestation of these physiological phenomena by the ratio of SDH and KDH activities. For standard investigations we have chosen the age of 8 weeks, however other ages can be used but in separate groups. Usually these details are not considered in experiments, which lead to the increase in variations of

Besides careful choice of animals, maintenance of conditions of treatment of isolated tissue preparation is of importance. Usual measurement of data in so called "parallel" samples for analysis taken from liquid preparation: suspension of mitochondria, tissue homogenate or liquid blood is indeed not simultaneous but subsequent in time. It is carried out subsequently, often during prolonged time of storage, while preparation is changed, which gives variations of results. These variations reflect not only changes in the state of the preparation itself but also its response to the pronounced changes in time of cosmophysical environment. This problem is fundamentally investigated in many year studies of S. Shnoll [49]. Many investigators confess that they meet such variations when make repeated measurements. However, observed deviations seem to be obscure and are neglected. Our method of preparation of fixed smears, which are more stable in time, allows one to carry out analysis much more close to simultaneous. Elimination of the considered sources of variations affords the possibility to observe practically completely identical results for different animals in the same physiological state within the selected group of 6-8 species. Such example is given at the upper panel of Figure 3. As shown under 120 min PES CBCh data for two animals are identical. Statistical probability of data for individual organism, considered in paragraph 1, allows one to compare the influence of investigated factors on individual animals in the same group. Comparison the results obtained in one group of animals (upper panel) completely agree with results in the other group (lower panel), selected in different time. In both cases we observe maximal activation of dehydrogenases in the initial stage of stress - alarm reaction at 30 and 60 min, their further decline approaching the initial level to 120 min, however in the both cases with a small excess, which is considered in the description. Prolongation of stress to 180 min, presented at lower panel, shows increase in inhibition in time. On the reasons considered for upper panel we did not show minor bars in the lower panel as well as in the further similar figures in the text. However the number of images measured and statistical significance are given in legend.

The attention to the creative search of causes of result variations and to the real possibilities of their diminishing will improve efficiency of studies. This affords making smaller number of experiments with more definite results. Such approach presents more deep penetration of statistical analysis into biomedical investigations.
