- Email: [email protected]
The reproducibility of resting and post exercise plasma β-endorphins
Life Sciences:, Vol. 43, pp. 787-791 Printed in the U.S.A.
Pergamon Press
THE REPRODUCIBILITYOF RESTINGAND POST EXERCISE PLASMA t]-ENDORPHINS D. S. Shepsl, G. Koch2, E. E. Bragdon3, M. N. Ballenger3 and R. G. McMurray4 1Department of Medicine and 3Center for Environmental Medicine, School of Medicine, 2Department of Biostatistics, School of Public Health, and 4Exercise Physiology Laboratory, Department of Physical Education, University of North Carolina, Chapel Hill, North Carolina (Received in final form July 7, 1988)
Summarv This investigation examined the reproducibility of resting and post exercise plasma 13-endorphin levels. Twenty subjects (10 men and 10 women) had their resting endorphin levels measured under controlled conditions on four separate occasions. Concomitantly, the endorphin response of eight trained runners completing three similar ten mile runs was also determined. For the resting data, there was no significant overall variation among trials, but the intra-subject variability was substantial; the within subject variance was 6.16, and it corresponded to an intra-class reliability coefficient of r = 0.239. No gender effect was noted for the average I~-endorphin values for the four occasions (men = 4.6+1.7; women = 4.4+2.1 pM/I); however, the males' within-subject variance of 8.548 (r = 0.080) was significantly larger than that of 3.719 (r -- 0.485) for females. Of the runners, one outlier subject had a uniquely high average I~-endorphin level of 85.67. Analysis including and excluding the outlier subject yielded within-subject variances of 29.61 (r = 0.960) and 34.47 (r = 0.176), respectively; variances for differences in confidence limits for random variation, they must exceed 7 pM/I at rest, 17 pM/l post exercise, and 20 pM/L difference from rest to post exercise. In order to interpret properly the results of I]-endorphin studies, investigators must know how much day-to-day and patient variation can be expected in resting 13-endorphin levels as well as in various measures of response (e.g., absolute difference from resting level, percent difference from resting level) to a given stimulus, such as exercise testing. A review of the literature pertaining to I~-endorphins and exercise elicits a wide range of resting and post-stress plasma 13-endorphin values. Krieger et al. (1) found resting levels which fell between 2 and 15 pg/ml, while Wardlow and Frantz reported resting levels that averaged 21 _+ 7.3 pg/ml (2). The use of assay methods exhibiting considerable crossreactivity with I~-Iipotropin or other endorphins and enkephalins may account in part for such discrepancies. In addition, Dent et al. (3) have demonstrated that 13-endorphin levels tend to follow a diurnal cycle, peaking between 4:00 and 10:00 a.m. and ebbing during evening hours; hence, differences in time of day at which blood samples were drawn may also confound drawing comparisons between studies.
Please direct requests for reprints to Dr. Sheps, Division of Cardiology, 338 Clinical Sciences Building 229 H/CB# 7075, University of North Carolina School of Medicine, Chapel Hill, NC 27599. 0024-3205/88 $3.00 + .00 Copyright (c) 1988 Pergamon Press plc
788
Reproducibility of Plasma Endorphins
Vol. 43, No. 9, 1988
Kruger et al. (4) examined intra-individual variations in plasma 13-endorphins at rest and at 7 minutes and 3 hours after a 10 km run on two separate occasions. Their test-retest reliability revealed no correlation at rest (r = 0.02) and only a moderate correlation after the 10 km run (r = 0.56). The purpose of this study was to evaluate intra- and inter-individual changes in rest and post exercise I~-endorphin levels in normal subjects studied on more than two occasions. Methods Two phases of investigation were undertaken. Ten men and ten women took part in the first phase of research. Eight male runners with at least five years training, who were medically screened prior to testing, participated in the second phase. Informed consent was obtained from all subjects. Restina Studv. Measurement of resting plasma 13-endorphin level was repeated for the twenty subjects on four separate visits. For each visit, the participant reported to the laboratory at 8:00 a.m. After the subject's weight and height were recorded, he or she rested, covered by a thin blanket, on a cot in a supine position for 20 minutes. A catheter was inserted into an antecubital vein prior to the rest period and kept patent with a heparin lock. At the end of rest a 10 ml blood sample was obtained and a portion immediately analyzed for hematocrit. The remainder was centrifuged, and the ethylene diamine tetraacetic acid plasma was aspirated and immediately frozen at -20°C until assayed. All samples were assayed in duplicate for I~endorphin concentration, within 12 weeks of being obtained, using a radioimmunoassay kit (Immuno Nuclear Corporation). This radioimmunoassay procedure uses affinity gel extraction (5), which is an excellent method of concentrating 13-endorphin from other plasma components that nonspecifically bind 1251-13-endorphins. The recovery yield from plasma determined using internal standards ranged between 90% and 104%. The specificity of the antibody cross-reacts 100% with human ~-endorphin and shows less than 5% cross-reactivity with 13-1ipotropin and has no cross-reactivity with related peptides and hormones such as ~-endorphin, y-endorphin, leucine enkephalin, methionine enkephalin or ACTH. The sensitivity of the assay, defined as the apparent concentration at 3 standard deviations from the counts at maximum binding, is less than 3 pM/liter. The intra- and interassay coefficients of variation were 6.5% and 18.1%, respectively. Running Study. Each of the eight runners first completed a maximal oxygen uptake test using a running protocol; a speed and grade equal to 70% of the VO2 max was then extrapolated for use during three 10-mile runs. To undertake these runs, the subject reported to the lab at 8:00 a.m. on three separate occasions between one and three weeks apart. Resting data was obtained in the same manner as for phase one subjects. After the resting period, the subject on two occasions ran ten miles on a treadmill at the previously determined speed and grade. One of the three trials was run outdoors, with the first and last mile completed on a track to ensure proper pace. Oxygen uptake was measured during the tenth mile for all three trials, using open circuit spirometry techniques. The subject breathed through a Hans-Rudolph valve. Expired air from a mixing chamber was monitored for oxygen and carbon dioxide content using an Applied Electrochemistry 02 analyzer and a Beckman LB-2 CO2 Ventilization was measured using a Parkinson-Cowen dry gas meter. For the outdoor trial the subject's expired air was collected in a meteorological balloon and analyzed for volume and content using the above mentioned analyzers. Immediately following each run, a blood sample was obtained and a portion measured for hematocrit. The remainder was centrifuged and the plasma removed, frozen and assayed for 13-endorphin level by the method described above for phase one. Statistical Analysis. A subjects-by-trials repeated measures ANOVA was used to assess the variation due to subjects and to trials. The reliability of the measure was expressed in terms of within subject variance for trials and an intra-class reliability coefficient for the extent of variability which is not due to trials (6,7); that is, the intra-class reliability is the ratio of inter-subject variance versus the sum of inter-subject variance and intra-subject variance.
Vol. 43, No. 9, 1988
Reproducibility of Plasma Endorphins
789
TABLEI ~-Endorphin Levels for Resting Trials ~-endorohin Subiect
~x
levels
Trial 1
Trial 2
Trial 3
Trial 4
1
F
3.1
1.2
1.6
1.3
2
M
8.8
2.2
0.7
1.8
3
M
5.9
0.3
0.3
5.6
4
F
6.0
4.7
7.7
4.8
5
M
7.7
7.1
6.8
2.9
6
F
0.7
3.4
2.7
3.2
7
M
7.2
4.8
6.5
5.9
8
M
6.4
6.8
6.8
2.0
9
F
6.4
4.5
5.1
0.5
10
M
4.8
4.8
6.0
6.2
11
M
1.6
6.5
7.2
1.5
12
F
4.9
4.7
5.6
5.0
13
F
5.3
6.2
6.4
3.4
14
F
2.9
0.9
2.4
7.9
15
F
2.1
2.7
2.0
3.5
16
F
4.0
4.9
4.5
4.5
17
F
10.0
13.3
8.1
5.0
18
M
3.2
3,2
5.9
3.0
19
M
17.1
3,5
5.5
2.2
20
M
0.9
3.3
1.9
0.3
Results and Discussion Resting Study. Data from the resting trials are reported in Table I. The mean response for plasma 13-endorphin levels obtained from resting trials ranged from 5.5+3.7 pM/I for the first trial to 3.5+2.0 pM/I for the fourth trial, with individual values ranging from 0.3 to 17.1 pM/I (see Figure 1). Analysis of variance indicated that the trials did not differ significantly overall (p = 0.118). The calculated reliability coefficient was r = 0.239. Although the trials
790
Reproducibility of Plasma Endorphins
Vol. 43, No. 9, 1988
did not differ over all, significant inter-subject variability existed (p = 0.010). subject variance for trials was 6.16, which is indicative of substantial variability.
The within
No significant gender-related differences in mean plasma 13-endorphin levels were found either at the first trial (men = 6.3+4.6; women = 4.5 + 2.6 pM/I) or over all four trials (men = 4.6+1.7; women = 4.4+2.1 pM/I). However, the within subject variance of 8.548 for males was significantly larger than that of 3.719 for females. Similarly, the reliability coefficient for males (r = 0.080) was substantially smaller than that for females (r = 0.485). This suggests that intra-subject variability may be a greater concern in testing male than in testing female subjects. Runnip_a Study. The data from the running trials appear in Table II. The oxygen uptake and heart rate responses to all three 10 mile runs were similar (41.3 ml/kg/min and 157+5 beats/rain, respectively; p > 0.05). The three 10-mile runs resulted in mean plasma 13endorphin levels of 19.0+27.9, 21.8+27.7 and 16.5+27.1 pM/I. The ANOVA comparison of plasma 13-endorphin response following the three trials demonstrated no significant difference among trials (p = 0.190). The within subject variance was 29.61. Since there was one outlier subject with a uniquely high average value of 85.67, the reliability coefficient r -- 0.96 was unusually high. When this one subject was excluded from the analysis, the within subject variance was 34.47 and the reliability coefficient was a relatively low r = 0.176. Kruger has reported that difference between resting and postexercise I}-endorphin levels may be more reliable than absolute post-exercise values (4). For these differences in concentration from baseline, the within subject variance was 40.28 for the analysis which included all eight subjects and 46.45 for that which excluded the outlier subject; the respective reliability coefficients from these analyses were r = 0.927 and r = -0.131. As
TABLE II 13-Endorphin Levels for Running Trials ~-endorohin
levels
Absolute values Subiect
Chanae from rest values
Sex
Trial 1
Trial 2
Trial 3
1
M
85.0
89.0
2
M
5.3
3
M
4
Trial 1
Trial 2
83.0
69.3
73.3
66.3
14.1
5.0
0.8
9.2
0.8
4.5
4.7
4.5
1.0
1.9
0.5
M
6.9
10.5
12.5
0.8
4.4
8.6
5
M
29.0
11.6
11.1
20.5
0.6
6.3
6
M
4.0
8.4
3.1
3.3
2.5
0.4
7
M
12.0
13.2
4.0
0.5
9.1
9.5
8
M
5.6
22.6
8.8
4.0
21.9
1.1
*Outdoor trial
*
Trial 3 *
Vol. 43, No. 9, 1988
Reproducibility of Plasma Endorphins
791
indicated by these results, the presence of the outlier subject had a relatively small effect on the within subject variance, but a relatively large effect on the reliability coefficient. The reason for this is that the reliability coefficient is inherently sensitive to the representativeness of the subjects whereas the within subject variance only presumes representativeness of trials for subjects. Our findings demonstrate that intra-subject plasma 13-endorphin levels at rest and following exercise may be quite variable. It should be noted that the presence in our runners' sample of one individual with extreme values relative to the other study subjects may have rendered that sample unrepresentative of the larger population. In addition, prior conditioning may affect subjects' endorphin production; thus, responses of untrained subjects to exercise may not correspond to those reported for our runners. Extrapolation from our results suggests that if one wishes to consider changes for an individual subject in resting or post exercise 13-endorphin levels as being due to more than what would be encompassed by 95% limits for random variation, then one should observe a change of at least 7 pM/I at rest and a change of at least 17 pM/I at post-exercise. For the difference between resting and post exercise conditions, the change would need to be at least 20 pM/I to be beyond random variability. Alternatively, our findings can be used to specify necessary numbers of individuals to evaluate whether resting or post-exercise 13-endorphin levels change across two experimental conditions which are to be compared through a two period changeover design. For two-sided statistical significance at the a = 0.05 Type 1 error level and 0.80 power probability, a sample size of n = 25 subjects is sufficient to detect a change of >2 pM/I for resting 13-endorphin levels, a change of >5 pM/I for post-exercise 13-endorphin levels, and a change of _>6 pM/I for the difference between resting and post-exercise conditions. Acknowledaement This study was supported in part by grant 1-R01-HL38168 from the National Heart, Lung and Blood Institute, Bethesda, Maryland. References 1. 2. 3. 4. 5. 6. 7.
KRIEGER D.T., LIOTTA A.S., BROWNSTEIN M.J., ZIMMERMAN E.A., Rec. Prog. Hormone Res. 36 277-344 (1980). WARDLOW S.L., FRANTZ A.G., J. Clin. Endocrinol. Metab. 48 176-180 (1979). DENT R.P.M., GUILLEMINAULT C., ALBERT L.H., POSNER B.I., COX B.M., GOLDSTEIN A. J. Clin. Endocrinol. Metabo 52 942-947 (1981). KRUGER A., WlLDMANN J., SCHENK H., SCHMOLE M., NIEMANN J., MATTHAEI H., Ann. Sports Med. 2 111-116 (1985). BARON S.A., GINTZLER A.R., Brain Res. 321 341-346 (1984). FLEISS J.L., SLAKTER M.J., FISCHMAN S.L., PARK M.H., CHILTON N.W., J. Dent. Res. 58 604-609 (1979). THOMAS J.R., NELSON J.K., Introduction to Research in Health. Physical Education. Recreation and Dance. p. 260, Human Kinetics Publishers, Champaign IL (1985).
Pergamon Press
THE REPRODUCIBILITYOF RESTINGAND POST EXERCISE PLASMA t]-ENDORPHINS D. S. Shepsl, G. Koch2, E. E. Bragdon3, M. N. Ballenger3 and R. G. McMurray4 1Department of Medicine and 3Center for Environmental Medicine, School of Medicine, 2Department of Biostatistics, School of Public Health, and 4Exercise Physiology Laboratory, Department of Physical Education, University of North Carolina, Chapel Hill, North Carolina (Received in final form July 7, 1988)
Summarv This investigation examined the reproducibility of resting and post exercise plasma 13-endorphin levels. Twenty subjects (10 men and 10 women) had their resting endorphin levels measured under controlled conditions on four separate occasions. Concomitantly, the endorphin response of eight trained runners completing three similar ten mile runs was also determined. For the resting data, there was no significant overall variation among trials, but the intra-subject variability was substantial; the within subject variance was 6.16, and it corresponded to an intra-class reliability coefficient of r = 0.239. No gender effect was noted for the average I~-endorphin values for the four occasions (men = 4.6+1.7; women = 4.4+2.1 pM/I); however, the males' within-subject variance of 8.548 (r = 0.080) was significantly larger than that of 3.719 (r -- 0.485) for females. Of the runners, one outlier subject had a uniquely high average I~-endorphin level of 85.67. Analysis including and excluding the outlier subject yielded within-subject variances of 29.61 (r = 0.960) and 34.47 (r = 0.176), respectively; variances for differences in confidence limits for random variation, they must exceed 7 pM/I at rest, 17 pM/l post exercise, and 20 pM/L difference from rest to post exercise. In order to interpret properly the results of I]-endorphin studies, investigators must know how much day-to-day and patient variation can be expected in resting 13-endorphin levels as well as in various measures of response (e.g., absolute difference from resting level, percent difference from resting level) to a given stimulus, such as exercise testing. A review of the literature pertaining to I~-endorphins and exercise elicits a wide range of resting and post-stress plasma 13-endorphin values. Krieger et al. (1) found resting levels which fell between 2 and 15 pg/ml, while Wardlow and Frantz reported resting levels that averaged 21 _+ 7.3 pg/ml (2). The use of assay methods exhibiting considerable crossreactivity with I~-Iipotropin or other endorphins and enkephalins may account in part for such discrepancies. In addition, Dent et al. (3) have demonstrated that 13-endorphin levels tend to follow a diurnal cycle, peaking between 4:00 and 10:00 a.m. and ebbing during evening hours; hence, differences in time of day at which blood samples were drawn may also confound drawing comparisons between studies.
Please direct requests for reprints to Dr. Sheps, Division of Cardiology, 338 Clinical Sciences Building 229 H/CB# 7075, University of North Carolina School of Medicine, Chapel Hill, NC 27599. 0024-3205/88 $3.00 + .00 Copyright (c) 1988 Pergamon Press plc
788
Reproducibility of Plasma Endorphins
Vol. 43, No. 9, 1988
Kruger et al. (4) examined intra-individual variations in plasma 13-endorphins at rest and at 7 minutes and 3 hours after a 10 km run on two separate occasions. Their test-retest reliability revealed no correlation at rest (r = 0.02) and only a moderate correlation after the 10 km run (r = 0.56). The purpose of this study was to evaluate intra- and inter-individual changes in rest and post exercise I~-endorphin levels in normal subjects studied on more than two occasions. Methods Two phases of investigation were undertaken. Ten men and ten women took part in the first phase of research. Eight male runners with at least five years training, who were medically screened prior to testing, participated in the second phase. Informed consent was obtained from all subjects. Restina Studv. Measurement of resting plasma 13-endorphin level was repeated for the twenty subjects on four separate visits. For each visit, the participant reported to the laboratory at 8:00 a.m. After the subject's weight and height were recorded, he or she rested, covered by a thin blanket, on a cot in a supine position for 20 minutes. A catheter was inserted into an antecubital vein prior to the rest period and kept patent with a heparin lock. At the end of rest a 10 ml blood sample was obtained and a portion immediately analyzed for hematocrit. The remainder was centrifuged, and the ethylene diamine tetraacetic acid plasma was aspirated and immediately frozen at -20°C until assayed. All samples were assayed in duplicate for I~endorphin concentration, within 12 weeks of being obtained, using a radioimmunoassay kit (Immuno Nuclear Corporation). This radioimmunoassay procedure uses affinity gel extraction (5), which is an excellent method of concentrating 13-endorphin from other plasma components that nonspecifically bind 1251-13-endorphins. The recovery yield from plasma determined using internal standards ranged between 90% and 104%. The specificity of the antibody cross-reacts 100% with human ~-endorphin and shows less than 5% cross-reactivity with 13-1ipotropin and has no cross-reactivity with related peptides and hormones such as ~-endorphin, y-endorphin, leucine enkephalin, methionine enkephalin or ACTH. The sensitivity of the assay, defined as the apparent concentration at 3 standard deviations from the counts at maximum binding, is less than 3 pM/liter. The intra- and interassay coefficients of variation were 6.5% and 18.1%, respectively. Running Study. Each of the eight runners first completed a maximal oxygen uptake test using a running protocol; a speed and grade equal to 70% of the VO2 max was then extrapolated for use during three 10-mile runs. To undertake these runs, the subject reported to the lab at 8:00 a.m. on three separate occasions between one and three weeks apart. Resting data was obtained in the same manner as for phase one subjects. After the resting period, the subject on two occasions ran ten miles on a treadmill at the previously determined speed and grade. One of the three trials was run outdoors, with the first and last mile completed on a track to ensure proper pace. Oxygen uptake was measured during the tenth mile for all three trials, using open circuit spirometry techniques. The subject breathed through a Hans-Rudolph valve. Expired air from a mixing chamber was monitored for oxygen and carbon dioxide content using an Applied Electrochemistry 02 analyzer and a Beckman LB-2 CO2 Ventilization was measured using a Parkinson-Cowen dry gas meter. For the outdoor trial the subject's expired air was collected in a meteorological balloon and analyzed for volume and content using the above mentioned analyzers. Immediately following each run, a blood sample was obtained and a portion measured for hematocrit. The remainder was centrifuged and the plasma removed, frozen and assayed for 13-endorphin level by the method described above for phase one. Statistical Analysis. A subjects-by-trials repeated measures ANOVA was used to assess the variation due to subjects and to trials. The reliability of the measure was expressed in terms of within subject variance for trials and an intra-class reliability coefficient for the extent of variability which is not due to trials (6,7); that is, the intra-class reliability is the ratio of inter-subject variance versus the sum of inter-subject variance and intra-subject variance.
Vol. 43, No. 9, 1988
Reproducibility of Plasma Endorphins
789
TABLEI ~-Endorphin Levels for Resting Trials ~-endorohin Subiect
~x
levels
Trial 1
Trial 2
Trial 3
Trial 4
1
F
3.1
1.2
1.6
1.3
2
M
8.8
2.2
0.7
1.8
3
M
5.9
0.3
0.3
5.6
4
F
6.0
4.7
7.7
4.8
5
M
7.7
7.1
6.8
2.9
6
F
0.7
3.4
2.7
3.2
7
M
7.2
4.8
6.5
5.9
8
M
6.4
6.8
6.8
2.0
9
F
6.4
4.5
5.1
0.5
10
M
4.8
4.8
6.0
6.2
11
M
1.6
6.5
7.2
1.5
12
F
4.9
4.7
5.6
5.0
13
F
5.3
6.2
6.4
3.4
14
F
2.9
0.9
2.4
7.9
15
F
2.1
2.7
2.0
3.5
16
F
4.0
4.9
4.5
4.5
17
F
10.0
13.3
8.1
5.0
18
M
3.2
3,2
5.9
3.0
19
M
17.1
3,5
5.5
2.2
20
M
0.9
3.3
1.9
0.3
Results and Discussion Resting Study. Data from the resting trials are reported in Table I. The mean response for plasma 13-endorphin levels obtained from resting trials ranged from 5.5+3.7 pM/I for the first trial to 3.5+2.0 pM/I for the fourth trial, with individual values ranging from 0.3 to 17.1 pM/I (see Figure 1). Analysis of variance indicated that the trials did not differ significantly overall (p = 0.118). The calculated reliability coefficient was r = 0.239. Although the trials
790
Reproducibility of Plasma Endorphins
Vol. 43, No. 9, 1988
did not differ over all, significant inter-subject variability existed (p = 0.010). subject variance for trials was 6.16, which is indicative of substantial variability.
The within
No significant gender-related differences in mean plasma 13-endorphin levels were found either at the first trial (men = 6.3+4.6; women = 4.5 + 2.6 pM/I) or over all four trials (men = 4.6+1.7; women = 4.4+2.1 pM/I). However, the within subject variance of 8.548 for males was significantly larger than that of 3.719 for females. Similarly, the reliability coefficient for males (r = 0.080) was substantially smaller than that for females (r = 0.485). This suggests that intra-subject variability may be a greater concern in testing male than in testing female subjects. Runnip_a Study. The data from the running trials appear in Table II. The oxygen uptake and heart rate responses to all three 10 mile runs were similar (41.3 ml/kg/min and 157+5 beats/rain, respectively; p > 0.05). The three 10-mile runs resulted in mean plasma 13endorphin levels of 19.0+27.9, 21.8+27.7 and 16.5+27.1 pM/I. The ANOVA comparison of plasma 13-endorphin response following the three trials demonstrated no significant difference among trials (p = 0.190). The within subject variance was 29.61. Since there was one outlier subject with a uniquely high average value of 85.67, the reliability coefficient r -- 0.96 was unusually high. When this one subject was excluded from the analysis, the within subject variance was 34.47 and the reliability coefficient was a relatively low r = 0.176. Kruger has reported that difference between resting and postexercise I}-endorphin levels may be more reliable than absolute post-exercise values (4). For these differences in concentration from baseline, the within subject variance was 40.28 for the analysis which included all eight subjects and 46.45 for that which excluded the outlier subject; the respective reliability coefficients from these analyses were r = 0.927 and r = -0.131. As
TABLE II 13-Endorphin Levels for Running Trials ~-endorohin
levels
Absolute values Subiect
Chanae from rest values
Sex
Trial 1
Trial 2
Trial 3
1
M
85.0
89.0
2
M
5.3
3
M
4
Trial 1
Trial 2
83.0
69.3
73.3
66.3
14.1
5.0
0.8
9.2
0.8
4.5
4.7
4.5
1.0
1.9
0.5
M
6.9
10.5
12.5
0.8
4.4
8.6
5
M
29.0
11.6
11.1
20.5
0.6
6.3
6
M
4.0
8.4
3.1
3.3
2.5
0.4
7
M
12.0
13.2
4.0
0.5
9.1
9.5
8
M
5.6
22.6
8.8
4.0
21.9
1.1
*Outdoor trial
*
Trial 3 *
Vol. 43, No. 9, 1988
Reproducibility of Plasma Endorphins
791
indicated by these results, the presence of the outlier subject had a relatively small effect on the within subject variance, but a relatively large effect on the reliability coefficient. The reason for this is that the reliability coefficient is inherently sensitive to the representativeness of the subjects whereas the within subject variance only presumes representativeness of trials for subjects. Our findings demonstrate that intra-subject plasma 13-endorphin levels at rest and following exercise may be quite variable. It should be noted that the presence in our runners' sample of one individual with extreme values relative to the other study subjects may have rendered that sample unrepresentative of the larger population. In addition, prior conditioning may affect subjects' endorphin production; thus, responses of untrained subjects to exercise may not correspond to those reported for our runners. Extrapolation from our results suggests that if one wishes to consider changes for an individual subject in resting or post exercise 13-endorphin levels as being due to more than what would be encompassed by 95% limits for random variation, then one should observe a change of at least 7 pM/I at rest and a change of at least 17 pM/I at post-exercise. For the difference between resting and post exercise conditions, the change would need to be at least 20 pM/I to be beyond random variability. Alternatively, our findings can be used to specify necessary numbers of individuals to evaluate whether resting or post-exercise 13-endorphin levels change across two experimental conditions which are to be compared through a two period changeover design. For two-sided statistical significance at the a = 0.05 Type 1 error level and 0.80 power probability, a sample size of n = 25 subjects is sufficient to detect a change of >2 pM/I for resting 13-endorphin levels, a change of >5 pM/I for post-exercise 13-endorphin levels, and a change of _>6 pM/I for the difference between resting and post-exercise conditions. Acknowledaement This study was supported in part by grant 1-R01-HL38168 from the National Heart, Lung and Blood Institute, Bethesda, Maryland. References 1. 2. 3. 4. 5. 6. 7.
KRIEGER D.T., LIOTTA A.S., BROWNSTEIN M.J., ZIMMERMAN E.A., Rec. Prog. Hormone Res. 36 277-344 (1980). WARDLOW S.L., FRANTZ A.G., J. Clin. Endocrinol. Metab. 48 176-180 (1979). DENT R.P.M., GUILLEMINAULT C., ALBERT L.H., POSNER B.I., COX B.M., GOLDSTEIN A. J. Clin. Endocrinol. Metabo 52 942-947 (1981). KRUGER A., WlLDMANN J., SCHENK H., SCHMOLE M., NIEMANN J., MATTHAEI H., Ann. Sports Med. 2 111-116 (1985). BARON S.A., GINTZLER A.R., Brain Res. 321 341-346 (1984). FLEISS J.L., SLAKTER M.J., FISCHMAN S.L., PARK M.H., CHILTON N.W., J. Dent. Res. 58 604-609 (1979). THOMAS J.R., NELSON J.K., Introduction to Research in Health. Physical Education. Recreation and Dance. p. 260, Human Kinetics Publishers, Champaign IL (1985).