Seasonal Independence of Low Prolactin Concentration and High Spontaneous Eye Blink Rates in Unipolar and Bipolar II Seasonal Affective Disorder Richard A.
Depue, PhD; Paul Arbisi; Steven Krauss, PhD; William G. Iacono, PhD; Arthur Leon, MD; Rachel Muir; John Allen
\s=b\ Twenty-four subjects with seasonal affective disorder (SAD: bipolar II, n 14; unipolar, n 10) and 20 normal controls were assessed for early follicular basal serum prolactin (PRL) concen=
=
tration in winter and summer. Luteal basal PRL concentration was assessed in winter. The PRL values represented the mean of three values derived during a 45-minute period. A subset of 17 subjects with SAD and 11 controls were also assessed for spontaneous eye blinking via a polygraphic recording in winter and summer. In winter, compared with controls, subjects with SAD were characterized by significantly lower follicular (10.1 vs 4.5 \g=m\g/L,respectively)and luteal (14.4 vs 7.4 \g=m\g/L,respectively) PRL values and by significantly higher eye blink rates (30 vs 61 blinks per 3 minutes, respectively). In summer, controls and subjects with SAD showed similar significant differences in follicular PRL values (9.3 vs 3.9 \g=m\g/L,respectively) and eye blink rates (25 vs 67 blinks per 3 minutes, respectively). No significant differences in PRL values or eye blink rates were found between the bipolar II and unipolar forms of SAD in either season. Results were discussed in terms of dopamine functioning. (Arch Gen Psychiatry. 1990;47:356-364)
the of phototherapy for seasonal affective disorder has been actively researched during the the search for variables that characterize SAD 5 has received much less attention. A review of published and unpublished work on biologic variables3 in SAD re¬ vealed that consistent trends have yet to emerge. Relative to controls, cortisol values were found to be significantly in¬ creased, and prolactin (PRL) values were found to be signifi¬ cantly decreased in SAD when stimulated by 5-hydroxytryptophan,4 whereas an unpublished study3 observed signifi¬ cantly increased PRL values when stimulated by a serotonin agonist. Basal levels of PRL have been found to be both increased4 and decreased.3,5,6 Dexamethasone suppression and 24-hour secretion of cortisol, thyrotropin (thyroid-stimulating hormone) responsivity, and cerebrospinal fluid metabolite lev¬ els of serotonin, norepinephrine, and dopamine were all nor¬ mal.3,7 Lewy et al8 demonstrated that SAD is characterized by a significant delay of the onset of nighttime melatonin secre-
efficacy (SAD) Although past years,1,2
Accepted for publication September 19,1989. From the Department of Psychology (Drs Depue, Krauss, and Iacono and Messrs Arbisi and Allen and Ms Muir), the Division of Epidemiology, School of Public Health (Dr Leon), and the Department of Physiology, Medical School (Dr Leon), University of Minnesota, Minneapolis. Reprint requests to Department of Psychology, N218 Elliott Hall, University of Minnesota, 75 E River Rd, Minneapolis, MN 55455 (Dr Depue).
tion and that
morning light exposure advanced this onset in proportion to the antidepressant effects in some patients. Other work, however, suggests that this finding may not gen¬ eralize to all subjects with SAD.9"11 Currently, then, there are few positive findings in SAD that provide a framework for further research. Where positive results have been reported, findings are either balanced by negative results or have yet to be replicated. Moreover, we are aware of no studies that have systematically assessed whether the unipolar (UP) and bipolar (BP) II forms of SAD, observed
in the clinical literature,12,13 represent a valid distinction when tested against external criteria. It is within this framework of inconsistency that we pursue our initial positive findings in SAD. In several initial studies, we observed significant differences between subjects with SAD and controls in the thermoregulatory heat loss réponse to a thermal challenge,14 spontaneous eye blink rates,15 and the basal follicular PRL concentration.56 (These variables were studied because of their common link to dopamine [DA] mech¬ anisms, as described in the "Comment" section.) The latter two variables revealed particularly marked differences. Pa¬ tients with SAD showed a significantly lower basal follicular PRL concentration (less than half of the control values) in winter, and the PRL values remained equally low after suc¬ cessful phototherapy5 and on retest in summer6 when patients were euthymic. Low basal PRL values in SAD were also reported in one unpublished study3 but not in another report.4 We also observed spontaneous eye blink rates in SAD that exceeded the rate ofcontrols by a factor of two.15 Both the PRL and eye blink rate findings were based on relatively small samples of subjects with SAD that were composed solely of the BP II form of SAD. Therefore, the current report represents (1) a successful replication of the PRL and eye blink rate findings on a much larger sample of subjects with SAD, (2) a demonstration of the seasonal independence of both of these variables in SAD, and (3) the first comparison of subjects with UP and BP II SAD on PRL and eye blink variables. SUBJECTS AND METHODS
Subjects The participants in the study were premenopausal females (24 with SAD and 20 normal controls). Subjects were obtained by media advertisements, and all respondents were then screened by the use of a modified Schedule for Affective Disorders and Schizophrenia-Life¬ time version16 interview, a process that represents the typical sub¬ ject-selection procedure in outpatient studies of SAD.1,4,8,11,2 All sub-
Downloaded From: http://archpsyc.jamanetwork.com/ by a University of Missouri - St Louis User on 09/30/2013
jects with SAD met DSM-III-R" criteria for a major depression that developed during the fall, was continuous during the winter, and remitted the following spring—for at least the last 3 consecutive years; subjects with SAD also met the newly developed DSM-III-R criteria for a seasonal pattern to the major depression. The winter depressions were not in reaction to seasonally varying social Stress¬ ors. Subjects with SAD also had to be free of depressed periods during
the last three consecutive summers. Self-reports of 14 of the subjects with SAD, concerning their behav¬ ior during summer months, satisfied criteria for hypomanic episodes that were included in the DSM-III-R criteria for cyclothymic disorder (but never met DSM-III-R criteria for a manic episode), where each of these subjects reported that she typically had three or more such episodes during summer months. Therefore, these 14 subjects had the BP II form of SAD that has been observed by others.12,13 The 10 subjects with SAD who did not meet criteria for hypomanic epi¬ sodes had diagnoses of UP SAD, a form of SAD also observed by others.8,12,13 No other axis I psychiatric disorder existed among sub¬ jects with SAD. Controls were free of psychiatric disorder as deter¬ mined by the same structured interview. The values for indepen¬ dently assigned diagnoses of interest to the study on this sample, determined by a dual diagnosis from audiotapes, were as follows: major depression (.90), seasonal pattern (.96), and hypomania (.78), ie, values that were consistent for our group.6,1,15,1 Potential subjects (those with SAD and controls) who reported postmenopausal status, pregnancy, use of birth control pills during the previous 4 months, any prescribed medication during the past 6 months, any form of substance abuse, menstrual cycle irregularities or significant mood and physical symptoms suggestive of premenstru¬ al syndrome, endocrinopathies or other relevant medical conditions, ie, an eating disorder, oculomotor disorders, or epilepsy were ex¬ cluded from study. None of the subjects had used phototherapy previously nor directly before the study. PRL Assessment The basal follicular serum PRL value was assessed in winter be¬ tween January 2 and 31 and in summer between June 28 and August 10. The serum PRL concentration was assessed by three separate blood draws during a 45-minute period by averaging the value of the three blood samples, each of which was separated by a 15-minute interval. The timing of the PRL sampling was strictly controlled: (1) blood samples were collected in the early to middle follicular phase of the menstrual cycle, where all samples for all subjects were ob¬ tained between 2 and 7 days into the menstrual cycle (the menstrual phase was determined by subject self-reports that were obtained throughout three consecutive menstrual cycles, including the cycle that was assessed in the study); and (2) to control for the time of day (diurnal variation), and because SAD is associated with a marked lateafternoon blood samples were collected at 3:30, 3:45, and 4 PM for each subject. Subjects with SAD and controls were also assessed for PRL concen¬ tration in exactly the same manner 2 weeks later in winter (between January 15 and February 17) during the middle to late luteal phase of the menstrual cycle, ie, between 19 and 25 days into the menstrual
worsening,5,13
cycle.
graph (Beckman
R612 Dynograph) in a sound-deadened, dimly lit Silver-silver-chloride electrodes were attached at the superior and inferior orbital rims of the left eye to record blinks, and to the outer canthi for the horizontal electro-oculogram recording. A ground electrode was attached to the right earlobe. Alternating current coupling with a 3-second time constant was used. The electrodes were also attached to measure palmar electrodermal, heart rate, and electroencephalographic variables, to be reported elsewhere. An eye blink was operationally defined as a sharp, high-amplitude wave from the immediately preceding baseline equal to or greater than 100 µ and less than 400 milliseconds in duration (approximately =sl mm in width, measured at a vertical level that is the approximate midpoint ofthe blink amplitude on the baseline side ofthe blink, where chart speed 2.5 mm/s). This definition was consistent with, but more stringent than, that found in previous human blink rate studies.19 High-amplitude waves coincident with horizontal eye movements were not counted as blinks. The counting of eye blinks was performed, blind of the subject's group identity. Two raters independently counted eye blinks on two thirds of the subjects with SAD and controls and achieved an interrater agreement (r =. 96) similar to that found in room.
=
previous studies.19 The timing of eye blink assessments was strictly controlled, exactly as in the PRL assessment: (1) to control diurnal factors, blink record¬ ings occurred between 3 and 6 PM (but not on the same day as the PRL assessment); and (2) to control for the phase of the menstrual cycle, blink assessment occurred during the early to middle follicular period (days 2 through 7). A 10-minute period of adaptation to the recording room was pro¬ vided. Eye blinks were recorded for a 3-minute period, while the subject sat silent with the lights on. Subjects were told to keep their eyes open but were not instructed in any manner about blinking. Subjects did not wear contact lenses during the recordings, did not have acute eye infections, and were instructed not to sleep (which was verified in chart recordings). The dependent variable was the total number of blinks that occurred in the 3-minute period. Although previous blink rate studies have used 60- to 90-second assessment periods,19 a 3-minute blink assessment period was selected on the basis of our earlier study,15 which showed that the number of blinks per minute during the first 3 minutes of recording did not vary within either subjects with SAD or controls. However, after 3 minutes, the rate of blinking, while continuing to differ between groups, became variable for both groups and, hence, a less reliable estimate of a subject's blink rate.
Mood Assessment
Depression was assessed without the knowledge of group member¬ ship in subjects with SAD and controls by the use of the newly developed 29-item Hamilton Depression Rating Scale (HDRS). De¬ veloped by SAD researchers in collaboration with the Department of Biometrics Research at the New York (NY) State Psychiatric Insti¬ tute, this version of the HDRS incorporates symptoms of depression that are typical of SAD but not necessarily typical of other forms of major depression. This version includes the commonly used 21-item HDRS20 plus 8 additional items relevant to SAD symptoms. The interrater agreement for our group on this new version by the use of an audiotape is adequate (intraclass correlation .86). The HDRS was administered on the day of the PRL assessment on three occasions: (1) Follicular phase: lb ensure that the HDRS score =
On blood-sampling days, subjects consumed the same liquid protein drink at noon, before and after which they fasted except for water. They had abstained from caffeine and alcoholic beverages for no less than 24 hours. Subjects sat for 40 minutes before blood sampling to control the activity level in a temperature-controlled room with an overhead illumination of 300 lux that was measured at eye level. Blood samples were spun down immediately, and serum samples were assayed in duplicate within 1 to 2 days by using a doubleantibody radioimmunoassay. The assay sensitivity was 1.0 ng/mL, the interassay coefficient of variation was equal to 7.2%, and the intraassay coefficient of variation was equal to 6.1%.
Spontaneous Eye Blink Assessment Spontaneous eye blinking was assessed during winter between January 2 and 31 and in summer between June 28 and August 10 via a polygraphic recording. The vertical electro-oculogram (eye blinks), as well as the horizontal electro-oculogram, were recorded on a poly-
reflected a stable mood state at the time of the follicular PRL assess¬ ment, the HDRS interview covered the week before and the day of the PRL assessment. In all cases of SAD, the level of depression on the day of assessment was equal to or more severe than the 7-day overall level of functioning. (2) Lutealphase: To assess the stability of depres¬ sion further, subjects with SAD and controls were also administered a second HDRS approximately 2 weeks later (which was the time of the luteal PRL assessment for those subjects participating in this analy¬ sis). This HDRS assessment also covered a 1-week period. This means that, for all subjects, the level of depression was assessed throughout 2 of the 3 weeks that surrounded the PRL assessment times. (3) Summer: The HDRS assessment also covered a 1-week period at the time ofthe summer PRL analysis. Also in the summer, the periods of depression of any length during the month before the PRL assess¬ ment were to be rated on the HDRS, but no ratable periods of depression were reported by subjects with SAD or controls for that month, confirming our previous summer mood assessments in SAD.6
Downloaded From: http://archpsyc.jamanetwork.com/ by a University of Missouri - St Louis User on 09/30/2013
Age, Weight, HDRS Score, PRL Level, and Spontaneous Eye Blink Rates in Subjects With Bipolar II and Unipolar SAD and Normal Controls* Weight, kg Group
Age, y
W
HORS Score
PRL
Level, (ig/Lt
Blinks/3 min W
SAD
Bipolar II 1
36
64.5
62.5
59.9
58.3
36 27
47
30
25
11
32
60.2
58.2
39
37
10
31
53.1
51.2
29
31
21
67.1
65.5
36
32
33
59.0
58.0
30
29
33
60.3
58.1
33
32
32
68.5
66.2
29 34
78
80
98
46 11
86 12
74 36
25
58.1
56.3
31 37
10
23
76.8
74.1
38
26
11
35 33
53.2
53.0
30
36
41
49
53.2
52.1
29
42
129
28
68.2
65.7
37
41
106 44
28
52.3
51.2
38
43
40
30±4
61 ±7
59±7
34±4
35±7
26
62.2
60.1
24
26
21
70.0
68.2
34
47
33
61.8
60.1
38
38
21
100
97.0
21
21
30
65.4
63.1
32
29
34
56.3
54.2
35
52.7
51.5
38 27
39 27
32
67.2
66.4
30
38
26
52.2
50.6
26
28
36
84.0
81.4
35
37
29±2
67±8
65±8
31 ±6
33±9
2±1
4.1 ±2
30±3
64±7
62±7
31 ±5
34±8
2±1
4.5 ±2
35
66.4
66.1
34
74.8
71.3
28
63.2
62.4
34
62.6
62.0
32
68.6
67.8
34 26
70.3 69.5
68.2 67.4
27
54.4
53.2
12 13 14
10
47
Mean ± SD
Unipolar 1
10
2±1
4.8 ±2
4.1+2
7.8 ±3
10
67 ±27
60 ±40
71
50
18
140 40
50 53
39
56 69
108
13
112
3.5 ±1
7.0 ±3
55 ±32
75 ±46
3.9 ±2
7.4 ±3
61+29
67 ±41
14
21
39
18
50
50
13
15
15
29
14
11
Mean ± SD
Total Mean ± SD Controls 1
13 10 11
10 11
22
54.0
54.0
10 11
26
52.1
53.1
27
71.2
69.1
10
12
30
64.4
64.1
19
10
26
59.1
57.1
62.5
60.1
15
31 22
50.0
50.6
16
29
56.2
55.5
17
24
50.8
13 14
12
13
13
16
13 19 17
20
23 14
50.3
14 11
18
10
13
39
16
58
18
21
65.3
65.1
10
19
20
66.9
66.0
14
20
35
75.0
72.1
11
13
18
28±5
63±8
62±7
10.1+4
9.3 ±5
14.4±3
14
Mean ± SD 0.8 ±.5
0.9 ±.4
0.4 ±.2
30±16
25±19
*HDRS indicates Hamilton Depression Rating Scale; PRL, prolactin; SAD, seasonal affective disorder; W, winter; S, summer; F, follicular; L, luteal; F„, follicular winter; and F„ follicular summer. fRounded to the nearest whole number.
Downloaded From: http://archpsyc.jamanetwork.com/ by a University of Missouri - St Louis User on 09/30/2013
20
18 -
PRL The PRL data are shown in the Table and Fig 1. Because the PRL values may be affected by age and weight, all three groups (BP II SAD, 14; UP SAD, 10; and controls, 20) were matched for age (one-factor analysis of variance: F 0.62, df=2, 41, P
Depue, PhD; Paul Arbisi; Steven Krauss, PhD; William G. Iacono, PhD; Arthur Leon, MD; Rachel Muir; John Allen
\s=b\ Twenty-four subjects with seasonal affective disorder (SAD: bipolar II, n 14; unipolar, n 10) and 20 normal controls were assessed for early follicular basal serum prolactin (PRL) concen=
=
tration in winter and summer. Luteal basal PRL concentration was assessed in winter. The PRL values represented the mean of three values derived during a 45-minute period. A subset of 17 subjects with SAD and 11 controls were also assessed for spontaneous eye blinking via a polygraphic recording in winter and summer. In winter, compared with controls, subjects with SAD were characterized by significantly lower follicular (10.1 vs 4.5 \g=m\g/L,respectively)and luteal (14.4 vs 7.4 \g=m\g/L,respectively) PRL values and by significantly higher eye blink rates (30 vs 61 blinks per 3 minutes, respectively). In summer, controls and subjects with SAD showed similar significant differences in follicular PRL values (9.3 vs 3.9 \g=m\g/L,respectively) and eye blink rates (25 vs 67 blinks per 3 minutes, respectively). No significant differences in PRL values or eye blink rates were found between the bipolar II and unipolar forms of SAD in either season. Results were discussed in terms of dopamine functioning. (Arch Gen Psychiatry. 1990;47:356-364)
the of phototherapy for seasonal affective disorder has been actively researched during the the search for variables that characterize SAD 5 has received much less attention. A review of published and unpublished work on biologic variables3 in SAD re¬ vealed that consistent trends have yet to emerge. Relative to controls, cortisol values were found to be significantly in¬ creased, and prolactin (PRL) values were found to be signifi¬ cantly decreased in SAD when stimulated by 5-hydroxytryptophan,4 whereas an unpublished study3 observed signifi¬ cantly increased PRL values when stimulated by a serotonin agonist. Basal levels of PRL have been found to be both increased4 and decreased.3,5,6 Dexamethasone suppression and 24-hour secretion of cortisol, thyrotropin (thyroid-stimulating hormone) responsivity, and cerebrospinal fluid metabolite lev¬ els of serotonin, norepinephrine, and dopamine were all nor¬ mal.3,7 Lewy et al8 demonstrated that SAD is characterized by a significant delay of the onset of nighttime melatonin secre-
efficacy (SAD) Although past years,1,2
Accepted for publication September 19,1989. From the Department of Psychology (Drs Depue, Krauss, and Iacono and Messrs Arbisi and Allen and Ms Muir), the Division of Epidemiology, School of Public Health (Dr Leon), and the Department of Physiology, Medical School (Dr Leon), University of Minnesota, Minneapolis. Reprint requests to Department of Psychology, N218 Elliott Hall, University of Minnesota, 75 E River Rd, Minneapolis, MN 55455 (Dr Depue).
tion and that
morning light exposure advanced this onset in proportion to the antidepressant effects in some patients. Other work, however, suggests that this finding may not gen¬ eralize to all subjects with SAD.9"11 Currently, then, there are few positive findings in SAD that provide a framework for further research. Where positive results have been reported, findings are either balanced by negative results or have yet to be replicated. Moreover, we are aware of no studies that have systematically assessed whether the unipolar (UP) and bipolar (BP) II forms of SAD, observed
in the clinical literature,12,13 represent a valid distinction when tested against external criteria. It is within this framework of inconsistency that we pursue our initial positive findings in SAD. In several initial studies, we observed significant differences between subjects with SAD and controls in the thermoregulatory heat loss réponse to a thermal challenge,14 spontaneous eye blink rates,15 and the basal follicular PRL concentration.56 (These variables were studied because of their common link to dopamine [DA] mech¬ anisms, as described in the "Comment" section.) The latter two variables revealed particularly marked differences. Pa¬ tients with SAD showed a significantly lower basal follicular PRL concentration (less than half of the control values) in winter, and the PRL values remained equally low after suc¬ cessful phototherapy5 and on retest in summer6 when patients were euthymic. Low basal PRL values in SAD were also reported in one unpublished study3 but not in another report.4 We also observed spontaneous eye blink rates in SAD that exceeded the rate ofcontrols by a factor of two.15 Both the PRL and eye blink rate findings were based on relatively small samples of subjects with SAD that were composed solely of the BP II form of SAD. Therefore, the current report represents (1) a successful replication of the PRL and eye blink rate findings on a much larger sample of subjects with SAD, (2) a demonstration of the seasonal independence of both of these variables in SAD, and (3) the first comparison of subjects with UP and BP II SAD on PRL and eye blink variables. SUBJECTS AND METHODS
Subjects The participants in the study were premenopausal females (24 with SAD and 20 normal controls). Subjects were obtained by media advertisements, and all respondents were then screened by the use of a modified Schedule for Affective Disorders and Schizophrenia-Life¬ time version16 interview, a process that represents the typical sub¬ ject-selection procedure in outpatient studies of SAD.1,4,8,11,2 All sub-
Downloaded From: http://archpsyc.jamanetwork.com/ by a University of Missouri - St Louis User on 09/30/2013
jects with SAD met DSM-III-R" criteria for a major depression that developed during the fall, was continuous during the winter, and remitted the following spring—for at least the last 3 consecutive years; subjects with SAD also met the newly developed DSM-III-R criteria for a seasonal pattern to the major depression. The winter depressions were not in reaction to seasonally varying social Stress¬ ors. Subjects with SAD also had to be free of depressed periods during
the last three consecutive summers. Self-reports of 14 of the subjects with SAD, concerning their behav¬ ior during summer months, satisfied criteria for hypomanic episodes that were included in the DSM-III-R criteria for cyclothymic disorder (but never met DSM-III-R criteria for a manic episode), where each of these subjects reported that she typically had three or more such episodes during summer months. Therefore, these 14 subjects had the BP II form of SAD that has been observed by others.12,13 The 10 subjects with SAD who did not meet criteria for hypomanic epi¬ sodes had diagnoses of UP SAD, a form of SAD also observed by others.8,12,13 No other axis I psychiatric disorder existed among sub¬ jects with SAD. Controls were free of psychiatric disorder as deter¬ mined by the same structured interview. The values for indepen¬ dently assigned diagnoses of interest to the study on this sample, determined by a dual diagnosis from audiotapes, were as follows: major depression (.90), seasonal pattern (.96), and hypomania (.78), ie, values that were consistent for our group.6,1,15,1 Potential subjects (those with SAD and controls) who reported postmenopausal status, pregnancy, use of birth control pills during the previous 4 months, any prescribed medication during the past 6 months, any form of substance abuse, menstrual cycle irregularities or significant mood and physical symptoms suggestive of premenstru¬ al syndrome, endocrinopathies or other relevant medical conditions, ie, an eating disorder, oculomotor disorders, or epilepsy were ex¬ cluded from study. None of the subjects had used phototherapy previously nor directly before the study. PRL Assessment The basal follicular serum PRL value was assessed in winter be¬ tween January 2 and 31 and in summer between June 28 and August 10. The serum PRL concentration was assessed by three separate blood draws during a 45-minute period by averaging the value of the three blood samples, each of which was separated by a 15-minute interval. The timing of the PRL sampling was strictly controlled: (1) blood samples were collected in the early to middle follicular phase of the menstrual cycle, where all samples for all subjects were ob¬ tained between 2 and 7 days into the menstrual cycle (the menstrual phase was determined by subject self-reports that were obtained throughout three consecutive menstrual cycles, including the cycle that was assessed in the study); and (2) to control for the time of day (diurnal variation), and because SAD is associated with a marked lateafternoon blood samples were collected at 3:30, 3:45, and 4 PM for each subject. Subjects with SAD and controls were also assessed for PRL concen¬ tration in exactly the same manner 2 weeks later in winter (between January 15 and February 17) during the middle to late luteal phase of the menstrual cycle, ie, between 19 and 25 days into the menstrual
worsening,5,13
cycle.
graph (Beckman
R612 Dynograph) in a sound-deadened, dimly lit Silver-silver-chloride electrodes were attached at the superior and inferior orbital rims of the left eye to record blinks, and to the outer canthi for the horizontal electro-oculogram recording. A ground electrode was attached to the right earlobe. Alternating current coupling with a 3-second time constant was used. The electrodes were also attached to measure palmar electrodermal, heart rate, and electroencephalographic variables, to be reported elsewhere. An eye blink was operationally defined as a sharp, high-amplitude wave from the immediately preceding baseline equal to or greater than 100 µ and less than 400 milliseconds in duration (approximately =sl mm in width, measured at a vertical level that is the approximate midpoint ofthe blink amplitude on the baseline side ofthe blink, where chart speed 2.5 mm/s). This definition was consistent with, but more stringent than, that found in previous human blink rate studies.19 High-amplitude waves coincident with horizontal eye movements were not counted as blinks. The counting of eye blinks was performed, blind of the subject's group identity. Two raters independently counted eye blinks on two thirds of the subjects with SAD and controls and achieved an interrater agreement (r =. 96) similar to that found in room.
=
previous studies.19 The timing of eye blink assessments was strictly controlled, exactly as in the PRL assessment: (1) to control diurnal factors, blink record¬ ings occurred between 3 and 6 PM (but not on the same day as the PRL assessment); and (2) to control for the phase of the menstrual cycle, blink assessment occurred during the early to middle follicular period (days 2 through 7). A 10-minute period of adaptation to the recording room was pro¬ vided. Eye blinks were recorded for a 3-minute period, while the subject sat silent with the lights on. Subjects were told to keep their eyes open but were not instructed in any manner about blinking. Subjects did not wear contact lenses during the recordings, did not have acute eye infections, and were instructed not to sleep (which was verified in chart recordings). The dependent variable was the total number of blinks that occurred in the 3-minute period. Although previous blink rate studies have used 60- to 90-second assessment periods,19 a 3-minute blink assessment period was selected on the basis of our earlier study,15 which showed that the number of blinks per minute during the first 3 minutes of recording did not vary within either subjects with SAD or controls. However, after 3 minutes, the rate of blinking, while continuing to differ between groups, became variable for both groups and, hence, a less reliable estimate of a subject's blink rate.
Mood Assessment
Depression was assessed without the knowledge of group member¬ ship in subjects with SAD and controls by the use of the newly developed 29-item Hamilton Depression Rating Scale (HDRS). De¬ veloped by SAD researchers in collaboration with the Department of Biometrics Research at the New York (NY) State Psychiatric Insti¬ tute, this version of the HDRS incorporates symptoms of depression that are typical of SAD but not necessarily typical of other forms of major depression. This version includes the commonly used 21-item HDRS20 plus 8 additional items relevant to SAD symptoms. The interrater agreement for our group on this new version by the use of an audiotape is adequate (intraclass correlation .86). The HDRS was administered on the day of the PRL assessment on three occasions: (1) Follicular phase: lb ensure that the HDRS score =
On blood-sampling days, subjects consumed the same liquid protein drink at noon, before and after which they fasted except for water. They had abstained from caffeine and alcoholic beverages for no less than 24 hours. Subjects sat for 40 minutes before blood sampling to control the activity level in a temperature-controlled room with an overhead illumination of 300 lux that was measured at eye level. Blood samples were spun down immediately, and serum samples were assayed in duplicate within 1 to 2 days by using a doubleantibody radioimmunoassay. The assay sensitivity was 1.0 ng/mL, the interassay coefficient of variation was equal to 7.2%, and the intraassay coefficient of variation was equal to 6.1%.
Spontaneous Eye Blink Assessment Spontaneous eye blinking was assessed during winter between January 2 and 31 and in summer between June 28 and August 10 via a polygraphic recording. The vertical electro-oculogram (eye blinks), as well as the horizontal electro-oculogram, were recorded on a poly-
reflected a stable mood state at the time of the follicular PRL assess¬ ment, the HDRS interview covered the week before and the day of the PRL assessment. In all cases of SAD, the level of depression on the day of assessment was equal to or more severe than the 7-day overall level of functioning. (2) Lutealphase: To assess the stability of depres¬ sion further, subjects with SAD and controls were also administered a second HDRS approximately 2 weeks later (which was the time of the luteal PRL assessment for those subjects participating in this analy¬ sis). This HDRS assessment also covered a 1-week period. This means that, for all subjects, the level of depression was assessed throughout 2 of the 3 weeks that surrounded the PRL assessment times. (3) Summer: The HDRS assessment also covered a 1-week period at the time ofthe summer PRL analysis. Also in the summer, the periods of depression of any length during the month before the PRL assess¬ ment were to be rated on the HDRS, but no ratable periods of depression were reported by subjects with SAD or controls for that month, confirming our previous summer mood assessments in SAD.6
Downloaded From: http://archpsyc.jamanetwork.com/ by a University of Missouri - St Louis User on 09/30/2013
Age, Weight, HDRS Score, PRL Level, and Spontaneous Eye Blink Rates in Subjects With Bipolar II and Unipolar SAD and Normal Controls* Weight, kg Group
Age, y
W
HORS Score
PRL
Level, (ig/Lt
Blinks/3 min W
SAD
Bipolar II 1
36
64.5
62.5
59.9
58.3
36 27
47
30
25
11
32
60.2
58.2
39
37
10
31
53.1
51.2
29
31
21
67.1
65.5
36
32
33
59.0
58.0
30
29
33
60.3
58.1
33
32
32
68.5
66.2
29 34
78
80
98
46 11
86 12
74 36
25
58.1
56.3
31 37
10
23
76.8
74.1
38
26
11
35 33
53.2
53.0
30
36
41
49
53.2
52.1
29
42
129
28
68.2
65.7
37
41
106 44
28
52.3
51.2
38
43
40
30±4
61 ±7
59±7
34±4
35±7
26
62.2
60.1
24
26
21
70.0
68.2
34
47
33
61.8
60.1
38
38
21
100
97.0
21
21
30
65.4
63.1
32
29
34
56.3
54.2
35
52.7
51.5
38 27
39 27
32
67.2
66.4
30
38
26
52.2
50.6
26
28
36
84.0
81.4
35
37
29±2
67±8
65±8
31 ±6
33±9
2±1
4.1 ±2
30±3
64±7
62±7
31 ±5
34±8
2±1
4.5 ±2
35
66.4
66.1
34
74.8
71.3
28
63.2
62.4
34
62.6
62.0
32
68.6
67.8
34 26
70.3 69.5
68.2 67.4
27
54.4
53.2
12 13 14
10
47
Mean ± SD
Unipolar 1
10
2±1
4.8 ±2
4.1+2
7.8 ±3
10
67 ±27
60 ±40
71
50
18
140 40
50 53
39
56 69
108
13
112
3.5 ±1
7.0 ±3
55 ±32
75 ±46
3.9 ±2
7.4 ±3
61+29
67 ±41
14
21
39
18
50
50
13
15
15
29
14
11
Mean ± SD
Total Mean ± SD Controls 1
13 10 11
10 11
22
54.0
54.0
10 11
26
52.1
53.1
27
71.2
69.1
10
12
30
64.4
64.1
19
10
26
59.1
57.1
62.5
60.1
15
31 22
50.0
50.6
16
29
56.2
55.5
17
24
50.8
13 14
12
13
13
16
13 19 17
20
23 14
50.3
14 11
18
10
13
39
16
58
18
21
65.3
65.1
10
19
20
66.9
66.0
14
20
35
75.0
72.1
11
13
18
28±5
63±8
62±7
10.1+4
9.3 ±5
14.4±3
14
Mean ± SD 0.8 ±.5
0.9 ±.4
0.4 ±.2
30±16
25±19
*HDRS indicates Hamilton Depression Rating Scale; PRL, prolactin; SAD, seasonal affective disorder; W, winter; S, summer; F, follicular; L, luteal; F„, follicular winter; and F„ follicular summer. fRounded to the nearest whole number.
Downloaded From: http://archpsyc.jamanetwork.com/ by a University of Missouri - St Louis User on 09/30/2013
20
18 -
PRL The PRL data are shown in the Table and Fig 1. Because the PRL values may be affected by age and weight, all three groups (BP II SAD, 14; UP SAD, 10; and controls, 20) were matched for age (one-factor analysis of variance: F 0.62, df=2, 41, P
