- Email: [email protected]
Stability of plasma GABA at four-year follow-up in patients with primary unipolar depression
Stability of Plasma GABA at Four-Year Follow-up in Patients with Primary Unipolar Depression Frederick Petty, Gerald L. Kramer, Mark Fulton, Lori Davis, and A. John Rush
The biology of mood disorders involves ~-aminobutyric acid (GABA), a neurotransmitter whose levels in plasma likely reflect brain GABA activity. Previous research has shown that a subset of patients with primary unipolar major depression have low plasma GABA levels, which parallels findings from studies of cerebrospinal fluid. We have completed a 4-year follow-up on 46 male patients with primary unipolar depression. Plasma levels of GABA were stable over this time. For the group, mean plasma GABA levels on follow-up did not change significantly J?om entry levels. Plasma GABA levels measured on follow-up were significantly (p < .001) correlated with entry levels. Patients with low plasma GABA levels (< 100 pmol/ml) on entry into the study were likely to remain low on follow-up, and patients with plasma GABA levels in the control range (>- lOO pmol/ml) at entry similarly remained in this category (X: = 7.23, p = .007). This was true whether or not the patient had recovered from depression on follow-up. Levels o f plasma GABA did not significantly correlate with severity of depression at either entry (p = .40) or follow-up (p = .52), nor was there a significant correlation between change in plasma GABA and change in the 17-item Hamilton Depression Rating Scale score from entr), to follow-up (p = .89). These data are consistent with the notion that plasma GABA is independent of clinical state in patients with primary unipolar depression. Low plasma GABA may be a trait marker of illness in a subset ~¢patients with mood disorder.
Key Words: GABA, depression, biological markers, mood disorder, follow-up study
Introduction Several lines of evidence implicate the "y-aminobytric acid (GABA) system in mood disorders. GABA agonists appear to be effective in the treatment of both depression and mania (Lambert 1984; Lloyd et al 1985; McElroy et al 1992; Bowden el al 1994), although a negative study has also been
From the Veterans Affairs Medical Center (FP, GLK, MF, LD) and Department of Psychiatry, University of Texas Southwestern Medical School /FP, MF. LD. AJR), Dallas, TX. Address reprint requests to Frederick Petty, Ph.D.. M.D., Psychiatry Service (116A). Veterans Affairs Medical Center. 451}11South Lancaster Road. Dallas, TX 75216: Fax (214) 372-7987. Received April 15, 1994; revised August 9. 1994.
© 1995 S~xziety of Biological Psychiatry
reported (Paykel et al 1991). For bipolar disorder, GABA agonists also appear to be useful prophylactic agents (Emrich et al 1980; Calabrese et al 1993). Considering its ubiquity as a neurotransmitter (about one third of central nervous system (CNS) neurons use G A B A as a neurotransmitter), it would be surprising to not find an involvement of G A B A in mental illnesses such as depression. G A B A can be measured in cerebrospinal fluid (CSF). CSF levels (120 nM) are comparable to those in brain extracellular space (1 O0 riM) (Drew et al 1989). A meta-analysis of the seven published studies of CSF GABA in depression confirms that G A B A levels are low in patients with depression compared to controls (p = .005; Petty et al 1993a). Additional, albeit indirect, evidence from animal models strongly sup(X)06-3223/95/$09.50 SSDI 0006-3223194)00226-S
Stability of Plasma GA BA
16
BIOLPSYCHIATRY
1995:37:80~810
807
--
]4
I~ -
• '\"x
12
/
! >- 10 0z ~
8
6
Figure I. Distribution of plasma GABA in healthy controls (solid bars) and in male patients with primary unipolar depression (open bars). Best fit to data are two curves with five parameters: 1 mean and SD for all normals and 60% of depressives; 1 mean and 1 SD for 40% of depressives. (Data from Petty et al 1992.)
///
// /
/
'
,
[
d 2
~.~ j =
GABA
ports a central role for GABA in the biochemical pathophysiology of depression (Petty 1986; Lloyd et al 1985). Plasma GABA provides a useful index of brain GABA activity, since virtually all GABA production occurs in the CNS, and since plasma levels of GABA accurately parallel brain GABA levels (Petty et al 1993a for review). Low levels of plasma GABA are found in 30%-40% of patients with major mood disorder, whether manic or depressed (Petty et al 1992a, 1993b). In patients with primary unipolar depressive disorder, 40% have levels of plasma GABA below 100 pmol/ml, compared to only 6% of healthy controis. This difference in distributions is highly significant (p < .0001 ), it can best be accounted for by the existence of two distinct patient populations in depression, one with low GABA and one with normal G A B A (Figure 1). In healthy subjects, plasma GABA is stable with time and does not exhibit appreciable circadian or circumannual fluctuations (Petty et a] 1987, 1992b). If low plasma GABA is a valid biological trait marker for depression in a subset of patients, follow-up studies should demonstrate temporal stability of plasma GABA. Low plasma GABA should not revert to normal over the course of time on follow-up, regardless of clinical state. Nor should patients with initial levels of GABA in the normal range change appreciably over time. On the other hand, if low plasma GABA in depression is a state marker for low mood, revision toward the mean plasma G A B A of healthy controls should be seen with the passage of time in patients with depression. In a previous study, we found that plasma GABA levels did not change appreciably following 1 month of treatment with desipramine, although symptoms of depression improved (Petty et al 1993c). These findings are compatible with the possibility that low plasma GABA may be a traitlike marker for mood disorders in a subset of patients, since
plasma GABA does not seem to be dependent on clinical state, whether manic or depressed. Nor does plasma G A B A normalize with improved mood during acute-phase treatment of depression. We here present a 4-year follow-up of outpatients with a diagnosis of primary unipolar nonpsychotic major depressive disorder, reporting on their plasma G A B A levels over time. Methods
Subjects participating in this study were initially in the patient group of 107 males with a diagnosis of primary unipolar major depressive disorder previously described and reported (Petty et al 1992a). Of this sample, 51 were available for follow-up study 4 years later. Treatment was uncontrolled between entry and follow-up; however, most patients received standard clinical treatment during the intervening period, including antidepressant medications. At follow-up, the Structured Interview for DSM-III-R (SCID) (Spitzer et al 1987) was readministered by a trained psychiatric research nurse, and the diagnosis was reviewed by the first author. Severity of depressive symptoms was assessed with the 17item Hamilton Rating Scale for Depression (HRS-D; Hamilton 1967), and note was made of current medications. Of the subjects available at lbllow-up, 4 had become bipolar and 1 was rediagnosed with a primary anxiety disorder, leaving 46 whose diagnosis of primary unipolar depression was confirmed and stable 4 years after the initial assessment. At both entry (T1) and follow-up (T2), blood samples for plasma GABA determination were obtained by venipuncture into ethylenediamine tetraacetic acid (EDTA) vacutainer, with the plasma quickly removed and stored at - 7 0 ° C until time of assay. Quantitative analysis of plasma
808
toOLPSYCHIATRY
F. Petty et al
1995;37:806 810
G A B A was conducted with high-performance liquid chromatography using the procedure of Hare and Manyam (1980) with minor modifications, including use of y-vinyl G A B A as an internal standard. Both intra- and interassay coefficients of variation of the assay are 6% and the limit of detection (5:1 S:N) is 10 pmol/ml. With these procedures, healthy control subjects (n = 71 ) have plasma G A B A levels of 123 + 20 pmol/ml (SD) (Petty et al 1992a).
Results All patients in the present study were male, with a mean age at TI of 52.2 (SD = 18.8). Age was not significantly related to plasma G A B A at T 1 ( r = .03, p = .83) or at T2 (r = 092, p = .54). For the 46 patients, mean levels of plasma G A B A on entry (T l) were not significantly different from levels at 4-year (T2) follow-up (106 + 24 vs. 108 _+ 27 pmol/ml, respectively, paired t = .39, p = .70). There was a significant correlation between plasma G A B A levels at T I and T2 (r = .45,p < .001; Figure 2). On follow-up, 30 patients were classified as recovered or improved, defined as a Hamilton Rating Score for Depression (HRS-D) score of -<9 at T2 (n = 25) or a -> 50% drop in HRS-D score between T1 a n d T 2 (n = 5), and 16 continued to be depressed. There were no significant changes in plasma G A B A levels between entry and follow-up for either group (recovered: T I - - 1 0 9 _+ 24 pmol/ml vs. T 2 - - 1 1 0 _+ 28 pmol/ml, paired t =. 15,p = .89, depressed: T I - - 1 0 1 + 23 pmol/ml vs. 104 _+ 25 pmol/ml, paired t = 0.48, p = .64). As noted, our previous research indicates plasma G A B A levels < 100 pmol/ml to be a biological marker in a subset of patients with depression, while plasma G A B A levels >-100 pmol/ml are seen in most depressives and in 95% of healthy controls (Petty et al 1992a). Of the present sample, 18/46 (39%) had low plasma G A B A ( < 100 pmol/ml) at T 1, while 28/46 had plasma G A B A in the normal range at T1. Eleven patients had low plasma G A B A at both TI and T2, and 23
160 140
s,L,,*- <
~ 120 <~ 100 8O 6O 40 20
Figure 2. Levels of plasma GAB A in patients with primary unipolar depression at entry into study and at 4-year follow-up. See text for statistical analysis.
had normal plasma G A B A levels at both points in time, while seven patients who were low at T1 were in the normal range at T2, and five who were normal at T 1 were low at T2. This distribution was significant (X2 = 7.23, p = .007). On entry into the study, all patients had HRS-D scores of -> 10 (range 10-37), while on follow-up 27 had score of -< 9 (range 0-28), and mean HRS-D scores at T1 were significantly higher than at T2 (23 + 7 vs. 10 + 8, respectively, paired t = 9.86, p < .001). There was a significant correlation between T I and T2 HRS-D scores (r = 0.308, p = .03); however, the correlation between plasma G A B A levels and HRSD was not significant at either T 1 (r = .27, p ---.40) or at T2 (r = .097, p = .52). Nor was the correlation between change in plasma G A B A (Y2 level minus T1 level) and change in HRSD score (T2-T 1) significant (r = -.02, p = .89). At entry, 12 patients were taking psychotropic medications and 34 were not; plasma G A B A levels were not significantly different between these two groups (107:___26 pmol/ml vs. 105 +_ 18 pmol/ml, respectively, t = .29, p = .39). For the 21 patients who were not taking psychotropic medications at either T1 or T2, there was no significant change in plasma G A B A levels (111 -+ 26 pmol/ml vs. 119 +- 28 pmol/ml, respectively, paired t = 1.31, p = .20). Of the 21 patients who were off psychotropic medication at both T I and T2, 15 had improved or recovered from depression at T2 (defined as --> 50% drop in HRS-D score or final HRS-D -< 9). Plasma G A B A levels did not change significantly for these 15 nonmedicated patients who experienced a change in clinical state (113 +_ 23 pmol/ml vs. 120 + 31 pmol/ml, TI vs. T2, t = 1.31, p = .44), nor did it change significantly for the 7 non-medicated patients who did not recover from depression at T2 ( 103 -+ 12 pmol/ml vs. 110 _+ 25 pmol/ml, T l vs. T2, t = .78,p = .46). Family history for mood disorder in first degree relatives could be determined for 42 cases, while 4 patients either were adopted or had inadequate knowledge of their first-degree relatives. Plasma G A B A at TI did not significantly differ between the patients with a positive family history ( 101 +_ 26 pmol/ml, n = 17) and those with a negative family history (104 _+ 29 pmol/ml, n = 25, t = 1.10, p = .30). Similarly, at T2 these groups were not significantly different (107_+20 pmol/ml vs. 108 _+ 32 pmol/ml, respectively, t = .00, p = .99). Of the 46 patients, 16 had a single episode of depression and 30 had recurrent depression with two or more episodes, determined at T2. Plasma G A B A did not differ significantly between these groups (103 _+ 26 pmol/ml vs. 108 _+ 22 pmol/ml, respectively, t = .35, p = .55).
Discussion The major finding of the present work is that plasma levels of G A B A are stable with time over a 4-year follow-up in male patients with primary unipolar major depression. This
Stability of Plasma GABA
BIOLPSYCHIATRY
809
1995:37:806-81[)
was true for the sample as a whole, and for the 21 patients who were not taking psychotropic medications at either initial evaluation or at follow-up. Additionally, plasma GABA levels were not significantly influenced by age, psychotropic medication, or recovery of depression. These data replicate and extend our previous work demonstrating no significant change in plasma GABA in patients with major depression treated with desipramine for 4 weeks with clinical improvement (Petty et al 1993c). We have also found plasma GABA levels in healthy controls to be stable over several years (unpublished data). Interpretation of the present work should be made with great caution, particularly since this was a naturalistic study. No attempt was made to control for treatment during the 4 years of follow-up; nor did we obtain detailed data on clinical status from the patients during the course of the study. Therefore, we do not have a precise picture of the pattern of symptoms or medications during the entire 4-year followup period. The stability of plasma GABA seen in this follow-up study is compatible with the notion that plasma GABA is a biological trait marker of primary unipolar depression. The daily certainly indicate that low plasma G A B A is n o t a biological marker for the state of depression, since plasma G A B A levels did not significantly change in patients who had improved or recovered from depression on follow-up. Furthemore, there was no correlation found between change in G A B A and change in depression rating scale scores. The issue of "state" versus "trait" in biological markers is complex and has been discussed recently (Kraemer et al in press). Basically, comprehensive analysis of this problem involves a determination of the precise contribution of the variance in plasma GABA levels due to chance, state of illness, time, and so forth. What is left can be considered the proportion of variance due to trait of illness. In the present work, we can only account for 20% (r 2 = .20) of the variance in plasma GABA. We have demonstrated little variability with time, gender, exercise, diet, or season in healthy controls (Petty 1987, 1992b); however, we do not presently know if plasma GABA in mood-disordered patients is similarly unaffected by these lectors. Considerably more research in mood-disordered patients will be required for a definitive answer as to the exact proportion of the variance in plasma GABA due to the trait of having a major depression. Specifically, a prospective study is needed with multiple determinations of plasma GABA at multiple points in time during different phases of the illness. If plasma GABA is low in some depressives, and continues low at follow-up, what neurochemical mechanisms might account for this? First, it is important to realize that GABA is a very brain-specific neurotransmitter, with high concentrations in brain, and with 99% of total body GABA in the CNS (Zachman et al 1966). Based on this consideration alone,
it is plausible that a peripheral measure of GABA might reflect brain GABA activity. While it is true that GABA does not cross the blood-brain barrier, going from blood to brain, the converse is not true, since experimental manipulation of brain GABA levels in laboratory animals are followed by similar changes in plasma GABA (Ferkany et al 1978). Furthermore, there is good evidence that levels of GABA in the extracellular fluid of brain reflect synaptic GABA levels, and that GABA in CSF and plasma is in equilibrium with brain extracellular fluid GABA (reviewed in Petty et al 1993b). Therefore, we can look to regulation of extracellular (or synaptic) GABA as a model for regulation of plasma GABA. Low extracellular GABA can result from at least four possible causes: slower synthesis, less release, more uptake, and/or faster metabolism. These processes for GABA are fairly well understood in brain. If plasma GABA is an accurate reflection of synaptic G A B A concentration, similar mechanisms should regulate both. In other words, low plasma GABA in patients with depression could be caused by one or more of the following brain mechanisms. Slower synthesis of G A B A might be due to decreased activity of the GABA-synthesizing enzyme, glutamate decarboxylase. Less neuronal release of GABA might be due to presynaptic inhibition, by GABA itself, or by another neurotransmitter. More neuronal or glial uptake of GABA might be due to differences in conformation of the G A B A uptake carrier. Faster metabolism of GABA might occur with changes in activity of the GABA-metabolizing enzymes, GABA-transaminase and succinyl semialdehyde dehydrogenase. These represent a finite series of possibilities and biochemical studies could be undertaken to determine whether an alteration in any of these molecular mechanisms in brain are responsible for low plasma GABA in depression, using human postmortem tissue or animal models. Finally, it is also conceivable that patients with depression have greater hepatic clearance of GABA, or that another, perhaps unknown, peripheral mechanism regulates plasma GABA levels differently in patients with major depression than in healthy controls, though no data support this theoretical possibility. In summary, plasma GABA levels in patients with primary unipolar major depression were stable at 4-year follow-up, and did not change with clinical improvement. Low plasma G A B A may be a stable biological trait marker for this depressive illness in a subset of 40% of patients. Supported by the Department of Veterans Affairs Medical Research Service: NIH Research Grants MH37899, AA07234and MH41115; and the John SchermerhornFund. We thank Jim Mintz, Ph.D.. lk~rhelpful discussions; Glynda Mekonen and Dinah Turner-Knight fnr superb secretarial support in preparing this manuscript; and Kenneth Z. Altshuler, M.D., Stanton Sharp Distinguished Professor and Chair, fl)rhis administrative support.
810
BIOLPSYCHIATRY
F. Petty et al
1995:37:806-810
References Bowden CL, Brugger AM, Swarm AC, et al. ( 1994): Efficacy of divalproex sodium vs. lithium and placebo in the treatment of mania. JAMA 271:918-924. Calabrese JR, Woyshville MJ, Kimmel SE, Rapport DJ (1993): Predictors of valproate response in bipolar rapid cycling. J Clin Psychopharmacol 13:280-283. Drew KL, O'Connor WT, Kehr L Ungerstedt U ( 1989): Characterization of gamma-aminobutyric acid and dopamine overflow following acute implantation of a microdialysis probe. Life Sci 45:1301-1317. Emrich HM, Zerssen DV, Kissling W, Moller HJ, Windorfer A (1980): Effect of sodium valproate on mania. Arch Psychiatr Nervenkr 229:1-16. Ferkany JW, Butler IJ, Enna SJ ( 1978): Effect of drugs on rat brain, cerebrospinal fluid and blood GABA content. J Neurochem 33:29-33. Hamilton M (1967): Development of a rating scale for primary depressive illness. Br J Soc Clin Psychol 6:278-296. Hare TA, Manyam NVB (1980): Rapid and sensitive ion-ex change fluorometric measurement of 3,-aminobutyric acid in physiological fluids. Anal Biochem 101:349-355. Kasa K, Otsuki S, Yamamoto M, Sato M, Kuroda H, Ogawa N (1982): Cerebrospinal fluid ~/-aminobutyric acid and homovanillic acid in depressive disorders. Biol Psychiatry 17:877883. Kraemer HC, Gullion CM, Rush AJ, Frank E, Kupfer DJ (1994): Can state and trait variables be disentangled? A methodological framework for psychiatric disorders. Psychiatry Res (in press). Lambert PA (1984): Acute and prophylactic therapies of patients with affective disorders using valpromide (dipropylacetamide). In Enrich HM, Okuma T, Muller AA (eds), Anticonvulsants in Affective Disorders. Amsterdam: Excerpta Medica, Amsterdam. pp 3344. Lloyd KG, Thuret F, Pilc A (1985): Upregulation of ~/-aminobutyric acid (GABA)H binding sites in rat frontal cortex: A common action of repeated administration of different classes of
antidepressants and electroshock. J Pharmaeol Exp Ther 235:191-199. McElroy SL, Keck PE, Pope HG, Hudson JI (1992): Valproate in the treatment of bipolar disorder: literature view and clinical guidelines. J Clin Psychopharmacol 12:42S-52S. Paykel ES, van Worerkom AE, Walters DE (1991): Fengabine in depression: A placebo controlled-study of GABA agonist. Hum Psychopharmacol 6:147-154. Petty F (1986): GABA mechanisms in learned helplessness. In Bartholini G, Lloyd KG, Morselli PL (eds), L.E.R.S. Monograph Series: GABA and Mood Disorders: Experimental and Clinical Research. New York: Raven Press, pp 61-66. Petty F, Kramer G, Feldman M (1987): Is plasma GABA of peripheral origin? Biol Psychiatry 22:725-735. Petty F, Kramer GL, Gullion CM, Rush AJ (1992a): Low plasma GABA in male patients with depression. Biol Psychiatry 32:354-363. Petty F, Kramer G (1992b): Stability of plasma ~-Aminobutyric acid (GABA) with time in healthy controls. Biol Psychiatry 31:743-745. Petty F, Kramer GL, Hendrickse W (1993a): GABA and depression. In Mann JJ, Kupfer DJ (eds), Biology of Depressive Disorders, Part A: A Systems Perspective. New York: Plenum Press, pp 79-108. Petty F, Kramer GL, Fulton M, Moeller FG, Rush AJ (1993b): Low plasma GABA is a trait-like marker for bipolar illness. Neuropsychopharmacology 9:125-132. Petty F, Steinberg J, Kramer GL, Fulton M, Moeller FG (1993c): Desipramine does not alter plasma GABA in patients with major depression. J Affect Disord 29:53-56. Spitzer RL, Williams JBW, Gibbon M, First MB (1987): Structured Clinical Interview for DSM-111-R. Washington DC: American Psychiatric Press. Zachman M, Tocci P, Nyhan WL (1966): The occurrence of ~/aminobutyric acid in human tissues other than brain. J Biol Chem 241:1355-1358.
The biology of mood disorders involves ~-aminobutyric acid (GABA), a neurotransmitter whose levels in plasma likely reflect brain GABA activity. Previous research has shown that a subset of patients with primary unipolar major depression have low plasma GABA levels, which parallels findings from studies of cerebrospinal fluid. We have completed a 4-year follow-up on 46 male patients with primary unipolar depression. Plasma levels of GABA were stable over this time. For the group, mean plasma GABA levels on follow-up did not change significantly J?om entry levels. Plasma GABA levels measured on follow-up were significantly (p < .001) correlated with entry levels. Patients with low plasma GABA levels (< 100 pmol/ml) on entry into the study were likely to remain low on follow-up, and patients with plasma GABA levels in the control range (>- lOO pmol/ml) at entry similarly remained in this category (X: = 7.23, p = .007). This was true whether or not the patient had recovered from depression on follow-up. Levels o f plasma GABA did not significantly correlate with severity of depression at either entry (p = .40) or follow-up (p = .52), nor was there a significant correlation between change in plasma GABA and change in the 17-item Hamilton Depression Rating Scale score from entr), to follow-up (p = .89). These data are consistent with the notion that plasma GABA is independent of clinical state in patients with primary unipolar depression. Low plasma GABA may be a trait marker of illness in a subset ~¢patients with mood disorder.
Key Words: GABA, depression, biological markers, mood disorder, follow-up study
Introduction Several lines of evidence implicate the "y-aminobytric acid (GABA) system in mood disorders. GABA agonists appear to be effective in the treatment of both depression and mania (Lambert 1984; Lloyd et al 1985; McElroy et al 1992; Bowden el al 1994), although a negative study has also been
From the Veterans Affairs Medical Center (FP, GLK, MF, LD) and Department of Psychiatry, University of Texas Southwestern Medical School /FP, MF. LD. AJR), Dallas, TX. Address reprint requests to Frederick Petty, Ph.D.. M.D., Psychiatry Service (116A). Veterans Affairs Medical Center. 451}11South Lancaster Road. Dallas, TX 75216: Fax (214) 372-7987. Received April 15, 1994; revised August 9. 1994.
© 1995 S~xziety of Biological Psychiatry
reported (Paykel et al 1991). For bipolar disorder, GABA agonists also appear to be useful prophylactic agents (Emrich et al 1980; Calabrese et al 1993). Considering its ubiquity as a neurotransmitter (about one third of central nervous system (CNS) neurons use G A B A as a neurotransmitter), it would be surprising to not find an involvement of G A B A in mental illnesses such as depression. G A B A can be measured in cerebrospinal fluid (CSF). CSF levels (120 nM) are comparable to those in brain extracellular space (1 O0 riM) (Drew et al 1989). A meta-analysis of the seven published studies of CSF GABA in depression confirms that G A B A levels are low in patients with depression compared to controls (p = .005; Petty et al 1993a). Additional, albeit indirect, evidence from animal models strongly sup(X)06-3223/95/$09.50 SSDI 0006-3223194)00226-S
Stability of Plasma GA BA
16
BIOLPSYCHIATRY
1995:37:80~810
807
--
]4
I~ -
• '\"x
12
/
! >- 10 0z ~
8
6
Figure I. Distribution of plasma GABA in healthy controls (solid bars) and in male patients with primary unipolar depression (open bars). Best fit to data are two curves with five parameters: 1 mean and SD for all normals and 60% of depressives; 1 mean and 1 SD for 40% of depressives. (Data from Petty et al 1992.)
///
// /
/
'
,
[
d 2
~.~ j =
GABA
ports a central role for GABA in the biochemical pathophysiology of depression (Petty 1986; Lloyd et al 1985). Plasma GABA provides a useful index of brain GABA activity, since virtually all GABA production occurs in the CNS, and since plasma levels of GABA accurately parallel brain GABA levels (Petty et al 1993a for review). Low levels of plasma GABA are found in 30%-40% of patients with major mood disorder, whether manic or depressed (Petty et al 1992a, 1993b). In patients with primary unipolar depressive disorder, 40% have levels of plasma GABA below 100 pmol/ml, compared to only 6% of healthy controis. This difference in distributions is highly significant (p < .0001 ), it can best be accounted for by the existence of two distinct patient populations in depression, one with low GABA and one with normal G A B A (Figure 1). In healthy subjects, plasma GABA is stable with time and does not exhibit appreciable circadian or circumannual fluctuations (Petty et a] 1987, 1992b). If low plasma GABA is a valid biological trait marker for depression in a subset of patients, follow-up studies should demonstrate temporal stability of plasma GABA. Low plasma GABA should not revert to normal over the course of time on follow-up, regardless of clinical state. Nor should patients with initial levels of GABA in the normal range change appreciably over time. On the other hand, if low plasma GABA in depression is a state marker for low mood, revision toward the mean plasma G A B A of healthy controls should be seen with the passage of time in patients with depression. In a previous study, we found that plasma GABA levels did not change appreciably following 1 month of treatment with desipramine, although symptoms of depression improved (Petty et al 1993c). These findings are compatible with the possibility that low plasma GABA may be a traitlike marker for mood disorders in a subset of patients, since
plasma GABA does not seem to be dependent on clinical state, whether manic or depressed. Nor does plasma G A B A normalize with improved mood during acute-phase treatment of depression. We here present a 4-year follow-up of outpatients with a diagnosis of primary unipolar nonpsychotic major depressive disorder, reporting on their plasma G A B A levels over time. Methods
Subjects participating in this study were initially in the patient group of 107 males with a diagnosis of primary unipolar major depressive disorder previously described and reported (Petty et al 1992a). Of this sample, 51 were available for follow-up study 4 years later. Treatment was uncontrolled between entry and follow-up; however, most patients received standard clinical treatment during the intervening period, including antidepressant medications. At follow-up, the Structured Interview for DSM-III-R (SCID) (Spitzer et al 1987) was readministered by a trained psychiatric research nurse, and the diagnosis was reviewed by the first author. Severity of depressive symptoms was assessed with the 17item Hamilton Rating Scale for Depression (HRS-D; Hamilton 1967), and note was made of current medications. Of the subjects available at lbllow-up, 4 had become bipolar and 1 was rediagnosed with a primary anxiety disorder, leaving 46 whose diagnosis of primary unipolar depression was confirmed and stable 4 years after the initial assessment. At both entry (T1) and follow-up (T2), blood samples for plasma GABA determination were obtained by venipuncture into ethylenediamine tetraacetic acid (EDTA) vacutainer, with the plasma quickly removed and stored at - 7 0 ° C until time of assay. Quantitative analysis of plasma
808
toOLPSYCHIATRY
F. Petty et al
1995;37:806 810
G A B A was conducted with high-performance liquid chromatography using the procedure of Hare and Manyam (1980) with minor modifications, including use of y-vinyl G A B A as an internal standard. Both intra- and interassay coefficients of variation of the assay are 6% and the limit of detection (5:1 S:N) is 10 pmol/ml. With these procedures, healthy control subjects (n = 71 ) have plasma G A B A levels of 123 + 20 pmol/ml (SD) (Petty et al 1992a).
Results All patients in the present study were male, with a mean age at TI of 52.2 (SD = 18.8). Age was not significantly related to plasma G A B A at T 1 ( r = .03, p = .83) or at T2 (r = 092, p = .54). For the 46 patients, mean levels of plasma G A B A on entry (T l) were not significantly different from levels at 4-year (T2) follow-up (106 + 24 vs. 108 _+ 27 pmol/ml, respectively, paired t = .39, p = .70). There was a significant correlation between plasma G A B A levels at T I and T2 (r = .45,p < .001; Figure 2). On follow-up, 30 patients were classified as recovered or improved, defined as a Hamilton Rating Score for Depression (HRS-D) score of -<9 at T2 (n = 25) or a -> 50% drop in HRS-D score between T1 a n d T 2 (n = 5), and 16 continued to be depressed. There were no significant changes in plasma G A B A levels between entry and follow-up for either group (recovered: T I - - 1 0 9 _+ 24 pmol/ml vs. T 2 - - 1 1 0 _+ 28 pmol/ml, paired t =. 15,p = .89, depressed: T I - - 1 0 1 + 23 pmol/ml vs. 104 _+ 25 pmol/ml, paired t = 0.48, p = .64). As noted, our previous research indicates plasma G A B A levels < 100 pmol/ml to be a biological marker in a subset of patients with depression, while plasma G A B A levels >-100 pmol/ml are seen in most depressives and in 95% of healthy controls (Petty et al 1992a). Of the present sample, 18/46 (39%) had low plasma G A B A ( < 100 pmol/ml) at T 1, while 28/46 had plasma G A B A in the normal range at T1. Eleven patients had low plasma G A B A at both TI and T2, and 23
160 140
s,L,,*- <
~ 120 <~ 100 8O 6O 40 20
Figure 2. Levels of plasma GAB A in patients with primary unipolar depression at entry into study and at 4-year follow-up. See text for statistical analysis.
had normal plasma G A B A levels at both points in time, while seven patients who were low at T1 were in the normal range at T2, and five who were normal at T 1 were low at T2. This distribution was significant (X2 = 7.23, p = .007). On entry into the study, all patients had HRS-D scores of -> 10 (range 10-37), while on follow-up 27 had score of -< 9 (range 0-28), and mean HRS-D scores at T1 were significantly higher than at T2 (23 + 7 vs. 10 + 8, respectively, paired t = 9.86, p < .001). There was a significant correlation between T I and T2 HRS-D scores (r = 0.308, p = .03); however, the correlation between plasma G A B A levels and HRSD was not significant at either T 1 (r = .27, p ---.40) or at T2 (r = .097, p = .52). Nor was the correlation between change in plasma G A B A (Y2 level minus T1 level) and change in HRSD score (T2-T 1) significant (r = -.02, p = .89). At entry, 12 patients were taking psychotropic medications and 34 were not; plasma G A B A levels were not significantly different between these two groups (107:___26 pmol/ml vs. 105 +_ 18 pmol/ml, respectively, t = .29, p = .39). For the 21 patients who were not taking psychotropic medications at either T1 or T2, there was no significant change in plasma G A B A levels (111 -+ 26 pmol/ml vs. 119 +- 28 pmol/ml, respectively, paired t = 1.31, p = .20). Of the 21 patients who were off psychotropic medication at both T I and T2, 15 had improved or recovered from depression at T2 (defined as --> 50% drop in HRS-D score or final HRS-D -< 9). Plasma G A B A levels did not change significantly for these 15 nonmedicated patients who experienced a change in clinical state (113 +_ 23 pmol/ml vs. 120 + 31 pmol/ml, TI vs. T2, t = 1.31, p = .44), nor did it change significantly for the 7 non-medicated patients who did not recover from depression at T2 ( 103 -+ 12 pmol/ml vs. 110 _+ 25 pmol/ml, T l vs. T2, t = .78,p = .46). Family history for mood disorder in first degree relatives could be determined for 42 cases, while 4 patients either were adopted or had inadequate knowledge of their first-degree relatives. Plasma G A B A at TI did not significantly differ between the patients with a positive family history ( 101 +_ 26 pmol/ml, n = 17) and those with a negative family history (104 _+ 29 pmol/ml, n = 25, t = 1.10, p = .30). Similarly, at T2 these groups were not significantly different (107_+20 pmol/ml vs. 108 _+ 32 pmol/ml, respectively, t = .00, p = .99). Of the 46 patients, 16 had a single episode of depression and 30 had recurrent depression with two or more episodes, determined at T2. Plasma G A B A did not differ significantly between these groups (103 _+ 26 pmol/ml vs. 108 _+ 22 pmol/ml, respectively, t = .35, p = .55).
Discussion The major finding of the present work is that plasma levels of G A B A are stable with time over a 4-year follow-up in male patients with primary unipolar major depression. This
Stability of Plasma GABA
BIOLPSYCHIATRY
809
1995:37:806-81[)
was true for the sample as a whole, and for the 21 patients who were not taking psychotropic medications at either initial evaluation or at follow-up. Additionally, plasma GABA levels were not significantly influenced by age, psychotropic medication, or recovery of depression. These data replicate and extend our previous work demonstrating no significant change in plasma GABA in patients with major depression treated with desipramine for 4 weeks with clinical improvement (Petty et al 1993c). We have also found plasma GABA levels in healthy controls to be stable over several years (unpublished data). Interpretation of the present work should be made with great caution, particularly since this was a naturalistic study. No attempt was made to control for treatment during the 4 years of follow-up; nor did we obtain detailed data on clinical status from the patients during the course of the study. Therefore, we do not have a precise picture of the pattern of symptoms or medications during the entire 4-year followup period. The stability of plasma GABA seen in this follow-up study is compatible with the notion that plasma GABA is a biological trait marker of primary unipolar depression. The daily certainly indicate that low plasma G A B A is n o t a biological marker for the state of depression, since plasma G A B A levels did not significantly change in patients who had improved or recovered from depression on follow-up. Furthemore, there was no correlation found between change in G A B A and change in depression rating scale scores. The issue of "state" versus "trait" in biological markers is complex and has been discussed recently (Kraemer et al in press). Basically, comprehensive analysis of this problem involves a determination of the precise contribution of the variance in plasma GABA levels due to chance, state of illness, time, and so forth. What is left can be considered the proportion of variance due to trait of illness. In the present work, we can only account for 20% (r 2 = .20) of the variance in plasma GABA. We have demonstrated little variability with time, gender, exercise, diet, or season in healthy controls (Petty 1987, 1992b); however, we do not presently know if plasma GABA in mood-disordered patients is similarly unaffected by these lectors. Considerably more research in mood-disordered patients will be required for a definitive answer as to the exact proportion of the variance in plasma GABA due to the trait of having a major depression. Specifically, a prospective study is needed with multiple determinations of plasma GABA at multiple points in time during different phases of the illness. If plasma GABA is low in some depressives, and continues low at follow-up, what neurochemical mechanisms might account for this? First, it is important to realize that GABA is a very brain-specific neurotransmitter, with high concentrations in brain, and with 99% of total body GABA in the CNS (Zachman et al 1966). Based on this consideration alone,
it is plausible that a peripheral measure of GABA might reflect brain GABA activity. While it is true that GABA does not cross the blood-brain barrier, going from blood to brain, the converse is not true, since experimental manipulation of brain GABA levels in laboratory animals are followed by similar changes in plasma GABA (Ferkany et al 1978). Furthermore, there is good evidence that levels of GABA in the extracellular fluid of brain reflect synaptic GABA levels, and that GABA in CSF and plasma is in equilibrium with brain extracellular fluid GABA (reviewed in Petty et al 1993b). Therefore, we can look to regulation of extracellular (or synaptic) GABA as a model for regulation of plasma GABA. Low extracellular GABA can result from at least four possible causes: slower synthesis, less release, more uptake, and/or faster metabolism. These processes for GABA are fairly well understood in brain. If plasma GABA is an accurate reflection of synaptic G A B A concentration, similar mechanisms should regulate both. In other words, low plasma GABA in patients with depression could be caused by one or more of the following brain mechanisms. Slower synthesis of G A B A might be due to decreased activity of the GABA-synthesizing enzyme, glutamate decarboxylase. Less neuronal release of GABA might be due to presynaptic inhibition, by GABA itself, or by another neurotransmitter. More neuronal or glial uptake of GABA might be due to differences in conformation of the G A B A uptake carrier. Faster metabolism of GABA might occur with changes in activity of the GABA-metabolizing enzymes, GABA-transaminase and succinyl semialdehyde dehydrogenase. These represent a finite series of possibilities and biochemical studies could be undertaken to determine whether an alteration in any of these molecular mechanisms in brain are responsible for low plasma GABA in depression, using human postmortem tissue or animal models. Finally, it is also conceivable that patients with depression have greater hepatic clearance of GABA, or that another, perhaps unknown, peripheral mechanism regulates plasma GABA levels differently in patients with major depression than in healthy controls, though no data support this theoretical possibility. In summary, plasma GABA levels in patients with primary unipolar major depression were stable at 4-year follow-up, and did not change with clinical improvement. Low plasma G A B A may be a stable biological trait marker for this depressive illness in a subset of 40% of patients. Supported by the Department of Veterans Affairs Medical Research Service: NIH Research Grants MH37899, AA07234and MH41115; and the John SchermerhornFund. We thank Jim Mintz, Ph.D.. lk~rhelpful discussions; Glynda Mekonen and Dinah Turner-Knight fnr superb secretarial support in preparing this manuscript; and Kenneth Z. Altshuler, M.D., Stanton Sharp Distinguished Professor and Chair, fl)rhis administrative support.
810
BIOLPSYCHIATRY
F. Petty et al
1995:37:806-810
References Bowden CL, Brugger AM, Swarm AC, et al. ( 1994): Efficacy of divalproex sodium vs. lithium and placebo in the treatment of mania. JAMA 271:918-924. Calabrese JR, Woyshville MJ, Kimmel SE, Rapport DJ (1993): Predictors of valproate response in bipolar rapid cycling. J Clin Psychopharmacol 13:280-283. Drew KL, O'Connor WT, Kehr L Ungerstedt U ( 1989): Characterization of gamma-aminobutyric acid and dopamine overflow following acute implantation of a microdialysis probe. Life Sci 45:1301-1317. Emrich HM, Zerssen DV, Kissling W, Moller HJ, Windorfer A (1980): Effect of sodium valproate on mania. Arch Psychiatr Nervenkr 229:1-16. Ferkany JW, Butler IJ, Enna SJ ( 1978): Effect of drugs on rat brain, cerebrospinal fluid and blood GABA content. J Neurochem 33:29-33. Hamilton M (1967): Development of a rating scale for primary depressive illness. Br J Soc Clin Psychol 6:278-296. Hare TA, Manyam NVB (1980): Rapid and sensitive ion-ex change fluorometric measurement of 3,-aminobutyric acid in physiological fluids. Anal Biochem 101:349-355. Kasa K, Otsuki S, Yamamoto M, Sato M, Kuroda H, Ogawa N (1982): Cerebrospinal fluid ~/-aminobutyric acid and homovanillic acid in depressive disorders. Biol Psychiatry 17:877883. Kraemer HC, Gullion CM, Rush AJ, Frank E, Kupfer DJ (1994): Can state and trait variables be disentangled? A methodological framework for psychiatric disorders. Psychiatry Res (in press). Lambert PA (1984): Acute and prophylactic therapies of patients with affective disorders using valpromide (dipropylacetamide). In Enrich HM, Okuma T, Muller AA (eds), Anticonvulsants in Affective Disorders. Amsterdam: Excerpta Medica, Amsterdam. pp 3344. Lloyd KG, Thuret F, Pilc A (1985): Upregulation of ~/-aminobutyric acid (GABA)H binding sites in rat frontal cortex: A common action of repeated administration of different classes of
antidepressants and electroshock. J Pharmaeol Exp Ther 235:191-199. McElroy SL, Keck PE, Pope HG, Hudson JI (1992): Valproate in the treatment of bipolar disorder: literature view and clinical guidelines. J Clin Psychopharmacol 12:42S-52S. Paykel ES, van Worerkom AE, Walters DE (1991): Fengabine in depression: A placebo controlled-study of GABA agonist. Hum Psychopharmacol 6:147-154. Petty F (1986): GABA mechanisms in learned helplessness. In Bartholini G, Lloyd KG, Morselli PL (eds), L.E.R.S. Monograph Series: GABA and Mood Disorders: Experimental and Clinical Research. New York: Raven Press, pp 61-66. Petty F, Kramer G, Feldman M (1987): Is plasma GABA of peripheral origin? Biol Psychiatry 22:725-735. Petty F, Kramer GL, Gullion CM, Rush AJ (1992a): Low plasma GABA in male patients with depression. Biol Psychiatry 32:354-363. Petty F, Kramer G (1992b): Stability of plasma ~-Aminobutyric acid (GABA) with time in healthy controls. Biol Psychiatry 31:743-745. Petty F, Kramer GL, Hendrickse W (1993a): GABA and depression. In Mann JJ, Kupfer DJ (eds), Biology of Depressive Disorders, Part A: A Systems Perspective. New York: Plenum Press, pp 79-108. Petty F, Kramer GL, Fulton M, Moeller FG, Rush AJ (1993b): Low plasma GABA is a trait-like marker for bipolar illness. Neuropsychopharmacology 9:125-132. Petty F, Steinberg J, Kramer GL, Fulton M, Moeller FG (1993c): Desipramine does not alter plasma GABA in patients with major depression. J Affect Disord 29:53-56. Spitzer RL, Williams JBW, Gibbon M, First MB (1987): Structured Clinical Interview for DSM-111-R. Washington DC: American Psychiatric Press. Zachman M, Tocci P, Nyhan WL (1966): The occurrence of ~/aminobutyric acid in human tissues other than brain. J Biol Chem 241:1355-1358.