Back to Index of Medical Articles.

Annals New York Academy of Sciences p388-

Hans Ågren,† Lars Terenius,‡ and Agneta Wahlström‡

† Department of Psychiatry
‡ Department of Pharmacology
University of Uppsala
Uppsala, Sweden

Introduction

Brain endorphins may serve as transmitters in neuronal systems mediating reward and satisfaction. It is possible that disease states characterized by dysphoria or euphoria are accompanied by endorphin dysfunction.

To investigate this possibility, we studied the effect of the opiate antagonist naloxone in five clinically depressed patients.¹ Two patients responded with a worsening of symptoms. Endorphin content was measured in a receptor assay, and found to be elevated in comparison with levels seen in healthy volunteers.

We now report a larger series of 55 depressive patients fulfilling RDC (Research Diagnostic Criteria) for Major Depressive Disorder, diagnosed and rated following the SADS (Schedule for Affective Disorders and Schizophrenia²). The same receptor assay³ has been maintained in thee present series. Several practical considerations have lead to continued use of this assay rather than a specific radioimmunoassay. First, a radioimmunoassay may also detect biologically inert material. Secondly, potent endorphins with longer chain-lengths than the enkephalins have been and are being isolated and characterized from biological tissues.^{4, 5} These peptides usually have higher metabolic stability and they may be functionally significant. It therefore seemed prudent at the time to use the receptor assay, in expectation of the chemical identity of the measured activity. Thus, there is evidence that Fraction I activity is confined within intermediate-sized (1000-2000 MW) hydrophilic peptides whereas Fraction II contains peptides of the same or lower molecular weight, which are more hydrophobic.

Analytical Methods

The patients were prepared for spinal taps in a standardized manner. A total of 13 ml lumbar CSF were withdrawn at 8-9 am with the patient in a supine position. The fluid was kept frozen at -20ºC or lower until analysis. A 4-5 ml aliquot was filtered through an Amicon PM-10 ultrafilter (nominal cut- off 10,000 daltons) and the eluate was passed over a Sephadex G-10 column 50 X 2 cm) with 0.2 M acetic acid at a flow rate of 1 ml/min. Forty 5 ml samples were collected and elution position of the salt peak was determined by flame photometry. Fraction I (defined as three samples of 5 ml prior to the salt peak) and Fraction II (four samples of 5 ml after the peak)³ were collected and lyophilized. Combined samples were reconstituted in one ml of water and 100 m l aliquots were assayed for displacing activity against tritium-labelled dihydromorphine being allowed to interact with rat brain synaptic plasma membranes. Each sample was tested in triplicates. Details of the assay are given elsewhere.⁷ Displacement activity was read against a standard curve and expressed as equivalents of methionine-enkephalin.

The assay of cortisol in 24-hour urine and the dexamethasone suppression test performed on the patients investigated have been described earlier (Ågren and Wide.¹⁹)

Methods of Interviewing Patients

All patients were interviewed using the SADS,² for which some reliability data have been publised.⁹ The interviews lasted between 1 and 2 hours and were all conducted by the same rater in Swedish with the SADS manual in English as a guide. All patients were admitted to a research ward for a minimum of 5 days, during which a number of psychobiological investigations were performed. No diagnostic medication was given prior to the lumbar puncture. All patients were off neuroleptic or antidepressant medication for at least 10 days before investigation.

Consent had been obtained from the Ethics Committee at the Medical Faculty of Uppsala University to perform lumbar puncture as part of a routine investigatory program.

Statistical Methods

Rating scores from the SADS interview, diagnostical dummy codes, patient bodily characteristics, sex, illness duration variables as well as biological measures were stored in a computer file at the Uppsala University Computing Center, operating an IBM 4341.

Statistical programs from the SAS (Statistical Analysis System, Inc.) package like the Correlation, Stepwise, and Plot procedures were employed.

In performing a multiple regression analysis between the two endorphin fractions in turns as the independent variables and the scored depressive symptoms and other patient variables as a large number of independent variables, or predictors, certain precautionary measures have to be taken in order to reduce the so-called capitalization of chance that is always a problem in these kinds of multivariate approaches (discussed in Ågren ⁸).

Using a split-half technique, we correlated endorphin values univariately (Pearson r with all patient variables at hand in the undivided sample and in two randomly split halves (odds and evens in terms of serial number) of the patient population. Those variables displaying a trend correlation (p<0.10) when correlating the endorphin measures with the undivided group, and retaining their r values reasonably well in both split sample halves, were selected for further analysis. For instance, if a variable showed a good correlation (even p<0.01) in the undivided group, but was highly unstable after splitting (for instance, p<0.01 in one group but p = 0.50 in the other), it was not considered trustworthy as representing anything other than a chance fluctuation.

The independent variables selected in this way are thought to play some role in explaining the total variance in the CSF endorphin measures, and not to represent wholly fluctuations due to chance and the so-called Type I statistical errors. These variables were used as the independent terms (x’s) in backward stepwise multiple regression analyses with Fraction I and II, in turns, as the dependent variables (y). The final selection would retain those independent variables that showed a probability of less than 0.10 that their b coefficient did not significantly differ from null.

Results

The concentrations (mean ± 1 SD) of CSF Fraction I and II were 2.1 ± 1.3 and 10.1 ± 5.0 pmoles/ml CSF (median values 1.7 and 9.1 pmoles/ml), respectively. The distributions were skewed with outliers in the upper tails. The Kolmogorov-Smirnov test for deviation from a Gaussian distribution revealed non-normal distributions for both fractions (D = 0.14 and 0.15; p<0.01 for both).

These values should be compared with results obtained from healthy volunteers (means ± 1 SD): Fraction I 1.0 ± 0.4 and Fraction II 2.5 ± 1.5 pmoles/ml (n = 19).¹⁰ These values are not significantly different from those in the present study.

Fraction I and II displayed a highly significant intercorrelation (Pearson r – 0.525, n = 54, p = 0.0001; Spearman r₈ = 0.488, p = 0.0002), as depicted in Figure 1.

Figure 1. Intercorrelation between the two endorphin fractions.

Figure 2. Linear correlation between endorphins Fraction I (pmoles/ml CSF) and urinary free cortisol (mmoles/24h urine).

Endorphins Fraction I, but not Fraction II, correlated positively and highly significantly with various measures of corticosteroid output, most clearly with urinary free cortisol (UFC). Correlation with plasma cortisol measured at 8 am after 1 mg dexamethasone given orally at 11:30 PM the night before reached the 5% level of significance (r = 0.355, n = 33, p = 0.043). Correlation with UFC was higher, as seen in Figure 2, and here the number of cortisol analyses matched those of the endorphin analyses (Pearson r = 0.412, n = 54, p = 0.0020; Spearman r₈ = 0.344, n = 54, p = 0.011).

There was no correlation between any endorphin fraction and other biological variables measured on the patients investigated (HVA, MHPG and 5HIAA in CSF, MHPG in 24-hour urine, MAO activity in blood platelets, plasma TSH response to TRH injection).

Correlating Fraction I and II with the diagnostic dichotomies unipolar/bipolar depression (n = 37/18), primary/secondary depression (n = 38/17), pure unipolars/all others (n = 12/43), spectrum unipolars/all others (n = 6/49), sporadic unipolars/all others (n = 7/48), and with-schizotypal/without- schizotypal features (n = 7/48) revealed a significant tie between Fraction I and the unipolar/bipolar dichotomy (Pearson r = 0.285, n = 55, p = 0.036; Spearman r₈ = 0.347, p = 0.0095). This difference with higher Fraction I levels in unipolar depressions is shown graphically in Figure 3, where a nonparametric median test for differences between the two groups is shown to confirm the significant result (x² = 4.01, df = 1, p = 0.045). Fraction II displayed no diagnostic relations at all.

A backward multiple regression analysis was then performed, with independent variables to be included in the initial regression selected in the way described above.

As for Fraction I, five variables withstood the univariate selection procedure (for undivided group: p < 0.10; r in the "worst" group less than 27% smaller than that in the undivided group).

Columns A and B in Table 1 show the univariate correlations (Pearson r) between five of those nine variables inserted in the initial equation, for the undivided sample (A) as well as both split groups (B). Columns C and D show the final result in both splits after the backward stepwise procedure had eliminated most of the variables. It can be seen that the end result in each split retained one suicide measure correlating positively with Fraction II (Seriousness + Medical Lethality of Worst Suicidal Attempt in one group and Number of Suicidal Attempts in the other) and one anxiety measure correlating negatively (Somatic Anxiety Worst Week in one group and Phobias, yes/no, in the other). The overall multiple correlation was quite high in both splits (odds: R² = 0.269, F = 4.41, df = 2,24, p = 0.023; evens: R² = 0.369, F = 4.48, df = 3,23, p = 0.013).

When those five independent variables that remained in the two splits were inserted in a stepwise regression for the whole sample, the finally selected variables were those listed in Column E. Exactly the same selection was the result if the original nine variables were used in a backward stepwise regression with the whole sample. Thus, the independent variables Seriousness of Intent + Medical Lethality at Worst Suicide Attempt during Present or Recent Depressive Episode as well as Somatic Anxiety Worst Week correlated positively and Overt Anger Worst Week negatively with endorphins Fraction II. These three variables explained 28% of the total variance in Fraction II (R² = 0.281, F = 6.52 df = 3,50, p = 0.0008).

Figure 3. Difference between unipolar and bipolar depression (all patients in present or recent depressive phase) in regard to endorphins Fraction I (pmoles/ml CSF)

The Suicide score is plotted against Fraction II in Figure 4. Individuals scoring 0 were those 37 patients who had never attempted suicide. The univariate regression line shows a significant correlation ( r = 0.309, n = 54, p = 0.023). If the 17 patients who had tried to commit suicide were viewed together and compared with the no-suicide group, the correlation actually strengthened (ANOVA: F = 6.69, p = 0.013), and was evident in the very robust nonparametric median test as well (x² = 4.91, df = 1, p = 0.027).

Figure 4. Summed scores on Seriousness of Intent at Worst Suicidal Attempt during Present Depression (0-6) and Medical Lethality at the same attempt (0-6) correlated with endorphins Fraction II (pmoles/ml CSF). Broken line denotes the comparison between the no-suicidal-attempt group (score 0) and the suicide group (scores > 2).

The correlation between Somatic Anxiety Worst Week and Fraction II is shown in Figure 5 (r = 0.274, n = 54, p = 0.045). If scores 1-3 were compared with scores 4-6 the correlation once again strengthened (ANOVA: F = 9.42, p = 0.0034), and replicated in a median test (x² = 6.12, df = 1, p = 0.013).

Phobia had correlated better than Somatic Anxiety in one of the splits. If this variable was correlated with Fraction II in the undivided sample, the result was significant (r = 0.305, n = 54, p = 0.025). In Figure 6, this difference in Fraction II is clearly shown, as evidenced by a two-tailed t-test (t = 2.31, n = 54, p = 0.025) as well as a median test (x² = 4.59, df = 1, p = 0.032). Of the 22 patients with phobic symptoms during their depression nearly all were agoraphobics.

Figure 5. Correlation between rated Somatic Anxiety Worst Week of Present Depression (scores 1-6) and endorphins Fraction II (pmoles/ml CSF). Broken lines denotes a comparison described in the text between patients with anxiety scores £ 3 and scores ³ 4.

The item Overt Anger Worst Week correlated univariately and negatively close to the 5% level of significance with Faction II, as shown in Figure 7 (r = -0.264, n = 54, p = 0.054).

Discussion

The broad actions of morphine on mood and behaviour suggest that endogenous opioids may be associated with psychiatric disturbances. The field has been reviewed. ¹¹ Inaccurate sensitivity to pain has also been reported to occur in depression.^{12, 13} Fink and co- workers reported¹⁴ that the narcotic antagonist cyclazocine had antidepressant activity in an open clinical trial. Terenius et al. observed¹ that whereas naloxone given intramuscularly at 0.4-0.8 mg t.i.d. for 1 to 2 weeks had no therapeutic effects abrupt discontinuation of treatment led to worsening of symptoms in two of the six trials. In another study, chronic naltrexone treatment induced a depression-like syndrome.¹⁵ Our early pilot study¹ indicated that CSF endorphins as assayed with the present procedure were sometimes elevated in depression but there were also cases with low levels. This suggested to us that this biological variable might relate to some form of behaviour rather than a traditional diagnostic category.

Figure 6. Difference in CSF levels of endorphin Fraction II (pmoles/ml) between patients without and with phobic symptoms (almost all were agoraphobics). N = 31 and 23.

Figure 7. Correlation between Overt Anger Worst Week of Present Depression.

One general problem of interpretation of CSF analyses of chemical markers is the relevance of the measured activities for central events. In our particular case another problem is the lack of knowledge about the chemical identity of the measured activities. Until the chemical structures of the active agents are known, we cannot for instance attribute the observed activities to any particular endorphin system. Also the interrelation between Fraction 1 and II and their relative significance are unknown. We are fully aware of these limitations, and special emphasis is presently given to the identification of Fraction II endorphins, some progress already being made for Fraction I.⁶

Hypercortisolism has been found to be a concomitant of depressions with melancholic features in several recent investigations, and the dexamethasone suppression test has been hailed as a laboratory test for melancholia.¹⁶ One of us has found a significant correlation between poor cortisol suppression after dexamethasone and scored "endogeneity" of RDC Major Depressions.^{8, 19} Increased urinary free cortisol (UFC) occurred in depressives with little overt anger.^{8, 19} The ties between endorphins Fraction I and cortisol measures found in the present study, the most evident being a positive correlation with UFC, was surprising but very compatible with the known stimulatory influence of endorphins on the secretion of hypothalamic- releasing hormones and the concomitant increase in ACTH and corticosteroid output.

Endorphins as analyzed in the CSF do seem to play roles in some brain mechanisms involved in affective disorders. The difference in Fraction I between uni- and bipolar patients is particularly interesting in view of the paucity of other findings of biochemical differences between these affective subpopulations. The results could tentatively be interpreted to the effect that unipolar depression is a disorder involving endorphin dysfunction.

The ties between Fraction II and suicide, anxiety, and anger (irrespective of affective subdiagnosis) are also of great theoretical interest, and can be compared with other findings of ties between CSF monoamine metabolites and suicidal behaviour and anxiety. For example, lower levels of the serotonin metabolite 5HIAA have been correlated with suicidal attempts^{8, 17} and anxiety,^{8, 18} and lower excretion of the noradrenaline metabolite MHPG in 24-hour urine has been connected with suicidal behaviour.^{8, 20} However, no connections between endorphin and monoamine metabolite levels in CSF have as yet been discovered.

Acknowledgment

The technical assistance of Ms. Inga Hansson is gratefully acknowledged.

References

1. Terenius, L., A. Wahlström. & H. Ågren. 1977. Naloxone (Narcan^®) treatment of depression. Clinical observations and effects on CSF endorphins and monoamine metabolites. Psychopharmacology 54: 31-33.
2. Endicott, J. & R. L. Spitzer. 1979. Use of the Research Diagnostic Criteria and the Schedule for Affective Disorders and Schizophrenia to study affective disorders. Am. J. Psychiatry 136: 52-56.
3. Terenius, L & A. Wahlström. 1975. Morphine-like ligand for opiate receptors in human CSF. Life Sci. 16: 1759-1764.
4. Stern, A. S., B. N. Jones, J. E. Shively, S. Stein & S. Udenfriend. 1981. Two adrenal opioid polypeptides: Proposed intermediates in the processing of …

This is where my copy ends, sorry. There are more references obviously but I don’t have them.

Back to Index of Medical Articles.

LinkExchange Member

Free Home Pages at GeoCities

---

This page hosted by Get your own Free Home Page