Biological psychiatry has pursued hypothesized relationships between the endogenous opioid system and behaviour to a large degree by focusing on possible endogenous opioid (endorphin) diatheses in psychiatric illness.¹ Partly on the basis of the behavioural effects of exogenous opiates in man and those of endogenous opioids in animals, alteration in the endogenous opioid system have been hypothesized to be related to schizophrenia^{2, 3} as well as to the affective disorders.^{4, 5} Three principal strategies have been used clinically to test endorphin hypotheses: the administration of the pure narcotic antagonist, naloxone,^{3, 6, 7} the administration of opioid peptide agonists,^{8, 11} and cerebrospinal fluid (CSF).^15-19 Each of these strategies has inherent strengths and liabilities; the measurement of endorphins themselves is the only nonpharmacologic approach and the one that is based on laboratory methodologies.

Two major assay techniques have been employed to measure endorphins in clinical studies: the radioimmunoassay (RIA)^{12, 14, 16, 18, 19} and the radioreceptor assay (RRA). ^{13, 15-17, 20} RIA determinations have the advantage that they detect levels of specific opioids (eg., b -endorphin, or Met-enkephalin, etc). The major limitation with the RIA method lies in the fact that the antibodies commonly used show cross-reactivity to other peptide molecules while they themselves may not possess opiate-like activity (e.g., b lipotropin); results of RIA, therefore, usually represent a composite of substances, with levels expressed as immunoreactivity.^{21, 22} A further limitation with the RIA is that there are a number of endogenous opioid peptides already discovered, and despite differing anatomic locations of these compounds, there are little data indicating that any single one (e.g., b -endorphin, Met-enkephalin, Leuenkephalin, dynorphin, etc.) is more closely related to behaviour or psychiatric illness than another. The principal theoretical advantage to the RRA is that determinations are based on biological activity, i.e., stereospecific binding to the opiate receptor.²⁰ This RRA approach, while limited in providing information regarding an individual opioid peptide, does permit an assessment of "functional" opioid activity, which may then reflect overall tone or activity of the endogenous opioid system. Biochemical separation techniques have been used with both RIA ¹⁹ and RRA^{15, 17} to enhance specificity for specific opioids or groups of opioids.

While the measurement of endorphins has been applied to both SCF and plasma, SCF analysis may more directly reflect activity of CNS endorphin systems. Evidence that opioid activity in SCF reflects brain opioid system activity may be gained from the results of several experiments. It has been shown, for example, that electrical stimulation of the periaqueductal grey matter produces both naloxone- reversible analgesia in patients with chronic pain and increases levels of b -endorphin-like ²³ and enkephalin-like²⁴ material in ventricular CSF. Electroacupuncture, a treatment hypothesized to be mediated through the endogenous opioid system, has been shown to produce elevations in lumbar CSF enkephalin- immunoreactivity²⁵and in RRA-determined opioid activity.²⁶ Recently, levels of certain fractions of CSF taken prior to general surgery have been found to be predictive of the amount of post-operative morphine required by patients to relieve pain.²⁷ A diurnal rhythm of CSF opioid activity has been demonstrated in non-human primates with a pattern of increased morning and decreased afternoon levels,²⁸ a rhythm similar to that of cortisol and ACTH²⁹ as well as to the reported diurnality in human pain sensitivity.³⁰

In this paper we describe our research testing endorphin hypotheses in schizophrenia and affective disorders^16,31 and in anorexia nervosa³² using the strategy of measuring opioid activity and b -endorphin-immunoreactivity (ir) in the CSF.

All psychiatric patients in this study were inpatients on clinical research wards of the N..H (?? my copy of this text is unreadable just here) and granted informed consent to participate in this study of levels of CSF endorphins. Patients met Research Diagnostic Driteria (DRC)³³ for either schizophrenia, major depressive disorder, mania or DSM III criteria³¹ for anorexia nervosa. In addition to psychiatric patients, a group of normal volunteers granted informed consent for participation and served as a control group. These normal subjects were free from medical or psychiatric illness. All patients and normals were maintained medication-free for at least 14 days prior to study. Demographic data of these subject populations are presented in Table 1.

For all patients and control subjects CSF was obtained by lumbar punctures performed between 8:00 and 9:00 a.m. All subjects were at bedrest since awakening and fasting since the previous evening’s meal. Immediately following collection, CSF samples were frozen at -70° C. Analysis was performed on samples obtained from a pool of the first through twelfth millilitres of withdrawn CSF.

Levels of CSF opioid activity were determined by the RRA methodology of Naber et al.²⁰ This technique is based on the competition between [³H]-[D- Ala²Jenkephalin-(L-Leu-amide)⁵ and biologically active opioid ligands using crude rat brain membranes. This assay methodology has been presented previously in detail including results from gel chromatography and opioid specificity analysis.²⁰ Non- opioid peptides such as ACTH, b -lipotropin and MSH have been shown to produce negligible radioligand displacement at physiologic concentrations. All samples analyzed by the RRA were assayed in triplicate blind to clinical information; samples from the different diagnostic groups and control subjects were assigned in a balanced fashion to individual assays by nonlaboratory personnel. CSF from all subjects were analyzed by the RRA for opioid activity.

After completing RRA analysis, additional CSF was available from some subjects to perform RIA for b -endorphin. This assay was performed with reagents and antibody supplied by New England Nuclear: rabbit antiserum was prepared against synthetic human b -endorphin. The antibody demonstrates <50% cross-reactivity with b -lipotropin, <.004% with Met- or Leu-enkephalin, <.01% with a -endorphin or a -MSH. Samples were assayed blind to clinical information, in duplicate and assigned in a balanced fashion throughout one assay.

Patients who met RDC for schizoaffective type schizophrenia were considered separately from other schizophrenic types (i.e., undifferentiated, catatonic, etc.) since they had shown significant mood- related symptomatology as part of their illness. Overall, schizophrenics showed significantly less CSF opioid activity than did the normal control group: mean ± SEM were 2.92 ± 1.9 pmol/ml, and 4.01 ± 2.3 pmol/ml, respectively; p<.05, independent t-test, two tailed-Welch’s method for comparing groups of unequal variances, while schizoaffective patient (3.96 ± 2.5 pmol/ml) showed no significant deviation from normals or other schizophrenics. Further analysis revealed that the schizophrenic-normal difference was accounted for primarily by sex: a nearly twofold decrease in CSF opioid activity was found in male schizophrenics in comparison to normal male subjects (p< .005), whereas levels in female schizophrenics and normal female subjects were similar. Analysis of CSF by RIA for b -endorphin in representative subsamples of these groups, however, revealed no significant or near significant differences between any of the schizophrenic groups and normal subjects in b -endorphin (ir) (Fig. 1), suggesting that the observed difference between male schizophrenics and normal male subjects may be related to opioids rather than b -endorphin.

There were no significant correlations between CSF opioid activity or b -endorphin (ir) and age, number of previous hospitalizations, or nurses’ ratings of psychosis. Levels of CSF opioid activity and b -endorphin (ir) were not significantly related.

In its most simple form the endorphin-schizophrenia hypothesis states that an excess in endogenous opioid system activity is related to symptomology of schizophrenia. The basis for this hypothesis is indirect: cyclazocine, a mixed agonist/antagonist is known to produce naloxone- reversible dysphoria and auditory hallucinations in normal subjects, intraventricularly administered b -endorphin in rats produces an unusual behavioural syndrome reminiscent of catatonia and preliminary studies reported that the administration of the "pure" narcotic antagonist, naloxone, was associated with reductions in auditory hallucinations in schizophrenic patients.³ Over the last half-decade there have been numerous studies addressing endorphins in schizophrenia; results from this work, however, have not consistently supported the "excess endorphin" hypothesis.^{3, 7} Many studies have now used the naloxone strategy; while some groups have found results suggestive of therapeutic effects of naloxone, others have not.³ Recently a World Health Organization collaborative study⁷ reported that schizophrenic patients who were concurrently treated with neuroleptics showed significant naloxone-associated reductions in physician-rated symptomatology whereas medication-free schizophrenics showed significant naloxone-associated worsening in the BPRS subscale, "withdrawal-retardation." The results from double-blind studies of the intravenous administration of the opioid peptide, b -endorphin, have been inconclusive. Berger et al.⁹ reported significant but clinically nonapparent improvement in ratings of schizophrenic patients following b -endorphin administration; Pickar et al.¹¹ and Gerner et al.¹⁰ reported no significant behavioural effects.

The RRA method of Terenius et al. ¹⁵ has been used most extensively in studying levels of opioids in the CSF of medication-free schizophrenics. This method differs from the RRA employed in our study primarily in that, prior to RRA analysis, CSF is separated into two fraction (I and II) by gel chromatography. Although further biochemical specification of these fractions is needed, reported elution profiles suggest that Fraction I is composed of opioid(s) intermediate in size between b -endorphin and the enkephalins which, in turn, co-elute with Fraction II. Using this method, individual medication-free chronic ³⁵ and "symptom rich"^{15, 17} schizophrenics have been reported to have elevated Fraction I levels beyond the "normal range." More recently, Rimon et al.³⁶ have reporteed that acute medication-free schizophrenic patients showed a statistically significant group elevation in Fraction I levels compared to normal subjects, while chronic schizophrenics were found to have significantly lower levels than acute schizophrenics. All previous studies using b -endorphin RIA analysis of CSF from schizophrenics have studied patients who were concurrently receiving neuroleptics. In one study, acute patients were found to have elevated levels of b -endorphin (ir) while chronic patients had decreased levels, each in comparison with controls.¹⁸ In another study, no differences between schizophrenics and normal subjects were found.¹²

The results of our study do not support the notion of excess endogenous opioid system activity in schizophrenia. On the contrary, our data indicate the possibility of decreased endogenous opioid system activity, at least in male schizophrenics. The fact that we found no differences between schizophrenics and normal subjects with regard to b -endorphin (ir) may suggest that the observed decrease in CSF opioid activity in the male schizophrenics may be related to opioid(s) other than b -endorphin. In this regard our data are consistent with other studies that have used the RRA of Terenius et al.^{15, 17, 35, 37} since deviations in opioid activity in non-b -endorphin CSF fractions have been found in both acute and chronic schizophrenic patient groups. The lack of correlation between b -endorphin (ir) and RRA determined opioid activity suggest that intact b -endorphin may not be a major contributing factor to total opioid activity.

Our data may be consistent with the results of the WHO collaborative project in which naloxone administration produced worsening in medication-free schizophrenics, rather than improvement. While further analysis of our data with regard to clinical variables in progress, a current working hypothesis includes the notion that decreased endogenous opioid activity in some patients may be related to anhedonic features of the schizophrenic illness such as emotional withdrawal and poor interpersonal relatedness.

Results of RRA and b -endorphin RIA analyses of CSF revealed no significant differences between depressed or manic patients and normal volunteers, or significant differences between depressed and manic groups for each variable (Table 2). In four manic-depressive patients in whom paired samples were available from closely associated depressed and manic periods, however, CSF opioid activity has higher in mania than during depression in each subject: mean ± SD were 3.81 ± 0.68 pmol/ml and 1.91 ± 0.51 pmol/ml for mania and depression, respectively (p<.05, paired t-test, two tailed).

In examining possible relationships between CSF opioid activity and symptomatology in depressed patients, a significant correlation was found between research ward nurses rating ³⁷ of anxiety on the day prior to LP and CSF opioid activity (r = .46, p<.05). These ratings reflect assessment of anxiety from the perspective of observed behaviour in a research ward setting.

As part of studies investigating abnormalities of the hypothalamic-pituitary-adrenal (HPA) axis in affective illness, determinations of urinary free cortisol were made by RIA in subgroups of depressed patients and normal volunteers (means of two 24 hr urines). These determinations were made from urine samples during the same medication-free period as were the lumbar punctures. We observed that CSF opioid activity was significantly related to mean urinary free cortisol excretion (MUFC) in the depressed patients (r=.47, p<.05) but not in the normal volunteers (r=-.01, NS) (Fig 2). In the subgroup of depressed patients in whom both b -endorphin (ir) and MUFC was available for analysis, a direct relationship was also found (r=.32), which, although not reaching statistical significance, was in contrast to the slight negative relationship between normals and CSF b -endorphin (ir) and MUFC (r=-.10). MUFC was significantly higher in depressed patients (80.0 ± 8.5 mg/24-hr) than in normal volunteers (56.4 ± 4.3 mg/24-hr) (p<.05) (Fig. 2)

Figure 2. The relationship between mean urinary free cortisol (MUFC) and CSF opioid activity in depressed subjects and normal control subjects.

The notion that the endogenous opioid system may be involved in affective disorders stems largely from the considerable mood-altering properties of exogenous opiates coupled with animal experimentation relating the endogenous opioid system to reinforced behaviour.^{4, 5, 38} Specifically, the ability or inability to experience pleasure has been suggested to be related to relative increases and decreases in endogenous opioid system activity. In this regard, enhanced opioid activity might be expected to be related to mania while decreased activity related to depression. Reflecting this view, most clinical stuudies of mania have involved naloxone administration, while those in depression have used the strategy of administering opiate agonists (exogenous as well as endogenous).³⁹

Following initial clinical studies in which naloxone was found to have therapeutic effect in manic patients,⁶ there have been several studies that have found no significant behavioural effects of naloxone in manic patients.^{7, 40} There have been two reports suggesting that opioid activity differs with state change in manic-depressive illness. Lindström et al.¹⁷ reported CSF Fraction I opioid activity to be elevated during mania in comparison to depression in several manic-depressive patients. Pickar et al.¹³ found significant elevations in plasma opioid activity during mania compared with depression in a cycli medication-free manic-depressive patient. Although these has been little study of depressed patients with either naloxone or endorphin measurement strategies, several studies have reported the effects of opiate agonist administration in depression. Antidepressant effects of intravenously administered b -endorphin in open or single blind studies have been reported.^{8, 41} Of the two studies using double-blind methodologies, one reported significant b -endorphin-associated improvement in depressed patients,¹⁰ while the other¹¹ reported no significant behavioural effect. The acute administration of exogenous opiates has been tried in individual depressed patients;⁴² data to date do not support antidepressant effects. Chronic oral opiate (methadone)administration has been reported in a refractorily ill depressed patient with some antidepressant effects ovserved.³⁸

The results of our CSF study do not point to an abnormality per si in the endorphin system in either depression or mania. The increased opioid activity observed during mania compared to depression in individual patients is consistent with some previous reports, and also suggests the possibility of a relative increase and decrease in the endogenous opioid system activities across state change. The relationship between nurses’ ratings of anxiety and SCF opioid activity in depressed patients is of interest, although it is difficult to know the significance of this finding. It is possible that this relationship may be part of a common stress response. It is also possible that relative alterations in endogenous opioid activity are related directly to the biological basis of anxiety, perhaps through interrelationship with other neurotransmitters such as the noraadrenergic system.

The observed direct relationship between MUFC excretion and CSF opioid activity is of interest for two reasons. First, there are considerable data from basic science experiments that suggest physiological relationships between the HPA axis and the endogenous opioid system.^{43, 44} Second, activation of the HPA axis has been a major focus of research in depression for a number of yeas.^{45, 46} Our data suggest that while the endogenous opioid system itself may not be abnormal in depression, it may be related to abnormality of the HPA axis found in this illness.

We used three patient groups in our study of CSF opioid activity in anorexia nervosa: anorectic patients currently hospitalised for treatment of weight loss (N = 5), recovered anorectics (at least 80% of ideal body weight) brought into the hospital for study (N = 8), and normal female control subjects studied during the first week of their menstrual cycle (N = 8). Ill anorectics were studied initially at minimum weight and again following refeeding prior to hospital discharge (80% of ideal body weight). Each anorectic showed greater CSF opioid activity when at minimal weight in comparison with levels following refeeding (p<.02, paired t-test, two tailed). The mean CSF opioid activity at minimal weight was also significantly greater than the mean levels of recovered anorectics and normal control subjects (p<.01, independent t-test, two tailed), while recovered anorectic patients had levels comparable to those of controls.

There is considerable evidence from animal experimentation suggesting that the endogenous opioid system may play a rode in eating behaviour. b -endorphin has been found to stimulate food intake in satiated rats when injected into the ventro-medial hypothalamus.⁴⁷ Naloxone given intraperitoneally reduced food intake in starved rats,⁴⁸ and abolished overeating in genetically obese mice and rats.⁴⁹ To date there have been few clinical studies of anorexia with regard to a possible endorphin diathesis, although there is a large body of work relating anorexia to abnormalities in neuroendocrine systems.

We observed pronounced elevations in CSF opioid activity in the anorectic patients when at minimal weight in comparison to levels found after refeeding, as well as those of recovered anorectics and controls. These data do not support a simple relationship between opioid activity and eating, since levels were highest when subjects were at minimum weight. It is well known clinically, however, that even in the starved condition anorectic patients demonstrate major preoccupation with food and eating behaviour, although the caloric intake is decreased. Furthermore, periods of increased food intake as well as gorging of food (bulimia) are known to occur in these patients. It is a possibility that the high levels of CSF opioid activity observed in the anorectics at minimal weight may reflect some alterations in endogenous opioid system-eating behaviour relationship. It is also a possibility that the elevations in CSF opioid activity during starvation represent a stress response, since the opioid system has been hypothesized to play a role in aiding survival in famine by conservation of nutrients and water and decreasing energy-expending activities.⁵⁰ Recently, abnormalities of the HPA axis were reported to occur during periods of minimum weight in anorectics.⁵¹ Further investigations might focus on relationships between the endogenous opioid system and the HPA axis in anorectic patients.

In this paper we have reported the results of studies in psychiatric patient groups using the strategy of measuring opioid activity and b -endorphin (ir) in CSF. Our findings do not lend support to the notion of excess endorphin activity in schizophrenia, but rather suggest the possibility of a decrease in endogenous opioid activity in some schizophrenic patients. In affectively ill patients our data suggest that there may be a relative change in endogenous opioid system activity across state change in manic-depressive illness. We also found a relationship between nurses’ ratings of anxiety and CSF opioid activity in depressed patients, although it is unknown whether this directly relates to the pathophysiology of this symptom, or is related to stress response. The relationship between CSF opioid activity and HPA axis activity, as reflected by urinary free cortisol excretion, supports the notion of important physiologic relationships between these systems and raises the issue of a role for the endogenous opioid system in the abnormal activation of this system in depression. Finally, the finding of increased CSF opioid activity in anorexia nervosa patients when a minimum weight coupled with data relating endogenous involvement of the endogenous opioid system involvement in this illness.