Endogenous opioids in cerebrospinal fluid of opioid-dependent humans

BIOL PSYCKJATRY 1988;24:649-662

i$$g

Endogenous Opioids in Cerebrospinal Fluid of Opioid-Dependent Humans Charles P. O’Brien, Lars Y. Terenius, Fred Nyberg, A. T. McLellan, and Ingrid Eriksson

Endogenous opioid systems may be altered as a consequence of addiction, but evidence to support this idea is meager so far. We obtained I36 cerebrospinal$uid (CSF) samples from 72 opioid addicts during four distinct states: methadone maintenance, detoxtjkation from methadone, opioid antagonist treatment, and drug-free status. CSF endorphins were measured in 86 patient samples using a radioreceptor assay ~~), and ~-endorphin levels were measured in 85 patient samples using a radioimmuno assay (RIA). During detoxification, both BRA fraction I and B-endorphin showed a generally similar pattern of changes. Both were lowest when measured 40-50 hr after the last opioid dose, and both showed an apparent rebound to higher than methadone maintenance values at 60-70 hr following the last dose. During methadone maintenance and drug-free states, the addicts’ levels offraction I IHA en~rphins in the CSF were higher than levels found in a normal control group. Fraction II endorphins were also elevated in the addicts who were drug free. In contrast, CSF B-endorphin during both methadone maintenance and drug-free states was lower in the addicts as compared to the normal, drug-naive group. Except for the pattern found during detoxification, there were no consistent changes in endorphin levels across dierent states of addiction.

Introduction More than a decade ago, specific opioid receptors (Pert and Snyder 1973; Simon et al. 1973; Terenius 1973) and endogenous opioids (Terenius and Wahlstr6m 1974; Hughes et al. 1975; Pastemak et al. 1975) were discovered Although there has been considerable progress in ~de~~ding the ~s~bution, biosynthesis, and physiology of endogenous opioids, it is as yet unclear how, if at ail, they are affected by tolerance and dependence on exogenous opioid drugs. Studies of animals after acute or short-term treatment with morphine have failed to find consistent changes in endorphin concentrations (Childers et al. 1977; Fratta et al. 1977; Wesche et al. 1977). Animal studies of specific brain regions during withdmw~ (Bergs~~m and Terenius 1979) or after chronic high-dose mo~hine (Przewlocki et al. 1979; Wuster et al. 1980) have found reduced levels of certain opioid

From the Veterans Administration Medical Center, University of Pennsylvania, Philadelphia, PA (C.P.O’B., A.T.McL.), and UppsaIa University, UppaIa, Sweden (L.Y.T., F.N., I.E.). Address reprint quests to Dr. Charles P. O’Brien, Psychiatry Service, Veterans Adm~s~don Medical Center (II@, University and Woodland Avenues, ~~ei~ia, PA 191W. Sum in part by NIDA Grant DA 01503 and by a grant from the VA Medical Research Service. Received October 1, 1987; revised January 6, 1988. 8 1988 Society of Biological Psychiatry

QOO&3223/88/$03.50

650

BIOL PSYCHIATRY 1988;24:649-662

C.P. O’Brien et al.

peptides, but the effects may have been produced by stress rather than by a specific opioid effect. Studies of human subjects have also been conducted, although the large number of variables and small sample sizes involved make interpretation of the data difficult. Emrich et al. (1983) reported increased P-endorphin immunoreactivity in the plasma of addicts in withdrawal. Clement-Jones and colleagues (Clement-Jones et al. 1979) found cerebrospinal fluid (CSF) P-endorphin activity to be elevated in five of six addicts studied during mild withdrawal. Met-enkephalin levels, however, were not elevated. Twelve patients in the Clement-Jones study who were in withdrawal gave 22 plasma samples, and 19 of these showed elevated plasma Q-endorphin-like immunoreactivity. In contrast, Ho and colleagues (Ho et al. 1980) found greatly decreased P-endorphin-like immunoreactivity in plasma from 19 men with unspecified levels of dependence of heroin. Holmstrand and associates (Holmstrand et al. 1981) studied CSF of 17 formerly dependent subjects at various times after stopping opiates. Using a radioreceptor assay (RRA), these investigators found great variance (both high and low levels) in their patients as compared to controls. When these authors reexamined the CSF endorphin levels of their subjects after 3 weeks of daily methadone, some showed marked increases in CSF endorphin concentrations. A major problem in relating endorphin activity to human opioid addiction is the question of what are the appropriate means to assess the endogenous opioid system. The opioid peptides derive from three separate genes, and at least 20 peptides with opioid-like activity have been identified. Immunoassays are available for many of these peptides, and these assays have the advantage of specificity for at least a part of the molecular structure of each peptide. The radioimmunoassay (RIA) approach has disadvantages because, depending on the antiserum, cross-reactivity with other opioid or nonopioid peptides may occur, and this would cause a misleading measurement. Also, it would be impossible to assay a biological sample for all known opioid peptides due to limitations of sample quantity and available antisera. Even if assaying for all known opioid peptides were practical, substantial endogenous opioid activity in CSF is not accounted for by known immunoassayable peptides. A second common assay approach, the RRA, has the potential for identifying all opiate receptor-active material in a sample of body fluid. In the RRA, one measures the degree to which unknown endogenous opioids in a biological sample compete with the binding of a radiolabeled index agent in a standard tissue preparation containing opiate receptors. A major disadvantage of the RRA method is that although total opioid activity in the sample can be measured, the identity of the substance producing the activity at the receptor is unknown. The specificity of the RRA can be increased by using a simple chromatographic separation procedure prior to the assay (Terenius and Wahlstrom 1975). In this article, we report CSF endogenous opioid activity using both types of assays: RRA values for two chromatographic fractions and RIA measures for one peptide, P-endorphin.

Methods Subjects Subjects were 72 men in treatment at the Drug Dependence Treatment Center of the Philadelphia Veterans Administration Medical Center. Sixty percent were black; the average age was 32.2 t 4 years (variances are expressed ?SD), with an average of

CSF EndogenousOpioids and Addiction

BIOL PSYCHLURY 1988;24549-662

651

11.8 + 4 years of opioid dependence, including an average of 5.8 + 2.6 years on methadone maintenance treatment at a modal dose of 35-45 mg. All subjects were fully informed of the experimental procedure, and then they were required to score 100% on a written quiz concerning the nature of the study. Subsequent to the quiz, the patients gave their consent to participate in writing. A total of 142 spinal taps were performed on these 72 subjects while they were hospitalized. Subjects were encouraged to participate repeatedly, especially when they made the transition from one stage of addiction to another (i.e., methadone maintenance to detoxification). All subjects were required to provide an observed urine specimen for drug screening prior to the spinal tap. Six of these were positive for illicit drugs, and the corresponding data were therefore discarded. In addition, six subjects who were anxious about the spinal tap procedure were given benzodiazepine medication the night prior to the spinal tap. As systematic differences in endorphin levels have been observed after benzodiazepines (Wuster et al. 1980), specimens from these six subjects were omitted from the report. The remaining 136 samples included in the present analyses were divided into four groups based on the following criteria. Methadone Maintenance. Subjects were included in this group if the spinal tap had been performed 24-30 hr since their last methadone dose ana’ the dose had remained constant for at least 1 week. Detoxification. Subjects were included in this group if they were in the process of detoxification, if there had been at least a 30% reduction from their stabilized methadone dose over the preceding week, and if they had received this reduced dose 96 hr or less prior to the spinal tap. This group included subjects who agreed to abruptly stop methadone at doses up to 50 mg/day and then have a sample of CSF collected 48 to 96 hr later while acute withdrawal symptoms were present. Drug-Free. Subjects were included in this group if they had been detoxified and had ingested no methadone or other drug since detoxification. The verified drug-free period prior to the spinal tap averaged 95 days ( + 58) and was more than 30 days in all subjects. Naltrexone. Subjects included in this group had been on 50 mg/day of this long-acting opioid antagonist for at least 2 weeks prior to the spinal tap. Normal Controls. CSF was obtained from nonaddicted subjects using a procedure identical to that used for the patients. The control samples were obtained from Swedish volunteers who were white and had no history of drug abuse or of any neurological or psychiatric disorder. The 38 samples for RRA control were obtained from 18 men and 20 women. The 16 samples for RIA control were obtained from 6 men and 10 women.

Lumbar Puncture

Procedure

Cerebrospinal fluid (CSF) was obtained between 8:30 and 9:30 AM by lumbar puncture with the patient in the lateral decubitus position. A 22-gauge spinal needle was inserted into the L3-L4 interspace after local anesthesia with 2% lidocaine; 16 ml of CSF was collected in 4 tubes, centrifuged in a refrigerated centrifuge, and the supematant was frozen at - 60°C.

652

BIOL PSYCHIATRY 1988;24:649-662

C.P. O’Brien et al.

Assay Procedures RRA. After thawing, a 4.0-ml portion of the CSF was extracted with methylene chloride to remove methadone and methadone metabolites if present. The recovery of the extraction was checked by mass spectrometry using an LKB 2091 GUMS equipped with a multiple ion detector. More than 98% of methadone was recovered in methylene chloride. Any remaining methadone is retained on the chromatographic column and would elute as a broad peak, partly overlapping with fraction 11. Samples from subjects receiving naltrexone were extracted at room temperature, as this has been found necessary to remove naltrexone metabolites that may interfere with the RRA. Next, the extracted sample was separated on a Sephadex GlO column into fractions described below and then assayed in a mu-receptor assay (Terenius and WahlStrom 1975). The receptor source was a synaptic plasma membrane fraction obtained by sucrose gradient centrifugation from synaptosomes prepared from whole rat brain minus cerebellum. The labeled index ligand was 3H-dihydromorphine (nominally 1,7,8-3Hdihydromorphine, specific activity 60 Ci/mmol, from Amersham, Arlington Heights, IL). Incubations occurred at 25°C for 20 min. At each experimental occasion, a standard curve with Met-enkephalin was included. The separation procedure yields CSF opioid peptides in two major fractions---I and II-which account for more than 75% of the total opioid activity. The biochemical characteristics of these two fractions have been investigated intensively (Nyberg et al. 1983, 1986). Figure 1 presents an overview of the characteristics and concentrations of the two fractions and their relationship to opioid peptides of known structures. Receptor activity is present in three areas: (1) in fractions 43-47, close to the void volume, representing high molecular weight material (~3,000 daltons) that probably interacts nonspecifically in the receptor assay (such material is routinely removed in the analytical procedure by ultrafiltration); (2) fraction I containing hydrophilic peptides with more than eight amino acids, such as dynorphin or its fragments; and (3) fraction II, containing enkephalyl-hexapeptides, the major component of which is Met-enkephalin-Lys6 (Hughes et al. 1975) and enkephalyl-heptapeptides and octapeptides. Figure 1 also shows results obtained with radioimmunoassay for the “common” endorphins P-endorphin, dynorphin A, and Leu- and Met-enkephalin in normal human CSF samples. Note that the levels of these known peptides are very low (less than 0.2 pmol/ml by radioimmunoassay). One might argue that it is not permissible to make direct quantitative comparisons between immunoassay and receptor assay results because both the concentration and the affinity of the peptides are important variables. However, after separation of the individual components in fraction I, we have confirmed that dynorphin A is a minor component and that P-endorphin is not included in this fraction. For fraction II, the various enkephalyl peptides have affinities close to those of the enkephalins. Thus, because of significant receptor activity not measured by immunoassay we find it to be reasonable to continue the receptor assay for clinical correlations until sufficient chemical information about the various peptides has accumulated and precise measurement of all opioid components can be achieved. @EndorphinRIA. Two milliliters of CSF was diluted with a pyridine formate buffer (pyridine 0.018 M, formic acid acid 0.1 M) and run through a l-ml SP-Sephadex C-25 column equilibrated in the same buffer. The column was washed with pyridine formate buffers (8 ml pyridine 0.1 M, formic acid 0.1 M, then 4 ml pyridine 0.35 M, formic acid 0.35 M), and finally eluted with 4 ml pyridine 1.6 M, formic acid 1.6 M buffer. The eluate was evaporated in vacua in a Savant-Vat evaporator. The residue was dissolved

CSF Endogenous Opioids and Addiction

BIOL PSYCHIATRY 3988;24549-662

I I t

2

d : a

0.3

: = .

r,

: so.2 80 cE f ._ e; t

al =

-

0.1

Fracllon

Figure 1.

number

_

653

654

BIOL PSYCXLKi’RY 1988;24:649-662

C.P. O’Brien et al.

in 100 ~1 methanol:O. 1 M HCl(1: 1) and was divided into 25p.1 aliquots that were measured by radioi~unoassay in triplicate. Samples with a known content of ~-endo~hin were run in parallel with the samples with unknown content. Interassay variation was 10%. All values are corrected for recovery. @Endorphin was assayed using the charcoal-absorption technique, as previously described (Nyberg and Terenius 1983). The ~tise~m 42F was raised against human pendorphin. It showed no cross-reaction (CO. 1%) with enkephalins, dynorphins, or substance P; it showed full cross-reaction with Beta lipotropin (P-LPH). Antiserum, final dilution 1:30,000, and standard or sample were incubated for 24 hr in 200 ~1 buffer (Naphosphate 0.05 M, pH 7.5, NaCl 0.15 M, 0.02% Na-azide, 0.1% gelatin, 0.1% bovine serum albumin, 0.1% Triton X-100); then 8000 cpm ‘251-B-endorphin was added in 25 ~1 buffer and incubation continued for 20 hr. Incubation was terminated by treatment for 10 min with dextran-coated charcoal to separate anti~y-Lund and free peptide. After cent~fugation for 1 min in a Beckman Microfuge, 300 $ of the supematant was counted in a gamma spectrometer. The isotope was prepared by the chloramine-T procedure using carrier-free Na1125.A monoi~inated product was isolated by high-buoyance liquid c~omatography and was used in radioimmunoassay.

Results Between-Group Comparison The initial analyses were designed to determine if differences existed in the CSF levels of fractions I and II among the three addict samples and between each of these groups and the group of normal controls [no sex differences (p > 0.10 fraction I, p > 0.10 fraction II) were found, so male and female results in the control groups were pooled]. Analysis of Variance (ANOVA) was used to compare the mean values of fractions I and II across the four groups. When the overall analysis indicated a significant ~twee~-soup effect (p < 0.05), Duncan’s Multiple Range Test (Dixon and Brown 1979) was used to make paired comparisons. The mean values ( -I-SE) for both RRA fractions are presented for all groups in Table 1, with the results of the statistical analyses presented in the last column. [Nure: the detoxification group will be shown separately, as this was a group in flux, and there was evidence that endogenous opioid levels varied over the four-day period following the subject’s last dose of methadone (see below), thus precluding these subjects from being analyzed as a single group.1 When the methadone maintenance, drug-free and naltrexone groups were compared with the normal subjects, several differences emerged. First, all three addict groups showed higher mean endogenous opioid levels on the RRA measures (fraction I, fraction II) than the normal control group. In the case of fraction I, all paired comp~sons between the addict groups and normals reached statistical significance (p < 0.05), whereas there were no significant differences among the three addict groups (p > 0.10). In the case of fraction II, paired comparisons showed that both drug-free and naltrexone patients had levels si~ific~tly higher than the normal subjects. Subjects on me~adone main~nance, however, were not significantly different from the control group (p > 0.10). The fraction II levels of patients on naltrexone were significantly higher than those of patients who were on methadone maintenance or patients who were drug-free Cp < 0.02). ~-Endo~~n levels, as measured by RIA (Table l), were signific~tly tower in the

655

BIOL PSYCHIATRY

CSF EndogenousOpioids and Addiction

1988;24:649-662

Table 1. BBA and Beta-EndotphinBL4 Measures in CSF Methadone maintenance

ANOVAb Drug-Free

RRA of endogenous opioid levels in CSF” n 29 22 Fraction I 1.8 + 0.3 1.4 f 0.2

Fraction II

7.2 f 5.7

10.6 ” 2.0

&endorphin RL4 measures in CSF’ n 26 20 P-endorphin 13.3 f 0.9 11.6 ~fr0.8

Naltrexone

Normals

11 1.4 f 0.3

38 0.9 f 0.06

30.2 ” 10.2

5.3 f 0.8

8 14.2 + 0.9

16 21.8 f 0.3

paired comparisons

Meth Mnt > Norm, p < 0.002 Drug-free > Norm, p < 0.001 Naltrexone > Norm, p < 0.005 Drug-free > Norm, p < 0.006 Naltrexone > Norm, p < 0.0001 Naltrex > Drug-free, p < 0.02 Naltrex > Meth Mm, p < 0.001

Norm > Meth Mnt, p < O.ocCll Norm > Drug-free, p < O.oool Notm > Nakrexone, p < 0.02

“Mean fraction I or II levels in pm&ml Met-enkephalin equivalents f SE. Detoxification group (n = 26) is Figure 2. %ne-way ANOVA. ‘Mean p-endotphin levels, fmoYml k SE. Detoxification group (n = 30) is shown in Figure 2.

shown

in

three stable addict groups (methadone maintenance, drug-free, and naltrexone) than in the normal subjects, and there were no differences among the three stable addict groups.

W~r~~n-~ruu~A~lyses The second set of analyses were designed to explore relationships between each endogenous opioid value and a range of patient baek~und and current status variables in order to identify those factors that could account for the observed intragroup variability in CSF results. As indicated, great care had been taken to keep the conditions surrounding the spinal tap as standard as possible. Thus, factors such as absence of medications, verification of drug status by frequent urine screens, hospitalization before the test, time of the spinal tap, and the spinal tap @ocedure itself were held relatively constant. A full range of patient variables was collected by an experienced interviewer and by the physician (CO’B.) prior to each spinal tap on the addicts. Table 2 presents correlations between the endogenous opioid values and a range of these patient variables for the methadone main~n~ce, detoxification, and drug-free groups. The naltrexone group was not included in this analysis due to the relatively small sample size. Two patient variables are composite measures derived by combining observer- and subject-rated measures. For example, the measure of ~ir~r~~~ ~~pr~~ has been derived from the ~mbination of a self-rating,

0.05

0.20 0.20

-0.09 0.15 0.37”

0.13

0.26

0.07 0.14 0.30

-0.03

0.14

0.15

Fraction I Fraction II (n = 29) 0.09

0.18 0.46h 0.24

0.25

0.39”

0.07

0.35” 0.21 0.21

-0.10

0.03

0.20

- 0.03

0.22 0.42” 0.10

0.41&

0.25

0.25

0.29

P-endorphin fn = 30)

0.38“ 0.00 0.23

0.22 0.25 0.29 0.00 - 0.02 0.02 0.23

-0.23 -0.14 -0.20 0.04

Fraction I Fraction II Cn = 22)

Drug-free

and drug-free groups, althoughsymptom scores are much lower than in detoxification group.

0.08 -- 0.34* - 0.26

- a.02

-0.09

0.20

- 0.21

@-endorphin (n = 26)

Detoxification

Opioid Levels

Fraction I Fraction II (n = 26)

Status Variables and Endogenous

Methadone maintenance

between Addiction

cWi&&walsymptoms ratedin methadone maintenance

Age Race (0 = W, 1 = B) Years addicted Mg last methadone dose Time since methadone Withdrawal symptoms Beck Depression Inventory

Variable

Table 2. Correlations

0.00 0.04 0.30

-- 0.43” 0.27 -- 0.25 -0.23

p-endorphin in = 20)

BIOL PSYCHIATRY 1988;24549-662

CSF Endogenous Opioids and Addiction

657

CSF Fraction I during Withdrawal 0

=

2.5-

i V

2.0-

';1 1.50 ‘3

c! R

l.O'0.50.0

t :

0

I

l t I

0 I

1

1

20

100

Hours sin: last mzt)hadone i&e y = 5.1304

- 0.3125x

+ 0.0075~~2

- 6.394e-5xA3

+ 1 .784e-7xA4

R = 0.54

Figure 2A.

the Strong Opiate Withdrawal Scale, the Addiction Research Center Inventory (Haertzen 1974), and a physician rating of withdrawal signs. The depression symptom measure has been derived from the Beck Depression Inventory (Beck and Beamesderfer 1974) (selfrating} and an observer rating of depression signs. The combined results from the correlation analyses showed few patient variables that were consistently related to any of the three endogenous opioid measures. These results suggest that the patient variables selected for study, despite their close conceptual relationship to addiction and to the hypothesized effects of endogenous opioids, were generally not well correlated with the endogenous opioids measured during specific stages of the addiction cycle. The only potential exceptions were the modest general ~lationships seen between the composite measure of depressive symptoms and the RRA measures of fractions I and II. Another notable correlation was the positive correlation between severity of withdrawal symptoms and level of fraction I in detoxifying subjects (r = 0.4&p < 0.01). As the subjects were likely to be undergo~g the greatest changes in their endogenous opioid systems during the detoxification period, and as this was also the period in which we observed the greatest variability in levels of CSF endogenous opioids, the detoxification period was examined in more detail. To this end, each set of endogenous opioid values for the group of patients studied during detoxification was plotted separately according to the hours since last dose of methadone (Figures 2A-C). Increased variance is observed during detoxification, even during the first 24 hr after methadone, possibly because the patients’ dose had already been significantly reduced. The values (+-SE) for the other groups (maintenance, naltrexone, drug-free, and controls), who were in a stable state mtber than detoxification, are indicated‘on each graph for comparison.

658

BIOL PSYCHIATRY 1988;24:649-662

C.P. O’Brien et al.

CSF Fraction II during withdrawal

50

l

l

0

I

0

.

y = 30.1006

20

1

40

I

60

l

P I

80

1 100

Hours since last methadone dose - 2.3751x + 0.073~~2 - 7.886e-4xA3 + 2.725e-6xA4 R = 0.48

Figure 2B.

In order to assess the nature of the relationship between the hours since last methadone dose and each of the three endogenous opioids, we calculated a series of regression equations to determine the line of “best fit.” In the initial analyses, linear regression of hours-since-methadone was performed on each of the CSF endogenous opioid measures. Analysis of the residuals from these analyses indicated that the line of “best fit” (nature of the relationship) was not linear across the full range of hours. As is standard in these situations (see Dixon and Brown 1979), we turned to second-order, curvilinear regression analyses to develop a more representative characterization of the relationships seen. Results of these curvilinear functions are presented in Figure 2A-C. There appears to be increased variance during the first 24 hr (compared to maintenance) and then an increase (?rebound) for all three endogenous opioid measures during the interval 60-80 hr. The levels of all endogenous opioids measured recovered to approximately a predetoxification range during the drug-free state, suggesting that the reduction/recovery cycle is complete by 100 hr.

Discussion To our knowledge, this study involves the largest series of human opioid addicts whose endogenous opioid activity has been measured. In some aspects, the results are surprising in that large changes in opioid intake (methadone maintenance state versus drug-free state) are reflected in relatively small changes in CSF endogenous opioid levels. Clearly.

CSF Endogenous Opioids and Addiction

BIOL PSYCHIATRY 1988;24:649462

659

CSF B-endorphin during withdrawal 40 -r .

3 *a E

. . .

30 -

“0

20

40

60

60

Hours since last methadone dose

100

y = 7.4977 - 0.0728x + 0.0103~~2 - 9.926e-5x*3 R = 0.52 Figure 2C.

there is no evidence from these data for general suppression of endogenous opioid activity as a result of large doses of exogenous opioids. The assays performed on these 72 patients, with a mean of almost 12 years of opioid dependence, show significant differences from the 54 Swedish normals. The stable addict patients (methadone maintenance, drug-free, naltrexone) have signi&antly higher mean CSF levels of the two RRA fractions than the normals. In contrast, CSF 6-endorphin levels were significantly Lower in the addicts as compared to the normals. The normals were approximately the same age as the addicts, but they differed from the addicts in sex ratio, race, and ethnic background. The sex ratio difference may not be a serious problem, as men and women in the normal group did not differ significantly in levels of @endorphin, but the other differences prevent clear interpretations between the Swedish controls and the Philadelphia addicts. Clearly, a control group matched in all demographic characteristics is needed to make valid comparisons of endogenous opioid levels. The finding of lower 3-endorphin values in all addict groups except the detoxification group is in contrast to the findings of Kosten and colleagues (1987), who found significantly elevated CSF ~-endo~hin values in 11 methadone-m~n~ined patients in New Haven, CT. Our 26 methadone maintenance patients who had CSF 6-endorphin levels determined had values much lower than controls. The major differences between the two groups of methadone maintenance patients were that the Philadelphia patients were receiving half as much daily methadone and a greater proportion of the Philadelphia patients were black (60% versus 27%) and had been on methadone longer (5.8 versus 2.5 years). In the samples obtained during detoxification, the results were particularly interesting.

660

BIOL PSYCHIATRY t988;24649-662

C.P. O’Brien et al.

The values for the three endogenous opioid measures are shown in Figures 2A-C. The subjects were in the process of detoxification during the period prior to the spinal tap, which might explain why they showed greater variance in the first 24 hr than does the methadone maintenance group. For all three measures of endogenous opioid activity, there appears to be an increase at 40-80 hr after the last dose of methadone. presumably, this falls again, as the drug-free values are back to the range seen in the methadone maintenance groups. The finding of a roughly similar pattern for the three separate measures of endogenous opioid activity suggests that there may be some release of suppression of the endogenous opioid system when the methadone is removed. The finding of low P-endorphin levels in the stable methadone group is in keeping with this hypothesis. Against this simple hypothesis is the finding of increased levels of fractions I and II in the stable methadone patients. The fact that P-endorphin and the RRA values for fractions I and II can move in opposite directions is not contradictory, as they represent separate peptides from separate biosynthetic pathways. As shown in Figure 1, in the pre-RRA Sephadex separation procedure, P-endorphin elutes prior to fraction I, and thus, it is not measured in either RRA fraction. The data suggest that the relationship between exogenous and endogenous opioids is quite complex, and there is a possibility that individual “families” of opioid peptides may be affected in different ways during opioid dependence. A clearer explanation might be obtained by the use of more specific measures of peptides from the prodynorphin and proenkephalin systems, which are the main contributors to fractions 1 and II. Measures of receptor activity and turnover rates of endogenous opioids are also needed. During detoxification, fraction I levels correlated significantly with severity of withdrawal responses, a composite measure of signs and symptoms (Table 2). These changes in fraction I are unlikely to be a simple stress response, as they were not found during alcohol withdrawal (Borg et al. 1982). The drug-free subjects (mean 95 days) are of interest because they were significantly different from the normal group in all three measures of endogenous opioids, despite their being verified as abstinent by urine tests. Although these patients were not symptomatic, they were within the time frame of the protracted abstinence syndrome (Martin and Jasinski 1969), which lasts about 6 months. One cannot say, therefore, whether the endogenous opioid levels in the drug-free patients are a CSF reflection of protracted abstinence or whether they represent a trait difference between the addicts and normals. During naltrexone treatment, fractions I and II were elevated over normals, but only the fraction II levels of the naltrexone subjects were higher than those of patients who were in the drug-free or methadone maintenance states. Fraction II levels were reported to be greatly elevated by naltrexone in a preliminary study (O’Brien et al. 1982), but the values we obtained from subsequent samples were not as high, possibly because of more effective removal of receptor-active naltrexone metabolites from the CSF samples. Although animal studies of naltrexone effects on CSF endogenous opioid levels have not been reported, there have been several reports (Schulz et al. 1979; Zukin et al. 1982) of reversible increases in rat brain opioid receptor number and sensitivity in response to opioid antagonist treatment (both naloxone and naltrexone). Further studies are needed to determine whether or not there are any physiological consequences to long-term naltrexone blockade of opioid receptors. Although none of the patients in this study met diagnostic criteria for clinical depression at the time of the spinal tap, the presence of depression in opioid addicts is well documented

BIOL PSYCHLQRY 1988;%649-&62

CSF Endogenous Opioids and Addiction

66 1

(O’Brien et al. 1984), and many patients in our study had some depressive symptoms. There was a consistent, if modest, relationship between both fraction I and fraction II

values (but not p-endorphin values) and our composite measure of depression across all three patient groups. This was most p~found in the me~adone main~n~ce sample where fraction I (r = 0.30) and fraction II (r = 0.37) were significantly related to our composite measure of depression symptoms. This correlation is consistent with findings of increased CSF fraction I in unipolar depressed nonaddicted patients (Agren and Terenius 1984).

References &en H, Terenius L (1984): Depression and CSF endorphins fraction I--Circannua.l variation and higher levels in unipolar than in bipolar patients. Psychiatry Res 10~303-3 11. AT, Beamesderfer A (1974): In Pichot P (ed), Assessment ufDepression: The Depression Inventoq in P~c~p~r~co~~y, ~017. Karger: Basel, pp 151-169.

Beck

Berg&&m L, Terenius L (1979): Enkephalin levels decrease in rat striatum during morphine abstinence. Eur J Pharmacol60:349. Borg AT, Kvande H, Rydberg U, Terenius L, Wahlstr6m A (1982): Endorphin levels in human cerebrospinal fluid. Psychopharmacology 78:101-103. Childers S, Simantou R, Snyder SH (1977): Enkephalin: Ra~o~unoassay assay in morphine dependent rats. Ear J PharrnacoI 46:289.

and radioreceptor

Clement-Jones V, McLaughfin L, Lowry LP, Besser GM, Rees LH (1979): Acupuncture in heroin addicts; Changes in met-enkephalin and endorphin in blood and cerebrospinal fluid. Lancer ii:380. Dixon WV, Brown MB (1979): BIUDP-79: Biomedical Computer Programs. Los Angeles: University of California Press. Emrich HM, Nusselt L, Gramsch C, John S (1983): Heroin addiction: Beta endo~~n reactivity in pIasma increases during withdrawal. P~~op~chi~riu 16:93-96.

i~uno-

Fratta W, Yand HYT, Hong J, Costa E (1977): Stability of metenkephalin content in brain structures of morphine dependent or foot-shock stressed rats. Nature 268:452. Haertzen CA (1974): An Overview of Addiction Research Center Inventory Scales (ARCH). Rockville, MD: National Institute on Drug Abuse, p 127. Ho WKK, Wen HL, Ling N (1980): Beta endo~hin-lye i~unoactivi~ addicts and normal subjects. Neurophurmacology 19:117.

in the plasma of heroin

Holmstrand J, Gunne LM, Wahlstriim A, Terenius L (1981): CSF-endorphins in heroin addicts during methadone maintenance and during withdrawal. Phurmacopsychiatriu 14:126-128. Hughes J, Smith TW, Kosterlitz HW, Fothergill L, Morgan BA, Morris HR (1975): Identification of two related pentapeptides from the brain with potent opiate agonist activity, Nature 258: 577-579. Kosten TR, Kreek M, Swift C, Carney MK, Ferdinands L (1987): Beta endorphin levels in CSF during methadone maintenance. Life Sci (in press). Martin WR, Jasinski DR (1969): Physiological parameters of morphine in man-tolerance, early abstinence, protracted abstinence. J Pqchiatr Res 7:9-16. Nyberg F, Terenius L (1983): Electrophoresis of opioid peptides on columns with agarose suspension. A~fytic Biochem 134:245-253. Nyberg F, Wahlstriim A, Sjiilund B, Terenius L (1983): Characterization of electrophoreticaily separable endorphins in human CSF. Bruin Res 259:267-274. Nyberg F, Nylander I, Terenius L (1986): Enkephalin-containing spinal fluid. Brain Res 371:278-286.

polypeptides in human cerebro-

662

BIOL PSYCHL4TRY 1988;24:649-662

C.P. O’Brien et al.

O’Brien CP, Terenius L, Wahlstrom A, McClellan AT, Krivoy W (1982): Endorphin levels in opioid-dependent human subjects. Ann NY Acad Sci 398:378-387. O’Brien CP, Woody GE, McLellan AT (1984): Psychiatric disorders in opioid dependent patients. J C/in Psychiatry 45:9-13.

Pastemak GW, Goodman R, Snyder SH (1975): An endogenous morphine-like factor in mammalian brain. Life Sci 16: 1765-I 769. Pert CB, Snyder SN (1973): Opiate receptor: ~monstration 101 l-1014.

in nervous tissue. Science 179:

Przewlocki R, Hollt V, Duka T, Lkeber G, Gramsch C, Haarmann I, Herz A (1979): Long term morphine treatment decreases endorphin levels in rat brain and pituitary. Brain RPS 174:3.57. Schulz R, Wuster M, Herz A (1979): Supersensitivity to opioids following chronic blockade of endorphin activity by naloxone. Naunyn Schmiedeberg Arch Pharmacol 306:93-96. Simon EJ, Hiller JM, Edelman I (1973): Stereospecific binding of the potent narcotic analgesic H3-etorphine to rat brain homogenate. Proc Nat1 Acad Sci USA 70:1947-1949. Terenius L (1973): Characteristics of the “receptor” for narcotic analgesics in synaptic plasma membrane fractions from rat brain. Acta Pharmacol ToxicoE 33:377-384. Terenius L. W~ls~~rn A (1974): I~ibito~s) of narcotic receptor binding in brain extracts and in cerebrospinal fluid. Actu P~rrnuco~ (Kbh) 33fSuppl 1):55. Terenius L, Wahlstrom A (197.5): Morphine-like Iigand for opiate receptors in human CSF. Life Sci 16: 1759-1764. Wesche D, Hollt V, Herz A (1977): Radio-immunoassay of enkephalins: Regional distribution in rat brain after morphine treatment and hypophysectomy . Naunyn Schmiedeberg Arch Pharmakol 301:79.

Wuster M, Schulz R, Herz A (1980a): Inquiry into endorphinergic feedback mechanisms during the development of opiate tolerance/dependence. Bruin Res 189:403-441. Wuster M, Duka T, Herz H (198Ob): Di~ep~-induced release of opioid activity in the rat brain. Neurosci Lett 16:335-337. ‘&&in RS, Sug~~n JR, Fitz-Syage ML, Garner EL, Zukin SR, Gintzler AR (1982): Naltrexoneinduced opiate receptor supersensitivity. Brain Res 245:285.