The selective serotonin reuptake inhibitor paroxetine is effective in the treatment of diabetic neuropathy symptoms

135

Pain, 42 (1990) 135-144 Elsevier PAIN 01616

Clinical Section The selective serotonin reuptake inhibitor paroxetine is effective in the treatment of diabetic neuropathy symptoms Serren H. Sindrup a, Lars F. Gram a, Kim Brersen a, Ole Eshcij b and Erik F. Mogensen b a Departments of Clinical Pharmacology and b Internal Medicine, Odense University, School of Medicine, Odense (Denmark)

(Received 4 October 1989, revision received 5 February 1990, accepted 9 February 1990)

The effect of the selective serotonin reuptake inhibitor paroxetine on diabetic neuropathy symptoms was examined Sin comparison to i~pr~ne and placebo in a randomised, double-blind, cross-over study. Paroxetine was given as a fixed dose of 40 mg/day, while the dose of imipramine was adjusted to yield optimal plasma levels of imipramine plus desipramine of 400-600 nM. Paroxetine significantly reduced the symptoms of neuropathy as measured by both observer- and self-rating, but was somewhat less effective than imipramine. However, patients showing a weaker response to paroxetine than to imipramine had lower plasma concentrations of paroxetine than patients with similar response to the 2 drugs. On imipramine 5 patients dropped out because of intolerable side effects and 4 of 19 patients completing the study reported withdrawal symptoms after discontinuing imipramine. On paroxetine no patients dropped out due to side effects and no withdrawal symptoms were reported. Self-rating showed no depressive symptoms at baseline, and no changes during the study. Neither paroxetine nor imipramine caused changes in objective measures of peripheral nerve function. In conclusion, 40 mg paroxetine/day significantly reduced the symptoms in peripheral diabetic neuropatby, and it was suggested that by dose adjustment on the basis of drug level monitoring, paroxetine may become as effective as imipramine. Paroxetine was devoid of the often disturbing autonomic side effects limiting the use of imipramine in several patients.

Key words: Diabetic neuropathy; Imipramine; Paroxetine

Introduction Various tricyclic antidepressants have been shown to relieve the symptoms of peripheral diabetic neuropathy. We have previously found that irnipramine was superior to placebo in 2 doubleblind cross-over studies [19,28]. ~t~pty~ne and nortryptyline have, in similarly designed studies, also proved efficacious [12,22], although the be-

Correspondence to: Siren H. Sindrup, Department of Clinical Pharmacology, Odense University, J.B. Winslews Vej 19, DK-5000 Odense C, Denmark. 0304-3959/90/$03.50

nefits of amitriptyline have been questioned in 1 smaller study [23]. Tricyclic antidepressants block ari-adrenergic, II,-histaminergic and muscarinergic receptors, and block reuptake of noradrenaline and serotonin [13]. It is not known which of these mechanisms, if any, are responsible for the effect on peripheral diabetic neuropathy. Action mediated by blockade of a-receptors in sprouting diseased peripheral nerve fibers has been proposed by Young et al. [30], whereas we have suggested f28] action through the endogenous pain suppressing system dependent on serotonergic and possibly also noradrenergic receptors described by Fields and Bas-

0 1990 Elsevier Science Publishers B.V. (Biomedical Division)

baum [lo]. Our assumption is supported by the fact that the selective serotonin reuptake inhibitor zimelidine has been shown to be superior to placebo in the treatment of chronic pain patients 1171. In order to further elucidate the mechanism of action, we conducted a double-blind cross-over study with a selective serotonin reuptake inhibitor (paroxetine) and compared this to imipramine and placebo. Treatment with imipramine is frequently accompanied by troublesome side effects and the search for more selective drugs with fewer side effects is therefore also clinically justified. The effect of treatments on a range of neurophysiological measurements and on depressive symptoms was also recorded.

Methods and patients Study design and medication The treatment period comprised 10 days for dose finding on a fixed dose of imipramine (50 or 75 mg/day), followed by 1 week on placebo for baseline studies, and then a double-blind randomised treatment for 2 + 2 + 2 weeks with placeboimipramine-paroxetine, paroxetine-placeboimipramine-paroxetine-placebo or imipramine, paroxetine-imiprarnine-placebo. On the basis of the dose/plasma level relationship established from blood samples collected after the 8th, 9th and 10th day of the imipramine dose finding period, a final dose of imipramine was determined [2] that would yield a target concentration of imipramine plus desipramine of 400-600 nM, which is considered an optimal drug level for this condition [19,28]. The precision of this dose prediction can be expected to be limited by the dose-dependent kinetics of imipramine [2]. An initial sparteine test [3] identified 3 patients (nos. 4, 17 and 20) as poor metabolisers of sparteine (metabolic ratio (MR) > 20), and the remaining patients as extensive metabolisers (MR < 20) (Table II). The dose finding period was omitted in 2 elderly poor metabolisers of sparteine and a fixed dose of 25 mg imipramine/day was given in the imipramine treatment period. In the remaining patients, the final imipramine dose ranged from

50 to 350 mg/day (Table II). In all patients paroxetine (Ferrosan/Beecham) was given as II fixed dose of 40 mg/day. The paroxetine (10 mg). imipramine (25 mg) and placebo tablets were ol identical size and colour and the same number ot tablets was given to each patient by use of additional placebo tablets if needed in each of the 3 double-blind treatment periods as a single dose at 8 p.m. No washout periods were intended in extensive metabolisers of sparteine, but whenever the patients’ time schedules made it necessary, pauses of 2-4 weeks separated the treatment periods. Due to the very long elimination half-life of desipramine in poor metabolisers of sparteine [4], 2 or 3 week washout periods followed each treatment period in this phenotype. On the 13th and 14th day of each treatment period, blood for drug level measurements was collected at 8-9 a.m. Imipramine, desipramine and paroxetine were assayed by quantitative thin-layer chromatography 1141. Patients Patients were recruited from 2 diabetic outpatient clinics, from general practitioners, and from an epidemiological survey comprising approximately 675 patients with insulin-dependent diabetes mellitus. They all had neurological signs of peripheral neuropathy and had been troubled for at least 1 year by several of the following symptoms: pain, paraesthesia, dysaesthesia, numbness, nightly exacerbation and sleep disturbances. None had renal or cardiac dysfunction, a diagnosis of pernicious anaemia, reduced levels of vitamin B,, or folic acid, or untreated hypothyroidism, or a recent weight loss/major change in metabolic control. Twenty-nine patients were included. Three patients already withdrew because of side effects (nausea, fatigue, dry mouth) during the initial imipramine dose finding period, and 7 patients did not complete the double-blind part of the study. Of these, 4 patients withdrew because of side effects during imipramine (tiredness, dizziness, vomiting - 2 patients) or for personal reasons (2 patients). Three patients were excluded because of compliance problems (1 patient) or because they needed analgesic drugs to treat pain

137

TABLE

I

PATIENT

CHARACTERISTICS

No/sex/age/ DM debut age

Duration of neuropathy symptoms

AT THE TIME

OF INCLUSION

Pain quality

IN THE STUDY

Additional

symptoms

*

Paraesthesia

Dysaesthesia

Hypaesthesia

++ ++ ++ ++ ++ ++ ++ ++ ++ ++ + ++ ++ ++ +++ ++ ++ ++ ++ ++

++ ++ ++ + + ++ ++ + _ _ ++ _ _ ++ ++ +++ ++ ++

++ ++ ++ ++ ++ ++ ++ ++ + + _ ++ _ _ _ ++ + +++ ++

Distribution of symptoms

**

Reflex loss

(years) l/F/41/5 2/F/39/10 3/F/40/6 4/F/42/22 5/M/61/58 6/M/58/20 7/M/40/23 8/M/37/16 9/M/45/20 10/F/28/20 11/F/41/27 12/F/38/19 13/F/36/6 14/M/42/28 15/M/41/19 16/M/62/57 17/M/72/41 18/M/54/39 19/F/46/31 20/F/75/57

12 2 1 IO 3 3 10 2 8 5 2 1 2 3 5 6 2 3 10 5

aching, tightness aching, lancinating aching, stinging lancinating lancinating (mild) lancinating (mild) lancinating aching, laminating lancinating aching, lancinating, cramps aching (severe), lancinating aching (severe), burning aching (mild), burning aching, lancinating, burning stinging (very mild) aching (severe), laminating aching, lancinating aching, lancinating aching, laminating aching, lancinating (mild)

* Symptom absent (-), mild (+). moderate ** H = hands, A = arms, F = feet, L = legs.

H, A, F, L

no

H, A, F, right L

yes yes yes yes yes yes yes yes no

F, L F, L H, A, F, L F, L H, F, L F, L H, A, F, L H, A, F, L F, L H, A, F, L L (fem. nerves) H, A, F, L, pelvis F, F, H, H, H, H,

L L F, A, A, A,

L F, L F, L F, L

yes yes no no yes yes yes no yes yes

(+ + ), or severe (+ + + ).

symptoms not related to the peripheral neuropathy (2 patients). Thus, 19 patients completed the study. One patient, who withdrew because of imipramine side effects, did complete both the placebo and paroxetine period, and data from this patient were included in the data analysis whenever present. Clinical data on these patients are given in Table I. Seventeen patients were insulin-treated, while 3 (nos. 5, 18, 20) were treated with glibenclamide. In 2 patients (nos. 5 and 12) alcohol abuse was suspected to be a concurrent aetiology of neuropathy and in 1 patient (no. 2) ankle/arm systolic blood pressure index was 0.7 bilaterally and arterial insufficiency could thus partly be responsible for the symptoms in the legs. Effect recording and evaluation During each treatment period, patients performed daily self-ratings by use of a 100 mm vertical visual analogue scale (VAS) [27] for each of the symptoms pain, paraesthesia, dysaesthesia,

nightly aggravation and sleep disturbances. The mean of the VAS score (in mm for 5 items, maximum value 500) of the last 10 days of each treatment period was used for data analysis. At the end of each treatment period, the neuropathy symptoms were assessed by 1 physician (SHS) by structured interviews with the patients using a 6-item neuropathy observer scale including items for pain, paraesthesia, dysaesthesia, numbness, nightly deterioration and sleep disturbances as described earlier [19,28]. Symptoms were scored as not present (= 0), very mild (= 0.5) mild (= l), moderate (= 1.5) or severe (= 2). The scoring on the 6-item scale was carried out independently of the VAS score and served as the primary measure of effect. Likewise tests of peripheral and central nerve function were carried out at the end of each treatment period. Vibration threshold was determined on 1st finger, wrist, 1st toe and medial malleol with a hand-held biothesiometer (Bio-

1.38

medical Instruments, Ohio. U.S.A.) [21]. A Marstack stimulator (Somedic AB, Stockholm, Sweden) [ll] was used to record heat and cold detection limit at the wrist and at the dorsal aspect of the foot, as well as heat pain threshold on the wrist, using a slightly modified version of the method of Claus et al. [5]. Evoked potentials were recorded along the somatosensory pathway after stimulation of the median nerve at the wrist and the tibia1 nerve at the ankle. Stimulation and recording technique was as described by Pedersen et al. [24] and measurements were made using a Neuromatic 2000 C (Dantec, Copenhagen, Denmark). Self-rating of depression [l] and side effects followed each treatment period. For side effects a VAS covering dry mouth, sweating. visual disturbances, tinnitus, palpitations, dizziness, head-

TABLE

II

TREATMENT PAROXETINE NO.

1 2 3 4 5 6 I x 9 10 11 12 13 14 15 I6 17 18 19 20 Median

ache. fatigue. micturition difficulties, nausea and diarrhea was used. Glycaemic control was assessed by postprandial blood glucose and fructosamine at baseline. and at the end of the placebo, paroxetlne and imipramine treatment periods. Neither postprandial blood glucose (median 16.6 vs. 15.0 vs. 14.7 vs. 12.5 mM) nor fructosamine (3.33 vs. 3.42 vs. 3.41 vs. 2.43 mM) showed any significant changes throughout the study ( P > 0.05, Wilcoxon’> test for pair differences). Statistical analyses were carried out by the Mann--Whitney test. Kruskal- Wallis test, Wilcoxon’s test for pair differences, Page test, and Spearman rank correlation using the MEDSTAT program package version 2.1 [29]. The study was approved by the Regional Ethics

Treatment sequence

pl pl pl pl pl pl

SEQUENCES, IMIPRAMINE DOSES, AND IMIPRAMINE TREATMENT *

+ ip + pa + ip + pa + ip --f pa + ip --) pa + ip + pa + ip * pa pa + pl + ip pa ---t pl + ip pa + pl + ip Pa + pl + ip pa + pl + ip ip * pa --) pl ip * pa + pl ip + pa + pl ip + pa + pl ip + pa + pl pa * ip -+ pl pa + ip + pl pa --f ip + pl pa + ip + pl

Sparteine ** pheno-

EM EM EM PM EM EM EM EM EM EM EM EM EM EM EM EM PM EM EM PM

DRUG

LEVELS

AND

SYMPTOM

Plasma Imipra. + desipra.

h-item neurop. Placebo

Parox.

Imipra.

(nM)

Imipra. dose (mg/day)

205 21x 160 381 263 126 66 x4 155 192 174 97 155 47 134 168 576 203 271 660

175 150 300 50 250 350 200 300 100 200 125 250 150 350 300 300 25 200 150 25

615 685 1043 541 663 192 522

7.0 6.0 6.5 3.5 3.5 7.5 3.5 7.5 5.9 3.0 5.0 11.5 2.0 3.0 3.0 5.? 9.5 6.0 11 .o 9.5

5.5 3.5 4.0 3.5 2.0 4.5 5.0 4.0 I .o 0.0 6.0 11.0 0.5 3.0 6.0 0.5 2.0 2.0 11.5 4.0

4.5 2.5 3.0 0.0 1.o 0.0 2.0

171

200

615

Plasma parox.

240 768 529 263 340 406 673 701 485 635 804 475

* pl = placebo, ip = imipramine, pa = paroxetine. ** EM = extensive metaboliser of sparteine; PM = poor metaboliser

5.75

of sparteine.

scores

3.75

RATINGS

DURING

Visual analogue

I .o 0.0 3.5 2.0 I .o 2.0 1.5 I .o 2.5 1.5 11.0 2.5

1.97

Placebo

Parox.

203 145 256 71 50 146 103 165 143 34 128 448 23 62 35 66 307 140 452 392

133 63 I57 78 17 75 113 95 37 40 I 70 413 1 x5 284 8 IX 26 441 101

141.5

x1.5

PLACEBO,

scales lmipra

139 700

Committee and patients consented to participate on the basis of verbal and written information.

600

1

.

-I

.

a,’

,’

Results Table II lists treatment sequences, the imipramine doses, the paroxetine and imipramine plus desipramine plasma concentrations (mean of 2 measurements), the scores on the B-item neuropathy observer scale and VAS self-rating during placebo, paroxetine and imipramine treatment. Periodical effect for all sequences (placebo-imipramine vs. imipramine-placebo, placebo-paroxetine vs. paroxetine-placebo, and paroxetineimipramine vs. imipramine-paroxetine) was tested separately for the neuropathy observer scale and the VAS self-rating, and no significant differences were found (P = 0.27-0.91, Mann-Whitney test). Likewise, residual effect was tested for all treatment combinations (Mann-Whitney test) and for all 3 treatments together (Kruskal-Wallis test). Again no significant effects were found (P = 0.33-1.00). During placebo neither imipramine, desipramine or paroxetine could be traced in the patients’ plasma. The effect of treatment was tested using Wilcoxon’s test for pair differences. The scores on both the neuropathy observer scale and the VAS selfrating showed significantly better effect (lower score) with paroxetine (P = 0.0039 and P = 0.0121) and with imipramine (P < 0.00005 and P = 0.0002) than with placebo (Table II). On both the neuropathy observer scale and on the VAS there was a significantly better effect (lower score) with imipramine compared with paroxetine (P = 0.0007 and P = 0.0071). The reduction in scores and the reduction in per cent of placebo score on the neuropathy observer scale did not correlate with the plasma drug concentration either in the paroxetine (rs = 0.31, P > 0.05 and r, = 0.24, P > 0.20) or in the imipramine (rs = 0.06, P > 0.50 and r, = 0.03, P > 0.50) treatment period. As shown in Fig. 1 the response to paroxetine and imipramine was quite similar in the majority of patients. However, in 7 patients the paroxetine reponse was less than 50% of that with imipramine, and these patients ap-

.

<50 %

>50 %

Response on paroxet1ne rn comparison with thot on lmpramme

Fig. 1. Reduction in the scores on the neuropathy observer scale with paroxetine compared to the reduction with imipramine. n, patients with a paroxetine response less than 50% of that with imipramine; 0, patients with a paroxetine response more than 50% of that with imipramine. A plot of plasma concentrations of paroxetine in these 2 groups of patients is inserted on the right.

parently had lower plasma paroxetine levels as shown in the right part of Fig. 1. The effect of each treatment on single items on the neuropathy observer scale is detailed in Table III. Imipramine treatment caused a significant reduction in all items except hypaesthesia, whereas paroxetine only caused a significant reduction in scores of pain, paraesthesia and nightly aggravation and not in scores of dysaesthesia, hypaesthesia and sleep disturbances. TABLE

III

SINGLE ITEM SCORES ON THE NEUROPATHY OBSERVER SCALE DURING PLACEBO, PAROXETINE AND IMIPRAMINE Medians are given and significant are indicated.

Pain Paraesthesia Dysaesthesia Hypaesthesia Nightly aggravation Sleep disturbance : Paroxetine Imipramine ’ Imipramine

differences

(Wilcoxon’s

Placebo

Paroxetine

Imipramine

1.47

0.52 0.54 0.48 0.03 0.52 0.47

0.49 0.49 0.03 0.02 0.04 0.02

1.48 0.75 0.04 1.49 0.75

significantly significantly significantly

different from placebo. different from placebo. different from paroxetine.

a,bs a*b b.c a,b=c bs

test)

The onset of uction of paroxetine and imipramine was studied from the daily VAS self-ratings in patients treated with either paroxetine or imipramine immediately after a placebo period or the baseline period (Fig. 2). It appears that both drugs started to act from the 1st day of treatment and the action seemed to be maximal within 4-5 days. This was indirectly confirmed by the Page test, as a significant decrease in total score was seen from day 0 to 4 with both paroxetine and imipramine treatment (P < 0.01and P < O.OOl), whereas no significant further change in total score was seen from day 8 to 12 with either paroxetine or imipramine treatment (P > 0.05 and P > 0.10). Apart from the apparently better sleep restoring effect of imipramine. the response profiles of the 2 drugs were quite similar (Fig. 2). The results of the neurophysiological measurements are detailed in Table IV. Of 17 variables tested, only one (biothesiometry on wrist during

TABLE

imipramine) showed a significant difference hetween active and placebo treatment (P = 0.0093). Scores on the depression self-rating inventory were not indicative of depression in any of the patients at baseline, whereas one (no. 1) displayed moderate symptoms of depression during paroxetine. The scores did not change during the double-blind part of the study (placebo: median 4.8 vs. paroxetine: 2.8 vs. imipramine: 3.0. P b 0.05). Five patients dropped out because of side effects during imipramine treatment, whereas none withdrew for this reason during paroxetine. This fact is expected to bias side effect evaluation in favour of the imiprarnine treatment. Side effect self-ratings showed a tendency towards higher scores during paroxetine and imipramine, than during placebo (Table V). However, only self-rating of dry mouth was significantly higher during imipramine than during placebo treatment,

IV

MEDIANS

OF NEUROPHYSIOLOGICAL

MEASURES

DURING

THE TRIAL Treatment

period

Placebo

Paroxetine

Imipramine

Biothesiometry

(V)

Finger 1 Wrist Toe 1 Med. malleolus

8.2 X.5 19.5 17.2

7.2 8.3 22.9 19.0

7.8 8.3 * 15.3 17.0

Heat detection

limit (r”C)

wrist Foot

4.6 14.2

14.0

3.3 13.1

wrist Foot

1.3 4.6

1.4 5.1

1.2 10.6

wrist

47.1

48.7

47.4

Forearm Leg

57.3 59.9

55.1 62.7

57.6 63.2

Med. n.. N20 Tib. n.. onset

21.8 39.8

21.6 39.0

21.4 38.0

1.6 0.7

1.5 1.0

1.5 0.8

6.0 11.0

6.0 11.0

6.0 11.0

-

Cold detection

limit (r”C)

Heat pain (“C) Mixed peripheral

Latency

to cortical

Amplitudes

Central

NCV+

SEPs (msec)

of cortical

conduction

(m/set)

SEPs (PV)

times (msec)

Med. n.. onset-N20 Tib. n.. onset-peak Med. n., C7-N20 Tib. n., TH12-onset

1

4.0

SEP = somatosensory evoked potentials. + nerve conduction velocities were corrected to a standard skin temperature of 35°C [16]. * P -c 0.01 (imipramine different from placebo, Wilcoxon’s rank sum test, paired observations). (P > 0.01).

All other comparisons

._

--

not significant

141

Sleep

Disturbance

~

Days

of active

lidocaine [IS] and oral mexiletine [7] relieve the symptoms of neuropathy, However, the most commonly accepted symptomatic treatment appears to be tricyclic antidepressants [12,19,22,28,30]. The use of tricyclic antidepressants is often limited by side effects, especially of an anticholinergic nature. We have now shown that paroxetine, a selective serotonin reuptake inhibitor without the usual side effects of tricyclic antidepressants, reduces the symptoms of diabetic neuropathy. Whereas the patients during imipramine treatment were treated with optimal doses determined by plasma level monitoring of active compounds, the paroxetine treatment was given as a standard dose. This difference in treatment regimens could

treatment

and imipraFig. 2. Onset of action of paroxetine (0 -@) 0). The total and single item VAS scores given mine (Oas per cent of the placebo or baseline score (medians indicated and only data from active treatment periods following immediately after the placebo or the baseline period are included).

whereas none of the other side effects was rated si~fic~tly higher during active tr~tment, than during placebo. Ratings of dry mouth, palpitation and dizziness were significantly higher during imipramine than during paroxetine treatment. Patients with plasma paroxetine con~ntrations above 150 nM did not have higher VAS scores of side effects (single items or total) than patients with lower drug levels (P > 0.05, Mann-Whitney test). ~s~ntinuation of i~pr~ne did cause moderate to pronounced withdrawal symptoms (nausea, vomiting, tremor [8]) in 4 patients (nos. 7, 15, 16, 17), while such symptoms were not reported after discontinuation of paroxetine.

Discussion

During the last decade, several approaches to treat diabetic neuropathy have been proposed. Aldose reductase ~bitors seem to bring a slight improvement in nerve function [9,15], but the effect on the symptoms of neuropathy is less clear [20]. It has been reported that intravenous

TABLE V MEDIANS AND RANGES OF VAS SCORES OF SIDE EFFECTS DURING PLACEBO, PAROXETINE AND IMIPRAMINE TREATMENT Significant differences (Wilcoxon’s test) are indicated.

Dry mouth Sweating Visual disturbances Tinnitus Palpitation Dizziness

Placebo

Paroxetine

0.3 (O-96) 0.4 (O-65) 0.4 (O-58) 0.2 (O-66) 0.3 (O-75) 1.0 (O-86) 0.4 (O-95) 12.5 (O-97)

4.5 (O-96) 3.5 (O-89) 1.5 (O-63) (::3) (tz8,

Micturition difficulties Nausea

&9)

Diarrhea

(ZO) 0.1 (O-71)

(::7) 0.3 (O-58) 21.5 (O-95) 0.4 (O-36) 0.3 (O-53) 0.2 (O-9)

55.5 (O-614)

85.5 (O-396)

Headache Fatigue

Total

Imipramine 55.0

b*c

(O-96) 31.0 (O-94) 5.0 (O-89) 0.2 (O-49) 0.4 c (O-62) 29.0 ’ (O-59) (0%) 30.0 (O-84) 0.4 (O-97) 0.4 (O-78) (0%) 224.0 ’ (5-445)

b Imipramine significantly different from placebo. ’ Paroxetine significantly different from imipramine.

largely explain the somewhat weaker therapeutic response to paroxetine seen in certain patients. These patients had lower plasma concentrations of paroxetine than patients with the same responses to the 2 drugs. Apparently optimal response to paroxetine requires plasma levels in excess of 150-200 nM. Confirmation of these assumptions requires studies on the response to paroxetine at doses yielding such higher plasma drug levels, and it cannot be excluded that different pharmacodynamic properties of paroxetine and imipramine are responsible for at least some of the difference in effect. Imipramine possesses an inhibitory effect on reuptake of both serotonin and noradrenaline, and therefore imipramine may exert a dual action on the endogenous pain suppressing system. Even if appropriate dosing of paroxetine reduces the difference in effect of the 2 drugs, it cannot be ruled out that noradrenergic mechanisms are involved in the effect of imipramine. The inhibitory effect of noradrenaline and serotonin on nociception occurs in the spinal cord and at this level there is an intimate interaction between noradrenergic and serotonergic neurones [25], and paroxetine may exert a secondary effect on adrenergic synapses at higher plasma levels. No measureable changes appeared in objective measures of peripheral nerve function either with paroxetine or with imipramine. This is fully in line with our earlier findings with imipramine [28]. It appears that blockade of a-receptors in diseased peripheral nerves [30] cannot be solely responsible for the effect of imipramine on diabetic neuropathy symptoms. Taken as a whole. our data support the view that imipramine and related drugs act through the endogenous pain suppressing system. A weaker effect of paroxetine than of a tricyclic antidepressant has been demonstrated when the drug is used to treat depression [6], but in that study the plasma level measurements did not convincingly suggest that insufficient dosing of paroxetine could explain the difference in antidepressant effect. Our data give no support to the view that the effect of imipramine and paroxetine is mediated through an antidepressant action. Firstly, none of our patients was depressed at baseline and no significant changes in self-ratings of depression occurred during the study. Secondly,

the present and previous studies [19,30] indicate that the full effect on diabetic neuropathy is achieved much faster than seen for the antidepressant effect. Thirdly. the effective plasma levels cjf imipramine plus desipramine are lower than those required for antidepressant effect [19.26.28]. Paroxetine treatment on a fixed dose schedule of 40 mg/day resulted in ;I nearly 1%fold variation in steady-state plasma concentrations (47 -660 nmol/l). The highest plasma levels were seen 111 the 3 poor metabolisers of sparteine (Table II). indicating that paroxetine metabolism co-segregates with the oxidation of sparteine. The well known co-segregation of imipramine and sparteine metabolism [3,4] was reflected in a difference in therapeutic dose range between poor metabolisers (25. 50 mg/day) and extensive metabolisers (100 3.50 mg/day). A methodological problem in our study is the possible unblinding of both the patients and the investigators due to the frequent occurrence of side effects during imipramine treatment. Max et al. [22] added benztropine and diazepam in the placebo period to mimic amitriptyline side effects in their double-blind cross-over study. This could invalidate the study as blockade of cholinergic receptors by tricyclic antidepressants could be responsible for some of the effects and we therefore avoided such additions. However, many of the side effects of imipramine and related drugs are mimicked by common conditions in diabetic patients. e.g., hyperglycaemia (dry mouth. visual disturbances), autonomic neuropathy (sweating, dizziness, palpitations. nausea. micturition difficulties) and retinopathy (visual disturbances). Thus, monitoring of side effects of the active drugs 1s also obscured in these patients, which is also clear from our study showing few differences in side effects. No patients dropped out because of side effects with paroxetine, whereas 5 withdrew hecause of intolerable side effects during imipramine treatment. In addition, 4 patients reported withdrawal symptoms after discontinuation of imipramine [ 81. In conclusion, this study shows that the selective serotonin reuptake inhibitor paroxetine is effective in the treatment of diabetic neuropathy symptoms. The data suggest that by appropriate

143

dose adjustment, paroxetine may become as efficacious as imipramine without losing the advantage of being clearly better tolerated. This study investigated the effect during a 2 week treatment period; longer-term studies are needed to evaluate whether the beneficial effect persists during prolonged treatment.

Acknowledgements This study was financially supported by the Danish Diabetes Association, the Foundation of 1870, Ferrosan/Beecham and Ciba-Geigy, Copenhagen, Denmark. Ferrosan/Beecham kindly provided the paroxetine, imipramine and placebo tablets, and the Neuromatic 2000 C electromyograph was provided by Dantec (Copenhagen, Denmark). Our thanks are due to Mrs. Karin B. Nielsen, Mrs. Anne&e Casa, and Mr. Anders Tiedje for skilful technical assistance, and to Mr. Henrik Horneberg for excellent secretarial assistance.

References 1 Beck, A.T., Ward, C.H., Mendelson, M., Mock, J. and Erbaugh, J., An inventory for measuring depression, Arch. Gen. Psychiat., 4 (1961) 53-63. 2 Bresen, K., Gram, L.F., Klysner, R. and Beth, P., Steadystate levels of imipramine and its metabolites: significance of dose-dependent kinetics, Eur. J. Clin. Pharmacol., 30 (1986) 43-49. 3 Bresen, K., Klysner, R., Gram, L.F., Otton, S.V., Beth, P. and Bertilsson, L., Steady-state concentrations of imipramine and its metabolites in relation to the sparteine/debrisoquine polymorphism, Eur. J. Clin. Pharmacol., 30 (1986) 679-684. 4 Brosen, K., Otton, S.V. and Gram, L.F., Imipramine demethylation and hydroxylation: impact of the sparteine phenotype, Clin. Pharmacol. Ther., 40 (1986) 543-549. 5 Claus, D., Hilz, M.J., Hummer, I. and Neundorfer, B., Methods for measurement of thermal thresholds, Acta Neurol. Stand., 76 (1987) 288-296. 6 Danish University Antidepressant Group, Paroxetine: a selective serotonin reuptake inhibitor showing better tolerance, but weaker antidepressant effect than clomipramine in a controlled multicenter study, J. Affect. Dis., 18 (1990) 289-299. 7 Dejgaard, A., Petersen, P. and Kastrup, J., Mexiletine for treatment of chronic painful diabetic neuropathy, Lancet, i (1988) 9-11.

8 Disalver, S.C., Greden, J.F. and Snider, R.M., Antidepressant withdrawal syndromes: phenomenology and pathophysiology, Int. Clin. Psychopharmacol., 2 (1987) 1-19. 9 Fagius, J. and Jameson, S., Effects of aldose reductase inhibitor treatment in diabetic polyneuropathy - a clinical and neurophysiological study, J. Neurol. Neurosurg. Psychiat., 44 (1981) 991-1001. 10 Fields, H.L. and Basbaum, A.I., Endogenous pain control mechanisms. In: P.D. Wall and R. Melzack (Eds.), Textbook of Pain, Churchill Livingstone, London, 1984, pp. 142-152. 11 Fruhstorfer, H., Lindblom, U. and Schmidt, W.G., Method for quantitative estimation of thermal thresholds in patients, J. Neurol. Neurosurg. Psychiat., 39 (1976) 1071-1075. 12 Gomez-Perez, F.J., Rull, J.A., Dies, H., Rodriguez-Rivera, G., Gonzalez-Barranco, J. and Lozano-Castaneda, O., Nortriptyline and fluphenazine in the symptomatic treatment of diabetic neuropathy. A double-blind cross-over study, Pain, 23 (1985) 395-400. 13 Gram, L.F., Antidepressants: receptors, pharmacokinetics and clinical effects. In: G.D. Burrows, T. Norman and B. Davies (Eds.), Antidepressants, Elsevier Science Publishers, Amsterdam, 1983, pp. 81-95. 14 Gram, L.F., Bjerre, M., Kragh-Sorensen, P., Kvinesdal, B., Molin, J., Pedersen, O.L. and Reisby, N., Imipramine metabolites in the blood of patients during therapy and after overdose, Clin. Pharmacol. Ther., 33 (1983) 335-342. 15 Jaspan, J.B., Herold, K. and Bartkus, C., Effects of sorbinil therapy in diabetic patients with peripheral neuropathy and autonomic neuropathy, Am. J. Med., 79. Suppl. 5A (1985) 24-37. 16 Jesus, P.V. de, Hausmanowa-Petrusewicz, I. and Bar&i, R.L., The effect of cold on nerve conduction of human slow and fast nerve fibers, Neurology, 23 (1973) 1182-1189. 17 Johansson, F. and Von Knorring, L., A double-blind controlled study of a selective serotonin uptake inhibitor (zimelidine) versus placebo in chronic pain patients, Pain, 7 (1979) 69-78. 18 Kastrup, J., Petersen, P., Dejgaard, A., Angelo, H.R. and Hilsted, J., Intravenous lidocaine infusion - a new treatment of chronic painful diabetic neuropathy, Pain, 28 (1987) 69-75. 19 Kvinesdal, B., Molin, J., Freland, A. and Gram, L.F., Imipramine treatment of painful diabetic neuropathy, JAMA, 251 (1984) 1727-1730. 20 Lewin, I.G., O’Brien, I.A.D., Morgan, M.H. and Corrall, R.J.M., Clinical and neurophysiological studies with the aldose reductase inhibitor, sorbinil, in symptomatic diabetic neuropathy, Diabetologia, 26 (1984) 445-448. 21 Lowenthal, L.M. and Hockaday, D.R., Vibration sensory thresholds depend on pressure of applied stimulus, Diabetes Care, 10 (1987) 100-102. 22 Max, M.B., Culnane, M., Schafer, SC., Gracely, R.H., Walther, D.J., Smoller, B. and Dubner R., Amitriptyline relieves diabetic neuropathy pain in patients with normal or depressed mood, Neurology, 37 (1987) 589-596. 23 Mendel, C.M., Klein, R.F., Chappell, D.A., Dere, W.H., Gertz, B.J., Karan, J.H., Lavin, T.N. and Grunfeld, C., A

144 trial of amitriptyline and fluphenazine m the treatment of painful diabetic neuropathy, JAMA, 255 (1986) 637-639. 24 Pedersen, L. and Trojaborg. W.. Visual, auditory and somatosenso~ pathway involvement in hereditary cerebeliar ataxia. Friedreich’s ataxia and familial spastic paraplegia, Electroenceph. Clin. Neurophysiol.. 52 (19X1) 28% 2Yl. 25 Proudfit. H.K., Pharmacological evidence for the modulation of nociception by noradrenergic neurons. In: H.L. Fields and J.-M. Besson (Eds.). Pain Modulation. Progress in Brain Research, Vol. 77, Elsevier Science Publishers. Amsterdam, 1988, pp. 357-370. 26 Reisby, N.. Gram, L.F., Beth, P., Nagy, A., Petersen, G.O., Ortmann, J., Ibsen, I., Dencker, S.J., Jacobsen, O., tidutwald, O., Ssndergaard, I. and Christensen, J., Imipramine:

27 2X

2Y 30

clinical effects and pharmacokmetic variablht!, Psvch~)pharmacology (Berl.). 54 (1977) 263”-272. Scott. J. and Husk&on. b.C.. Graphic representation t)l pain. Pain, 2 (I 976) 175 - 1X4. Sindrup. S.H.. Ejlertsen. B., Frurland. A., Sindrup. E.H.. Bresen, K. and Gram. L.F.. Imipramine treatment in dlabrtic neuropathy: rehef of subjective symptoms without changes in peripheral and autonomic nerve function, Eur. J. C’lin. Pharmacol., 37 {19X9) 151- 153. Wulff, H.R. and Schlichting, P., MEDSTA’I‘ Version 2.1 The ASTRA Group, Denmark, IYXX. Young. R.J. and Clarke. B.F.. Pain relief in diabetic neuropathy-: the effectiveness of imip~dmine and related drugs, Diabetic Med.. 2 (1985) 363-366.