MATERIALS AND METHODS

Study population

Between December 1992 and December 1995, 32 French dermatology departments prospectively included all consecutive patients who initiated thalidomide therapy for a severe dermatologic disorder. Each dermatologic center was associated with a neurophysiology department in charge of performing the electrophysiologic examinations in a standardized way.

Exclusion criteria were: ongoing pregnancy (a pregnancy test was systematically performed on women of childbearing age), current peripheral neuropathy, Hansen disease, human immunodeficiency virus seropositivity, and other risk factors of neurotoxicity. These factors were: recent treatment by other neurotoxic drugs; diabetes, if the patient was treated with insulin or oral agents, or in case of a glycosylated hemoglobin test above 8%; chronic renal failure with a creatinine level above 200 mol per liter; and excessive intake of alcohol, i.e., higher than 60 g per day. Women of childbearing age had to use a reliable method of contraception, and to sign the usual informed consent explaining the teratogenic risk.

Patient management

All patients included in the cohort study had a standardized neurologic examination and an electrophysiologic examination before the initiation of thalidomide therapy. Clinical examination was then performed monthly for a daily dose of 100 mg per day or higher, and every 3 mo for a daily dose below 100 mg per day. An electrophysiologic examination was systematically performed every 6 mo or in case of clinical suspicion of neuropathy. Each patient was followed until the occurrence of a neuropathy, the thalidomide withdrawal, or during at least 2 y.

The clinical management of patients and the choice of the thalidomide dosage was left to the responsibility of the local physician in each center. The treatment had to be withdrawn if both clinical and electrophysiologic signs of neuropathy were present. Investigators were allowed to withdraw the drug, however, in case of isolated clinical or electrophysiologic signs, of other troublesome adverse event, of success or failure of the treatment, or at the patient' request.

Data collection

During each follow-up visit, dermatologists at each center conducted direct interviews with a structured questionnaire and performed a standardized neurologic examination. Patients were questioned about possible subjective sensitive abnormalities. Clinical examination included search for distal hypoaesthesia, muscle weakness, and ankle jerk depression. Physicians also assessed the characteristics of thalidomide ingestion (dosages and treatment duration) and other risk factors of neuropathy, such as alcohol misuse and drug intakes.

Electrophysiologic studies included bilateral nerve conduction studies using noninvasive standard techniques of percutaneous stimulation and surface electrode recording. Motor conduction velocity, distal motor latency and compound muscle action potential amplitude (from baseline to negative peak) were measured on common peroneal nerve. Sensory nerve conduction velocity, and SNAP amplitude (from peak to peak) were determined on median and sural nerves.

Classification of patients and definitions

Patients with less than 1 mo of thalidomide therapy were excluded from the analysis. At the end of the follow-up study, i.e., after the thalidomide treatment had been stopped or after at least a 2 y follow-up period, all the medical charts and electrophysiologic examinations were simultaneously reviewed by four of us (SO, JR, PB, RKG), in order to classify patients according to prespecified criteria. Each patient had to be classified in only one group and the decision had to be unanimous. The members of this committee were unaware of the thalidomide dosage and treatment duration. Patients were classified into five groups:

1."Definite neuropathy": the diagnosis of thalidomide-induced neuropathy was accepted when typical subjective and/or objective clinical features were present and associated with electrophysiologic signs and after exclusion of another possible cause of neuropathy. These typical clinical features consisted of pins and needles in the feet, and/or tactile distal hypoesthesia, symmetrical and predominating in the lower limbs. Electrophysiologic signs were defined by the presence of a 50% decrease of sural SNAP amplitude as compared with previous electrophysiologic results, with relative conservation of sensory nerve conduction velocity. For these patients, the experts had also to determine which was the first significant sign (clinical and/or electrophysiologic) and its date.

2."No thalidomide neuropathy": patients with neither clinical, nor electrophysiologic criteria for thalidomide neuropathy were classified in this group.

3."Atypical neuropathy": patients with both clinical and electrical signs were considered as "atypical neuropathy" if clinical and/or electrical signs were not symmetrical, if they predominated at the upper limbs, or if a decrease of sural SNAP amplitude existed but was between 30 and 50%.

4."Conflicting cases": patients with isolated clinical or isolated electrical signs during a period of at least 3 mo were classified as "conflicting cases".

5."Missing electrophysiologic control": this group included patients who had either stopped their treatment because of clinical signs, or taken thalidomide for at least 6 mo, without electrophysiologic evaluation in contradiction with the protocol.

Analysis

Prevalence and incidence rates

In order to estimate the minimal and maximal prevalence rates of neuropathy we used two methods. The minimal rate was calculated by dividing the number of patients with definite neuropathy by the number of patients who received more than 1 mo of therapy. The maximal rate was calculated by dividing the number of patients with definite neuropathy plus those with atypical neuropathy, plus those with discordant clinical and electrophysiologic features (conflicting cases), and those with missing electrophysiologic data but in whom clinical signs were present, by the number of patients who received more than 1 mo of therapy. The 95% confidence intervals (CI) were calculated for both estimates.

The annual incidence rate of neuropathy during treatment was estimated by considering patients with definite neuropathy after dividing the follow-up period into years intervals. For that purpose, the date considered was the date of the diagnosis. Kaplan–Meier survival analysis (^{Kaplan and Meier, 1958}) was used to determine the probability of neuropathy over time, the starting point was the first day of the thalidomide therapy.

Risk factors

The critical endpoint was the existence of a definite neuropathy, as defined above; however, for the purpose of risk factors analyses, the date of this endpoint was the date of the first clinical or electrophysiologic sign considered as significant by the experts.

We compared the characteristics of patients with definite neuropathy (group 1) with those of patients without neuropathy (group 2) in order to identify potential risk factors. Besides age and sex, the variables considered for these analyses were indication for thalidomide treatment, alcohol consumption and characteristics of the thalidomide therapy, including: initial, cumulative and daily doses, and treatment duration. ² statistics and Mann–Whitney U test were used for qualitative and quantitative data, respectively. Then, Cox proportional-hazards modeling was used to take into account simultaneously all potential risk factors over time (^{Cox, 1972}), the starting point was the first day of the thalidomide therapy. Observations were censored when the treatment was discontinued for any reason other than the first significant sign of a definite neuropathy. Variables to include in the final multivariate models were those emerging from univariate models with p0.15. Hazard ratios and their two-sided 95% CI were calculated according to the following formula: exp (1.96 * SD ()). All significance tests were two-tailed.

Data were analyzed using the BMDP software (University of California, Berkeley, CA).

RESULTS

Population study

One hundred and sixty-four patients were initially included. Twenty-nine patients were excluded for the following reasons: 13 did not fulfill the inclusion criteria, and 16 received less than 1 mo of treatment. The lack of inclusion criteria were the following: lack of electrophysiologic examination before the initiation of treatment (n=2), SNAP below 10 V (n=3), clinical signs of neuropathy before treatment (n=6), and diabetes (n=2). The reasons why 16 patients received less than 1 mo of treatment were: recovery (n=1), ineffectiveness (n=1), lost of follow-up (n=9), and non-neurologic early side-effect (n=5). Therefore, we analyzed 135 patients (Figure 1), whose main characteristics are summarized in Table I. Indications for thalidomide therapy mainly included: severe aphthosis such as Behçet disease (37%) and lupus erythematosus (35.6%). Mean initial thalidomide dosage was 97.5 mg per day (25.6 SD). The mean follow-up duration was 69.5 wk (49.9 SD), ranging from 4 to 179 wk. One hundred and seventeen patients stopped the treatment during the 2 y follow-up period. The main causes of drug withdrawal were: recovery (n=24), ineffectiveness (n=12), suspicion of neuropathy (n=72), and other adverse events (n=9). Among the 18 patients who received the drug during more than 2 y, three finally stopped it because of a suspicion of neuropathy. Therefore, more than half of the patients (55.6%) stopped the drug because of a neurologic event.

Figure 1.

Flow chart and classification of patients

Full figure and legend (29K)

Table I - General characteristics of the patients included in the cohort (n=135).

Full table

Classification of the patients

Classification of the 135 patients is presented in Figure 1. Thirty-four patients were classified as "definite neuropathy", 58 as "no thalidomide neuropathy", 10 as "atypical neuropathy", 18 as "conflicting cases", and 15 in the group "missing electrophysiologic control". Thirteen of these 15 patients had typical clinical signs, they were considered as possible neuropathies and included as such in the computation of maximal incidence.

Prevalence rates of thalidomide neuropathy

The minimal prevalence rate, considering only patients with confirmed neuropathies (n=34) among the 135 patients who received more than 1 mo of therapy was 25.2% (95% CI: 17.9–32.5%). The maximal rate, included 34 patients with definite neuropathy, 10 with atypical neuropathy, 18 with discordant clinical and electrophysiologic features, and 13 with typical clinical signs but without electrophysiologic control. This rate was 55.6% (95% CI: 47.2–63.9%).

Annual incidence rate of thalidomide neuropathy

The annual incidence rate of definite thalidomide neuropathy was 20% during the first year of treatment, 10% and 11% the following years (Table II). Figure 2 shows the Kaplan–Meier plot of time until development of neuropathy. It shows that the patients were more likely to develop neuropathy early on, and that the risk decreases over time. The thalidomide mean daily dose significantly decreased across treatment duration, from 71.9 mg per day (33.5 SD) the first year, to 44.8 (29.4 SD) the second and 29.7 (18.3 SD) the third one (p<10^-3).

Figure 2.

Kaplan-Meier plots showing development of neuropathy

Full figure and legend (12K)

Table II - Annual incidence rate of neuropathy according to the treatment duration.

Full table

Table III - Comparison of patients with and without neuropathy.

Full table

Risk factors for thalidomide neuropathy, comparisons between groups 1 and 2 (Table III)

No statistically significant difference was observed between patients who did (group 1) vs those who did not (group 2) develop definite neuropathy when considering age, gender, indications for thalidomide therapy, initial thalidomide dosage, and cumulative doses (at the time of the first sign of neuropathy). By contrast, mean daily dose of thalidomide was significantly higher among patients with neuropathy, both at the occurrence of the first sign of neuropathy and at the time of the diagnosis, if the drug was not withdrawn before. A trend for a shorter treatment duration was observed in the group of patients having a neuropathy.

The association between the daily dose and the occurrence of thalidomide neuropathy remained significant in multivariate analysis when treatment duration was taken into account; for each additional mg per day, the relative risk (hazard ratio) for neuropathy was 1.03 (CI: 1.02–1.04) (p=10^-4). In order to estimate the level of the influence of daily dose on the occurrence of thalidomide neuropathy, patients were stratified according to the mean daily dose they received: less than 25 mg per day; 25–50 mg per day; 50–75 mg per day; more than 75 mg. The percentage of definite neuropathy significantly increased with the daily dose (Table IV). Considering the patients with a daily dose of 50 mg per day or lower as a reference class, the relative risk for thalidomide neuropathy was 8.2 for the patients with a daily dose of between 50 and 75 mg per day and 20.2 for patients with a daily dose higher than 75 mg per day (p<10^-3). Similar results were obtained when considering the whole population (Table V).

Table IV - Relative risk (hazard ratios) of neuropathy according to the daily dose when considering patients with (n=34) and without (n=58) thalidomide neuropathy.

Full table

Table V - Relative risk (hazard ratios) of neuropathy according to the daily dose when considering patients with definite neuropathy in one hand (n=34) and all the other patients in the other hand (n=101).

Full table

Early signs of neuropathy

The mean delay between the first sign and the confirmation of the diagnosis of neuropathy in the 34 patients with definite neuropathy was 11 wk (14 SD). The delay between the initiation of thalidomide and the first sign of neuropathy was 39 wk (28 SD), ranging from 4 to 110 wk.

The first sign detected was clinical alone in 14 patients (41%), electrophysiologic alone in eight (23.5%) and both clinical and electrophysiologic in 12 patients (35%). Thus, clinical initial symptoms existed for 26 of the 34 patients with a definite neuropathy. The first clinical abnormalities were subjective in 81% of cases (21 of 26), objective in 46% (12 of 26); seven patients had both subjective and objective signs. The early subjective signs included paresthesia (n=9), cramps (n=9), constriction (n=3), numbness (n=3), and burning sensation (n=2), two of these signs or more being frequently associated. The early objective signs were isolated motor signs in three cases, isolated sensory signs in two, and both sensory and motor in seven. At the time of definite diagnosis, 76% of patients had objective signs on neurologic examination (26 of 34).

In other respects, during the treatment, five patients had signs suggesting a carpal tunnel syndrome. Among these patients, two had no thalidomide neuropathy, two had an atypical neuropathy, and one was a conflicting case.

DISCUSSION

In this prospective cohort we found a prevalence rate of neuropathy of 25.2% when considering only definite cases, and of 55.6% when all potential cases were considered. The annual incidence rate was maximal during the first year of treatment. Besides neuropathy, nine patients stopped thalidomide because of another adverse event. Thus, 81 (60%) patients stopped their treatment because of a side-effect. The single most important risk factor for thalidomide neuropathy was the daily dose. A strong concentration-dependent relationship was observed, even after 1 y of treatment. Other potential risk factors such as gender, age, or cumulative doses were not observed in this study. The neuropathy was discovered by a systematic electrophysiologic examination for nearly a quarter of patients with such an adverse event.

The validity of the results relies upon several factors, including the unbiased recruitment of patients, their classification according to the outcome measures and accuracy information of the thalidomide use. The prospective nature of the study allowed an accurate collection of data. The patients included are fairly representative of those treated by thalidomide for dermatologic conditions as, according to French regulations, all patients who need such therapy are referred to a hospital dermatology department. In fact, the percentages of the different indications, i.e., aphthosis and lupus were those that could be expected. We did not include in the analyses patients with only 1 mo of thalidomide treatment as the occurrence of a neuropathy is very unlikely after such a short duration of drug administration. In fact none of the definite or atypical cases of neuropathy had any sign before 1 mo of treatment. Inclusion of these patients in the analyses would have led to a systematic underestimation of the incidence rate.

Clinical and electrophysiologic criteria taken into consideration for the diagnosis of definite thalidomide neuropathy were those considered by most authors in thalidomide guidelines (^{Gardner-Medwin et al, 1994; Calabrese and Fleischer, 2000}). A 50% decrease of sural SNAP amplitude is considered the choice electrophysiologic criterion (^{Lagueny et al, 1986; Gardner-Medwin et al, 1994}) and has been reported to be closely related to the clinical sensory abnormalities (^{Ochonisky et al, 1994; Grover et al, 2000}).

The classification of patients was done by a panel of experts after blinding the characteristics of the therapy, thus avoiding major bias. Some patients could not be clearly classified as "definite neuropathy" or "no neuropathy". We chose to compute two incidence rates, using as numerator definite cases only for the first one, and all potential cases for the second one. In order to estimate the role of the potential risk factors, however, the consideration of definite neuropathies on the one hand and clear absence of neuropathy on the other hand appeared as the best choice. The observation of similar relative risks when the model was applied to the whole cohort (definite neuropathy vs others) is an approach of the validation of the model.

The prevalence of thalidomide-induced neuropathy has been variously estimated in retrospective studies, from less than 1% in 34 patients treated for lepra reactions (^{Sheskin and Yaar, 1979}) to more than 70% in small series of patients (four to eight) treated for prurigo nodularis (^{Aronson et al, 1984; Clemmensen et al, 1984; Wulff et al, 1985}). A prevalence of 25% was reported in a series of 60 patients treated for discoid lupus erythematosus during a 2 y period (400 mg per day, then 50–100 mg per day) (^{Knop et al, 1983}). This variability in prevalence has been interpreted as reflecting a disease-related susceptibility. In fact the retrospective nature of the studies, the large range of daily doses used, and the heterogeneity of the clinical and electrophysiologic criteria considered for the diagnosis of neuropathy preclude a real estimation of the prevalence rate and risks factors of thalidomide neuropathy. In recent series of patients with refractory multiple myeloma treated with high doses of thalidomide, i.e., 200–800 mg per day, neuropathy was reported to occur in 10% and 30% of treated patients (^{Singhal et al, 1999; Singhal and Mehta, 2001}), but the clinical and electrophysiologic criteria for neuropathy were not clearly specified. Thalidomide-induced neuropathy was recently prospectively evaluated among 67 elderly men with metastatic prostate cancer (^{Molloy et al, 2001}); however, only 26 patients were evaluable, among whom six developed a neuropathy (23.1%).

In the present cohort of middle-aged patients treated with thalidomide for various dermatologic conditions and without pre-existing neuropathy or risk factor for peripheral neuropathy, the prevalence of thalidomide-induced neuropathy was prospectively surveyed with standardized clinical and electrophysiologic criteria. The stringent criteria used for the diagnosis of definite neuropathy gives strength to our conclusion of 25% as a low estimate of incidence. This percentage is close to the 21% of 42 dermatologic patients previously reported in a retrospective study (^{Ochonisky et al, 1994}) and to that reported by^{Molloy et al (2001)}. A high estimate, however, taking into account all patients with clinical or electrophysiologic signs of neuropathy and patients with atypical features could be as high as 55.6%.

Most thalidomide neuropathies occurred during the first year of treatment. The lower rates observed afterwards may be explained by the decrease in daily doses or by the fact that susceptible patients, i.e., patients having an individual predisposition of unknown nature had already stopped the drug because of neurologic symptoms.

The role of the dose of thalidomide has been suspected from the first reports. Cumulative doses were generally thought to be the main risk factor (^{Wulff et al, 1985}). In hematologic series, a relationship between the daily dose and the occurrence of a neuropathy was suggested but not proven (^{Dimopoulos et al, 2001; Singhal and Mehta, 2001}). In the series reported by (^{Molloy et al, 2001}), the cumulative dose was significantly higher among patients with neuropathy, but the treatment duration was not taken into account. A previous retrospective study failed to identify a concentration-dependent risk for neuropathy (^{Ochonisky et al, 1994}).

This study exemplifies the role of the daily dose in toxicity of thalidomide in the absence of any influence of the cumulative dose and/or the duration of therapy. This effect was particularly striking as any additional milligram per day had a significant impact on the risk of developing a neuropathy. No neuropathy was observed with less than 25 mg daily doses, which is probably safe whatever the duration of therapy. We could not evaluate other risk factors such as diabetes, alcoholism, or previous chemotherapy as these patients were excluded from the study. Age, gender, and indications for thalidomide, were not significantly associated with a greater occurrence of thalidomide neuropathy.

These results do not rule out the potential role of an individual predisposition to thalidomide neuropathy, but no data support this hypothesis.

Guidelines have been proposed for the early detection of neuropathy during thalidomide treatment: a neurophysiologic examination before treatment and every 3 mo (^{Molloy et al, 2001}) or 6 mo (^{Gardner-Medwin et al, 1994; Powell and Gardner-Medwin, 1994}) during treatment; however, it was also noted that electrophysiologic testing sometimes failed to anticipate clinical symptoms. Indeed, in the present study, isolated electrophysiologic features were the initial signs of neuropathy in only 25% of cases and clinical signs were detected earlier in 41%. The first electrophysiologic feature was a decrease in sural SNAP amplitude. According to previous studies, only a decrease of at least 50% was considered significant (^{Gardner-Medwin et al, 1994}). We considered the diagnosis of established neuropathy only and consequently discontinued thalidomide when both clinical and electrophysiologic signs were present. It may be justified to stop earlier, i.e., at the first sign—clinical or electrical—as there have been reports of incomplete recovery of the neuropathy after delayed cessation of treatment (^{Clemmensen et al, 1984; Wulff et al, 1985}). Admittedly, the decision also depends on the severity of the underlying disease.

References

1. Aronson, IK, Riley, Y, West, DP, Van den Broek, H, Antel, J: Thalidomide-induced peripheral neuropathy. Arch Dermatol 1984 120: 1466–1470,  | Article | PubMed | ISI | ChemPort |
2. Calabrese, L, Fleischer, AB: Thalidomide. Current and potential clinical applications. Am J Med 2000 108: 487–495,  | Article | PubMed | ISI | ChemPort |
3. Clemmensen, OJ, Zander Olsen, P, Andersen, KE: Thalidomide neurotoxicity. Arch Dermatol 1984 120: 338–341,  | Article | PubMed | ISI | ChemPort |
4. Cohen, S: Thalidomide polyneuropathy. N Engl J Med 1962 266: 1208,
5. Cox, DR: Regression models and life-tables. J R Stat Soc[B] 1972 34: 187–220,
6. Dimopoulos, MA, Zervas, F, Kouvatseas, G, et al: Thalidomide and dexamethasone combination for refractory multiple myeloma. Ann Oncol 2001 12(2)991–995,  | Article | PubMed | ISI | ChemPort |
7. Gardner-Medwin, JMM, Smith, NJ, Powel, RJ: Clinical experience with thalidomide in the management of severe oral and genital ulceration in conditions such as Behçet' disease use of neurophysiological studies to detect thalidomide neuropathy: Ann Rheum Dis 1994 53: 828–832,  | PubMed | ChemPort |
8. Grover, JK, Vats, V, Gopalakrishna, R, Ramam, M: Thalidomide: A Relook: Natl Med J India 2000 13: 132–141,  | PubMed | ChemPort |
9. Kaplan, EL, Meier, P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958 53: 457–481,  | ISI |
10. Knop, J, Bonsmann, G, Happle, R, Ludolph, A, Matz, DR, Mifsud, EJ, Macher, E: Thalidomide in the treatment of sixty cases of chronic discoid lupus erythe-matosus. Br J Dermatol 1983 108: 461–466,  | PubMed | ISI | ChemPort |
11. Laffitte, E, Revuz, J: Thalidomide. Ann Dermatol Venereol 2000 127: 596–602,  | PubMed |
12. Lagueny, A, Rommel, A, Vignolly, B, Taieb, A, Vendeaud-Busquet, M, Doutre, MS, Julien, J: Thalidomide neuropathy: An electrophysiological study. Muscle Nerve 1986 9: 837–844,  | Article | PubMed | ISI | ChemPort |
13. Mellin, GW, Katzenstein, M: The saga of thalidomide (Concluded). Neuropathy to embryopathy, with case reports of congenital anomalies. N Engl J Med 1962 267: 1238–1244,  | PubMed | ISI | ChemPort |
14. Molloy, FM, Floeter, MK, Syed, NA, et al: Thalidomide neuropathy in patients treated for metastatic prostate cancer. Muscle Nerve 2001 24: 1050–1057,  | Article | PubMed | ISI | ChemPort |
15. Ochonisky, S, Verroust, J, Bastuji-Garin, S, Gherardi, R, Revuz, J: Thalidomide neuropathy incidence and clinicoelectrophysiological finding in 42 patients. Arch Dermatol 1994 130: 66–69,  | Article | PubMed | ISI | ChemPort |
16. Powell, RJ, Gardner-Medwin, JMM: Guideline for the clinical use and dispensing of thalidomide. Postgrad Med J 1994 70: 901–904,  | PubMed | ISI | ChemPort |
17. Radomsky, CL, Levine, N: Thalidomide. Dermatol Clin 2001 19: 87–103,  | PubMed | ISI | ChemPort |
18. Sheskin, J: Thalidomide in the treatment of lepra reactions. Clin Pharmacol Ther 1965 6: 303–306,  | ISI |
19. Sheskin, J, Yaar, I: Motorische Leitungsgeschwindigkeit der Kubitalnerven bei Patienten mit Leprareaktion. Der Hautarzt 1979 30: 376–379,  | ChemPort |
20. Singhal, S, Mehta, J: Thalidomide in cancer potential uses and limitations. Biodrugs 2001 15: 163–172,  | PubMed | ISI | ChemPort |
21. Singhal, S, Mehta, J, Desikan, R, et al: Antitumor activity of thalidomide in refractory multiple myeloma. N Engl J Med 1999 341: 1565–1609,  | Article | PubMed | ISI | ChemPort |
22. Tseng, S, Pak, G, Washenik, K, Keltz Pomeranz, M, Shupack, J: Rediscovering thalidomide. A review of its mechanism of action, side effects, and potential uses. J Am Acad Dermatol 1996 35: 969–979,  | Article | PubMed | ISI | ChemPort |
23. Vogelsang, GB, Farmer, ER, Hess, AD, et al: Thalidomide for the treatment of chronic graft-versus-host disease. N Engl J Med 1992 326: 1055–1058,  | PubMed | ISI | ChemPort |
24. Wulff, CH, Hoyer, H, Asboe-Hansen, G, Brodthaden, H: Development of polyneuropathy during thalidomide therapy. Br J Dermatol 1985 112: 475–480,  | PubMed | ISI | ChemPort |

Acknowledgments

This study would not have been possible without the cooperation of dermatologists and neurophysiologists of the Thalidomide Neuropathy Study Group; we are deeply indebted to them. Dermatologists included: Frances C., Chosidow O., Morel P., Dubertret L., Flageul B., Cricxk B. (AP-HP, Paris), Bonnetblanc J.M. (Limoges), Souteyrand P. (Clermont-Ferrand), Joly P. (Rouen), Beylot C. (Pessac), Bonerandi J.J. (Marseille), Grosshans E., Chartier C (Strasbourg), Weber M. (Nancy), Humbert Ph. (Besançon), Sassolas B. (Brest), Ortonne J.P., Rodot S. (Nice), Guillot B. (Nîmes), Cambazard F., Perrot J.L. (St Priest), Lambert D. (Dijon), Schubert B., Juillard J. (Mulhouse), Chevrant-Breton J., Patoux-Pibouin M. (Rennes), Cuny J.F., Truchetet F. (Thionville), Amblard P. (Grenoble), Moulin G., Augey F. (Lyon), Plantin P. (Quimper), Guilhou J.J. (Montpellier), Vaillant L., Machet L. (Tours), Denoeux J.P., Lok-Charles C. (Amiens), Stalder J.F. (Nantes); Neurophysiologists included: Bouche P., Diverrez J. (Paris), Clavelou P. (Chamalières), Tabaraud F. (Limoges), Samson D. (Rouen), Burbaud J.M. (Pessac), Tognetti D. (Marseille), Wasser Ph. (Strasbourg), Brichet B. (Nancy), Monnier G. (Besançon), Esnault S. (Brest), Delpont R. (Nice), Prat D. (Nîmes), Lapras J. (St Etienne), Soichot P. (Dijon), Cohen-Khallas E. (Mulhouse), Toulouse P. (Rennes), Scherer C. (Thionville), Godet E. (Metz), Reymond F. (Grenoble), Diverrez J. (Paris), Kahter-Boidin X. (Amiens), Gonnaud P.M. (Pierre Benite), Sergeant T. (Quimper), Georgescu M. (Montpellier), Kiffer A. (Tours), and Mathe J.F. (Nantes). We are indebted to Isabelle Durand-Zaleski for her helpful review of the manuscript, and to Philippe Amiel for technical assistance. The study was sponsored by grants from Laboratoire Laphal.