Dexamethasone-based regimens versus melphalan-prednisone for elderly multiple myeloma patients ineligible for high-dose therapy

In multiple myeloma (MM), melphalan and prednisone (MP) has been used since the 1960s and still remains the most widely accepted treatment option in elderly patients ineligible for high-dose therapy.¹  Slightly different dosages and schedules have been used over the years without any demonstrated impact on survival. More complex alkylating agent combinations have often added toxicity and inconvenience without providing a survival advantage.²  Since the 1990s, pulse dexamethasone (DEX) alone is also frequently used, both at relapse and in untreated patients, although very limited trial results are available.^3,4  In the major study of 112 untreated patients, DEX had been considered to account for most of the responses to VAD (vincristine, adriamycin, dexamethasone), with a response rate of 43% and a lower incidence of serious complications (27% for VAD versus 4% for DEX).⁴  Nevertheless, the DEX toxicity, especially neurologic disturbances, psychiatric complications, secondary diabetes mellitus, gastrointestinal problems, and infection susceptibility, could be a primary concern in elderly patients. Only one recent study has compared melphalan and dexamethasone (M-DEX) with MP and has concluded that MP should remain the reference for comparison of new treatments involving innovative drugs.⁵  Limited trial data are also available regarding the dexamethasone-interferon alpha (DEX-IFN) combination. In a small nonrandomized study of newly diagnosed patients, the addition of IFN (3 million units [MU]/m²/d) to DEX achieved results similar to those with DEX alone.⁶  In another study, DEX-IFN was given to patients who failed induction chemotherapy, and the outcome of these patients was improved with a median survival of 48 months from the start of DEX-IFN.⁷  The tolerance of DEX, M-DEX, and DEX-IFN has not been fully evaluated in the context of a large randomized trial in elderly patients.

In June 1995, the Intergroupe Francophone du Myélome (IFM) group initiated a randomized clinical trial (IFM 95-01) for newly diagnosed MM patients aged 65 to 75 years comparing MP with dexamethasone-based regimens: M-DEX, DEX alone, and DEX-IFN.

Inclusion criteria were patients aged between 65 and 75 years and fulfilling a diagnosis of stage II or III MM according to the Durie and Salmon criteria.⁸  Durie and Salmon stage I MM patients could be enrolled if they met one of the following criteria (defining high-risk stage I patients): presence of one lytic lesion on skeletal radiographs; bone pain with a corresponding magnetic resonance imaging (MRI) lesion; hemoglobin less than 120 g/L (12 g/dL) for males or 110 g/L (11 g/dL) for females, associated to the presence of more than 25% plasma cells in bone marrow or to M component over 30 g/L (IgG) or 25 g/L (IgA) or 1 g per 24 hours (light chain). Patients were previously untreated (except the minimum dose of radiotherapy to localized lesions required to relieve symptoms).

Patients were excluded if they met the criteria of primary amyloidosis or had a prior history of another neoplasm or of seizure. Patients with significant cardiac, psychiatric (including mood alterations), or hepatic dysfunction were also excluded as well as patients who were considered to have a contraindication to high-dose steroids or who could be enrolled in IFM concomitant high-dose therapy programs. Written, informed consent was obtained from all patients before enrollment into the study.

This randomized, multicenter, parallel trial was carried out at 104 IFM centers in France, Belgium, and Switzerland. Recruitment of patients took place between June 1995 and September 1998. Patients were randomized per center to receive MP, M-DEX, DEX, or DEX-IFN in a 1:1:1:1 ratio. Randomization was performed per center by the biostatistical center using permutation tables of size 4 or 8 according to the expected number of enrollments within each center.

MP. The regimen consisted of 12 6-week cycles of chemotherapy. Melphalan (0.25 mg/kg) and prednisone (2 mg/kg) were given orally for 4 days. The neutrophil count must have reached 1.5 × 10⁹/L and the platelet count 100 × 10⁹/L before full-dose chemotherapy was given. If the patient did not meet these criteria, the treatment decision was reported 1 week later. At week 7 of the start of chemotherapy, full-dose chemotherapy was given if the previous criteria were met, a 50% melphalan reduction was performed if the neutrophil count was between 1.0 × 10⁹/L and 1.5 × 10⁹/L or the platelet count between 50 × 10⁹/L and 100 × 10⁹/L, and the decision was discussed with the trial coordinator if these levels had not been reached.

DEX. The treatment consisted of 12 6-week cycles of dexamethasone, 40 mg/d, for 4 days beginning on days 1, 9, and 17 for the first 2 cycles and 40 mg/d for 4 days beginning on day 1 for the next 10 cycles. The dose could be reduced by 50% (20 mg/d) in case of toxicity.

M-DEX. The doses of melphalan and dexamethasone and dose adjustments for side effects were the same as those presented for the MP and dexamethasone regimens.

DEX-IFN. IFN alfa-2b (Schering-Plough) was administered subcutaneously at the dose of 3.0 MU 3 times weekly. It was started with dexamethasone and stopped on day 42 of the last dexamethasone cycle. IFN was permanently discontinued in the case of an emergence of cardiac dysfunction or an occurrence of seizures or psychiatric complications. Protocol doses of IFN were reduced by 20% to 50% in patients who experienced significant fatigue or other symptoms suggesting significant toxicity. The dose was subsequently reescalated if this was feasible.

Pamidronate. Pamidronate was supplied by Ciba-Geigy (Basel, Switzerland) and then Novartis (Basel, Switzerland). Pamidronate was administered intravenously at the dose of 90 mg within a 4-hour infusion time at day 1 of each cyle of chemotherapy (12 injections). Patients who had at diagnosis a creatinine serum level of 0.45 mM/L (50 mg/L) or more could be enrolled in the trial but were excluded from pamidronate treatment.

At inclusion, usual clinical and biologic data were collected (Table 1). Visits were planned after inclusion at 3 months, 6 months, and every 6 months thereafter or up to treatment withdrawal due to treatment toxicity, MM progression, patient consent withdrawal, or death. After treatment withdrawal, patient status was regularly updated for MM progression and death. At the time of the interim and final analyses, a death certificate-based search was used to update the date of death of patients when necessary.

Primary end point was overall survival (OS). Secondary end points were progression-free survival (PFS) and survival after progression, response rates, and toxicities. A strategy was defined a priori to compare survival curves between the 4 treatment groups: step 1: comparison between M-DEX and MP groups; step 2: comparison between DEX-IFN and DEX groups; step 3: comparison between M (melphalan) groups (MP and/or M-DEX) and DEX without M groups (DEX and/or DEX-IFN). For step 3, if a previous comparison at step 1 or 2 was clearly not significant (P > .20), the corresponding groups were collapsed before the comparison at step 3. Based on the primary end point, the number of patients to be randomized was estimated to be 173 per group, as follows: median survival time of 24 months in the MP group, power of 80% to detect an increase in median survival time of 12 months in a 2-sided test, type I error of 2% to ensure approximately a global type I error of 5% taking into account the number of comparisons according to the prespecified strategy, accrual time of 3 years with additional follow-up of 3 years. Assuming that 70% of enrolled patients could be analyzed, it was planned to enroll 800 patients.

The achievement of any response required an improvement in bone pain and performance status, correction of hypercalcemia, and no increase in size or number of lytic bone lesions. A partial response required a reduction in the size of soft-tissue plasmacytomas, at least a 50% reduction in the level of the serum monoclonal protein, and a reduction in 24-hour urinary light chain excretion by 75% or more. A complete response required the absence of the original monoclonal protein in serum and urine by immunofixation, less than 5% plasma cells in a bone marrow aspirate, and disappearance of soft tissue plasmacytomas.

Progressive disease required one or more of the following: more than 25% increase in the level of the serum monoclonal protein, which must also be an absolute increase of at least 5 g/L and confirmed by at least one repeated investigation, more than 50% increase in the 24-hour urinary light chain excretion confirmed by at least one repeated investigation, definite increase in the size of existing bone lesions or soft-tissue plasmocytomas, development of new bone lesions or soft-tissue plasmacytomas, or development of hypercalcemia not attributable to any other cause. Patients not meeting the criteria of either partial or complete response or progressive disease were classified as having stable disease.

Distributions of parameters evaluated at inclusion were compared globally between treatment groups through the χ² test for categorical variables and Kruskal-Wallis rank test for continuous variables. Curves for OS, PFS, and survival after progression were calculated from randomization and from progression for the latter using the Kaplan-Meier method.⁹  Time to event (death, death or progression, and death after progression) was expressed as median plus or minus SE. Relative hazards of death (relative risk [RR]), progression or death, and death after progression, with 95% confidence interval (95% CI), were estimated through the proportional hazards model.¹⁰  The parameters evaluated at inclusion were analyzed for their prognostic value on OS from randomization, after stratification on treatment arm, through stepwise multivariate proportional hazards model, using forward selection with likelihood ratio test.¹¹  In these prognostic analyses, each continuous variable was first divided into 5 categories at approximately the 20th, 40th, 60th, and 80th percentiles. If the relative death rates (ratio of the observed number of deaths to the expected number of deaths in each category, assuming no variation of death rate across categories) in 2 or more adjacent categories were not substantially different, these categories were collapsed.^12,13  If no clear pattern was observed, the median was used as cutoff point. Usual limits (eg, 3 or 6 mg/L for c-reactive protein [CRP] level) were also tested. As a consequence, 2 to 3 categories were used for each continuous variable. After univariate analysis, all variables with a P value below .20 were proposed in several steps to multivariate analyses, first including all variables with no missing values and then proposing successively variables with an increasing number of missing values. At each step, stability of the previously derived model was checked and no further analysis was performed in case of instability. All analyses were performed on an intent-to-treat basis with SPSS software (Chicago, IL). The study was reviewed and approved by the institutional ethical committee of the University Hospital of Lille. Preliminary results of the trial analyzed to July 1, 1998, were reported in September 1999.¹⁴ 

Five hundred patients were entered into the IFM 95-01 trial between June 1995 and September 1998 with an interim analysis in July 1998. Following the interim analysis, the data safety monitoring board (DSMB) recommended stopping enrollment in the DEX arm based on a striking disadvantage in terms of progression-free survival (P < .001) of DEX as compared with M groups (MP and M-DEX) and a trend on OS (P = .03). Stopping enrollment in the DEX arm led to the decision to stop enrollment into the trial, because the probability of demonstrating an advantage of M-DEX or DEX-IFN as compared with MP was too low, taking into account the results available at the time of the interim analysis.¹⁵  Of these 500 patients, 12 were excluded from analysis because they did not fulfill the above-mentioned eligibility criteria. The clinical, radiologic, and biologic characteristics of the 488 evaluable patients are summarized in Table 1. Of those eligible patients, 122 were allocated to receive MP, 118 to receive M-DEX, 127 to receive DEX, and 121 to receive DEX-IFN. There was no difference in the distributions of pretreatment characteristics between treatment arms. No patients were lost to follow-up.

Complete responses were rare in all treatment arms, less than 2% at 6 months, with no difference between the 4 treatment groups. The achievement of at least a partial response was more frequent with M-DEX (70% at 6 months) as compared with other treatment groups (about 41% at 6 months), and this difference was statistically significant (P < .001) (Table 2).

After a median follow-up time of 82.8 ± 1.6 months, the median survival time for the whole series was 35.0 ± 1.6 months (415 deaths), and the median PFS time was 18.3 ± 0.8 months (473 progressions or deaths without progression). The median PFS time was significantly longer in M groups, 22.4 ± 1.2 months, than in DEX without M groups, 12.6 ± 1.3 months (RR 1.55, 95% CI 1.30 to 1.86, P < .001), even if this difference did not convert into a survival advantage for patients receiving M (Figure 1A-B). Indeed, as described in Figure 1A, there was no significant difference in survival among the 4 treatment groups; or between MP and M-DEX (RR = 1.17, P = .27), between DEX and DEX-IFN groups (RR = 1.05, P = .75), and between DEX without M groups (DEX and DEX-IFN) and M groups (MP and M-DEX), with a survival time (median ± SE) of 37.9 ± 2.3 and 32.8 ± 2.1 months, respectively, in the two latter groups (RR = 1.16, 95% CI 0.95 to 1.40, P = .14). This result was concordant with a longer survival after the first progression for patients receiving DEX without M, 19.9 ± 2.0 months, as compared with those receiving M, 14.2 ± 1.8 months (Figure 1C) (RR 1.30, 95% CI 1.04 to 1.62, P = .02). Second-line treatments are summarized in Table 3. Most patients enrolled in DEX without M groups received alkylating agent-based regimens at time of first relapse. In the late period of the study (after January 1999), some relapsed patients received thalidomide, which was equally distributed access arms (17 patients in MP, M-DEX, and DEX; 10 patients in DEX-IFN).

Independent factors having an adverse impact on survival were a high serum calcium level (2.75 mM or more [110 mg/L or more], P < .001), an elevated serum β₂-microglobulin level with 2 cutoffs (more than 2.5 and more than 4.0 mg/L, P < .001), a poor World Health Organization (WHO) index at inclusion (more than 1, P < .001), a low platelet count (150 × 10⁹/L or less, P = .002), a high white blood cell count (more than 7.5 × 10⁹/L, P = .02), and a low hemoglobin level (90 g/L [9 g/dL] or less, P = .02). When adjusting on these prognostic factors, the results obtained for the OS comparisons between treatment groups were confirmed, with no difference between MP and M-DEX (RR = 1.01, P = .93), DEX and DEX-IFN (RR = 1.10, P = .51), and DEX without M and M groups (RR = 1.12, 95% CI 0.92 to 1.38, P = .26).

Thirty-five deaths occurred in the first 3 months (early deaths), representing 7.2% ± 1.2% (mean ± SE of Kaplan-Meier estimate in the whole population). Early deaths were significantly more frequent (P = .004) in the DEX without M groups (10.5% ± 2.0%) than in the M groups (3.8% ± 1.2%). Thirteen (37%) of these 35 early deaths were related to myeloma progression. Causes of death in the first 3 months of treatment were analyzed according to treatment allocation (Table 4). Deaths related to MM progression were more frequent in patients receiving DEX without M than in patients receiving M (5% versus 0.5%, P = .003). Except for more cancers, including myelodysplastic syndrome/secondary leukemia, in M groups (4%) than in DEX without M groups (0.5%) (P = .02), no differences were found between treatment groups with respect to causes of death after 3 months.

No significant differences in severe hematologic toxicity were noted between MP and M-DEX; 18 patients (15%) and 20 patients (17%) displayed grade 3-4 hematologic toxicity in MP and M-DEX, respectively. No myelosuppression was observed with DEX alone. In the DEX-IFN arm, IFN had to be stopped or reduced (25% to 50% dose reduction) due to myelosuppression in 1 and 12 patients, respectively. In the DEX-IFN arm, 28 patients (23%) had to stop IFN because of IFN-related (or at least partially related) toxicity: arythmia in 2 patients, intolerance in 13 patients (with refusal in 2 patients), refusal in 2 other patients, depression in 3 patients, severe confusion with hallucinations in 4 patients (which could also be DEX related), seizure in 1 patient, hematotoxicity in 1 patient, hepatotoxicity in 1 patient, and miscellaneous in 1 patient. IFN was stopped between month 1 and month 12, at a median time of 3 months. In addition, 30 patients (24%) had to reduce by 20% to 50% at a median time of 3 months the IFN dosage in relation to toxicity.

Severe nonhematologic toxicities are presented according to treatment allocation in Table 5. Severe pyogenic infections were more frequent in the M-DEX arm compared with other arms: 19% versus 10%, 11%, and 9% among patients on MP, DEX, and DEX-IFN, respectively (P = .01), and this was mainly due to a higher incidence of pneumopathy. When hemorrhage, severe diabetes, perforated diverticulum, and psychiatric complications were considered together, they occurred less frequently in the MP group (3%) as compared with the DEX-containing groups (11%) (P = .007) but also when the comparison was restricted to DEX alone (13%) (P = .007). Deep venous thrombosis (DVT)/pulmonary embolism was equally distributed among arms (3% to 5%). When combining all severe nonhematologic complications, the incidence was also lower in the MP group (16%) than in the DEX-containing groups (28%) (P = .01) but also when the comparison was restricted to M-DEX (31%) and to DEX alone (27%) (P = .01 and .05, respectively). A total of 14 patients had a solid tumor during trial follow-up (4 patients in MP, 4 patients in M-DEX, 4 patients in DEX, and 2 patients in DEX-IFN groups), the direct cause of death in 6 patients.

In a large randomized trial comparing MP as the reference treatment in elderly newly diagnosed MM patients with 3 different treatments including DEX (M-DEX, DEX, and DEX-IFN), no difference in OS was evidenced between the 4 treatment groups, whereas the response rate appeared to be significantly higher in patients receiving M-DEX than in those receiving other treatments, and PFS was significantly better in patients receiving M than in those receiving DEX without M. In addition, severe nonhematologic toxicities were significantly more frequent in patients receiving DEX regimens than in patients receiving MP, severe pyogenic infections in patients receiving M-DEX, and hemorrhage, perforated diverticulum, psychiatric complications, or severe diabetes in those receiving a DEX-containing regimen.

At the time the IFM 95-01 trial was initiated, the MP regimen was still the reference treatment for elderly newly diagnosed MM patients ineligible for high-dose therapy, but DEX or DEX-containing regimens were considered of interest. Indeed, Alexanian et al had reported a 43% response rate in newly diagnosed MM patients treated with DEX alone, which was only 15% below the response rate achieved with the VAD regimen with no differences in survival.⁴  The study was not randomized, consisting of 112 consecutive patients with a median age of 60 years (the youngest patient was 30 years of age) and a relatively short follow-up. In 88 patients who had failed to achieve response with induction chemotherapy, Salmon et al had investigated the DEX-IFN regimen as a rescue treatment.⁷  These patients had a better average outcome in terms of survival than did patients responsive to the initial chemotherapy with a median survival of 48 months from start of DEX-IFN, providing a rationale for a large evaluation of DEX-IFN in newly diagnosed patients. In 1995 the M-DEX combination had not been evaluated in a randomized study, and only one recent publication has presented a comparison between MP and M-DEX.⁵ 

The aim of the IFM 95-01 study was to compare, in elderly patients ineligible for high-dose therapy, MP, M-DEX, DEX alone, and DEX-IFN. This study represents the largest randomized trial comparing MP with various DEX regimens before the availability of new drugs such as thalidomide, bortezomib, or lenalidomide. It is of importance to consider that these patients were between 65 and 75 years of age and that the study was performed in 104 centers in France, Belgium, and Switzerland, adequately reflecting an elderly unselected MM population referred to hospital in a recent period of time. Moreover, the follow-up was rather long for a clinical trial in MM, with a median of about 7 years.

The achievement of at least a partial response was significantly more frequent in the M-DEX arm (P < .001) (Table 2). Response rates at 6 months (and also at 12 months, data not shown) were similar in DEX without M and MP regimens (Table 2). A slight increase in response rate could not be excluded between 6 and 12 months for MP (41% at 6 months versus 50% at 12 months, data not shown), possibly reflecting the presence of few slow responders in the MP arm. In the recent study reported by the Spanish PETHEMA group (170 patients, median age 74 years, 87 receiving MP and 83 M-DEX), the partial response rates at 6 months were similar to those achieved in our study: 41% in both studies for MP, 59% and 70% for M-DEX in the Programa para el Estudio de la Terapeutica en Hemopathia Maligna (PETHEMA) study and the IFM 95-01 study, respectively. In both studies also, response rates achieved by M-DEX were superior at 12 months to those achieved by MP.⁵ 

The disease control achieved by DEX and DEX-IFN was clearly inferior to that achieved by M-containing treatments (MP and M-DEX) in terms of PFS and MM-related deaths in the first 3 months of treatment (Table 4; Figure 1B). Notably, 12 of 248 patients receiving DEX or DEX-IFN had MM progression and death in the first 3 months, whereas it was the case for only 1 of 240 patients receiving MP or M-DEX (P = .003). Nine of these early deaths occurred in patients treated with DEX-IFN, and the possibility that IFN could promote MM progression in some rare patients may not be ruled out. The PFS curves separated very early in a 2 by 2 pattern whether the patients received M or not (Figure 1B). The PFS advantage obtained in patients receiving M did not translate in a better survival, because the OS was similar among arms (Figure 1A). In fact, patients initially treated with DEX or DEX-IFN had a better average outcome after the first progression (Figure 1C). Second-line treatment consisted of alkylating agent-based regimens in most of these patients, and these regimens appeared able to rescue patients at relapse in the DEX and DEX-IFN arms (Table 3).

The IFM 95-01 provides the most complete database for DEX toxicity in elderly MM patients. The severe nonhematologic toxicities were significantly more frequent in all DEX-containing arms (Table 5). The incidence of serious DEX complications was lower in the initial report of Alexanian et al,⁴  but their patients were younger (median age, 60 years versus 70 years in the IFM 95-01 study), including some very young patients. DEX-related toxicity is in part age related, and our results are in line and extend the results in the recent PETHEMA study. Our 19% incidence (22 of 118 patients) of severe pyogenic infections in the M-DEX arm was similar to the 14% incidence of the PETHEMA study, which also considered an elderly population. In addition, we found approximately 12% more severe nonhematologic complications in DEX-containing arms than in the MP arm, a result comparable to the more limited difference observed in the PETHEMA study (grade 3-4 toxicity during cycles 1 to 6 was 3% in the MP arm versus 12.5% in the M-DEX arm). Even when considering DEX alone or M-DEX, the toxicity was higher to the one observed with MP. Of note, in a recent Eastern Cooperative Oncology Group (ECOG) study in elderly patients (median age, 65 years), DEX toxicity was also a concern, with 17% of patients having a toxicity of grade 4 or greater.¹⁶  The DEX-IFN arm had the same profile for DEX toxicity with a greater incidence of psychiatric complications and approximately one fourth of patients who had to stop IFN because of toxicity. The IFM 95-01 trial also provides useful information regarding the incidence of DVT in the context of MP- or DEX-containing regimens before the use or the addition of thalidomide. Overall, we had a 4% incidence of DVT, and this incidence was similar among the different treatments. Recent results from ECOG also found a 3% incidence of DVT using DEX alone¹⁶ 

Taking into account efficacy and toxicity results, we concluded that in the context of the IFM 95-01 trial the standard M-prednisone remained the best treatment choice in elderly patients. It does not exclude that the use of DEX could be an option in individual patients and selected situations such as renal failure, cord compression, or reduced blood count values. The addition of IFN to DEX was of no benefit in our trial, as also shown recently with maintenance IFN in the US Intergroup trial.¹⁷  In the recent period, DEX-containing regimens have been often designed for transplantation candidates to avoid M exposure prior to stem cell harvest, but they have been extrapolated to the nontransplantation population without any convincing data. The results of the IFM 95-01 trial are useful to design future combinations of innovative drugs, such as thalidomide, bortezomib, or lenalidomide. These new drugs will possibly add efficacy to the conventional treatment but will also add their own toxicity. Due to the toxicity of DEX or DEX-containing regimens in elderly patients, the combination of these innovative drugs with M should, in our opinion, be favored.

We gratefully acknowledge all the clinicians of participating hospitals: Pr Facon (Lille 1); Dr Pegourie, Pr Sotto (Grenoble 1); Pr Attal, Dr Payen (Toulouse 1); Pr Guilhot, Dr Renaud, Dr Sadoun (Poitiers 1); Dr Voillat (Besancon); Dr Dorvaux, Dr Hulin (Nancy); Dr Lepeu (Avignon); Pr Harousseau, Pr Moreau (Nantes 1); Pr Eschard (Reims 2); Pr Ferrant, Dr Straetmans (UCL Bruxelles, Belgique); Dr Blanc (Chambery); Dr Boehn, Dr Maloisel, Dr Neyrolles (Strasbourg); Dr Orfeuvre (Bourg-en-Bresse); Pr Rossi (Montpellier 1); Dr Azais (Poitiers 2); Dr Fruchart, Dr Monconduit (Rouen-Centre Becquerel); Dr Collet, Dr Pallot-Prades (Saint-Etienne 2); Dr Anglaret, Dr Peaud (Valence); Dr Morice, Dr Yakoub-Agha (Saint-Brieuc); Dr Wetterwald (Dunkerque); Dr Eghbali (Bordeaux-Institut Begonie); Dr Vekemans (Hornu-Belgique); Dr Maisonneuve, Dr Tiab (La Roche-sur-Yon); Pr Michallet, Dr Troncy (Lyon 1); Pr Grosbois (Rennes 1); Pr Bosly, Dr Doyen (Yvoir-Belgique); Dr Francois, Dr Gardembas (Angers); Pr Abgrall, Dr Autrand (Brest 1); Dr Bressot, Dr Salles (Chalon/Saone); Dr Stoppa, Dr Chabbert (Marseille); Dr Patoz, Dr Leloet (Rouen, Bois-Guillaume); Dr Cousin, Pr Marit, Pr Reiffers (Bordeaux-Haut Leveque); Dr Fouilhoux, Dr Muron (Clermont-Ferrand); Pr Rose (Lille 3); Dr Thyss (Nice, Centre Lacassagne); Dr Caulier (Sables-d'Olonne); Dr Jaubert (Saint-Etienne 1); Dr Baumelou (Suresnes); Dr Zarnitsky (Le Havre 1); Pr Fuzibet (Nice, Cimiez); Dr Brocq (Nice, Archet); Dr Kerneis (Pontoise); Pr Laroche (Toulouse 3); Dr Jardel (Vannes); Pr Casassus (Bobigny); Pr Fain (Bondy); Dr Da Silva (Elbeuf); Dr Boulet (Mons-Belgique); Dr Gallas, Dr Cusset (Montbrisson); Dr Hamidou (Nantes 2); Pr Najman (Parise, Saint-Antoine); Dr Merlet (Pau); Dr Plantier (Roubaix); Dr Chollet (Tarbes); Dr Lo Re (Tulle); Dr Frenkiel (Alencon); Dr Pierre (Arlon-Belgique); Dr Caulier (Arras); Dr Buyse (Charleville-Mezieres); Dr Depernet (Chaumont); Dr Juan (Corbeil); Dr Kara Slimane, Dr Solary (Dijon 1); Dr Valenza (Draguignan); Dr Jacquin (Feurs); Dr Massot (Grenoble 2); Dr Ravoet (Haine-Belgique); Dr Rosselet (Lausanne, Suisse); Dr Durand (Le Havre 2); Dr Bouabdallah (Libourne); Dr Berthier, Dr Martinon (Lyon 3); Dr Brantus, Dr Delmas (Lyon 4); Dr Riviere (Montpellier 3); Dr Baudard (Paris, Hotel Dieu); Dr Colomb (Romans); Dr Bernard, Dr Hauteville (Saint-Mande); Dr Kuntz (Strasbourg 2); Dr Cheron (Bligny); Dr Lassoued (Cahors); Dr Canon (Charleroi-Belgique); Dr Connan (Cognac); Dr Moriaux (Dieppe); Dr Flesch (Dijon 2); Dr Kentos (Erasme-Belgique); Dr Lang (Firminy); Dr Matthes (Geneve-Suisse); Dr Mineur (Gilly-Belgique); Dr Maisonneuve (La-Roche-sur-Yon); Dr Cance (Le-Puy); Dr Dupriez (Lens); Dr Dervite (Lille 4); Dr Houvenagel (Lomme); Dr Blay (Lyon 2); Dr Christian (Metz); Dr Gayraud (Montsouris); Dr Duprez (Ottignies-Belgique); Dr Schlaifer (Pau 2); Dr Vallantin (Perpignan); Dr Hutin (Quimper); Dr Azagury (Saint-Germain-en-Laye); Dr Maigre (Saumur); Dr Benboubker (Tours 1); Dr Bindi (Verdun); Dr Richard (Vernon); Dr Castaigne (Versailles).

We thank Myriam Cambié, Jacqueline Regnault, and Colette Geneix for excellent technical assistance. We thank Allen Ho for assistance in writing the manuscript. We thank Ciba-Geigy (Novartis) for supplying the pamidronate.