The risk of severe infection is a crucial factor in the assessment of the short-term risk:benefit ratio of biologic drugs in rheumatoid arthritis (RA). There is no increase in severe infections in RA patients treated with rituximab (RTX) in controlled trials, but this has not yet been assessed in daily practice. We undertook this study to investigate the occurrence of and risk factors for severe infections in off-trial patients using data from the AutoImmunity and Rituximab (AIR) registry.
The AIR registry was set up by the French Society of Rheumatology. The charts of patients with severe infections were reviewed.
Of the enrolled patients, 1,303 had at least 1 followup visit at 3 months or later, with a mean ± SD followup period of 1.2 ± 0.8 years (1,629 patient-years). Eighty-two severe infections occurred in 78 patients (5.0 severe infections per 100 patient-years), half of them in the 3 months following the last RTX infusion. Multivariate analysis showed that chronic lung disease and/or cardiac insufficiency (odds ratio 3.0 [95% confidence interval 1.3–7.3], P = 0.01), extraarticular involvement (odds ratio 2.9 [95% confidence interval 1.3–6.7], P = 0.009), and low IgG level (<6 gm/liter) before initiation of RTX treatment (odds ratio 4.9 [95% confidence interval 1.6–15.2], P = 0.005) were significantly associated with increased risk of a severe infection.
The rate of severe infections in current practice is similar to that reported in clinical trials. The risk factors for severe infections include chronic lung and/or cardiac disease, extraarticular involvement, and low IgG before RTX treatment. This suggests that serum IgG should be checked and the risk:benefit ratio of RTX discussed for patients found to have low levels of IgG.
Explanatory controlled clinical trials are usually performed on selected, restricted populations of patients and focus on short-term efficacy. For example, the 3 controlled trials carried out before the monoclonal anti-CD20 antibody rituximab (RTX) was approved for treating rheumatoid arthritis (RA) involved fewer than 1,000 patients with RA (1–3). Patients with other disorders, such as cancer (except skin basocellular cancer), recurrent infections, or severe concomitant illness, were excluded from these trials. In clinical practice, however, treatment is likely to be more prolonged and the patient population more heterogeneous, and many of the patients have concomitant diseases. There is, therefore, some uncertainty as to whether the tolerance or safety data obtained in randomized controlled trials (RCTs) can be extrapolated to patients in daily practice.
The risk of a severe infection is a crucial factor when assessing the short-term risk:benefit ratio of using biologic drugs to treat autoimmune diseases such as RA. RCTs have shown no increase or, in some cases, slight increase, in severe infections when biologic agents (tumor necrosis factor [TNF] antagonists, abatacept, RTX, tocilizumab) have been used to treat RA. However, data from some meta-analyses and most registries show that patients with RA treated with anti-TNF agents have a 1.5–2.5-fold greater risk of severe infection than those treated with disease-modifying antirheumatic drugs (DMARDs) (4–10). The RCTs that evaluated RTX showed a rate of severe infections of 1.9–5.9 per 100 patient-years (1–3). The rate of severe infections was 4.26 per 100 patient-years during the open-phase followup of these patients (11, 12). However, the occurrence of severe infections in the setting of daily practice has not yet been determined. The French Society of Rheumatology has therefore set up a national multicenter prospective registry of patients treated with RTX for autoimmune diseases. This registry is intended to determine the tolerance to and efficacy of RTX in daily practice. The present study analyzed the rates of severe infections in RA patients treated with RTX and the associated risk factors in a real-life setting.
PATIENTS AND METHODS
Organization of the AutoImmunity and Rituximab (AIR) registry.
The AIR registry is an ongoing nationwide prospective cohort study investigating the long-term safety and efficacy of RTX for treating RA and other refractory autoimmune diseases. This study was approved by the French authorities (“Comité Consultatif sur le Traitement de l'information en matière de Recherche dans le domaine de la Santé” and “Commission Nationale de l'Informatique et des Libertés”). Written informed consent was obtained from all patients. All of the 55 French university rheumatology units and all of the 71 nonuniversity rheumatology units were invited to take part in the registry by regular mail and electronic mail, using the mailing list of the French Society of Rheumatology. No centers declined to participate. Some centers did not respond despite iterative solicitations. No center that agreed to participate dropped out subsequently. To participate in the registry, clinicians had to send a signed form indicating their willingness to follow up patients prospectively for 5 years and to avoid losing patients to followup. Centers were informed that study nurses would help them collect data in order to increase the participation rate. In this noninterventional study, no laboratory examination was required at baseline or during followup, but results of all examinations ordered by the clinician at baseline or during followup were recorded. Newsletters and the protocol informed all the centers that biologic data and immunologic data assessed, such as levels of gamma globulins and IgG, would be collected when available.
There are currently 88 centers involved in the AIR registry, including 22 departments of internal medicine (10 university and 12 nonuniversity hospitals), which registered 126 patients (7.5%), and 66 departments of rheumatology (34 university hospital departments of rheumatology and 32 other centers, widely geographically distributed), which registered 1,555 patients (92.5%). See Appendix A for a list of investigators in the AIR registry. The participating rheumatology departments corresponded to 62% of the university rheumatology departments and 45% of the nonuniversity rheumatology departments. In each participating center, all consecutive eligible patients were included. With information from the hospital pharmacies on the total amount of RTX prescribed for nonhematologic, nononcologic indications, it has been estimated that the 88 centers participating in the AIR registry were writing >85% of the RTX prescriptions in nonhematology, nononcology departments in France. Thus, the patients included were representative of those for whom RTX was prescribed for nonhematologic, nononcologic indications and for RA.
Data were collected at baseline (at the time of the patient's first exposure to the drug) and at 3- and 6-month followup visits, then every 6 months or at disease relapse, using an electronic case report form. If a patient had no followup visits for ≥10 months, either the primary care physician or the private rheumatologist was telephoned. Thirteen research study nurses were specifically trained about RA, RTX, and the use of the electronic case report form by the coordinators of the study (J-EG and XM). Study nurses visited each center regularly to update the clinical and biologic data on the enrolled patients. The amount of missing data was minimized by providing the physician in charge of the patient and the study nurses with summaries of missing data for each patient in each center, and requesting that the missing data be provided whenever possible. Inconsistencies in data were noted, and corrected whenever possible. The charts of patients with severe infections were reviewed by 2 of the authors (J-EG and XM). In each patient with severe infection according to the electronic case report form, it was verified in the charts that patients were hospitalized due to their infection and/or received intravenous antibiotics or died. The main inconsistencies were represented by patients hospitalized for a reason other than infection (e.g., osteoporotic vertebral fracture) and by patients who received oral antibiotics during their hospitalization (e.g., amoxicillin for a sore throat); 11 infections were thus reclassified as nonsevere. Moreover, the charts of all hospitalized patients were reviewed by the 2 coordinators of the registry. This allowed the reporting of 23 additional severe infections not previously reported.
Taking into account the long duration of action of RTX, a severe infection was defined as an infection occurring during the 12 months following each infusion of RTX (initial treatment and/or each subsequent re-treatment infusion) and requiring hospitalization and/or intravenous antibiotics and/or resulting in death. The charts of patients with severe infections were reviewed by 2 of the authors (J-EG and XM).
Patients contributed person-years of followup between the first infusion of the first cycle and either death or the last followup visit. For calculation of the rate of severe infections, the time of severe infections was used as the right censoring rule. Patients who developed >1 infection (4 patients) were censored at the time of their first infection. Rates of severe infections are presented as events per 100 patient-years. We investigated the relationships between severe infections occurring during the 12 months following each infusion of RTX (initial treatment or re-treatment) and potential predictors, i.e., age, sex, disease severity (baseline Disease Activity Score in 28 joints [DAS28]  and disease duration), the number of previous DMARDs and TNF antagonists, rheumatoid factor (RF) positivity, comorbidities, extraarticular manifestations, concomitant use of steroids and DMARDs, and serum gamma globulin and immunoglobulin levels.
Both univariate and multivariate analyses were performed, using as the outcome of interest the occurrence of at least 1 severe infection either during the followup period or after the first cycle. Qualitative variables were compared using the chi-square test (or Fisher's exact test when appropriate), and quantitative variables were compared using one-way analysis of variance (or the Mann-Whitney U test when appropriate). All the variables showing univariate association with the dependent variable with a P value less than 0.20 were entered into a multivariable logistic regression model. A backward selection procedure was then used (with P values greater than 0.05 as the removal criterion, using Wald's test). The interactions between the remaining variables in the final model were tested. Results are expressed as odds ratios (ORs) with 95% confidence intervals (95% CIs).
Characteristics of the patients in the AIR registry.
RA patients (n = 1,681) from 88 centers, all enrolled at the time of their first exposure to the drug, were included in the AIR registry between September 2005 and April 2009. At the time of analysis of the results, 378 patients had just been recently enrolled and the duration since the first infusion of RTX was <3 months. Thus, these 378 patients had not reached their first followup visit at 3 months and could not be analyzed for the risk of severe infections, the main objective of the study. The analysis was therefore carried out on 1,303 patients (177 with followup visits between 3 and 6 months, 406 with followup visits between 6 and 12 months, 303 with followup visits between 12 and 18 months, and 417 with followup visits at >18 months), with a mean ± SD followup period of 1.2 ± 0.8 years, for a total of 1,629 patient-years. Of these patients, 712 were re-treated with RTX (466 with 2 cycles, 176 with 3 cycles, 45 with 4 cycles, and 25 with ≥5 cycles), and 31 (2.4%) were lost to followup.
Two hundred fourteen patients (20.3%) had experienced previous severe or recurrent infections, including 40 previous active tuberculosis infections, and 139 (13.0%) had a history of cancer (Table 1). The median time between the diagnosis of cancer and the first RTX infusion was 4 years (range 1–9 years).
Table 1. Baseline characteristics of the 1,303 patients with RA treated with rituximab in the AutoImmunity and Rituximab registry*
Except where indicated otherwise, values are the number (%) of patients. Calculated percentages often reflect a denominator of <1,303, since data for a given outcome were not always available for all 1,303 patients. RA = rheumatoid arthritis; DMARDs = disease-modifying antirheumatic drugs; TNF = tumor necrosis factor; RF = rheumatoid factor; ESR = erythrocyte sedimentation rate; CRP = C-reactive protein; DAS28-ESR = Disease Activity Score in 28 joints, calculated using the ESR.
Characteristics of the RA patients and concomitant medication at the beginning of RTX treatment (Table 1).
The mean ± SD disease duration was 15.5 ± 9.4 years. RF was present in 878 patients (78.5%). There was RA-related extraarticular involvement, including rheumatoid nodules, Sjögren's syndrome, scleritis, RA-related lung involvement, and Felty's syndrome, in 221 patients (17.3%). Patients had been given a mean ± SD of 3.2 ± 1.4 previous DMARDs; prior to RTX infusion, 267 patients (20.5%) had not been given any prior anti-TNF agents, 298 (22.9%) had been given 1 anti-TNF agent, 432 (33.2%) had been given 2 anti-TNF agents, and 304 (23.4%) had been given 3 anti-TNF agents. Prior to RTX infusion, 152 patients (11.9%) had received anakinra and 49 (3.9%) had received abatacept.
The mean ± SD DAS28 was 5.7 ± 1.2. Most of the patients (1,033 [79.8%]) were still receiving oral corticosteroids at the beginning of RTX treatment (mean ± SD 10.1 ± 10.5 mg/day). One-third of the patients (416 [33.6%]) were treated with RTX alone, and 821 (66.4%) were treated with RTX plus a nonbiologic DMARD. Most patients (1,247 [95.7%]) were treated at the first cycle with 2 infusions of 1 gram of RTX each, separated by an interval of 2 weeks, and 1,004 patients (82.8%) had received premedication with methylprednisolone.
Of the 1,303 patients, baseline data on serum gamma globulin levels were available for 829 (63.6%), and baseline data on immunoglobulin levels were available for 634 (48.7%). Of these patients, 44 (5.3%) had hypogammaglobulinemia (<6 gm/liter) and 29 (4.6%) had low IgG levels (<6 gm/liter) before RTX treatment. The baseline B cell count (CD19+ cells) was available for only 29.5% of patients.
Rate of severe infections.
Ninety-two severe infections, which required hospitalization and/or intravenous antibiotics and/or resulted in death, occurred in 88 patients. During the 12 months following any infusion of RTX (first or subsequent cycles), there were 82 severe infections in 78 patients (5.0 severe infections per 100 patient-years). These infections resulted in 4 deaths. Thirty-four severe infections were bronchopulmonary (41.5% of the infections), 13 were of the skin/soft tissue (15.9%), 11 were urinary (13.4%), 11 were of the gastrointestinal tract (13.4%), 10 were osteoarticular (12.2%), 2 were of the eye, nose, and throat (2.4%), and 1 was septicemia (1.2%).
Only 1 opportunistic infection was reported (fungal septic arthritis), and there were no tuberculosis infections. The infections occurred after the first cycle of RTX (n = 56), the second cycle (n = 22), the third cycle (n = 3), and the fourth cycle (n = 1). Among these infections, 65 (79.3%) occurred in the 6 months following the last RTX infusion (42 [51.2%] within 3 months and 23 [28.0%] between 3 and 6 months), and 17 (20.7%) occurred between 6 and 12 months (Figure 1). A pathogen was identified in 29 (35.4%) of these infections (Pseudomonas aeruginosa [n = 6], Escherichia coli [n = 6], Staphylococcus aureus [n = 5], Enterobacterium [n = 5], Streptococcus pneumoniae [n = 2], Clostridium difficile [n = 1], Scedosporum apospermi [n = 1], Enterococcus faecalis [n = 1], Corynebacterium [n = 1], varicella-zoster virus [n = 1]).
Risk factors for severe infections.
Factors predictive of the severe infections that occurred during the 12 months following any RTX infusion (first or subsequent cycles).
Univariate analysis showed that age, chronic lung disease and/or cardiac insufficiency, fewer previous anti-TNF agents, extraarticular involvement, previous severe infection, dosage of corticosteroids, hypogammaglobulinemia, or low IgG level before initiation of RTX treatment were associated with a higher risk of severe infection (Table 2). Thus, 6 of 29 patients (20.7%) with low baseline IgG levels subsequently developed a severe infection, compared with 31 of 605 patients (5.1%) with a normal baseline IgG levels (P = 0.005). Multivariate analysis showed that chronic lung disease and/or cardiac insufficiency (OR 3.0 [95% CI 1.3–7.3], P = 0.01), RA-related extraarticular involvement (OR 2.9 [95% CI 1.3–6.7], P = 0.009), and low IgG level (<6 gm/liter) before RTX treatment (OR 4.9 [95% CI 1.6–15.2], P = 0.005) were significantly associated with an increased risk of severe infection after RTX infusions.
Table 2. Univariate analysis of risk factors for severe infections that occurred during the 12 months following any RTX infusion (after first or subsequent cycles)*
Except where indicated otherwise, values are the percent of patients. Multivariate analysis showed that RA-related extraarticular involvement (P = 0.009), chronic lung disease and/or cardiac insufficiency (P = 0.01), and low IgG level (P = 0.005) were significantly associated with an increased risk of severe infection after rituximab (RTX) infusions. Anti-CCP = anti–cyclic citrullinated peptide (see Table 1 for other definitions).
Within the year following the last infusion of RTX.
Dosage in patients treated with corticosteroids, mean ± SD mg/day‡
10.8 ± 12.4
7.7 ± 8.0
Hypogammaglobulinemia (<6 gm/liter) before RTX
Low IgG (<6 gm/liter) before RTX
Low IgM (at least 1 value <0.5 gm/liter) before RTX
Neutropenia (1 value <1,000 neutrophils/mm3) before or after RTX
Number of cycles, mean ± SD
1.8 ± 1.0
1.8 ± 1.0
Factors predictive of the severe infections that occurred during the 12 months following the first cycle of RTX.
Multivariate analysis indicated that age (OR 1.05 [95% CI 1.01–1.09], P = 0.01), RA-related extraarticular involvement (OR 2.6 [95% CI 1.1–6.1], P = 0.03), and low IgG level (<6 gm/liter) before RTX treatment (OR 3.7 [95% CI 1.1–12.1], P = 0.03) were significantly associated with an increased risk of severe infection in the 12 months following the first cycle of RTX.
To our knowledge, this is the largest prospective study, based on a national registry, of severe infections in RA patients treated with RTX in the clinical setting. Many of the patients had comorbidities (13% had a history of cancer, 12.7% had chronic lung disease and/or cardiac insufficiency, and 10.2% had diabetes), unlike patients included in clinical trials. The occurrence of severe infections in the RA patients in the AIR registry (5.0 per 100 patient-years) was similar to that reported in controlled trials (1–3), despite these other disorders. Lung and cardiac diseases, extraarticular involvement, and low IgG level before RTX treatment were independently associated with an increased risk of a severe infection.
The limitations of the study include the absence of comparison with patients treated with DMARDs or another biologic drug, missing data, and the relatively short followup period. Unfortunately, we could not compare patients treated with RTX and those treated with DMARDs or other biologic agents, since no registry of patients treated with anti-TNF agents is available in France. Regarding missing data, some of them were related to the observational nature of the study and reflect current real-life practice. Thus, we had no DAS28 value at enrollment for 19% of the patients. This indicates that the clinical assessment of disease activity in RA could be collectively improved. In this study, we focused on severe infections, which by definition result in hospitalization, intravenous antibiotics, and/or death and can be easily identified. Reporting every 6 months by clinical nurses after systematic reviews of charts at the site where the patients are treated might be more efficient and less biased than declarative reporting by clinicians. The central reviewing of charts enabled us to correctly identify severe infections. Moreover, the percentage of loss to followup (2.4%) was very low, which supports the chosen methodology. The followup time is currently limited, so that we could not analyze the long-term safety of RTX, including the impact of repeated cycles of RTX, its efficacy, or therapeutic maintenance.
The AIR registry contains a high percentage of RTX-treated patients with comorbidities (e.g., 13% with a history of cancer). Thus, this registry is useful to determine the safety of RTX in patients with concomitant disorders, for whom reported data are very limited. This high proportion of other disorders and past severe infections, including tuberculosis, probably led to the prescribing of RTX in 20% of patients not previously treated with anti-TNF agents and to one-third of the patients being prescribed RTX alone, without any DMARD. Similar proportions have been reported from other European registries; among the 794 patients included in CERERRA (Collaborative European Registries for Rituximab in Rheumatoid Arthritis), a collaborative registry of 9 European countries, 15.5% had not received anti-TNF agents before RTX (14), and this was also the case for 26% of the 322 patients in the German RABBIT (German acronym for Rheumatoid Arthritis–Observation of Biologic Therapy) registry (15).
The definition of the at-risk period has led to differences in the evaluation of the occurrence of severe infections in patients receiving anti-TNF agents (6). The at-risk period for RTX is even harder to define, since RTX is currently not prescribed at fixed time intervals. Since time to B cell repopulation (data not available in the registry) might be 6–12 months, the at-risk period was defined as 12 months following each RTX infusion. The other reason for this choice was that most patients are either re-treated with RTX 6–12 months after the last infusion (in these patients, the overall 5-year period from the first cycle of RTX is therefore analyzed) or stop RTX and start a new biologic agent after RTX discontinuation. In this latter population, inclusion of severe infections occurring >12 months after the last infusion of RTX in the analysis seems debatable. Risk factors for all severe infections, irrespective of the period since the last RTX infusion, were nonetheless the same as those for severe infections occurring in the 12 months following the last RTX infusion (data not shown). The rate of severe infections after RTX appears to be similar to that after anti-TNF agents (1.7–6.0 per 100 patient-years) (4–10). The greatest risk of severe infections occurs within the first 6 months after RTX, the same as with anti-TNF agents (4, 5, 9), and especially within the first 3 months, when half of the severe infections occurred. This early occurrence of severe infections could have been partially due to the methylprednisolone infusions (2 × 100 mg) given concomitant with RTX to reduce infusion reactions.
There was only 1 opportunistic infection. Although 40 patients had had a tuberculosis infection, there was no reactivation or new active tuberculosis. There was also no progressive multifocal leukoencephalopathy, but collaborative pooled analysis of registries is needed to analyze very rare events such as progressive multifocal leukoencephalopathy (16–18). There were only 2 pneumococcal infections in patients without pneumococcal vaccination.
The main objective of the study was to determine the risk factors for severe infections in patients treated with RTX. Some of the risk factors for severe infections identified were similar to those reported for RA patients treated with DMARDs or other biologic drugs—coexistence of other diseases, RA extraarticular involvement, and corticosteroid dosage (19, 20). The last remaining factor associated with severe infections in multivariate analysis was a low baseline IgG level (<6 gm/liter before RTX treatment).
Low IgG is a well-recognized risk factor for infections in diseases such as genetic and acquired immunodeficiency syndromes and multiple myeloma. IgG concentrations <6 gm/liter have been associated with severe infections in acquired immunodeficiencies,such as transplantation, chronic lymphocytic leukemia, or nephrotic syndrome (21). However, Ig concentrations are not routinely measured in RA, which explains why the role of low IgG in the occurrence of severe infections is usually not studied with other biologic drugs. The gamma globulin and IgG data for patients in the AIR registry were frequently missing, perhaps because there is no international consensus on biologic monitoring of serum Ig before RTX treatment and in RTX-treated patients. Nonetheless, we have these data for >50% of the patients at baseline. We now need to determine the factors underlying the low IgG levels at enrollment, since they occurred in 4.6% of the patients. Only 1.7% of the RA patients included in clinical trials of RTX had low IgG (12), which is another difference between patients in clinical trials and in registries. The association between low IgG and severe infections should also be studied in patients treated with other types of biologic drugs, such as TNF antagonists.
In clinical trials of RTX, the frequency of low IgG is elevated after repeated treatment with RTX (4.7% of patients after 4 cycles), and there is a nonsignificant increase in severe infections in these patients with low IgG levels after repeated cycles of RTX (11, 12). We did not analyze the association between multiple treatments with RTX and the development of low IgG levels and severe infections because data on IgG values after RTX treatment were available for only a small number of patients (320 of 1,303 [24.6%]), and the followup of patients who received repeated RTX treatments was limited. However, the fact that low baseline IgG, present in ∼5% of the patients, is a risk factor for infection after RTX and the fact that acquired low IgG levels can occur in a significant subset of patients after repeated cycles of RTX should be taken into account. In the future, we should assess the frequency of low IgG levels after several re-treatments and the possible association of acquired low IgG with the occurrence of severe infections. In patients with low IgG, especially those who already have had a severe infection and/or have severe comorbidities, use of a different therapy should be considered.
In conclusion, the rate of severe infections in unselected RA patients treated with RTX is similar to that reported in clinical trials. Half of severe infections occurred in the first 3 months following RTX infusion. Factors predicting severe infections include age, lung and cardiac comorbidities, extraarticular involvement, and low IgG levels. We suggest measuring serum Ig concentrations and checking for vaccination against pneumococcal infections before the first infusion of RTX, since a minority of patients might already have a low serum IgG level. The Ig concentrations should also be assessed before each new cycle of RTX, given the cumulative risk of decreased Ig levels with repeated treatments with RTX. Finally, the risk:benefit ratio of RTX should be evaluated for each individual patient with a low level of IgG.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Drs. Gottenberg and Mariette had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Gottenberg, Ravaud, Bardin, Cacoub, Cantagrel, Combe, Dougados, Flipo, Godeau, Guillevin, Le Loët, Hachulla, Schaeverbeke, Sibilia, Mariette.
Acquisition of data. Gottenberg, Bardin, Cacoub, Cantagrel, Combe, Dougados, Flipo, Godeau, Guillevin, Le Loët, Hachulla, Schaeverbeke, Sibilia, Mariette.
Analysis and interpretation of data. Gottenberg, Ravaud, Bardin, Cacoub, Cantagrel, Combe, Dougados, Flipo, Godeau, Guillevin, Hachulla, Schaeverbeke, Sibilia, Baron, Mariette.
ROLE OF THE STUDY SPONSOR
The AIR registry receives financial support (an unrestricted educational grant) from Roche. However, Roche was not involved in the design, protocol, data collection, or statistical analysis of the study. Roche was not involved in writing of the manuscript, their agreement was not required to submit the manuscript for publication, their approval of the content of the submitted manuscript was not required, and publication of the manuscript was not contingent upon their approval.
We thank all the investigators of the AIR registry, the French Society of Rheumatology, The French Society of Internal Medicine, Dr. Jamila Filipecki (previously at Roche), Dr. Nadine Mackenzie (Roche), Isabelle Pane (data manager, working with Philippe Ravaud), the 13 research study nurses (Emilie Blanchard, Aude Bourgeois, Françoise Carmet, Marie-Hélène Da Costa Silva, Sylvie Delmas, Geovana Meneses, Fatiha Medkour, Nathalie Minot, France Marie Ange Ouattarra, Valérie Pinosa, Christèle Szatanec, Hélène Thibault, Irena Vukusic), and Philippe Meme (Euraxipharma) for their contributions. The English text was edited by Dr. Owen Parkes.
INVESTIGATORS IN THE AIR REGISTRY
Investigators in the French AIR registry, in addition to the authors of this article, are as follows: Dr. Allanore (Paris), Dr. Amoura (Paris), Dr. Ardizzonne (Mulhouse), Dr. Balblanc (Belfort), Dr. Benmansour (Châateauroux), Dr. Berenbaum (Paris), Dr. Bergeron-Lafaurie (Paris), Dr. Berthelot (Nantes), Dr. Berthier (Lyon), Dr. Bolla (Cannes), Dr. Bonnet (Limoges), Dr. Boumier (Amiens), Dr. Bouvard (Angers), Dr. Breban (Paris), Dr. Briançon (Aix les bains), Dr. Brocq (Monaco), Dr. Carli (Toulon), Dr. Chales (Rennes), Dr. Chapurlat (Lyon), Dr. Claudepierre (Paris), Dr. Collet (St. Etienne), Dr. Cormier (La Roche-sur-Yon), Dr. Costedoat (Paris), Dr. Couret (Valence), Dr. Damade (Chartres), Dr. Debandt (Aulnay sous bois), Dr. Delacroix (Créteil), Dr. Deprez (Valencienne), Dr. Dernis (Le Mans), Dr. Desjonqueres (Bron), Dr. Duquesne (Bron), Dr. Eschard (Reims), Dr. Euller-Ziegler (Nice), Dr. Fain (Bondy), Dr. Fautrel (Paris), Dr. Gardin (Bonneville), Dr. Gaudin (Grenoble), Dr. Gayraud (Paris), Dr. Georges (Fort de France), Dr. Ghiringhelli (Bordeaux), Dr. Giraud-Morelet (Reims), Dr. Godde (Marseille), Dr. Gombert (La Rochelle), Dr. Goupille (Tours), Dr. Grados (Amiens), Dr. Grasland (Paris), Dr. Guyot (Roubaix), Dr. Hacene (Amiens), Dr. Hilliquin (Corbeil), Dr. Hoang (Vannes), Dr. Houvenagel (Lille), Dr. Khellaf (Paris), Dr. Lafforgue (Marseille), Dr. Larbre (Lyon), Dr. Larroche (Bobigny), Dr. Lecuyer (St. Claude), Dr. Legrand (Riaumont-Liévin), Dr. Leone (Reims), Dr. Le Parc (Paris), Dr. Lequerre (Rouen), Dr. Le Seaux (Draguignan), Dr. Loeuille (Nancy), Dr. Lohse (Belfort), Dr. Maillefert (Dijon), Dr. Marcelli (Caen), Dr. Martinon (Lyon), Dr. Messer (Colmar), Dr. Meyer (Paris) (Bichat), Dr. Moreau (Colmar), Dr. Pagnoux (Paris), Dr. Pallot-Prades (St. Etienne), Dr. Pasquali (Strasbourg), Dr. Pennaforte (Reims), Dr. Perrot (Paris), Dr. Pertuiset (Pontoise), Dr. Pham (Marseille), Dr. Puechal (Le Mans), Dr. Richette (Paris), Dr. Rist (Orléans), Dr. Roth (Marseille), Dr. Rouaghe, Dr. Rouidi (Dreux), Dr. Solau-Gervais (Poitiers), Dr. Sordet (Strasbourg), Dr. Soubrier (Clermont Ferrand), Dr. Thevenot (Laon), Dr. Thomas (Thionville), Dr. Thorel (Lorient), Dr. Toussirot (Besançon), Dr. Vignon (Lyon), Dr. Vittecoq (Rouen), Dr. Wendling (Besançon), Dr. Zarnitsky (Le Havre), Dr. Ziza (Paris).