Asthma severity and exposure to occupational asthmogens

Abstract

Rationale

Severe asthma is a public health problem with limited information regarding preventable causes. Although occupational exposures have been implicated as important risk factors for asthma and asthma exacerbations, associations between occupational exposures and asthma severity have not been reported.

Objective

To examine associations between occupational exposures and asthma severity.

Methods

The Epidemiological study on the Genetics and Environment of Asthma combines a case control study with a family study of relatives of asthmatic cases. Adult cases (n=148) were recruited in chest clinics and non-asthmatic controls (n=228) were population-based. Occupational exposures to non-asthmagenic irritants and asthmagens (classified as ‘any asthmagen’ and 3 broad groups (high molecular weight agents, low molecular weight agents, mixed environments)) were assessed by an asthma-specific job exposure matrix. Asthma severity was defined from a 7-grade clinical score (frequency of attacks, persistent symptoms and hospitalisation). Severe (score ≥ 2) and mild asthmatics were compared to controls using nominal logistic regression.

Main Results

Significant associations were observed between severe adult onset asthma and exposure to any occupational asthmagen (odds ratio 4.0 [95% CI 2.0–8.1]); high molecular weight agents (3.7 [1.3–11.1]); low molecular weight agents (4.4 [1.9–10.1]), including industrial cleaning agents (7.2 [1.3–39.9]); and mixed environments (7.5 [2.4–23.5]). No significant associations were found between non-asthmagenic irritants and asthma severity, nor between asthmagens and childhood onset asthma or mild adult onset asthma.

Conclusions

Our results suggested a strong deleterious role of occupational asthmagens in severe asthma. Clinicians should consider occupational exposures in patients with moderate to severe asthma.

Asthma prevalence has increased markedly in recent decades and is now a major public health problem (1). Trends for both hospital admission rates and asthma mortality also indicate an increase in severe asthma, to which the majority of asthma costs is attributable (1). It has been suggested that severe asthma may represent a different phenotype compared to mild asthma (2–5) and that adult onset asthma may also be a separate phenotype (2). Risk factors for asthma severity among adults have been scarcely studied (5) and include smoking habits, female gender, body mass index and possibly hormonal factors (4, 6–9). Given the burden of disease represented by severe asthma, especially in adults, it is particularly important to identify modifiable risk factors for this disorder.

Although occupational exposures have been implicated as important modifiable risk factors for asthma, associations between occupational exposures and asthma severity have not been reported (5). An American Thoracic Society Statement concluded that occupational exposures may account for about 15% of the population burden of adult asthma (10); a recent analysis of data from the US general population estimated that at least 27% of asthma among adults may be ‘work-related’ (11). Clinical and workplace studies have shown that among persons with work-related asthma, continued exposure is associated with a worsening of symptoms (12–13). Even removal from exposure can leave subjects with persistent asthma (12–13), with persistence linked to the severity of the disease before removal (13–14). The absence of population-based studies on the contribution of occupational exposures to severe asthma is likely due, in part, to difficulties in estimating the relevant occupational exposures (10,15), out of more than 250 incriminated occupational asthmagens (10,12–13), and in evaluating asthma severity in such studies (3–5).

In the French EGEA (Epidemiological study on the Genetics and Environment of Asthma, bronchial hyperresponsiveness and atopy) survey, considerable work has been done previously to define a score for current asthma severity based on clinical indicators in the past 12 months (6,16) and to develop a method for estimating relevant occupational exposure to asthmagens (17–18) suitable for case-control or population based studies.

The aim of the current analysis was to study the relationship between asthma severity and occupational exposures, in the EGEA population, using these previously developed methods. We hypothesize that the role of occupational exposure with respect to asthma severity may differ according to asthma phenotype, with a different role for asthmatics whose asthma first appeared in childhood (potentially work-aggravated asthma) compared to those whose asthma started in adulthood (potentially occupational asthma) (2). Some of the results have been previously reported in the form of an abstract (19).

METHODS

The EGEA survey (1991–1995) combines a case control study with a family study of relatives of asthmatic cases. The detailed protocol and descriptive characteristics have been published elsewhere (20–22) and some details are given in the online supplement. Asthmatic cases were recruited in chest clinics using a standardized procedure, including verification that all cases did actually have clinically relevant asthma (22–23). Controls were recruited from electoral rolls, a surgery department and one check-up centre (20–21). The present analyses were performed in 148 adult asthmatic cases and 228 non-asthmatic controls after exclusion of subjects without a job (n=62) or with uncertain estimates of exposure (n=19), with missing values for asthma (n=6), age of asthma onset (n=1) or asthma severity (n=22), and asthmatic controls (n=37).

Information was recorded for treatment, respiratory symptoms, environment, and smoking by a detailed interviewer-administered international standardized questionnaire (23). Main analyses were performed using a clinical severity score (severity classification I) previously described in detail (6–7, 16), based on frequency of asthma attacks (from 0 to 3), persistent symptoms between attacks (scored from 0 to 3), and hospitalisation (yes/no), all in the past 12 months (Table 1). The total score thus ranged from 0 to 7; a cut-off point of 2 or greater was used to classify cases with a high score as ‘severe’ and cases with a low score as ‘mild’ asthma. This cut-off point was chosen in order to have a similar number of subjects in both groups. In addition, a new severity score (severity classification II), combining the clinical severity score and treatment, as recently suggested in the GINA guidelines (24), was constructed taking into account clinical severity (classification I), asthma attacks in the last year, and treatment (none, treatment not including inhaled corticosteroids, treatment including inhaled corticosteroid) (described in the online supplement, Table E1). Analyses using this new score and using 3 other approaches, to classify cases by markers of severity (based on lung function (severity classification III), history of hospitalisation (severity classification IV), and using a ordinal logistic regression for the clinical score (severity classification V)), were also performed (online supplement).

Table 1.

Description of key measures: clinical severity score and occupational exposure

Asthma severity clinical score calculation			Asthma job exposure matrix: occupational exposure estimates
Characteristic in the last 12 months	Score	Number with this score	Exposure group	Specific exposure category	Number classified as exposed
			Asthmagens
Frequency of asthma attacks			High molecular weight	rodents, livestock	2
< 1 attack/month	0	60		fish, shellfish	0
≥ 1 attack/mo and < 1 attack / wk	1	23		arthopods, mites	1
≥ 1 attack /wk and < 1 attack /day	2	32		latex	17
≥ 1 attack /day	3	33		flour	4
				other plants	0
Symptoms between asthma attacks				bioaerosols, moulds	5
None	0	111		biological enzymes	4
Wheezing	1	5	Low molecular weight	highly reactive chemicals (any)	28
Wheezing and dyspnea	2	10		isocyanates	4
Activities limited by dyspnea	3	22		sensitizing drugs	4
				industrial cleaning agents	10
Hospitalisation for asthma				sensitizing wood dusts	1
No	0	126		metal sensitizers	10
Yes	1	22	Mixed environments	metalworking fluids	5
				agriculture	3
				textile production	6
Severity classification I :				irritant peaks/accidents	5
Severe asthma (total score of 2 or greater)		78	Non-asthmagenic irritants	low level chemicals	37
				irritant, not high peak	12
				exhaust fumes	14

Occupational exposure to each of 18 agents, groups of agents, or environments known to be associated with an increased risk for occupational asthma (referred here as “asthmagens”) and to non-asthmagenic irritants was estimated, using a Job Exposure Matrix (JEM) method described in detail previously (17). This method is now used in the European Community Respiratory Health Survey (ECRHS). Exposure estimates were derived for the current or most recent job code (25), in order to examine the association between ‘current’ asthma severity and ‘current’ exposure. Duration of the current or most recent job was also determined from the questionnaire. The asthmagens were classified in three groups as high molecular weight agents (HMW), low molecular weight agents (LMW) or mixed environments (Table 1).

Nominal logistic regression analyses adjusted for age, sex, and smoking habits were performed to compare mild and severe asthmatic cases to non-asthmatic controls. Secondary exploratory analyses were also carried out, first stratified on whether or not the case was taking inhaled corticosteroids, and second to include some other potential determinants of asthma (eg. duration of employment in the current or most recent job, and body mass index and age of menarche among women). To study the specificity of the associations, both asthmagens and non-asthmagenic irritants were considered. Childhood onset and adulthood onset cases were studied separately. Specific asthmagens were studied when at least five cases were exposed as described previously (17).

RESULTS

Subjects were, on average, about 43 years old (Table 2). Age of asthma onset did not differ between cases with mild and severe asthma overall (Table 2), nor among those with childhood onset (around 7) or adult onset asthma (around 35). All asthmatics without asthma attack in the last 12 months received inhaled corticosteroids treatment in the last year. Asthmatics classified as being ‘severe’ using the asthma severity clinical score (score 2 or greater) (Table 1) had more frequent attacks and more symptoms between attacks. However, there were no differences between severe and mild asthmatics regarding other markers of asthma (Table 2).

Table 2.

Description of the EGEA population by cases (based on the asthma severity clinical score) and controls

		Controls (n = 228)	Mild asthma * (n = 70)	Severe asthma * (n = 78)	p value
Men, %		50.4	54.3	52.6	0.84
Age, mean (SD)		43.5 (11.7)	42.3 (11.5)	42.6 (12.9)	0.71
Age of asthma onset, mean (SD)		-	24.7 (16.9)	23.1 (15.6)	0.55
Asthma onset > 18 years old, %		-	61.4	61.5	0.99
IgE, GM (95% CI) †		31 (2–525)	155 (8–2955)	211 (10–4316)	<0.0001
Positive skin prick test response (11 allergens), %		30.4	68.1	69.4	<0.0001
Peripheral blood eosinophils > 5%, %		6.6	30.3	31.0	<0.0001
FEV1‡ % predicted, mean (SD)		104.7 (14.4)	87.3 (21.2)	86.8 (21.8)	<0.0001
Methacholine test, n		161	32	21
PD20‡ < 4mg, %		19.2	87.5	90.5	<0.0001
Used of inhaled corticosteroids, %		0.9	77.1	79.8	<0.0001
Frequency of asthma attacks, past 12 months
≥ 1 attack per day		-	0.0	42.3	<0.001
≥ 1 attack per week		-	0.0	41.0
≥ 1 attack per month		-	24.3	7.7
< 1 attack per month		-	75.7	9.0
Persistent symptoms between attacks, %		-	2.9	44.9	<0.001
Hospitalisation, past 12 months, %		-	17.1	12.8	0.46
Smoking habits
Non smokers, %		46.7	47.8	36.7	0.007
Ex-smokers, %		24.7	40.6	36.7
Smokers, %		28.6	11.6	26.6
Asthma-specific Job Exposure Matrix
Exposed to asthmagens	All, %	14.9	17.1	30.8	0.008
Men, %		9.6	18.4	34.2	0.001
Women, %		20.3	15.6	27.0	0.50
Exposed to non asthmagenic irritants	All, %	21.1	13.8	14.8	0.33
Men, %		35.6	25.8	25.9	0.45
Women, %		4.4	0.0	3.7	0.54

^*Severity classification I : cases with a high clinical severity score (2 or greater) were classified as severe asthma and cases with a low score (less than 2) as mild asthma.

^†IgE, immunoglobulin E measured in IU/ml ; GM, geometric mean; 95% CI : 95% confidence interval.

^‡FEV₁, forced expiratory volume in one second; PD20, provocative dose of methacholine for a decrease of 20% in FEV₁.

Exposure to occupational asthmagens was significantly higher among severe asthmatics compared to controls and mild asthmatics, while the reverse trend was seen for exposure to irritants (highest in controls). These relationships differed considerably by gender (Table 2).

In childhood onset asthmatics, there was no relationship between any current occupational exposure and asthma (mild or severe asthmatics compared to controls) (Table 3). Comparing adult onset asthmatics with mild asthma to controls, no significant associations were found between current occupational exposure and asthma (Table 3); however, all odds ratios (OR) were greater than 1. Severe adult onset asthma was strongly and significantly associated with asthmagen exposure in the current or most recent job with an OR of four (for any asthmagen). Analyses by broad exposure groups showed consistently higher ORs for severe adult onset asthma than for mild asthma, with significant associations for the three broad exposure groups (LMW, HMW agents and mixed environments). For specific asthmagens, significant associations were observed between severe adult onset asthma and exposure to highly reactive chemicals, industrial cleaning agents, metal sensitizers and textile production jobs (Table 3).

Table 3.

Relationship between occupational exposure and the asthma clinical severity score^*

	Childhood onset asthmatics Odds ratio (95% CI)			Adult onset asthmatics Odds ratio (95% CI)
	Exposed cases (mild/severe) / controls, n	Mild asthma * (n = 27)	Severe asthma * (n = 30)	Exposed cases (mild/severe) / controls, n	Mild asthma * (n = 43)	Severe asthma * (n = 48)
Asthmagens
Any asthmagen	5/5/34	1.0 (0.3, 3.1)	0.9 (0.3, 2.6)	7/19/34	1.2 (0.5, 3.0)	4.0 (2.0, 8.1)
HMW asthmagen, any	3/2/13	1.7 (0.3, 8.9)	0.9 (0.2, 4.6)	3/6/13	1.6 (0.4, 6.2)	3.7 (1.3, 11.1)
HMW – latex	3/1/8	2.7 (0.5, 15.2)	0.6 (0.1, 5.7)	2/3/8	1.7 (0.3, 9.2)	3.3 (0.8, 14.1)
LMW asthmagens, any	1/4/19	0.4 (0.0, 3.4)	1.5 (0.4, 5.1)	4/12/19	1.2 (0.4, 3.8)	4.4 (1.9, 10.1)
LMW – highly reactive chemicals (any)	1/4/12	0.9 (0.1, 7.4)	2.4 (0.7, 8.9)	3/8/12	1.5 (0.4, 6.0)	4.8 (1.7, 13.2)
LMW – industrial cleaning agents	1/2/3	3.2 (0.3, 37.0)	5.6 (0.7, 42.4)	1/3/3	1.9 (0.2, 20.6)	7.2 (1.3, 39.9)
LMW – metal sensitizers	0/1/4	-	3.1 (0.3, 33.0)	1/4/4	1.6 (0.2, 15.7)	6.6 (1.5, 29.5)
Mixed environments – any	1/0/7	0.8 (0.1, 7.6)	-	3/8/7	2.3 (0.5, 9.6)	7.5 (2.4, 23.5)
Mixed environment – textile production jobs	0/0/1	-	-	1/4/1	4.7 (0.3, 80.8)	24.8 (2.6, 240.4)
Non-asthmagenic irritants	1/4/41	0.2 (0.0, 1.4)	0.9 (0.3, 3.3)	7/4/41	0.8 (0.3, 2.1)	0.4 (0.1, 1.5)

Analyses have been performed for specific asthmagens when at least 5 cases were exposed. Odds ratios adjusted on age, smoking habits and gender. Definitions of abbreviations : 95% CI, 95% confidence interval; HMW, high molecular weight; LMW, low molecular weight.

^*Severity classification I : cases with a high clinical severity score (2 or greater) were classified as severe asthma and cases with a low score (less than 2) as mild asthma.

Due to multiple exposures, exposure to highly reactive chemicals, cleaning agents, and latex were highly correlated and could not be examined in the same model. A model including a four-class variable (non-exposed to asthmagens, exposed to latex alone, exposed to highly reactive chemicals alone, exposed to at least cleaning agents), showed significant associations between severe adult onset asthma and industrial cleaning agents (8.1 (1.5–44.4), highly reactive chemicals alone (4.1 (1.2–13.5)) and non significant association for exposure to latex alone (2.6 (0.5–14.0)).

Secondary analyses for severe adult onset asthma led to similar results when duration of employment, or body mass index or early menarche among women were included in the models (data not shown).

Analyses stratified by inhaled corticosteroid treatment (not shown) showed that the associations between severe adult onset asthma and exposure to industrial cleaning agents, textile production jobs and latex exposure were stronger (and still significant) in asthmatics not being treated with inhaled corticosteroids (compared to those treated). No differences were found according to treatment for associations with highly reactive chemicals in general (similar OR for treated and untreated subjects) and metal sensitizers (all asthmatics treated).

Similar results were found when four other definitions of asthma severity (online supplement), a new severity score (severity classification II, Table E2) combining treatment and the first clinical score, a score based on airflow limitation (severity classification III, Table E3), ever being hospitalised for asthma (severity classification IV, Table E4), and the first clinical score with a cutoff point of 3 (not shown) were used. Analyses performed in cases only, showed significant associations between exposure to asthmagens and asthma severity using both logistic (data not shown) and ordinal logistic regressions (severity classification V, Table E5).

DISCUSSION

These results indicate that, in the EGEA case-control survey, with well defined asthmatics, exposure to asthmagens in the current or most recent job was strongly related to asthma severity among cases with adult onset asthma. The specificity of these results, the lack of association with non-asthmagenic irritants, the strength of the association for asthmagens, and the strength of the association for adult onset asthma, support a causal association between current occupational exposure to asthmagens and severe asthma (as defined here, based on a clinical severity score). The results indicate a strong deleterious role of both low and high molecular occupational asthmagens in severe asthma. For mild adult onset asthma, all OR were greater than one but associations were not significant. The associations were consistent before and after adjustment for other factors previously found related to asthma severity in the EGEA study such as smoking, body mass index and early menarche. Furthermore, consistent results were observed whatever the asthma severity definition used (as described in the online supplement) including a new composite severity score based on symptoms and treatment.

Occupational exposures and asthma severity

Our findings of a link between current occupational exposure (especially among cleaners and those in textile jobs) and markers of severe asthma are consistent with previous results from the ECRHS survey, in which higher OR were observed for asthmatics with bronchial hyperresponsiveness than for “current asthma symptoms” (compared to non asthmatics) for both textile jobs (26) and cleaners (26–28). An excess prevalence of low lung function in cleaners with asthma was observed comparing cleaning workers to office workers (29). In the French PAARC survey higher OR for exposure to asthmagens were observed for asthma with airflow limitation than for “ever asthma” compared to non asthmatics (18).

One strength of our approach is that we did not rely on self-reported exposure to occupational agents in this analyses. Self-reported exposure is not sufficiently accurate to be used as a sole exposure estimate, and has the potential to generate biased estimates, as previously underlined (15, 18). Our approach, which combines a job exposure matrix with a standardized review step has been suggested by some authors (15, 17) to be a more appropriate method, and has been shown to give reliable asthmagen exposure estimates (17–18). This combined method, easy to use, is now used in the ECRHS survey. To limit the errors of classification and increase the specificity (18), subjects with imprecise estimates of exposure were excluded from the present analyses.

For this analysis, we focused on occupational exposure in the current or most recent job in order to correspond to the fact that we based our definition of asthma ‘severity’ on the past 12 months. In a larger study, complete occupational and asthma histories, taking into account exposure windows according to the timing of asthma onset, are recommended (17–18). Results for specific asthmagens also need to be interpreted with caution because of the small number of subjects exposed in each category which induces OR with wide confidence intervals.

Our analyses that examined the association between current exposure to non-asthmagenic irritants and asthma displayed ORs lower than one, regardless of asthma severity or age of asthma onset. Such low ORs may reflect a true lack of causal association between current irritant exposure and asthma, or possibly, a selection bias. For example, subjects with asthma may not choose jobs exposed to irritants even at low level. In the general population from Spain, a higher asthma risk was observed in former cleaners than in current cleaners, suggesting self-selection from exposure (30). Selection bias could occur through self-selection (the choice of non-exposed jobs, not staying in exposed jobs) or as a result of physician advice (31). A follow-up of young asthmatics is necessary to estimate the impact of selection bias in association with occupational hazards.

Limitations regarding the definition of asthma severity

Asthma severity is difficult to define in epidemiology as previously underlined (16). The asthma severity score used here was based on clinical items, i.e. self report of frequency of asthma attacks, respiratory symptoms, and hospitalization summed over the past 12 months. This score has been used in previous EGEA analyses (6–8, 16), to assess the role of other risk factors (7–8) and has been shown to be related to familial resemblance (6) in the EGEA survey.

Asthma severity may include several dimensions and authors (32) as well as experts of the recent GINA guidelines (24) have pointed out the potential distinction between asthma control (recent symptoms) and asthma severity (treatment needed to control the symptoms). Unfortunately, in this study, asthma severity was not easily distinguished from asthma control because of limited information regarding asthma treatments and lack of data on asthma activity in the past weeks. It could be argued that our findings may reflect a relationship between occupational exposure and “uncontrolled asthma” rather than intrinsic asthma severity. However, when using a score combining clinical items and treatment, the associations between exposure to asthmagens and a severe combined score were very similar to the results obtained with the clinical score alone. This may suggest that asthmagen exposure in adult onset asthma is related to asthma severity and control. Further studies with more detailed information on asthma severity and control would be helpful to explore this issue.

It is also possible that the stronger associations observed with severe asthma in our study may be the effect of ‘severe’, as defined here, being a more specific definition of asthma. For example, in a previous population-based study (18) we found that combining airflow obstruction with ‘reported asthma’ was associated with stronger associations between asthma and occupational exposure; however, in that study asthma was based solely on self report of diagnosis, a method known to be highly non-specific. In contrast, in the EGEA survey, cases were well defined (all were recruited from chest clinics and had either a positive response to all four questions regarding asthma symptoms and diagnosis or a combination of positive responses and a medical record review). Therefore, it is unlikely that increased specificity alone would explain our results.

It should be noted that, as this is a case control epidemiological survey, it is subject to the limitations of epidemiology in general, and case-control studies in particular. For example, although asthmatic cases were recruited in chest clinics with specific criteria (they may represent slightly more severe asthma than the general population), the definition of asthma attacks and symptoms between attacks were still self-reported. It is not possible to determine whether this biased the findings reported here, although the use of exposure estimates based on job titles (and not self-report) would serve to minimize self-report bias. In addition, in case-control studies, the OR may be overestimated when the prevalence of the disease is high. However, the prevalence of asthma is around 10% and less for severe asthma, suggesting that the OR observed here are not likely overestimated.

Public health implications

Our results are consistent with the hypothesis of a progressive increase in severity or lack of control of asthma among persons currently exposed to asthmagens at work. This increase in severity is especially observed in jobs not widely known as leading to asthma such as textile production jobs or exposure to industrial cleaning agents. As previously underlined, factors associated with incidence may be different from those associated with persistence (33). In the same way, risk factors for asthma and asthma severity may differ. For example, smoking is a risk factor for asthma severity but less likely a risk factor for the development of asthma (8). The same could be true for some occupational exposures, especially if exposure continues despite the development of asthma symptoms. It is possible that the elevated but non-significant associations between occupational exposures and mild asthma seen here were simply due to low statistical power. However, another hypothesis is that asthma caused or exacerbated by work exposures quickly becomes severe due to persistent exposure, thus making it difficult to ‘detect’ an association between exposure and the mild phase of the disease. Our results suggest also that exposure to some specific asthmagens such as industrial cleaning agents and metal sensitizers may play a role in both exacerbation and development of asthma (OR greater than 1.5 in both childhood and adult asthma onset) as previously suggested for cleaning agents (29). Our results are also consistent with the hypothesis of sub-optimal treatment of subjects exposed to cleaning agents. A high risk of asthma in cleaners has been previously reported (11, 26–30, 34) especially for private home cleaners (28, 30), and health care cleaners (30). Unfortunately, the majority of cleaners have no training on the toxicity of the products used, especially private home cleaners. Furthermore, if this association is causal, many women, and especially housewives, are potentially at risk from this typically household exposure (30, 35). Further studies are needed to better understand the role of cleaning agents in asthma and to evaluate which products are implicated.

In summary, these results highlight the importance of identifying occupational exposures when investigating asthma in adults. The rate of asthma remission is very low in adults (33). Other research confirms that the best prognosis for occupational asthma is obtained by early removal from causative exposure when asthma is mild (12, 36). Furthermore, it has been shown that asthma severity predicts disability (13) and that changes in working conditions may prevent disability (37). Further studies on risk factors of asthma severity are warranted. Our findings underscore the necessity for clinicians to consider occupational exposures as a risk factor for severe asthma and to recommend preventive measures to limit exposures in patients with both mild and severe asthma. The early identification of occupational risk factors and their reduction at the individual level may be important for the prevention of severe disease (10).

EGEA Cooperative group

Respiratory epidemiology: INSERM U472, Villejuif: I Annesi-Maesano, F Kauffmann (coordinator), MP Oryszczyn; INSERM U408, Paris: M Korobaeff, F Neukirch. Genetics: INSERM EMI 00-06, Evry: F Demenais; INSERM U535, Villejuif: MH Dizier; INSERM U393, Paris: J Feingold; CNG, Evry: M Lathrop. Clinical centers: Grenoble : I Pin, C Pison; Lyon: D Ecochard (deceased), F Gormand, Y Pacheco; Marseille: D Charpin, D Vervloet; Montpellier: J Bousquet; Paris Cochin: A Lockhart, R Matran (now in Lille) ; Paris Necker : E Paty, P Scheinmann; Paris-Trousseau: A Grimfeld. Data management: INSERM ex-U155: J Hochez ; INSERM U472: N Le Moual.