Article Text

Extended report
A prospective study of the 6 min walk test as a surrogate marker for haemodynamics in two independent cohorts of treatment-naïve systemic sclerosis-associated pulmonary arterial hypertension
  1. Sébastien Sanges1,2,3,4,
  2. David Launay1,2,3,4,
  3. Rennie L Rhee5,
  4. Olivier Sitbon6,7,8,
  5. Éric Hachulla1,2,3,4,
  6. Luc Mouthon9,
  7. Loïc Guillevin9,
  8. Laurence Rottat6,7,8,
  9. David Montani6,7,8,
  10. Pascal De Groote1,10,
  11. Vincent Cottin11,
  12. Pascal Magro12,
  13. Grégoire Prévot13,
  14. Fabrice Bauer14,
  15. Emmanuel Bergot15,
  16. Céline Chabanne16,
  17. Martine Reynaud-Gaubert17,
  18. Sylvie Leroy18,
  19. Matthieu Canuet19,
  20. Olivier Sanchez20,
  21. Christophe Gut-Gobert21,
  22. Claire Dauphin22,
  23. Christophe Pison23,24,
  24. Clément Boissin25,
  25. Gilbert Habib26,
  26. Pierre Clerson27,
  27. François Conesa27,
  28. Jean-François Cordier10,
  29. Steven M Kawut5,
  30. Gerald Simonneau6,7,8,
  31. Marc Humbert6,7,8
  1. 1Université de Lille, UFR Médecine, Lille, France
  2. 2Département de Médecine Interne et Immunologie Clinique, CHRU Lille, Pôle Spécialités Médicales et Gérontologie, Lille Cedex, France
  3. 3Centre National de Référence Maladies Systémiques et Auto-immunes Rares (Sclérodermie Systémique), Lille Cedex, France
  4. 4LIRIC, INSERM UMR 995, EA2686, Lille, France
  5. 5Department of Medicine, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania, USA
  6. 6Faculté de Médecine, Université Paris-Sud, Le Kremlin-Bicêtre, France
  7. 7AP-HP, Service de Pneumologie, DHU Thorax Innovation, Hôpital Bicêtre, Le Kremlin-Bicêtre, France
  8. 8INSERM U999, Centre Chirurgical Marie-Lannelongue, LabEx LERMIT, Le Plessis-Robinson, France
  9. 9Service de Médecine Interne, Centre de Référence des Vascularites Nécrosantes et de la Sclérodermie Systémique, Université Paris Descartes, Hôpital Cochin, Paris, France
  10. 10Pôle Cardio-Vasculaire et Pulmonaire, Clinique de Cardiologie, CHRU de Lille, Lille, France
  11. 11Service de Pneumologie, Hospices Civils de Lyon, Centre de Compétence de l'Hypertension Pulmonaire, Centre de Référence des Maladies Pulmonaires Rares, Lyon, France
  12. 12Service de Pneumologie, Centre Hospitalier Régional Universitaire, Tours, France
  13. 13Pôle des Voies Respiratoires, Hôpital Larrey, Centre Hospitalier Universitaire, Toulouse, France
  14. 14Service de Cardiologie, Hôpital Charles Nicolle, Centre Hospitalier Universitaire, Rouen, France
  15. 15Service de Pneumologie, Centre Hospitalier Universitaire Côte-de-Nacre, Caen, France
  16. 16Service de Chirurgie Thoracique et Cardiovasculaire, Centre Hospitalier Universitaire Pontchaillou, Université de Rennes I, Rennes, France
  17. 17Service de Pneumologie, Centre Hospitalier Universitaire Nord, APHM, Université de la Méditerranée, Marseille, France
  18. 18Service de Pneumologie, Hôpital Pasteur, Centre Hospitalier Universitaire, Université de Nice Sophia Antipolis, Nice, France
  19. 19Service de Pneumologie, Nouvel Hôpital Civil, Centre Hospitalier Universitaire, Strasbourg, France
  20. 20AP-HP, Service de Pneumologie et Soins Intensifs, Université Paris Descartes, Hôpital Européen Georges-Pompidou, INSERM UMR-S 1140, Paris, France
  21. 21Service de Médecine Interne et Pneumologie, Centre Hospitalier Universitaire La Cavale Blanche, Brest, France
  22. 22Service de Cardiologie et Maladies Vasculaires, Hôpital Gabriel Montpied, Centre Hospitalier Universitaire, Clermont-Ferrand, France
  23. 23Clinique Universitaire de Pneumologie, Centre Hospitalier Universitaire, Grenoble, France
  24. 24Université Joseph Fourier, Grenoble, France
  25. 25Service des Maladies Respiratoires, Hôpital Arnaud-de-Villeneuve, Centre Hospitalier Universitaire, Montpellier, France
  26. 26Service de Cardiologie, Hôpital de la Timone, Centre Hospitalier Universitaire, Marseille, France
  27. 27Soladis Clinical Studies, Biostatistics, Roubaix, France
  1. Correspondence to Professor David Launay, Service de Médecine Interne, Hôpital Claude-Huriez, CHRU Lille, Rue Michel Polonovski, LILLE Cedex F-59037, France; david.launay{at}


Objectives Despite the wide use of the 6 min walk distance (6MWD), no study has ever assessed its validity as a surrogate marker for haemodynamics and predictor of outcome in isolated pulmonary arterial hypertension associated with systemic sclerosis (SSc-PAH). We designed this work to address this issue.

Methods Treatment-naïve patients with SSc-PAH were prospectively included from two sources: the French PAH Network (a prospective epidemiological cohort) (n=83) and randomised clinical trials submitted for drug approval (Food and Drug Administration) (n=332). Correlations between absolute values of the 6MWD and haemodynamics at baseline, as well as between variations of 6MWD and haemodynamics during follow-up, were studied in both populations.

Results In the French cohort, baseline cardiac output (CO) (R2=0.19, p=0.001) and New York Heart Association class (R2=0.10, p<0.001) were significantly and independently correlated with baseline 6MWD in multivariate analysis. A significant, independent, but weaker, correlation with CO was also found in the Food and Drug Administration sample (R2=0.04, p<0.001). During follow-up, there was no association between the changes in 6MWD and haemodynamic parameters in patients under PAH-specific treatments.

Conclusions In SSc-PAH, CO independently correlates with 6MWD at baseline, but accounts for a small amount of the variance of 6MWD in both study samples. This suggests that other non-haemodynamic factors could have an impact on the walk distance. Moreover, variations of 6MWD do not reflect changes in haemodynamics among treated patients. Our results suggest that 6MWD is not an accurate surrogate marker for haemodynamic severity, nor an appropriate outcome measure to assess changes in haemodynamics during follow-up in treated SSc-PAH.

  • Systemic Sclerosis
  • Arterial Hypertension
  • Outcomes research
  • Autoimmune Diseases

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Over the last decades, pulmonary involvement has emerged as the leading cause of mortality in systemic sclerosis (SSc).1 Indeed, pulmonary arterial hypertension (PAH) associated with SSc (SSc-PAH) is a frequent complication and is associated with a bad prognosis.2 ,3 This stresses the need for early detection and management of this complication,4 and for accurate assessment of severity and therapeutic efficacy. The gold standard method for evaluating severity and treatment effectiveness in PAH remains invasive right-heart catheterisation (RHC). However, in daily practice, patients are also evaluated by non-invasive measures like New York Heart Association (NYHA) functional staging, 6 min walk distance (6MWD) or echocardiography-derived biomarkers.

The 6MWD is a simple, non-invasive and inexpensive tool, initially developed to study submaximal exercise capacity in chronic obstructive pulmonary disease5 and chronic heart failure.6 Since then, its use has been widened to other conditions7 ,8 such as idiopathic PAH (iPAH), where its correlation with haemodynamic severity9 and prognosis10–12 was formally proven. Because of its usefulness in iPAH, the 6MWD has been applied to SSc-PAH, where it is now widely used as a surrogate marker of haemodynamic severity and a measure of treatment efficacy, in daily practice13 and in clinical trials.14–17

However, the generalisation of assumptions of the 6MWD in iPAH to SSc-PAH must be approached with caution. First, studies on the role of 6MWD as a prognostic factor in SSc have yielded conflicting results.3 ,18 ,19 Second, no work so far has ever quantified the correlation between the 6MWD and the gold standard RHC in SSc-PAH, a crucial requirement for its validation as a surrogate marker of haemodynamics severity and outcome according to the Outcome Measures in Rheumatology (OMERACT) filter.20 Third, SSc is a multiorgan disease, responsible for systemic manifestations (skin fibrosis, interstitial lung disease (ILD), musculoskeletal pain, heart involvement, digital ulcers, anaemia, depression)21 that could influence the results of the 6MWD and act as confounding factors.22–24

Using patients with SSc-PAH and without extensive ILD from two independent treatment-naïve samples, we assessed (1) whether a correlation existed between the 6MWD and the RHC parameters measured at baseline and during follow-up; and (2) if the change in 6MWD between baseline and follow-up accurately reflected the modification in haemodynamics.


Patient cohorts

Patients belonged to two separate independent prospective samples.

The first cohort consisted of 83 incident patients from the French PAH network, prospectively enrolled between January 2006 and November 2009 in 21 pulmonary vascular centres across France. They were included in the study if they fulfilled the following criteria: (A) a diagnosis of precapillary PAH, defined by a mean pulmonary arterial pressure (mPAP) ≥25 mm Hg and a pulmonary capillary wedge pressure ≤15 mm Hg, measured during a resting RHC; (B) a definite diagnosis of SSc, according to the 1980 American College of Rheumatology25 and/or Leroy's criteria26; (C) the absence of extensive ILD (extent of lung parenchymal involvement ≤20% on high resolution CT-scan and/or forced vital capacity ≥70% of the predicted value on pulmonary function tests (PFT))27; (D) no prior treatment with a PAH-specific medication at baseline. Patients were excluded if they had an overlap syndrome with another defined connective tissue disease.

The second cohort consisted of 332 patients with prevalent SSc-PAH from the US Food and Drug Administration (FDA) database, enrolled in 11 international multicentre therapeutic randomised controlled trials (RCTs)28–37 between 1998 and 2006 (see online supplementary table S1). All the patients recruited from this database and included in our study fulfilled the following criteria: (A) a diagnosis of precapillary PAH, proven by a resting RHC; (B) a diagnosis of SSc, determined by the participant's individual clinician; (C) the absence of extensive ILD, as defined in online supplementary table S1; (D) no prior treatment with a PAH-specific medication at baseline.

Data collection

French cohort

Data were first collected at baseline (defined as the time of the first 6MWD).

A global evaluation recorded demographic, clinical and immunological parameters, such as age, sex, body mass index (BMI), disease duration, NYHA functional class, SSc cutaneous subtype according to Leroy's classification,26 musculoskeletal involvement, serological status (anticentromere and antitopoisomerase I antibodies) and ongoing treatments.

Patients also underwent a non-encouraged 6MWD, performed as recommended by the American Thoracic Society,38 with measurement of the total 6MWD in absolute value (with calculation of the relative value as previously described39), modified Borg score, peripheral oxygen saturation and heart rate before and after the test.

The haemodynamic status was assessed within 2 days after the 6MWD, by performing a resting RHC (as previously described40) and measuring systolic (sPAP), diastolic (dPAP) and mean (mPAP) pulmonary arterial pressures, pulmonary capillary wedge pressure, cardiac output (CO) and index (CI), mean right atrial pressure (mRAP), pulmonary vascular resistance (PVR) and total pulmonary resistance, stroke volume (SV), pulmonary arterial pulse pressure (PP), pulmonary arterial capacitance (SV/PP) and right ventricular stroke work index.

Other data collected at baseline included results of PFT performed at rest (measuring forced expiratory volume during the 1st second, forced vital capacity, total lung capacity and diffusion coefficient of carbon monoxide (KCO), expressed as percentages of the predicted values), blood gases, serum brain natriuretic peptide levels and transthoracic echocardiography (TTE).

After baseline assessment, a standard-of-care PAH-specific medication41 could be introduced, either by endothelin receptor antagonists, phosphodiesterase type 5 inhibitors or prostacyclin analogues. A follow-up assessment was conducted at least 3 months after baseline evaluation (with no upper limit), and included a 6MWD and resting RHC, both performed within 2 days.

All data were prospectively recorded, except for data regarding musculoskeletal involvement, which were retrospectively retrieved from medical observations.

FDA cohort

Patient data were obtained from participants in phase III RCTs submitted to the FDA for drug approval (see online supplementary table S1).

Demographic, clinical, functional, haemodynamic and therapeutic data were collected at baseline (defined as time of randomisation) in all of the RCTs in similar conditions to the French cohort, with the exclusion of SSc characteristics, serological tests, detailed PFTs, blood gases, brain natriuretic peptide, TTE and ongoing treatments. In 4 of the 11 trials, the initial RHC was performed several months prior to baseline 6MWD; in the remaining 7 RCTs, the 2 tests were performed within a short period of time.

After baseline evaluation, subjects were randomly allocated to a PAH-specific medication (endothelin receptor antagonists, phosphodiesterase type 5 inhibitors or prostacyclin analogues) or placebo. Among the 11 RCTs, only 5 collected follow-up functional and haemodynamic data, both performed during the 12th week post randomisation.

Statistical analysis

Statistical analyses were performed using the SAS V.9.3 (SAS Institute, Cary, North Carolina, USA) and Stata V.12.1 (StataCorp, College Station, Texas, USA) software.

Baseline and follow-up characteristics of the French and FDA cohorts were compared using χ2 and z tests. Correlations between the baseline 6MWD and each of the collected parameters at baseline were assessed by performing univariate linear regression. As several variables were closely related, we built a multivariate model, based on augmented backward elimination, as previously described42 and detailed in online supplementary methods.

Correlations between follow-up 6MWD and follow-up haemodynamics, and between variation of the 6MWD and variation of the haemodynamic parameters, were investigated by linear regression.


Patient characteristics at baseline evaluation

Overall, 83 consecutive patients were included in the French cohort, and 332 in the FDA cohort (table 1). The proportion of the diffuse cutaneous disease was greater in the FDA cohort. When compared with French patients, FDA patients had a slightly more severe PAH (mean PVR: 10±6 vs 8.2±4.2 Wood units, p<0.001), but similar right ventricular function and NYHA functional class.

Table 1

Baseline characteristics of the 83 patients in the French cohort and the 332 patients in the FDA cohort

Correlation of the 6MWD with other parameters at baseline

As detailed in table 2, univariate analysis using linear regressions showed that the baseline 6MWD significantly correlated with several parameters, including age (p=0.005, r=−0.30), baseline NYHA classes III–IV (p<0.001, r=−0.44) and every baseline haemodynamic parameter, especially CO (p<0.001, r=0.39, R2=0.15) in the French cohort.

Table 2

Correlation between baseline 6MWD and other baseline SSc-PAH parameters in the French and FDA cohorts: univariate analyses

In multivariate analysis (detailed in online supplementary methods), baseline NYHA classes III–IV (vs classes I–II) (p<0.001, R2=0.10) and baseline CO (p=0.001, R2=0.19) were significant independent predictors of baseline 6MWD. These two parameters explained 28% of the variance.

We next assessed the same correlations in the FDA cohort. The univariate analysis performed on this population yielded similar results (table 2), especially for CO (p<0.001, r=0.23, R2=0.05) (figure 1). In multivariate analysis, age (p<0.001, R2=0.05), sex (p=0.032, R2=0.01), BMI (p<0.001, R2=0.05), mPAP (p=0.002, R2=0.03) and CO (p<0.001, R2=0.04) were significant independent predictors of baseline 6MWD. These parameters explained 16% of the variance.

Figure 1

Relations between baseline 6-min walk distance (6MWD) and baseline cardiac output in the French (A) and Food and Drug Administration (FDA) (B) cohorts. 6MWD is expressed in metres (m). Cardiac output is expressed in L/min. In the French and FDA cohorts, cardiac output significantly correlated with the 6MWD (p<0.001, r=0.39, R2=0.15 and p<0.001, r=0.23, R2=0.05, respectively).

Patient characteristics at follow-up evaluation

After baseline evaluation, a PAH-specific treatment was started in 73 (88%) French patients and 225 (68%) patients in the FDA cohort, while the remaining 107 FDA patients (32%) received a placebo.

The follow-up assessment was performed about 27±20 weeks and 12 weeks after baseline, in the French and FDA cohorts, respectively. The variations of the 6MWD and haemodynamic parameters in treated patients are presented in table 3. Data regarding placebo FDA patients are summarised in online supplementary table S2.

Table 3

Follow-up characteristics of the 73 treated patients in the French cohort and the 225 treated patients in the FDA cohort

Correlation of the 6MWD with haemodynamics at follow-up

In the French treated cohort, we found that the follow-up 6MWD significantly correlated with sPAP (p=0.049, r=−0.23) and mRAP (p=0.015, r=−0.31), but not with CO (p=0.401) and PVR (p=0.244) (table 4). The analysis of the treated patients within the FDA cohort showed that the follow-up 6MWD significantly correlated with sPAP (p=0.034, r=−0.23), dPAP (p=0.006, r=−0.30), mRAP (p=0.009, r=−0.28), CO (p<0.001, r=−0.44) and SV (p<0.001, r=−0.47), but not PVR (p=0.086) (table 4). In those randomised to placebo, the follow-up 6MWD significantly correlated with mPAP (p=0.011, r=−0.38), mRAP (p=0.004, r=−0.42), CO (p=0.002, r=0.44), PVR (p=0.029, r=−0.43) and SV (p=0.006, r=0.41) (see online supplementary table S3).

Table 4

Correlation between follow-up 6MWD and follow-up RHC parameters in the French and FDA treated cohorts

Correlation of 6MWD variations with haemodynamic changes during follow-up

We next investigated the associations between changes in the 6MWD (Δ6MWD) and changes in each haemodynamic parameter (ΔRHC) between baseline and follow-up.

In the French and FDA treated cohorts, no significant correlation was found between Δ6MWD and any of the ΔRHC parameters (table 5), including ΔCO (p=0.68). However, in the placebo group of the FDA cohort, we found significant correlations of Δ6MWD with ΔmPAP (p<0.005, r=−0.41) and ΔmRAP (p=0.003, r=−0.44), but not ΔCO (p=0.92) (see online supplementary table S4).

Table 5

Correlation between Δ6MWD and ΔRHC parameters in the French and FDA treated cohorts


To our knowledge, this is the first study of the association between 6MWD and haemodynamics in treatment-naïve SSc-PAH without extensive ILD. Our results suggest that: (1) absolute values of 6MWD are correlated with haemodynamic parameters, especially CO which is independently associated with baseline 6MWD in both cohorts; (2) CO explains only a small fraction of the variance of 6MWD in both cohorts (between 4% and 19%), highlighting the importance of extrahaemodynamic factors in this test; (3) changes in the 6MWD between baseline and follow-up do not accurately reflect modifications of haemodynamics in treated patients.

The associations we found at baseline between 6MWD and haemodynamics in SSc-PAH were suggested by several previous studies. Among patients with SSc with lung involvement, Villalba et al43 showed that a 6MWD <400 m was associated with age, dyspnoea and TTE sPAP >30 mm Hg in univariate analysis. Conversely, Deuschle et al23 found correlations between 6MWD and age, BMI (as in the FDA cohort), NYHA class (as in the French and FDA cohorts), disease activity and PFT values, but not with TTE measurements. All these studies, however, lacked RHC data.

In both samples, only a small fraction of the variance of the 6MWD was accounted for by CO, with an R2-value between 0.04 and 0.19. This suggests the presence of other factors influencing this test in SSc-PAH. Unfortunately, the design of our cohorts did not allow a precise identification of those confounders, besides age and BMI in FDA patients. Musculoskeletal involvement is often cited as a potent confounder in SSc. Schoindre et al22 previously have shown that C-reactive protein and calcinosis were significantly correlated in multivariable regression with 6MWD in SSc. In two studies, one of the main causes of limitation during the 6MWD was lower limb pain.23 ,24 We were able to retrospectively retrieve the frequency of musculoskeletal involvement in the French cohort, but prospectively collected data will be needed to assess its role in the distance walked in SSc-PAH.

We found no correlation between variations in 6MWD and changes in resting haemodynamics after initiation of PAH-specific treatment in both cohorts. In light of these data, 6MWD does not pass the criterion/construct validation of the OMERACT filter.20 It adds concerns about the relevance of 6MWD as an outcome measure in SSc-PAH, beyond the potential inherent value of its increase.

In contrast to treated patients, some correlations existed between variations of 6MWD and modifications of haemodynamics in the FDA placebo group, such as with mRAP and mPAP. To explain the lack of association in treated patients and its presence in untreated patients, one hypothesis could be that PAH-specific treatments also improve the 6MWD by extra-‘pulmonary vascular’ mechanisms. In their work focused on patients with iPAH, Savarese et al44 argued that, since PAH-specific medications are potent vasodilators, they might also enhance exercise capacity by directly increasing the perfusion of skeletal muscles. A recent study by Martin et al45 further highlighted the importance of the peripheral action of phosphodiesterase-5 inhibitors, by showing that tadalafil corrected the exercise-induced muscle ischaemia in patients with Becker muscular dystrophy. Moreover, we have shown that the impact of PAH therapy on outcome is not well explained by haemodynamics in iPAH.46

Similar studies have been conducted to assess the correlation between variations of 6MWD and other end points in iPAH. By performing a meta-analysis of 22 RCTs on aggregate data, Savarese et al44 found a significant correlation between Δ6MWD and ΔPVR, but not ΔCI. Sitbon et al10 showed that the increase of the 6MWD from its initial value did not reflect an improvement of survival. Finally, by conducting a pooled analysis of 10 placebo-controlled trials, we have shown that Δ6MWD explained only a small proportion (22%) of the treatment effect on short-term outcomes and, as such, could not be viewed as a valid surrogate end point for clinical events.47

Our study had several limitations. The FDA cohort was not similar to the French cohort. For example, a majority of FDA patients had a diffuse SSc whereas most French patients had a limited SSc. Selection bias due to strict inclusion criteria for clinical trials compared with a prospective cohort study is possible. Another potential source of selection bias stemmed from the lack of systematic follow-up haemodynamic evaluation in the French cohort, so that included patients might somehow differ from those without follow-up evaluation. As we found close results in both cohorts, the impact of these potential biases was probably limited. The similarity of the findings despite these differences improves the generalisability of our conclusions. In some patients of the FDA cohort, there was a several-month delay between baseline RHC and 6MWD, increasing variability (and weakening the associations) compared with the French cohort, which had simultaneously performed RHC and 6MWD. Finally, although bias is minimised by our statistical approach, unknown latent parameters could also impact 6MWD variations. Our analyses at this stage are not able to identify those parameters.

In conclusion, as it poorly reflects the haemodynamic severity at baseline, we suggest that the 6MWD is not an accurate surrogate marker for resting haemodynamics in SSc-PAH. Moreover, changes in 6MWD are not associated with changes in haemodynamics in treated patients. Altogether, our data suggests 6MWD has a limited utility to assess haemodynamic severity and evolution under treatment in SSc-PAH.


The authors thank Norman Stockbridge, MD, PhD, of the Division of Cardiovascular and Renal Products, Center for Drug Evaluation and Research, US Food and Drug Administration for providing the data for this study. The authors also thank the members of the French Pulmonary Arterial Hypertension Network that participated to this study: Claire Dromer, Marc-Alain Billes, Jean-Benoit Thambo, François Picard, Joel Constans, Virginie Hulot (Bordeaux, France); Irène Frachon, Yannick Jobic, Patricia Brize (Brest, France); Gerard Zalcman, Pascale Maragnes, Eric Saloux, Rémi Sabatier, Thérèse Lognone, Gilles Grollier, Natacha Sobolak (Caen, France); Aimé Amonchot, Bernard Citron, Jean-René Lusson, Isabelle Delevaux, Marc Ruivard, Denis Caillaud, Henri Marson, André Labbé, Benoît Leboeuf, Aurélie Thalamy (Clermont-Ferrand, France); Claudio Rabec, Sabine Berthier, Jean-Christophe Eicher, Caroline Bonnet, Nicolas Favrolt (Dijon, France); Christel Saint Raymond, Jean-Luc Cracowski, Stéphanie Douchin, Marie Jondot (Grenoble, France); Jean-Francois Bervar, Benoit Wallaert, Pascal de Groote, Nicolas Lamblin, Pierre-Yves Hatron, Marie Fertin, François Godart, Aurélie Noullez, Amandine Verhaeghe (Lille, France); Boris Melloni, Estelle Champagne, Francois Vincent, Elisabeth Vidal, Claude Cassat, Philippe Brosset, Stéphanie Dumonteil, Sandrine Nlend (Limoges, France); Sylvie Di Filippo, Sabrine Zeghmar (Lyon); Julie Haentjens-Sitri, Sébastien Renard, Alain Fraisse, Ana Nieves, Berengère Coltey, Carine Gomez (Marseille); Arnaud Bourdin, Pierre Fesler (Montpellier, France); François Chabot, Ari Chaouat, Emmanuel Gomez, Christine Suty-Selton, François Marçon, Anne Tisserant, Anne Guillaumot, Emmanuel Gomez (Nancy, France); Alain Haloun, Delphine Horeau-Langlard, Patrice Guérin, Annick Joly, Régine Valéro, Megguy Morisset (Nantes, France); Pierre Cerboni, Emile Ferrari, Charles-Hugo Marquette, Fernand Macone (Nice, France); Xavier Jais, Azzedine Yaici, Laurence Iserin (Paris, France); Pascal Roblot, Michèle-Laure Adoun, Jean-Claude Meurice, Anne-Claire Simon (Poitiers, France); Roland Jaussaud, Pierre Morand, Sonia Baumard, François Lebargy, Damien Metz, Sophie Tassan-Mangina, Fréderic Torossian, Pierre Nazeyrollas, Patrice Morville (Reims, France); Marcel Laurent, Jean-Marc Schleich, Philippe Delaval, Patrice Jego (Rennes, France); Catherine Viacroze, Charlotte Vallet (Rouen, France); Romain Kessler, Annie Trinh, Irina Enache, Paola Di Marco, Armelle Schuller (Strasbourg, France); Alain Didier, Laurent Tetu, Nathalie Souletie, Daniel Adoue, Philippe Acar, Yves Dulac, Joëlle Pauly (Toulouse, France); Patrice Diot, Elisabeth Diot, Alain Chantepie (Tours, France); Jocelyn Inamo (Fort-de-France, France); Patrice Poubeau (Saint-Pierre, France).


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Handling editor Tore K Kvien

  • SS and DL contributed equally.

  • Contributors Each individual named as an author meets the journal’s criteria for authorship (substantial contributions to the conception or design of the work, or the acquisition, analysis or interpretation of data; drafting the work or revising it critically for important intellectual content; final approval of the version published; agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved).

  • Funding SMK is supported by NIH K24 HL103844.

  • Competing interests None declared.

  • Patient consent All patients from the French cohort gave their informed written consent.

  • Ethics approval The study was approved by the local ethics committees and complied with the requirements of the French ‘Commission Nationale de l'Informatique et des Libertés’. For the FDA cohort, the study was exempt from review by the Institutional Review Board of the University of Pennsylvania due to the use of existing de-identified data, which is not considered part of the definition of human subjects research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement For the French cohort, relevant anonymised patient level data, full data set, technical appendix and statistical code are available upon reasonable request to the authors.