Factors associated with the 6-minute walk distance in patients with systemic sclerosis

Patient selection

We designed a cross-sectional study and recruited consecutive patients over 18 years old followed in the National Referral Centre for Systemic Sclerosis of Lille from November 2014 to August 2016. They were included in the study if they fulfilled the 2013 ACR/EULAR classification criteria for SSc [31]. There was no exclusion criterion.

Data collection

Data were collected systematically by a physician in a standardized case-report form for all patients referred to our day-patient clinic. All of them underwent a comprehensive evaluation performed within the same day (except for transthoracic echocardiography (TTE) that could be performed up to 6 months before or after the 6MWT).

Patients underwent a non-encouraged 6MWT as recommended [32]. The total 6MWD (in absolute and relative values [33]), modified Borg score, peripheral oxygen saturation (SpO₂), blood pressure (BP) and heart rate (HR) were recorded at the beginning and the end of the test. Variations of (Δ) Borg score, SpO₂, BP and HR were defined as the difference between their final and initial values.

Other data collected included patient and SSc characteristics, clinical parameters (including modified Rodnan skin score (mRSS) and New York Heart Association (NYHA) functional class), biological results (including C-reactive protein (CRP), Nt-pro-BNP and creatinin kinase (CK) levels), TTE (including estimated systolic pulmonary artery pressure (sPAP) and left ventricle ejection fraction (LVEF)), PFT (including FVC, total lung capacity (TLC) and diffusing capacity of the lung for carbon monoxide (DLCO)), composite scores (European Scleroderma Study Group Activity Index (EScSG-AI) [34], Medsger severity score [35], Health Assessment Questionnaire—Disability Index (HAQ-DI) [36]) and treatments (for a complete list of collected data, see Additional file 1).

Statistical analyses

Characteristics of the population were described using the mean ± standard deviation (SD) for quantitative variables, and number (percentage) for qualitative variables.

In order to study the associations between the 6MWD (in absolute value) as the dependent variable and the various explanatory variables, in a first step we conducted univariate analyses using linear regressions. Results were expressed using nonadjusted coefficients (95% confidence interval (CI)) expressed in meters. To better identify relevant explanatory variables, linear regressions were then adjusted on the main potential confounding factors (gender, age, body mass index (BMI), disease duration, smoking history and SSc subtype).

As a second step, we built a multivariate linear regression model using the following strategy. Firstly, we included a priori the six potential confounding variables already described. Secondly, we made a first selection of variables based on univariate analyses (p < 0.10). Thirdly, we ensured that among these variables, those of interest were present (i.e., the complications identified a priori as being able to influence the 6MWD: in particular PAH, ILD, joint and muscle symptoms, heart involvement using the left ventricle ejection fraction). Fourthly, we excluded the variables that mediated the relationship between the 6MWD and organ involvements (e.g., symptoms of dyspnea for PAH or ILD), so as not to attenuate the regression coefficients related to organ involvement variables. We also excluded collinear variables and composite scores. Fifthly, we ensured that the number of variables included in the model was consistent with the number of patients (i.e., close to one variable per 10 patients). Finally, we tested several potential interactions: between PAH or ILD and joint symptoms, muscle symptoms, smoking history or SSc subtype; and between ΔHR, initial HR and chronotropic drugs. None of the interactions tested was significant. Additionally, the HAQ-DI score was included in a separate regression model as a sensitivity analysis. Coefficients (95% CI) were expressed in meters. Regression diagnostics were performed.

All statistical analyses were performed using R software, version 3.2.5 [37]. The threshold for statistical significance was set to p < 0.05.

Patient characteristics

During the 6MWT, patients walked a mean distance of 438.3 ± 108.1 m (81.6 ± 17.8% of the predicted value), decreased their SpO₂ by a mean of 0.58 ± 2.72% and increased their HR by a mean of 10.4 ± 11.2 bpm. The mean Borg score was 0.4 ± 1.1 at the beginning of the test and rose to 2.5 ± 2.2 at the end.

Associations with the 6-minute walk distance: nonadjusted and adjusted analyses

Regarding general patient and SSc characteristics, the 6MWD was associated with gender (p < 10^–3), age (p < 10^–6) and BMI (p < 10^–6) both before and after adjustment. Associations were observed with smoking habits and disease duration in nonadjusted analysis, but did not remain significant after adjustment. Although we found no association with cutaneous subtype, immunological profile or capillaroscopic anomalies in nonadjusted analysis, adjustment on confounding variables revealed a significant association of the 6MWD with the cutaneous subtype (p = 0.008).

Regarding cardiopulmonary parameters, the 6MWD was significantly associated with PAH (p < 10^–6), ILD (p = 0.02) and history of cardiovascular events (venous thrombosis p = 0.001; arterial thrombosis p < 10^–3) in nonadjusted analysis. Association with ILD was no longer significant after adjustment (p = 0.08). We also found significant associations of the 6WMD with NYHA functional class (p < 10^–6), elevated Nt-pro-BNP levels (p < 10^–6), echocardiographic markers of PH (estimated sPAP p < 10^–3; right atrium area p = 0.01) and left-heart dysfunction (LVEF p = 0.03), and PFT results (TLC p = 0.007; DLCO p < 10^–3) in nonadjusted analysis. These associations remained significant after adjustment.

Other cardiopulmonary parameters measured during the 6MWT also displayed significant associations with the total distance, especially initial HR (p < 10^–3) and ΔHR (p < 10^–3), both before and after adjustment. Significant associations with chronotropic drug intake were observed only in nonadjusted analysis.

Regarding extracardiopulmonary parameters, we observed an association with articular involvement (joint symptoms p = 0.02) that became nonsignificant after adjustment on confounding variables. Similarly, there was no association with markers of muscular (muscle symptoms p = 0.08; CK levels p = 0.11) and skin (mRSS p = 0.74; digital ulcers p = 0.30) involvements in nonadjusted analysis.

Regarding global indicators of disease activity and severity, the 6MWD was significantly associated with hemoglobin (p < 10^–3), CRP (p < 10^–3), EScSG-AI (p = 0.001) and Medsger severity scores (p < 10^–3), both before and after adjustment. We also found a markedly large association of the 6MWD with the HAQ-DI score (p < 10^–3) in both analyses.

Associations with the 6-minute walk distance: multivariate models

Parameters that were significantly and independently associated with 6MWD were gender (β = 58.9 (22.6; 95.2) m for men vs women, p = 0.002), age (β = –2.8 (–3.9; –1.7) m per 1-year increment, p < 10^–3), BMI (β = –2.7 (–5.1; –0.2) m per 1-kg/m² increment, p = 0.03), history of arterial thrombosis (β = –39.6 (–77.9; –1.4) m vs no history, p = 0.04), PAH (β = –79.2 (–129.7; –28.8) m vs no PAH, p = 0.002), CRP (β = –4.3 (–6.7; –1.9) m per 1-mg/L increment, p < 10^–3), initial HR (β = –1.1 (–2.2; –0.1) m per 1-bpm increment, p = 0.03) and ΔHR (β = 2.8 (1.8; 3.8) m per 1-bpm increment, p < 10^–6) (Fig. 1). This model accounted for 59% of the variance of the 6MWD.

A sensitivity analysis including the HAQ-DI score revealed its independent association with the 6MWD (β = –77.1 (–101.3; –53.0) m per 1-unit increment, p < 10^–6) (Additional file 2).

Association of the 6MWD with ΔHR

Association between the 6MWD and ΔHR has never been studied previously in SSc. Interestingly, similar findings were made in IPF [38] and PAH [39] patients, where the variation in HR was found to be an important factor associated with the 6MWD as well as a valuable prognostic marker.

Interpretation of this result in the context of the 6MWT is challenging. On the one hand, the lower increase in HR observed in some SSc patients could simply indicate lower exercise intensity due to limited functional capacities or a lack of motivation [40]. On the other hand, it could also reflect an actual impairment of the chronotropic response to exercise. Chronotropic incompetence, defined as the inability to increase HR above 80% of its predicted peak value during a maximal exercise test, is a major cause of exercise intolerance in various cardiopulmonary diseases [40, 41]. It is a common finding in patients with heart diseases (sick sinus syndrome, atrial fibrillation, ischemic heart disease, chronic heart failure), neurological disorders (through autonomic dysfunction) or specific medication intakes (such as β-blockers and nondihydropyridine calcium-channel blockers) [40, 41].

Since the 6MWT is a submaximal exercise test, the diagnosis of chronotropic incompetence cannot be made here with certainty; and since detailed patient comorbidities are not available, the causes for the impaired chronotropic response in our population cannot be investigated fully. However, it is interesting to note that negative chronotropic drug intake was not independently associated with the 6MWD in our patients, making the hypothesis of a medication-induced chronotropic incompetence unlikely. Conversely, SSc is frequently associated with autonomic dysfunction [42] (especially in terms of HR regulation [43]) and with a specific myocardial disease [1], both of which could induce chronotropic incompetence.

Overall, this result suggests that an impaired chronotropic response could be an important factor associated with exercise intolerance in SSc. Interestingly, Someya et al. [44] showed in a previous work that the HR at the end of the 6MWT was higher in SSc patients than in controls to try and compensate for a decreased stroke volume. Thus, it seems reasonable to assume this compensation mechanism would be impaired in SSc patients with chronotropic incompetence, resulting in lower exercise capacities. In any case, this finding warrants further investigations, using maximal incremental exercise tests.

Association of the 6MWD with resting HR

Interestingly, we found that both ΔHR and resting HR are independently associated with the 6MWD. This result suggests that, during SSc, exercise capacities are conditioned not only by the patient’s ability to increase his or her HR during exercise, but also by his or her baseline HR value.

Only two studies have so far tested the association of the 6MWD with resting HR in SSc and they yielded discrepant results [19, 29]. Our observation is supported by similar findings in patients with nonidiopathic PAH and chronic heart failure [39, 45].

Here again, interpretation of this result in the context of our study is difficult. Heart rate is tightly regulated by the autonomous system in order to adapt cardiac output to situations of stress or exercise [41]. Thus, the reduction of physical performance in SSc patients with increased resting HR could be suggestive of advanced autonomic dysfunction (loss of balance between sympathetic and vagal basal tones) and/or severe myocardial disease (compensating mechanism for a decreased resting stroke volume), as implied by our previous result. Alternatively, this could also indicate global deconditioning or performance anxiety.

Either way, further studies are needed to better appreciate the pathophysiological implications of this result.

Association of the 6MWD with PAH but not with ILD

Our multivariate analysis also revealed a significant and independent association of the 6MWD with PAH, but not with ILD. Interestingly, all previous studies that tested the relation between the 6MWD and PH consistently found a shorter walked distance in PH patients [21, 30] or an association with echocardiographic markers of PH [16, 19, 24, 29] (with only one exception [18]). Conversely, data regarding the influence of ILD on the 6MWD are more conflicting, with some studies observing shorter 6MWD in SSc-ILD patients [16, 30] and others reporting no significant difference [20, 21]. However, accurate comparison with previous works is challenging because of heterogeneity in design and population (notably in PH diagnosis, ILD definition and severity).

It is interesting to note that our result is in line with several data pleading for a predominant importance of pulmonary hemodynamic alterations over lung parenchymal involvement. In patients with IPF and PH, the 6MWD is significantly associated with hemodynamic parameters measured during right heart catheterization (RHC), but not with lung volumes assessed by spirometry [46]. In SSc-ILD patients, gas transfer alteration has a stronger prognostic value than lung volume reduction [47, 48].

However, if this result confirms that PAH contributes to impaired exercise capacities during SSc, we also observed in a previous work that the 6MWD correlated poorly with RHC parameters in SSc-PAH patients [49]. Taken together, these results suggest an impact of confounding extracardiopulmonary factors on the walked distance.

Association of the 6MWD with HAQ-DI score

Another interesting finding of our study is the strong association observed between the 6MWD and the HAQ-DI score.

Several teams have already reported similar results, both in univariate and multivariate regressions [21, 23]. Remarkably, in a previous work, Chow et al. [50] showed that the HAQ-DI score was not associated with parameters of cardiopulmonary severity in SSc patients with PAH. This implies that the patient perception of disability and functional limitation may not be entirely explained by the actual severity of the disease. Moreover, the 6MWD has been shown to be associated with scales of depression, anxiety and quality of life in several cardiorespiratory diseases [45, 51].

Overall, these results suggest that if the 6MWD is influenced to some extent by the objective severity of organ involvements, it is also greatly conditioned by the patient’s subjective impression of disability and quality of life impairment.

Association of the 6MWD with CRP

An independent association between CRP levels and the 6WWD was observed in our population. A similar result was also found in previous work by Schoindre et al. [19] both in univariate and multivariate regressions, but not by Deuschle et al. [21]. Interestingly, CRP levels are associated with the EScSG-AI score, Medgser severity score and HAQ-DI score and with poorer prognosis in SSc patients [52]. In line with our prior results, this observation could suggest that the 6MWD also captures the overall disease activity and severity.

No major influence of the musculoskeletal involvement on the 6MWD

Contrary to a common idea [9], it is interesting to note that musculoskeletal involvement is not independently associated with the 6MWD in our population. However, we did not consider which sites were involved in patients with joint or muscle symptoms. Therefore, we could not test the association between the 6MWD and lower limb involvement, which could contribute to explain this negative result. The effect of joint and muscle symptoms has been suggested previously in univariate analyses [21, 25, 26] but has never been tested in multivariate regression. Interestingly, quadricipital strength and joint involvement have been found to correlate both with the 6MWD and with the HAQ-DI score in SSc patients [25, 26, 53], which suggests that this score may also account for the musculoskeletal-induced disability impacting on the 6MWD.

Our study draws strength from a large sample size and an important number of variables collected.

Our study also has some limitations. We did not perform incremental exercise testing to correlate the 6MWD with objective physiological parameters collected during effort. Similarly, we did not study autonomic dysfunction to help explain the results observed with HR variation and ΔHR.

In our multivariate regression, we chose to include only parameters relating to organ involvement (e.g., PAH and ILD) and exclude variables mediating them (e.g., NYHA class, PFT results and TTE data). Indeed, we believed that inserting the latter in our model could mask interesting associations of the 6MWD with organ involvement.