Abstract
Screening is important to determine whether patients with systemic sclerosis (SSc) have pulmonary hypertension because earlier pulmonary hypertension treatment can improve survival in these patients. Although decreased transfer factor of the lung for carbon monoxide (TLCO) is currently considered the best pulmonary function test for screening for pulmonary hypertension in SSc, small series have suggested that partitioning TLCO into membrane conductance (diffusing capacity) for carbon monoxide (DMCO) and alveolar capillary blood volume (VC) through combined measurement of TLCO and transfer factor of the lung for nitric oxide (TLNO) is more effective to identify pulmonary hypertension in SSc patients compared with TLCO alone. Here, the objective was to determine whether combined TLCO–TLNO partitioned with recently refined equations could more accurately detect pulmonary hypertension than TLCO alone in SSc.
For that purpose, 572 unselected consecutive SSc patients were retrospectively recruited in seven French centres.
Pulmonary hypertension was diagnosed with right heart catheterisation in 58 patients. TLCO, TLNO and VC were all lower in SSc patients with pulmonary hypertension than in SSc patients without pulmonary hypertension. The area under the receiver operating characteristic curve for the presence of pulmonary hypertension was equivalent for TLCO (0.82, 95% CI 0.79–0.85) and TLNO (0.80, 95% CI 0.76–0.83), but lower for VC (0.75, 95% CI 0.71–0.78) and DMCO (0.66, 95% CI 0.62–0.70).
Compared with TLCO alone, combined TLCO–TLNO does not add capability to detect pulmonary hypertension in unselected SSc patients.
Abstract
Compared with T LCO alone, combined T LCO–T LNO measurement does not improve detection of PH in unselected SSc patients http://ow.ly/ITHO30eldMk
Introduction
Systemic sclerosis (SSc) is a connective tissue disease characterised by widespread vasculopathy and excessive fibrosis in multiple organs, including the skin, digestive tract, kidneys, heart and lungs [1]. Since the early 1980s, with the advent of angiotensin-converting enzyme inhibitors that have improved the treatment of scleroderma renal crisis, lung diseases have become the leading cause of death in SSc [2]. Interstitial lung disease (ILD) and pulmonary hypertension are the two main forms of pulmonary involvement in SSc [3]. Interstitial abnormalities on high-resolution computed tomography (HRCT) are found in many scleroderma patients, but lesions are generally of limited extent [4], and early deaths attributable to ILD are uncommon in SSc [5, 6]. Although less frequent than ILD, pulmonary hypertension either isolated or associated with ILD has a poor prognosis in SSc [3]. Early diagnosis and prompt therapy of pulmonary hypertension are beneficial from a prognostic standpoint in SSc patients either with or without ILD [7, 8], and recommendations for active screening of pulmonary hypertension in SSc have therefore been made [9].
Several pulmonary function tests are widely used for screening of pulmonary hypertension. Among them, impaired transfer factor of the lung for carbon monoxide (TLCO) has been recognised as the most sensitive marker of pulmonary hypertension in SSc [10, 11]. Nevertheless, this test lacks specificity, at least in part because ILD contributes to altering TLCO [12]. In order to overcome this lack of specificity, a “partition” of TLCO into two transfer components, i.e. membrane conductance (diffusing capacity) for carbon monoxide (DMCO) and alveolar capillary blood volume (VC), has been proposed [13]. A decreased DMCO is interpreted as a “thickening” of the alveolo-capillary membrane or as a decrease in lung area, while a decreased VC is considered to represent reduced blood volume in the ventilated alveoli [14]. Partitioning TLCO can be calculated with the Roughton–Forster equation either by measuring TLCO at several oxygenation levels or by combining measurement of TLCO with transfer factor of the lung for nitric oxide (TLNO) [15–17]. Several studies have evaluated the diagnostic power of DMCO and VC to identify pulmonary hypertension in SSc patients. Yet, to date, only small, single-centre series of SSc patients have been studied and results have been predominantly contradictory [18–20]. In addition, the equations used for the calculation of DMCO and VC with combined TLCO–TLNO in these studies are currently viewed as inappropriate [16, 21].
Therefore, in a large, multicentre sample of unselected patients with SSc who had combined TLCO–TLNO measurements, we aimed to assess the usefulness of the TLCO partition into DMCO and VC for the identification of pulmonary hypertension, with DMCO and VC being calculated according to the latest equations proposed by the European Respiratory Society (ERS) Task Force [16]. We also aimed to study whether or not calculation of DMCO and VC helps to predict overall survival in patients with SSc.
Methods
Patients
Consecutive unselected patients with SSc were retrospectively recruited in seven French medical centres (Besançon, Bordeaux, Paris-Cochin, Lille, Nantes, Poitiers and Rouen). Patients were evaluated in stable conditions. The diagnosis of SSc was made or confirmed by a specialist according to current recommendations [22]. Inclusion date was the date of measurement of combined TLCO–TLNO. Clinical information was retrospectively collected for all participating patients, and included cutaneous disease subset (according to the latest recommendations [22]), disease duration (from the date of first disease manifestation other than Raynaud's phenomenon) and history of digital ulceration. Biochemical and serological blood tests, including haemoglobin, antinuclear antibodies, anticentromere antibodies and antitopoisomerase I antibodies, were also collected. ILD was assessed by the presence of subpleural ground-glass opacities and/or interstitial reticular pattern with or without traction bronchiectasis, and/or honeycomb cysts on HRCT [4, 20].
The study was approved by the Institutional Review Board of the French Language Society for Respiratory Medicine (Société de Pneumologie de Langue Française), which deemed it to be observational and therefore waived the need for informed consent (CEPRO 2015-025).
Assessment of pulmonary hypertension
All patients underwent Doppler echocardiography at the time of pulmonary function tests, for the measurement of left ventricular ejection fraction and tricuspid regurgitant jet velocity. Right heart catheterisation was performed in patients with tricuspid regurgitant jet velocity >3 m·s−1 and in patients with unexplained dyspnoea along with tricuspid regurgitant jet velocity >2.5 m·s−1 [11]. A diagnosis of pulmonary hypertension was retained in patients with a mean pulmonary artery pressure ≥25 mmHg together with a mean pulmonary arterial wedge pressure ≤15 mmHg measured during right heart catheterisation performed at rest [7].
Pulmonary function tests
Forced expiratory flow in 1 s, forced vital capacity, slow vital capacity, functional residual capacity, residual volume and total lung capacity were assessed with standard pulmonary function test equipment in each participating centre according to the ERS guidelines [23, 24]. References values for spirometry and lung volumes were from Quanjer et al. [25].
Combined TLCO–TLNO measurements were performed with a commercially available device following daily calibration (Medisoft, Dinant, Belgium), according to a procedure described elsewhere [14]. Calculations of DMCO and VC were made according to the most recent recommendations (see supplementary material) [16]. Reference values for TLCO, TLNO, VC and DMCO were from Zavorsky et al. [16].
Statistical analyses
Patients were first divided into four groups according to the results of ILD and pulmonary hypertension assessments: SSc patients without pulmonary involvement (Group A), SSc patients with ILD alone (Group B), SSc patients with both ILD and pulmonary hypertension (Group C), and SSc patients with pulmonary hypertension alone (Group D). Data are presented as mean±sd for quantitative variables and as percentages for categorical variables. Comparisons of categorical data between groups were performed using Fisher's exact test. ANOVA was performed to evaluate differences of quantitative variables between the four groups. Scheffe's tests were applied to evaluate pairwise comparisons when a significant difference was found by ANOVA. The Kaplan–Meier method was used to illustrate survival according to the group, with statistically significant differences in event rates assessed via the log-rank test.
Analyses were then performed between SSc patients without pulmonary hypertension (Groups A and B) and SSc patients with pulmonary hypertension (Groups C and D). The area under the receiver operating characteristic (ROC) curve (AUC) was used to assess the ability of TLCO, TLNO, VC, DMCO, TLNO/TLCO and DMCO/VC to discriminate between the presence and absence of pulmonary hypertension. The optimal threshold for the identification of pulmonary hypertension was assessed according to the Youden index of ROC curve analysis in order to maximise both sensitivity and specificity [26]. Comparisons of AUC values were performed with MedCalc Statistical Software version 16.8.4 (MedCalc, Ostend, Belgium). Univariate analysis based on the proportional hazards model was used to examine the relationship between survival and TLCO, TLNO, DMCO, VC and relevant clinical data. Multivariate analysis using Cox's proportional hazards regression model was used to examine the independent effect of each of these variables on survival. Variables with p<0.20 in the univariate analysis were included in the multivariate model.
A p-value of 0.05 was deemed to be statistically significant. All analyses except comparisons of AUC values were performed with SAS version 9.4 (SAS Institute, Cary, NC, USA).
Results
Patient characteristics
We identified a total of 572 SSc patients who had combined TLCO–TLNO measurements between November 2006 and April 2016 in the seven participating centres. The main characteristics of the four groups of patients, divided according to the results of pulmonary hypertension and of ILD assessments, are given in table 1.
Main characteristics at inclusion of patients with systemic sclerosis (SSc), divided into four groups: SSc patients without any pulmonary involvement, i.e. neither interstitial lung disease (ILD) nor pulmonary hypertension (Group A), SSc patients with ILD but without pulmonary hypertension (Group B), SSc patients with both ILD and pulmonary hypertension (Group C), and SSc patients without ILD but with pulmonary hypertension (Group D)
Pulmonary function tests
TLCO and TLNO were the lowest in SSc patients with pulmonary hypertension (Groups C and D), and were also lower in SSc patients with ILD (Group B) than in SSc patients without any pulmonary involvement (Group A) (table 2). In SSc patients with pulmonary hypertension (Groups C and D), VC was significantly lower than in patients without pulmonary hypertension (Groups A and B) (table 2). SSc patients with ILD (either with or without pulmonary hypertension; Groups B and C) had lower DMCO than patients without ILD (Groups A and D). It has to be noted that a total of 13 patients had a negative value of DMCO (these negative values were not taken into account in the current analysis). In addition, some patients had very high DMCO (a total of 11 patients had DMCO 200–300% predicted and nine patients had DMCO >300% predicted). These very high values were kept in the current analysis.
Pulmonary function tests of the studied population at inclusion, divided into four groups: systemic sclerosis (SSc) patients without any pulmonary involvement, i.e. neither interstitial lung disease (ILD) nor pulmonary hypertension (Group A), SSc patients with ILD but without pulmonary hypertension (Group B), SSc patients with both ILD and pulmonary hypertension (Group C), and SSc patients without ILD but with pulmonary hypertension (Group D)
Regarding TLNO/TLCO and DMCO/VC ratios, the only significant difference found was for DMCO/VC between SSc patients without pulmonary hypertension (Groups A and B) and SSc patients with pulmonary hypertension (Group D).
There was a strong correlation between TLCO and VC (supplementary figure S1). We also found a correlation between TLNO and TLCO, and, although weaker, between TLNO and DMCO. There was a strong nonlinear relationship between the TLNO/TLCO and the DMCO/VC ratios. However, we were unable to find any significant correlation between DMCO and VC.
Ability of pulmonary function tests to detect pulmonary hypertension
Analysis of AUC values showed that TLCO (% pred) had the highest value to detect pulmonary hypertension in our group of 572 SSc patients (figure 1 and table 3). TLNO had a slightly, but nonsignificantly lower AUC value. By contrast, VC and DMCO had a significantly lower diagnostic value than both TLCO and TLNO (figure 1 and table 3). The TLNO/TLCO and the DMCO/VC ratios had the worst diagnostic values for pulmonary hypertension (figure 1 and table 3).
Receiver operating characteristic (ROC) curves of transfer factor of the lung for carbon monoxide (TLCO, % pred), transfer factor of the lung for nitric oxide (TLNO), membrane conductance (diffusing capacity) for carbon monoxide (DMCO), alveolar capillary blood volume (VC), TLNO/TLCO ratio and DMCO/VC ratio for discrimination of pulmonary hypertension (n=58) in comparison with the absence of pulmonary hypertension (n=514) in patients with systemic sclerosis. Area under the ROC curve (AUC) values are also given.
Performance of the tested variables with different thresholds for the probability of pulmonary hypertension
The sensitivities and specificities of all parameters for the detection of pulmonary hypertension are given in table 3, as well as the positive and negative predictive values for different thresholds. For the optimal thresholds determined with the Youden methods, the negative predictive values of TLCO (% pred) and TLNO (% pred) were very high, indicating that a TLCO and/or a TLNO above these thresholds virtually excludes pulmonary hypertension. Nevertheless, the low positive predictive values indicate that “abnormal” TLCO or TLNO were often present in the absence of pulmonary hypertension.
Follow-up
The median follow-up time for the entire SSc cohort (n=572) was 24 months. The overall survival of the cohort is shown in figure 2. Patients with pulmonary hypertension (Groups C and D) had a significantly poorer survival than those without pulmonary hypertension (Groups A and B) (figure 2). Of note, also in SSc patients without pulmonary hypertension, the presence of ILD did not affect survival (figure 2).
Kaplan–Meier estimates of overall survival in patients with systemic sclerosis (SSc) but with no pulmonary involvement, i.e. neither interstitial lung disease (ILD) nor pulmonary hypertension (Group A), SSc patients with ILD but without pulmonary hypertension (Group B), SSc patients with both ILD and pulmonary hypertension (Group D), and SSc patients without ILD but with pulmonary hypertension (Group D). T0: date of inclusion, i.e. date of measurement of combined transfer factor of the lung for carbon monoxide and transfer factor of the lung for nitric oxide (TLCO–TLNO).
Univariate and multivariate analyses of factors associated with survival are presented in table 4. In univariate analysis, TLCO, TLNO, VC and DMCO were all associated with survival. In multivariate analysis, only TLCO, forced vital capacity, age at onset and pulmonary hypertension were associated with survival, with TLCO having the highest hazard ratio.
Univariate and multivariate analyses of selected biological, clinical and functional factors associated with survival
Discussion
The relevance of TLNO measurement and TLCO partitioning into DMCO and VC for the identification of pulmonary hypertension in SSc is debated, with sparse and conflicting results based on small, single-centre series of patients [18–20]. Here, we analysed a large, multicentre series of unselected patients with well-defined SSc. Although we used recently refined equations for the calculation of VC and DMCO, we found that TLNO was equivalent to TLCO, while DMCO and VC had a lower capability than TLCO to detect pulmonary hypertension. In addition, we found that in SSc patients, TLCO was a stronger predictor of overall mortality than TLNO, VC or DMCO.
TLCO has emerged as one of the most sensitive and specific routine pulmonary function tests that have been evaluated for identifying pulmonary hypertension in patients with SSc [10, 27, 28]. This has been demonstrated in selected SSc patients: in SSc patients with TLCO <60% predicted and disease duration >3 years, TLCO contributes to predicting pulmonary hypertension through the DETECT algorithm [29]. This has also been shown in unselected SSc patients, in whom TLCO contributes to predicting mean pulmonary artery pressure [30]. The analysis of our large series of unselected SSc patients confirms that TLCO is associated with the presence of pulmonary hypertension. A recent review recalled that an abnormally low TLNO, defined by a value lower than the lower limit of normal (below the 2.5th percentile (LLN2.5%)), is more sensitive than an abnormally low TLCO to identify the presence of “cardiopulmonary diseases” [31]. We also found that a TLNO <LLN2.5% was more sensitive than a TLCO <LLN2.5% for the identification of pulmonary hypertension (table 3), but at the price of much lower specificity for TLNO than for TLCO.
Our study confirms that although the negative predictive value of TLCO for the presence of pulmonary hypertension is very high, the positive predictive value is poor, reflecting the fact that many patients with “low” TLCO do not have pulmonary hypertension. This is largely related to the presence of ILD, a frequent complication of SSc, which is also associated with a low TLCO [12, 32].
Although the interpretation of VC and DMCO calculated with the Roughton–Forster model remains controversial [33], these physiological parameters are supposed to discriminate between vascular and alveolo-capillary membrane alterations. It was recently reported that VC is decreased in patients with pulmonary hypertension without SSc [34]. In accordance with this finding, we report that VC was lowest when pulmonary hypertension was present in SSc patients. Nevertheless, in our group of SSc patients, we also found that VC was lower in patients with ILD without pulmonary hypertension than in patients without any pulmonary involvement. Although SSc patients with ILD had preserved lung volumes and normal estimated systolic pulmonary artery pressure, our results suggest that they may have either pulmonary vascular destruction and/or pulmonary vasculopathy.
Several studies have reported a reduction in DMCO in pulmonary vascular diseases, either associated [18] or not [35] with SSc. Although thickening of the alveolo-capillary membrane due to remodelling of the pulmonary vasculature cannot be excluded, this has been previously interpreted as a loss of functional gas exchange area that accompanies the loss of VC, a hypothesis supported by a strong correlation between VC and DMCO [34]. Here, we used very recently refined equations, and we did not find any significant correlation between VC and DMCO [21]. In addition, we found that DMCO was significantly lower in patients with ILD (independently of the presence of pulmonary hypertension) than in patients without. This therefore reinforces the view that DMCO provides information regarding a “thickening” of the alveolo-capillary membrane or a decrease in lung area.
An increase in TLNO/TLCO and DMCO/VC ratios has been found in patients with pulmonary microvascular disease [36]. Although there was a tendency towards such a finding in our series of SSc patients, we found only slight differences in terms of TLNO/TLCO and DMCO/VC ratios between patients with pulmonary hypertension and those without. Regarding the DMCO/VC ratio, the absence of a significant difference between groups is likely due to the very high standard deviation, as DMCO was very high in some patients.
In agreement with previous reports, we also found here that survival of SSc patients with ILD but without pulmonary hypertension was similar to that of patients without any pulmonary involvement [6, 37]. Nevertheless, SSc patients with pulmonary hypertension either with or without ILD had a similar poor survival. In other words, survival was linked to pulmonary hypertension rather to ILD in SSc. As prompt therapy of pulmonary hypertension in SSc has been shown to be beneficial in patients without as well as with ILD [7, 8], this reinforces the pertinence of finding a test that can identify pulmonary hypertension independently of the presence of ILD in SSc patients. Our results demonstrate that TLCO remains better than VC and DMCO calculated with the modified one-step Roughton–Forster technique for that purpose.
Strengths of this study include the real-life nature of the patient series and the multicentre assessment. Our cohort represents the largest series of patients with SSc where diffusion impairments have been assessed by combined TLCO and TLNO measurements. The group of unselected SSc patients analysed here is likely to be representative of SSc patients followed in most referral centres because in the seven participating centres, each patient with SSc has an annually scheduled follow-up that comprises echocardiography and pulmonary function tests. 41% of our patients had some degree of ILD; this percentage is similar to that reported in previous SSc cohorts [38] and ILD was more than twice as common among patients with diffuse cutaneous subtype than among those with limited subtype. The prevalence of pulmonary hypertension (10.1%) found here is also close to that reported in previous series [3, 11, 39, 40].
Limitations of this study include the retrospective analysis and the observational nature of our cohort. We were therefore unable to identify in each of the seven participating centres the number of SSc patients who were not included in the current study because of missing data on echocardiography and/or right heart catheterisation and/or HRCT and/or pulmonary function tests. There are also limitations in terms of the assessments that were directed by clinical practice and did not follow a predefined study protocol. Some important complications were not ascertained (e.g. gastrointestinal tract involvement) and some others were impossible to collect in some centres (e.g. Rodnan score, which assesses skin thickness in the diffuse cutaneous subtype of SSc). We were unable to collect haemodynamic data for all patients who underwent right heart catheterisation; these data were therefore omitted and only the item “pre-capillary pulmonary hypertension according to right heart catheterisation” was collected in all patient files. Nevertheless, echocardiographic estimation of systolic pulmonary artery pressure was systematically reviewed and collected.
In conclusion, our study provides evidence that the lower TLCO in SSc patients with pulmonary hypertension, compared with SSc patients without pulmonary hypertension, is attributable to the lower VC in those with pulmonary hypertension. Nevertheless, we failed to demonstrate any clinical usefulness of the TLCO partition into DMCO and VC for the identification of pulmonary hypertension in patients with SSc. Our results are at variance with some results based on small and/or selected populations, thus highlighting the importance of cohort representativeness for the applicability of study results to a general population of SSc patients.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-01008-2017_Supplement
Footnotes
Conflict of interest: None declared.
This article has supplementary material available from erj.ersjournals.com
- Received May 17, 2017.
- Accepted July 20, 2017.
- Copyright ©ERS 2017
References
