Ultrasound evaluation of diaphragmatic function in patients with idiopathic pulmonary fibrosis: a retrospective observational study

In interstitial lung disease (ILD), dyspnea and quality of life are major issues in patient care [1‐3]. Several mechanisms can explain dyspnea, but among them, the diaphragmatic function is an important determinant because it is responsible for about 2/3 of the tidal volume at rest [4, 5]. The causes of diaphragmatic dysfunction in ILD may be multiple, namely, mechanical disadvantage due to restriction, chronic hypoxia, chronic inflammation, corticosteroids use, or exercise deconditioning [6‐8].

Different methods are used to assess diaphragmatic function including imaging by fluoroscopy, computed tomography to assess both the structure and function of the diaphragmatic muscle or the measurement of trans-diaphragmatic pressures, stimulation of the phrenic nerve, and electromyography exclusively evaluating neuromuscular function. However, all these methods have interpretation limits, and sometimes technical constraints related to their invasive nature [9‐12]. Thus, diaphragmatic ultrasound has a primordial place in this indication because of its accessibility, its non-invasive nature, and its inter- and intra-observer reproducibility [13‐19].

The measurement of diaphragmatic function by ultrasound has already been studied in many lung diseases such as asthma [19], COPD [20‐22], cystic fibrosis [23], or even in intensive care patients with respiratory failure [24‐26]. Norms of ultrasound diaphragmatic function in healthy subjects have recently been recently published by Boussuges et al. [29, 30]. However, there are only a few studies on ultrasound assessment of diaphragmatic function in ILD [27, 28, 31, 32], and they each present some limits: small population of patients, a very heterogeneous panel of ILD, no correlation with fibrosis extension on CT scan, or no correlation to the muscular mass of the patients. Thus, further studies are needed to assess diaphragm function in ILD, particularly in IPF.

This study aimed to describe the diaphragmatic structure and function by ultrasound in a homogeneous population of IPF and to compare them with healthy subjects.

This study demonstrates that diaphragmatic function assessed by ultrasound in IPF patients showed significant differences compared to healthy controls. First, we found that diaphragmatic amplitude at rest in IPF was significantly increased compared to controls. This finding might be due to diaphragmatic compensation, indicating the need for greater muscular work to maintain the same degree of hematosis. Such muscle work at rest can lead to a decrease in reserve strength and faster dyspnea. Second, a decrease in the maximal diaphragmatic excursion, which is consistent with available evidence-based medicine [27, 28, 31, 32], probably because, during exercise, the diaphragmatic capacities are limited by the lung, so the amplitude is weaker than in healthy subjects. These results are most likely the consequence of an alteration of the thoraco-pulmonary compliance and an impaired elastic recoil related to pulmonary fibrosis that are responsible for a mechanical constraint on the diaphragm. And last, like Santana et al. [27, 28], we found a lower diaphragm thickness in DB in IPF patients. However, the diaphragm thickness was also lower in QB, which is inconsistent with previous studies. These differences could be explained by the fact that other studies have focused on a heterogeneous panel of ILDs [27, 28], very few patients [31], or patients with less impaired lung function [32]. In our study, all patients were evaluated for lung transplantation in the setting of severe disease that could explain muscle deconditioning and diaphragmatic atrophy. Indeed, diaphragmatic atrophy might be induced by a chronic muscle injury related to work overload or even hypoxia, malnutrition, age, systemic inflammation, exercise deconditioning, or corticosteroid use [6‐8, 35]. Interestingly, even though diaphragm thickness was lower in DB and QB in our cohort, there was no difference in thickening fraction between patients and controls, meaning there was no muscular (or intrinsic) diaphragm dysfunction. On the contrary, Santana et al. [27, 28] showed a decreased thickening fraction in patients probably due to more advanced disease.

The diaphragm amplitude in DB was in our study well correlated with clinical features: positive correlation with FVC, DLCO, 6MWT, and negative correlation with dyspnea. Previous studies found also that amplitude was strongly and positively correlated with FVC and DLCO, which are strong predictive factors of mortality in IPF and ILDs [27, 28, 31]. Correlation with dyspnea at exercise (6MWT) and at rest (mMRC scale) could be explained by the increase of diaphragmatic work because of lung stiffness, to maintain the same level of exercise, eliciting early onset breathlessness [3]. The relationship between pulmonary volume and diaphragm excursion is debated and controversial in the literature. Some studies found a linear relationship between inspiratory lung volume and diaphragmatic excursion [36, 37], whereas others found only a weak correlation between lung volume and diaphragm amplitude [16, 38]. We think it is because inspiratory lung volumes are not only determined by diaphragmatic mobility but also by the recruitment of extra diaphragmatic muscles and thoraco-pulmonary compliance [16, 27, 39]. Walterspacher et al. showed a global respiratory muscle strength remains preserved in ILD patients [40] but not diaphragm force, which could explain the correlation between lung volumes and diffusion capacity with diaphragmatic function in our cohort. Indeed, the DLCO is positively correlated with the diaphragm function probably because it reflects the extension of the fibrosis as FVC and lung stiffness [7], but also because impaired diaphragm function may hinder ventilation throughout exercise causing additional mismatch on the ventilation to perfusion ratio.

Finally, like the other works studying diaphragm in IPF or ILD [27, 28, 31], we showed that diaphragmatic amplitude in DB was the ultrasound parameter most correlated with clinical and functional data. Nevertheless, we may notice differences between studies depending on the patient’s disease severity. The thickening fraction was altered compared to controls and correlated to clinical and functional outcomes only for Santana et al. [27, 28] where patients had the worse lung function. A decrease in diaphragm amplitude and a correlation to clinical and functional outcomes were shown in all studies [27, 28, 31], except for Kismet et al. [32], who had less impaired lung function and found no correlation. Thus, diaphragmatic amplitude seems to be reduced and correlated earlier with clinical and functional outcomes of IPF patients than the thickening fraction. In our opinion, the amplitude better reflects extrinsic diaphragmatic dysfunction related to lung fibrosis whereas the thickening fraction better reflects intrinsic (or muscle) diaphragmatic dysfunction secondary to chronic muscle injury. We found that DLCO was positively correlated with the thickening fraction but not FVC. The pathophysiology is multifactorial and may be related to pulmonary hypertension that was systematically present in our cohort of IPF patients but may also involve ventilation/perfusion phenomenon. Since the insertion of the diaphragm pillar is in West’s zone 3, where lung perfusion is greater than ventilation [41], even a small muscle weakness of the diaphragm could lead to decreased recruitment of pulmonary vessels and impaired diffusive lung capacity. Another way of putting it is that a decrease in the thickening fraction increases the ventilation/perfusion mismatch and can change the DLCO.

Regarding lung density assessed by CT, we found a negative correlation between the proportion of voxels greater than -600HU (as a surrogate of lung fibrosis) and the diaphragmatic amplitude in DB. This highlights the relationship between thoraco-pulmonary compliance and a mechanical constraint on the diaphragm in IPF patients. Kismet et al. [32] also analyzed the pulmonary parenchyma with the Total Fibrosis Score (TFS) but found no link with diaphragmatic function, perhaps due to a less precise assessment of fibrosis or patients with less severe lung disease.

This study has several strengths. Our cohort of IPF patients was compared to a large cohort of healthy controls used to define the normal value of diaphragm measurements by ultrasound [30]. For technical reasons, the diaphragm ultrasound of Santana et al. [27, 28], Kismet et al. [32], and Boccatonda et al. [31], was only analyzed on the right side. To our knowledge, this is the first bilateral ultrasound evaluation in ILDs. In addition, correlations were made with clinical and functional parameters, but also measurement of lung density was assessed by CT scan to obtain an objective assessment of the extent of fibrosis lesions.

Several limitations should nevertheless be noted. This is a monocentric study including a small number of IPF patients. Due to the retrospective nature of the study, data to quantify the patient's peripheral muscle mass were not available to assess the extent of the global muscle weakness. However, the cohort of IPF patients in this study had normal blood albumin levels, BMI around 25, and was not strongly deconditioned (median 6MWT at 524 m). We, thus, may hypothesize that patients had no strong sarcopenia. And last, we had no access to quality-of-life data, an essential element to characterize the impact of the disease in IPF patients [1‐3].

Our study shows that diaphragmatic amplitudes in QB and DB are altered in IPF compared to controls probably because of the change of the thoraco-pulmonary compliance responsible for a mechanical constraint on the diaphragm. Moreover, predictors of mortality such as FVC and DLCO, clinical outcomes such as 6MWT and dyspnea, and lung density are well correlated with the diaphragmatic amplitude in DB. Further studies are needed to know if the diaphragmatic amplitude could be a prognostic factor in IPF and is associated with exacerbations, hospitalizations, or mortality.