Clinical factors predicting the severity of obstructive sleep apnea in interstitial lung disease

Background and Objectives: Obstructive sleep apnea (OSA) is known to be one of the common complications of interstitial lung disease (ILD). Although it is suspected that the incidence of OSA and the progression of ILD are closely related to each other, the clinical features of ILD with such complications are not fully understood. The aim of this study is to clarify speci�c clinical factors in ILD patients that predict the complication and the severity of OSA. Study Design and Method: ILD patients in our institute were prospectively investigated for the incidence of OSA by polysomnography (PSG). Results: All 33 patients were diagnosed with OSA. Univariate regression analysis showed a lower Diffusing capacity of the lung for carbon monoxide (DLco) and a lower respiratory rate (RR) predicted a higher HI and a higher AI, respectively. Multivariate regression analysis showed a lower DLco and a lower RR also predicted a higher apnea hypopnea index (AHI). Conclusions: A high prevalence of OSA was exhibited in our study of ILD patients. Lacking rapid breathing pattern and impaired diffusing capacity were predictive factors for the severity of OSA. The examination of sleep disorders should be actively considered for these patients.


Introduction
Interstitial lung disease (ILD) is a diffuse lung parenchymal disease that presents restrictive disorders derived from various etiologies 1 .Several complications, including lung cancer, pulmonary hypertension, and gastroesophageal re ux, are known to have impacts on their clinical course and prognosis 2,3 .
The underlying mechanisms of how ILD develops the complication of OSA have not been fully clari ed.However, there has been some evidence that the incidence of OSA affects the progression of ILD.For example, reactive oxygen species are known to be generated by nocturnal intermittent hypoxia of OSA.
The excessive production of reactive oxygen species leads to cellular dysfunction and tissue damage, resulting in the deterioration of ILD 9 .Forced respiratory efforts against air ow obstruction in OSA possibly cause recurrent tractional injury to the periphery of the lung, which is also expected to cause lung damage and aggregate ILD 10 .Moreover, it is known that nocturnal hypoxia is strongly related to the presence of pulmonary hypertension 11,12 .Therefore, early detection of OSA is crucial for preventing the progression of ILD.OSA is sometimes asymptomatic in ILD patients and goes undiagnosed.Herein, our study aimed to clarify the clinical features predicting the incidence and severity of OSA.

Study subjects
Patients who were admitted for the examination of ILD in our institute between January 1st in 2019 and June 31st in 2021 were prospectively evaluated.Eligible patients were recruited regardless of whether they were symptomatic of sleep disorders, while those who needed urgent hospitalization or needed oxygen supply during sleep were excluded.All eligible patients underwent PSG using SOMNOtouch™ RESP (SOMNOmedics).

Patients' characteristics
Age, sex, and obesity, which were evaluated by body mass index (BMI), were investigated.Respiratory rate (RR), spirometry data, blood gas analysis, serum biomarkers such as Krebs von den Lungen-6 (KL-6) and surfactant protein D (SP-D), results of the 6-minute walk test, and the diagnosis were included in the ILD characteristics.The respiratory rate (RR) was determined by the average rate during the 5 minutes just after setting up PSG and before the patient slept.

De nition of the sleep disorder
The diagnosis of OSA was made by more than 5 apnea hypopnea index (AHI).Apnea was de ned as cessation of breathing for more than 10 seconds.Hypopnea was de ned as a more than 30% reduction in air ow with more than 4% desaturation, which lasted more than 10 seconds.An oxygen desaturation index 4% (ODI4%) was de ned as a frequency of desaturation greater than 4% per hour.The percentage of total sleep time spent with SpO2 < 90% (%TST90) was de ned as the percentage of the time of SpO 2 under 90% per sleep time.

Statistical analysis
Data are presented as the mean ± standard deviation for continuous variables.Univariate and multivariate regression analyses were undertaken to explore the potential risk factors for the increase in apnea index (AI), hypopnea index (HI), AHI, and ODI4%.All statistical analyses were carried out using Stata 17.0 (Stata Corp., College Station, TX, USA), and p values of < 0.05 were considered signi cant.

Ethics approval and participant consent
The authors conducted this research in full accordance with the Declaration of Helsinki.The study protocol was approved by the Institutional Review Board of Tokyo Medical and Dental University (approval number; M2018-189) and supported by a grant from Fukuda Foundation for Medical Technology.Informed consents were obtained from all the patients.
a Per 100 increase.

Predictive factors of the HI
The same explanatory factors were included in the univariate regression analysis for the HI (Table 4), which showed that a lower DLco signi cantly predicted a higher HI.Multivariate regression analysis also showed that a lower DLco predicted a higher HI (OR = -0.15,95% CI -0.32 to -0.02, p = 0.08).

Discussion
In our study, all 33 ILD patients were diagnosed with OSA.A lower RR predicted a higher AI, and a lower DLco predicted a higher HI.These parameters were also predictive factors for AHI and ODI4%.These results indicate that ILD patients who lack rapid breathing pattern and who reveal impaired diffusion capacity are at a high risk of severe OSA.For such patients, clinicians should actively consider examinations of sleep disorders.Moreover, two other signi cant ndings were present.
First, the high occurrence of OSA was not signi cantly related to BMI, while the high prevalence of this complication is consistent with previous reports [4][5][6][7][8] .The mean BMI of the patients was 23.9 ± 4.6, and either underweight or normal weight patients were the major population of this study (23 of 33 patients).
It is noteworthy that the group with such a background exhibited 100% prevalence of OSA.Moreover, BMI was not a signi cant predictive factor for the severity of OSA.These results suggest that factors other than obesity are implicated in the occurrence of OSA in ILD patients.Previous reports have presented some hypotheses for the relationship of these diseases, such as craniocaudal traction of the trachea resulting from decreased lung volume, which increases the collapsibility of the upper airway lesion 13,14 .
Ventilatory control system instability also results from intermittent hypoxia, which possibly exacerbates this complication 15,16 .However, no results supporting these hypotheses were obtained from our study.
Second, lacking rapid breathing pattern and impaired diffusion capacity were signi cant predictive factors for the severity of OSA.Advanced ILD typically shows rapid and shallow breathing 17 , which means that RR rises in severe ILD.On the other hand, diffusing capacity is likely to be impaired along with the progression of ILD, resulting in a lower DLco.Namely, these parameters are likely to change in the opposite direction in a clinical course of ILD progression, as an increase in RR and a decrease in DLco.
Therefore, the results of our study indicate that AI is likely to decrease with the progression of ILD, while HI is likely to increase.In other words, hypopnea may be the major constitutive factor of OSA in severe ILD (Figure 1).These results indicate the possibility that minimal pressure may be reasonable as an initial treatment with continuous positive airway pressure (CPAP) therapy for OSA complicated with ILD.
The underlying mechanism of the inverse relationship between RR and AI is unclear.However, progression of ILD leads to a decrease in FVC, which is known to exhibit a rapid and shallow breathing pattern responding to stretch receptor afferents from the periphery of the lung sensing mechanical load of increased lung elastance 17 .Under the condition that the afferent signal is strongly stimulated, transient cessation of respiration may be less likely to occur.Such a mechanism may explain the relationship between RR and AI, but this speculation lacks support from scienti c evidence.
The causal relationship between lower DLco and higher HI suggests that impaired diffusing capacity is likely to exhibit a drop in PaO 2 only by a weak air ow limitation.Namely, craniocaudal traction of the trachea caused by restrictive lung disease or a shallow breathing pattern of ILD may lead to subtle air ow limitation.This may not be enough for a complete cessation of breathing but may lead to a more prominent drop in PaO 2, especially in patients with lower DLco.Moreover, if the diffusing capacity is severely impaired, the baseline PaO 2 is also likely to be lower.Such a condition means they are sitting on the steep slope of the oxygen dissociation curve, which also increases the likelihood of SpO 2 dropping.
Previously, several authors have investigated the relationship between OSA and ILD, while very few of them have clearly distinguished AI and HI.However, considering the preciseness of their de nition, AI and HI are strictly not synonymous parameters.Namely, AI re ects complete air ow cessation, HI re ects a likelihood of desaturation with subtle air ow limitation, and it seems that they correspond to different etiologies.Therefore, we investigated the relationship between ILD and OSA by distinguishing AI and HI, which was a unique point in this study.
There are some limitations of our study.First, this was a single-center study with a small sample size.Second, the study did not examine whether the treatment of ILD affects the severity of OSA.A future randomized control study is needed.
In conclusion, there was a high incidence rate of OSA in patients with ILD.RR and DLco were predictive factors of OSA severity.ILD patients should be actively evaluated for the presence of sleep disorders, especially when they lack rapid breathing pattern and reveal lower DLco.Data availability; Our secured database contains several information used in other research projects and patients' identi ers.Therefore, we cannot share it completely.However, if requested, we will consider providing raw data. Figures AbbreviationsILD interstitial lung disease

Table 3
Predictive factors of the AI increase 2were not included in the explanatory variables in the multivariate regression analysis because their signi cant correlations with RR were seen (Not shown).

Table 4
Predictive factors of the HI increase a Per 100 increase.