PRISm and the risk for all-cause mortality in relation to sleep disordered breathing: a community-based study

Extract

Lung function and sleep disordered breathing (SDB) are associated with an increased risk for all-cause mortality [1]. Preserved ratio impaired spirometry (PRISm) is characterised by a predicted forced expiratory volume in one second (FEV₁) <80% and FEV₁ to forced vital capacity (FVC) ratio ≥70%. The reported prevalence of PRISm in population-based studies ranges between 7.1% and 20.3% [2]. Older age, smoking history and obesity are related to an increased risk of PRISm [3]. PRISm might be interpreted as a precursor condition of COPD, a view that was recently confirmed by the 2024 Global Initiative for Chronic Obstructive Lung Disease report [4]. In addition, PRISm has emerged as a potential marker for an increased risk of cardiovascular and all-cause mortality [5].

Shareable abstract

Survival probability in the SHHS shows a complex dependency on PRISm status and SDB severity. For a comprehensive understanding of PRISm related health risks, a differentiation of the aetiology and underlying pathological processes is needed. https://bit.ly/40E569D

To the Editor:

Lung function and sleep disordered breathing (SDB) are associated with an increased risk for all-cause mortality [1]. Preserved ratio impaired spirometry (PRISm) is characterised by a predicted forced expiratory volume in one second (FEV₁) <80% and FEV₁ to forced vital capacity (FVC) ratio ≥70%. The reported prevalence of PRISm in population-based studies ranges between 7.1% and 20.3% [2]. Older age, smoking history and obesity are related to an increased risk of PRISm [3]. PRISm might be interpreted as a precursor condition of COPD, a view that was recently confirmed by the 2024 Global Initiative for Chronic Obstructive Lung Disease report [4]. In addition, PRISm has emerged as a potential marker for an increased risk of cardiovascular and all-cause mortality [5]. Obstructive sleep apnoea is the most frequent form of SDB and has a high prevalence in patients with cardiovascular disease. It is also reported as a risk factor for myocardial infarction and heart failure [6]. However, the association between PRISm and SDB remains under-explored in current literature. We investigated the interaction between these two conditions using the Sleep Heart Health Study (SHHS) – a prospective cohort study (>6600 participants aged 40 years and older) – and assessed both spirometry and polysomnography (PSG) [7, 8]. After excluding obstructive airway disease (defined by a FEV₁ to FVC ratio<0.70), participants were dichotomised into Non-PRISm (n=2892) and PRISm (n=523) groups to analyse all-cause mortality based on varying severity grades of SDB. Adjustments for potential confounding variables, including age, sex, body mass index (BMI), smoking status, pack years and cardiovascular comorbidity were implemented.

A total of 3415 participants were analysed for all-cause mortality, with a mean follow-up of 15.9 years. Predicted FEV₁ and FVC were calculated using the Global Lung Function Initiative calculator, incorporating race [9]. Apneoa–hypopnea index (AHI) was scored in the PSGs according to the American Academy of Sleep Medicine recommended rule, i.e. including arousals and according to a 3% oxygen desaturation threshold for defining hypopneas. SDB was classified into four groups depending on AHI: AHI <5.0 h⁻¹ (n=552, no SDB), AHI ≥5.0–<15.0 h⁻¹ (n=1326, mild SDB), AHI ≥15.0–<30.0 h⁻¹ (n=993, moderate SDB) and AHI ≥30.0 h⁻¹ (n=625, severe SDB). Statistical analyses were performed using the R package “survminer”. Differences between groups were assessed using Welch's t-test for normally distributed data, the Wilcoxon rank-sum test for non-normal distributed data, and the Chi-square test for gender distribution. Statistical significance was set at p<0.05. Cox proportional hazards models were used to evaluate the association between SDB severity and all-cause mortality.

The participant selection flow chart and demographics characteristics are depicted in figure 1a and b. For the all-cause mortality analysis see figure 1c.

FIGURE 1.

a) Participant selection flow chart. b) Characterisation of participants and key parameters of spirometry and polysomnography. Smoking burden is reported in pack years for current and former smokers. There were no significant differences between non-PRISm and PRISm participants. Cardiovascular comorbidity was reported in three categories (0, 1 and 2) and defined by an additive score of five criteria, including prior major adverse cardiovascular events (MACE: heart failure, myocardial infarction and stroke) and a history of diabetes or self-reported hypertension before study visit one at baseline. Data are presented as mean±sd or median (interquartile range), unless otherwise stated. c) All-cause mortality for non-PRISm and PRISm stratified by different groups of SDB (no/mild/moderate/severe). The survival curves are depicted without the adjustment for confounding factors. SHHS: Sleep Heart Health Study; GLI: Global Lung Function Initiative; FEV₁: forced expiratory volume in 1 s; FVC: forced vital capacity; PRISm: preserved ratio impaired spirometry; FEV_1pp: FEV₁ % predicted; BMI: body mass index; AHI: apnoea–hypopnea index; ODI: oxygen desaturation index; S_pO2: peripheral oxygen saturation; T₉₀: cumulative time with oxygen saturation below 90% in total sleep time. ^#: missing due to any of FEV₁, FVC, height, gender, age, ethnicity, BMI, smoking status, cigarette pack-years, nadir S_pO2, average systolic blood pressure, average diastolic blood pressure and ESS score; ^¶: ethnicity considered.

We found a prevalence of 12.2% of PRISm in the SHHS, which is consistent with other population-based studies. Univariate analysis showed that compared to non-PRISm participants, those with PRISm were both significantly older and had a higher BMI. For smoking history and cardiovascular comorbidity at baseline we detected no significant differences between subgroups. AHI (15.13 h⁻¹ versus 13.69 h⁻¹, p=0.0023) and oxygen desaturation index (7.65 h⁻¹ versus 6.81 h⁻¹, p=0.0036) as hallmarks of SDB were significantly higher in the PRISm group. Moreover, the percentage of cumulative time with oxygen saturation below 90% in total sleep time as a parameter for nocturnal hypoxemia in SDB was significantly increased in PRISm versus non-PRISm (0.44% versus 0.30%, p=0.006). However, daytime sleepiness, as measured by the Epworth Sleepiness Scale, did not significantly differ between the two subgroups.

In PRISm participants, Cox regression analysis indicated that only severe SDB was significantly associated with all-cause mortality, with a hazard ratio of 2.32 (1.12, 95% CI 4.83, p=0.024) (see figure 1 part c). In non-PRISm participants, survival probability was significantly linked to various SDB severities compared to no SDB: mild SDB had a hazard ratio of 1.48 (1.11, 95% CI 1.98, p=0.008), moderate SDB showed 2.22 (1.66, 95% CI 2.96, p<0.0001), and severe SDB had 2.54 (1.88, 95% CI 3.45, p<0.0001). Notably, after accounting for potential confounders age and high cardiovascular comorbidity (score of ≥2 in the five criteria) were the only factors with significant associations to overall survival in both the entire study population and within each subgroup.

The relevance of PRISm in SDB and its association with all-cause mortality was first examined in this study. PRISm is recognised as a heterogeneous condition [10] and appears to influence the relationship between SDB and all-cause mortality when compared to individuals with normal spirometry (see figure 1c). Unlike other studies that reported an increased risk of all-cause mortality in PRISm [11], our findings in the SHHS did not support this after adjusting for confounding factors. This discrepancy may be primarily due to the shared association of both SDB and PRISm with age, which is a significant driver of all-cause mortality. Such a relation was also detected in the original analysis of SDB and mortality using the SHHS data set by Punjabi et al. [12], where a mortality risk for SDB could only be shown for severe SDB after focusing on an age group of 40–70 years. Coronary artery disease was reported by Punjabi et al. [12] as a major contributor to mortality in severe SBD, particularly in males. Similarly, we found a high cardiovascular comorbidity strongly predicting all-cause mortality in the SHHS. The interplay between respiratory conditions, such as PRISm and SDB, and cardiovascular health is an important area of investigation.

Interestingly, in the whole cohort of 3415 participants PRISm exhibited a marginally significant association with improved survival (p=0.046), primarily driven by the significant survival benefit observed in the moderate SDB subgroup (AHI ≥15.0–<30.0 h⁻¹). Conversely, no significant differences in survival were observed in the other SDB subgroups when comparing PRISm to Non-PRISm. Caution is advised interpreting these findings because data on initiation of and adherence to continuous positive airway pressure therapy were not available. Keeping in mind the general difficulties in demonstrating reduced mortality as a benefit of continuous positive airway pressure therapy in SBD [13] and acknowledging the existing evidence for the effectiveness of PAP therapy in overlap syndromes of obstructive sleep apnoea and obstructive lung disease [14], further investigation into the behaviour of PRISm as a specific endotype in this context would be valuable.

The significance of PRISm aetiology in prognostic assessments is increasingly acknowledged within the scientific community, highlighting the need to distinguish between restrictive and obstructive causes [15]. The SHHS reports a low prevalence of active smokers, with less than 10% currently smoking and nearly 50% identified as never smokers. This contrasts with the COPDGene Study [16], which required a minimum of 10 pack-years of smoking history for participant inclusion. PRISm, as a lung function abnormality, is dynamic; it can decrease, remain stable, or progress to chronic obstructive pulmonary disease. The clinical spectrum of PRISm ranges from benign ventilation-perfusion mismatches due to obesity, which are typically reversible, to more severe lung pathologies that are often irreversible. Both forms of PRISm are associated with increased hypoxemia, underscoring the importance of the underlying mechanisms and potential for reversibility in clinical assessments. In the SHHS, PRISm is characterised more by overweight than by active smoking status or tobacco exposure, separating it from the COPDGene Study. This difference may facilitate beneficial clinical outcomes over time, as suggested by findings from the Copenhagen City Heart Study, which indicated a potential trajectory from PRISm to normal spirometry [17]. Moreover, the Rotterdam Study reported different subsets of PRISm with a remarkable impact on mortality and emphasised the role of PRISm as a restrictive pattern associated with increases in BMI and heart failure [18]. Therefore, it is essential to distinguish subgroups within PRISm, as these variations appear to influence the risk of all-cause mortality related to SDB.

Our study has several strengths including its large, community-based population. Both spirometry and PSG were performed at baseline and the relevance of lung function for the mortality risk of SDB was a predefined outcome parameter in the SHHS. Additionally, in contrast to other cohorts which investigated PRISm, the SHHS is not enriched for lung patients. Our study also has some limitations. The study population was primarily middle-aged and elderly white individuals which raises questions about its representativeness for other ethnic groups and younger individuals. The sleep evaluation was performed only once during the study, therefore potentially underestimating the prevalence of SDB. In addition, follow-up lung function testing was not done in the SHHS, meaning a change in the PRISm constellation could not be assessed. Also, the introduced cardiovascular score is only an approximation of preexisting cardiovascular health and the clinical reality.

In conclusion, survival probability in the SHHS shows a complex dependency on PRISm status and SDB severity, as not all SBD subgroups are identically affected by PRISm. For a comprehensive understanding of the PRISm related health risks a differentiation of the aetiology and underlying pathological processes is needed.