On the imperative of including 24-hour and longitudinal multidimensional physiological phenotyping in rodent models of sleep apnea

Comprehensive longitudinal and multidimensional phenotyping in obstructive sleep apnea: Implications for clinical management and research

Obstructive sleep apnea (OSA) is a major global health issue, linked to multimorbidity and increased mortality ^1–3. OSA disrupts cardiovascular, metabolic, and hormonal systems, affecting patients not only during sleep, when apneas and hypopneas occur, but also throughout the entire 24-hour cycle ^4–7. These disturbances manifest acutely and chronically, emphasizing the need for comprehensive multidimensional assessments conducted over multiple days and nights to adequately characterize OSA phenotypes and inform care strategies.

A critical example is OSA-related hypertension, which is characterized primarily by a diastolic, nocturnal (non-dipping) pattern and a morning rise in blood pressure (BP) ⁸. This specific pattern of OSA-related hypertension is poorly depicted by single isolated office BP measurements, and requires inclusion of 24-hour blood pressure monitoring (Ambulatory Blood Pressure Monitoring) and/or repeated home self- measurements of blood pressure in OSA patients ^9,10. Similarly, OSA’s effects on type 2 diabetes (T2DM) incidence and glycaemic control remain unclear, but 24-hour continuous glucose monitoring (CGM) is now used in T2DM to track glycaemic variability, a key predictor of complications, with promising applications for OSA patients ^11–15. Moreover, food intake and physical activity, critical to evaluating OSA’s cardiometabolic impacts, can now be longitudinally assessed by wearable technology ^6,16–19.

Combinations of wearable and nonwearable sensors are now widely used in cardiometabolic health management, enabling continuous, multi-day monitoring of physiological data linked to cardiovascular risk. Such comprehensive data collection surpasses traditional polysomnography, potentially enhancing insight into OSA’s cardiometabolic impacts, improving routine care, and lowering clinical trial costs²⁰. Multimodal telemonitoring, integrating blood pressure, physical activity and CPAP data, has been adopted in targeted interventions to reduce cardiovascular risk in multimorbid OSA patients ²¹. High-resolution, multi-day datasets are expected to become essential in OSA management and research, offering a clearer understanding of the respective roles of OSA and health behaviours in cardiometabolic complications. Appropriate pipelines of data analysis should be implement to address the challenge of interpretation of these datasets ²².

Rodent sleep apnea models must adapt to better capture complex physiological interactions occurring in response to intermittent hypoxia or sleep fragmentation

In the field of sleep apnea, observational studies involving patients with OSA are often criticized due to uncontrolled confounders, such as lifestyle behaviors and comorbidities. The role of sleep apnea, along with intermittent hypoxia (IH) and sleep fragmentation (SF), as independent cardiovascular risk factors remain a topic of debate. To clarify these relationships, establishing robust experimental data is crucial. Animal models, particularly those utilizing rodent exposure to IH, are the most widely used in recent decades to mimic OSA and provide valuable insights for isolating and investigating the specific contributions of IH and SF to cardiovascular and metabolic diseases ^23–29. Systematic reviews and meta-analyses including data from these rodents models have effectively summarized the impact of IH on cardiovascular and metabolic health, including key factors such as blood pressure, heart function, arterial structure and glucose regulation ^30–32. In rodent models, IH lead to elevated systemic blood pressure, cardiac remodelling, and contractile dysfunction, closely resembling conditions observed in humans ^33–36. Additionally, IH exacerbates infarct size following ischemia-reperfusion procedures and significantly impacts arterial health ^31,35–37. There is a dose relationship between severity and duration of IH exposure and these deleterious consequences.

Current methodologies in rodent models often rely on single temporal measurements taken from anesthetized or restrained animals, which can introduce inter-observer variability and significant stress, potentially skewing results. These single-point assessments pose a substantial risk of under- or over-estimating physiological changes, as they fail to capture the dynamic nature of parameters across the circadian cycle and over multiple days and nights. Consequently, important variations that may occur throughout this timeframe of IH exposure—particularly those affecting blood pressure, contractile function, glucose regulation and the impact of ischemia-reperfusion—may not be adequately assessed^38–40. While IH typically occurs during specific periods, its effects can extend beyond these exposure times, leading to acute and long-term consequences that remain poorly understood.

Telemetric recordings offer an excellent opportunity to improve the relevance of OSA rodent models by enabling automated, wireless physiological measurements and extensive data collection over extended periods. This approach reduces experimenter labour and variability while minimizing animal stress and human interactions, resulting in more accurate and naturalistic data ^41–43. These systems effectively monitor various circadian parameters, including sleep, blood pressure, locomotor activity, brain activity, cardiovascular and metabolic metrics ^43–46(Figure Panel A). Once implanted, animals can move freely in their cages while measurements are taken at regular intervals. Current technology allows real-time monitoring during hypoxic sleep episodes or nighttime activity cycles without disturbing the animals. This continuous monitoring offers valuable insights into cardiovascular and metabolic parameters, highlighting the temporal dynamics of how the effects of intermittent hypoxia (IH) and sleep fragmentation (SF) manifest over extended periods in the day-to-day lives of rodent models of sleep apnea. Furthermore, these data are essential for enhancing preclinical models and designing pre-clinical studies for drug development and safety assessments, ensuring a thorough evaluation of potential therapeutic interventions and their efficacy profiles ⁴⁷.

Figure:

Panel A. Recordings of physiological parameters in conscious and freely moving mice exposed in home cage to an intermittent hypoxia paradigm mimicking sleep apnea

This study investigated the effects of IH during the rest period on physiological parameters including core body temperature (CBT), heart rate (HR), blood pressure (BP), and locomotor activity. Male C57BL/6JRj mice were equipped with DSI telemetry implants (ETA-F10, n=5; HD-X10, n=3).

Panel B. Single time point measurement to 24-hour monitoring

This panel illustrates the added value of continuous monitoring compared to single time-point measurements (eye pictogram). Violin plots represent grouped data for CBT, mean BP, HR, and RMSSD (a marker of parasympathetic tone) between Circadian Times 8 to 12, derived from recordings after 20 days of chronic IH exposure. The remaining points on the 24-hour graph display raw, unaveraged data. This combined visualization highlights significant differences in physiological parameters while providing a detailed view of complex circadian responses, which would otherwise be missed by single time-point methods.

Panel C. Longitudinal multi-dimensional measurements depicting progressive acute and long-term responses to chronic IH

Longitudinal data were averaged over 30-minute (CBT, BP, activity) and 60-minute (HR, HRV) intervals, smoothed for graphical clarity, and presented as group averages (n=5 for ETA-F10; n=3 for HD-X10). This analysis captured the progression of acute and long-term physiological responses to IH over time, revealing distinct circadian patterns linked to the duration of IH exposure. Differences between normoxia and IH conditions, including progressive acclimation and recovery effects, are clearly represented.

All experimental procedures were carried out in accordance with European Directive 2010/63/UE. They were reviewed by the Institutional Ethics Committee for Animal Care and Use (Cometh 12) and authorized by the French Ministry of Research (APAFIS# 15156-2018051615245109).

To quantify the individualized physiological responses of mice under IH in free-living conditions, we conducted a comprehensive longitudinal study utilizing a telemetry system with implantable sensors from DSI (Data Sciences International, St Paul, MN, USA, Figure Panel A). Specifically, we utilized ETA-F10 sensors for monitoring locomotor activity, core body temperature (CBT), heart rate (HR), heart rate variability (HRV) (n=5 mice) and HD-X10 sensors for monitoring arterial blood pressure (BP) (n=3 mice). These sensors provided high data acquisition reliability and could be turned on and off as needed, offering a total battery life of ~2 months. This feature enables targeted recordings and supports long-term studies over several months. Mice were housed on a 24-hour light/dark cycle (lights on at 8:00 am = Circadian Time “0”, lights off at 8:00 pm = Circadian Time “12”) and provided with ad libitum food and water. Following recovery from surgery, they underwent sequential cycles of normoxia (4 weeks), IH (4 weeks), and normoxic recovery (1 day). IH exposure was limited to the sleep period (8:00 am = Circadian Time “0” to 4:00 pm = Circadian Time “8”) and consisted of alternating 30-second hypoxia (5% O₂) and normoxia (21% O₂) every minute using nitrogen-air injection cycles. This widely used protocol replicates the physiological impacts of repetitive hypoxia, eliciting arterial oxyhemoglobin levels (75–80%) comparable to severe OSA in humans^24,31. Control mice experienced similar air-air cycles to avoid bias from the noise and gas flow turbulence imposed by the gas changes. Continuous data were collected for ECG at 2000 Hz, BP at 1000 Hz, CBT at 10 Hz, and locomotor activity at 1 Hz. Values were averaged over 30-minute (CBT, activity, BP) and 60-minute (HR, HRV) windows and smoothed for graphical representation. This approach enabled the detection of nuanced circadian variations that are often missed by single time-point measurements, particularly those taken during the light period and outside hypoxic conditions. The longitudinal dataset spanning several weeks uncovered distinct circadian physiological responses to varying IH durations (ranging from 1 day to 3 weeks, Figure Panel C) and highlighted recovery dynamics during the first normoxic day following chronic IH exposure. For example, during the dark phase, mice exposed to room air exhibited increased CBT, locomotor activity, BP, HR, and reduced parasympathetic activity (RMSSD), consistent with normal circadian rhythms. Hypoxic exposure led to significant reductions in CBT, locomotor activity, and RMSSD, coupled with an increase in heart rate (HR). These changes reflect metabolic suppression and autonomic response, consistent with findings reported in other hypoxic models^48–50.

Interestingly, BP showed a progressive increase in circadian amplitude over the course of IH exposure, with a drop during hypoxic episodes and a rise during the dark phase. This progressive adaptation suggests acclimation to chronic IH. CBT and locomotor activity were lowest during the initial IH exposure, intermediate during acclimation (3 weeks), and partially recovered during normoxia, paralleling changes in RMSSD. HR and BP, however, displayed persistent alterations post-IH exposure, highlighting lasting cardiovascular adaptations. Ideally, extended monitoring over several weeks of normoxic recovery would better assess long-term adaptations and recovery dynamics.

These findings emphasize the importance of longitudinal and continuous monitoring. Temperature alterations during IH reflect shifts in metabolic and thermoregulatory processes, while reduced locomotor activity signals fatigue or behavioral adaptations. Non-dipping BP patterns, strongly linked to cardiovascular risk, underscore the importance of these parameters in disease progression. Similarly, reduced HRV signals autonomic imbalance, with increased sympathetic drive and impaired parasympathetic tone contributing to cardiovascular morbidity.

Continuous monitoring reveals trends and patterns that single time-point measurements overlook, offering valuable insights into both acute and long-term responses to IH. This variability is critical for understanding the comprehensive impact of IH and for informing the development and timing of targeted interventions. Exploratory analyses of this multi-signal data unveiled intricate interactions among variables such as hypoxic periods, activity patterns, blood pressure, heart rate (HR) and heart rate variability (HRV). To further enhance our understanding of preclinical sleep apnea models, it will be crucial to collect additional physiological parameters typically measured in clinical settings, including electroencephalogram (EEG), respiratory rate, oxygen saturation and 24-hour glycemia. Integrating these comprehensive datasets will require the development of advanced mathematical models to accurately capture the complex interactions among these variables. Additionally, correlating circadian physiological measurements with molecular data—such as transcriptomic, proteomic and metabolomic information—will provide deeper insights into the pathophysiological responses associated with sleep apnea. This integrative approach will not only advance our understanding of the mechanisms driving sleep apnea but also facilitate comparisons with circadian and longitudinal datasets across various disease models, both in rodents and other organisms ⁵¹. Ultimately, this work may uncover universal patterns and identify key biomarkers for the physiological mechanisms affected by chronic intermittent hypoxia, guiding future research directions in the field.