Upper airway microcirculation remodeling in obstructive sleep apnea is not driven by endothelial activation

Abstract

Microcirculation contributes significantly to blood flow resistance, with upper airway microcirculation in obstructive sleep apnea (OSA) affected by endothelial activation, perturbed blood flow shear stress, and snoring-induced tissue vibration. The relevance of these mechanisms on microcirculation response and remodeling remains largely unknown but may influence management decisions in OSA. This study analyzed pharyngeal muscle tissue from non-obese, young adult patients with OSA and chronic heavy snoring. We assessed arteriole morphometry and quantified the expression of endothelial activation markers: 8-isoprostane, vascular cell adhesion molecule-1, E-selectin, vascular endothelial growth factor, endothelin-1, and endothelial cell specific molecule-1. Morphometric analysis of 319 arterioles (mean of 8 vessels per patient) revealed thicker walls in severe OSA compared to mild OSA without lumen reduction, indicating outward hypertrophy, and a positive correlation between the apnea–hypopnea index (a measure of OSA severity) and arteriole wall thickness. However, analysis of 1872 arterioles showed no increase in endothelial activation markers with disease severity, either in the arteriole walls or muscle tissue. This suggests that, in young non-obese adults, severe OSA likely leads to adaptive, mechanically driven microcirculation outward hypertrophy, potentially due to perturbed shear stress, with potential implications for OSA management.

Introduction

Obstructive sleep apnea (OSA) is associated with increased risk of cardiovascular and cerebrovascular disease^1–3. The hallmark of OSA is intermittent hypoxia (IH) during sleep, which promotes increased oxidative stress, leading to endothelial activation and dysfunction and sympathetic activation, with systemic and vascular inflammation⁴. Oxidative stress is the excess production of reactive oxygen species and, in the endothelium, it decreases the availability of nitric oxide (called endothelial dysfunction), activates nuclear factor kappa B (NF-κB) and contributes to increase inflammatory cytokines^5,6. A good marker of oxidative stress is 8-isoprostane, a stable product of the peroxidation of arachidonic acid. Endothelial inflammation increases because NF-κB promotes the transcription of cell-surface adhesion molecules (called endothelial activation), such as E-selectin, P-selectin, vascular cell adhesion molecule-1 (VCAM-1) and intercellular adhesion molecule-1 (ICAM-1)^7,8. The excess of reactive oxygen species also drives another transcription factor, called hypoxia-inducible factor-1 (HIF-1), which induces the transcription of proteins that mediate adaptive responses to hypoxemia, such as vascular endothelial growth factor (VEGF), endothelin-1 (ET-1) and endothelial cell specific molecule-1 (ESM-1, also known as Endocan)^5,6. It is recognized that endothelial dysfunction depends on the severity of OSA^9,10.

Microcirculation is a network of small-diameter blood vessels (less than 100 μm) that transport oxygen and nutrients to the cells and remove carbon dioxide and metabolic waste from the cells^11,12. This network plays a relevant role in the vascular system because small arteries and arterioles provide around 45–50% of the total resistance to blood flow in the body¹³. The key component of microcirculation is the endothelium, which contributes to the local balance between pro and anti-inflammatory mediators and is responsible for the vasodilation balance^12,14. Little is known about the expression of endothelial activation molecules in direct tissue evaluation of the microcirculation in patients with OSA. Furthermore, evidence suggests that IH causes different oxygen swings according to the tissue studied¹⁵ and systemic measures of inflammatory mediators in patients with OSA and obesity are confounded by fat cells already releasing those mediators¹⁶. Previous studies on this issue left some questions unanswered, as they included a very limited number of patients with OSA and did not evaluate the upper airway tissues^17,18.

Apart from IH, OSA may affect microcirculation and cause vessel wall remodeling due to swings in shear stress¹⁹, which is the frictional force of blood flow parallel to the vessel wall. A constant shear stress maintains microcirculation endothelium homeostasis, while perturbed shear stress leads to changes in secretion of vasodilator and vasoconstrictor agents¹². Repetitive forced inspirations against an obstructed upper airway develop shear and wall stresses on intrathoracic blood vessels²⁰. Mimicking OSA in rats increases wall thickness and lumen diameter (called outward hypertrophic remodeling) of the descending thoracic aorta²¹. Because of repetitive pharyngeal pressure swings during sleep, it may be expected that OSA, at some stage and independent from comorbidities, promote remodeling of microcirculation of the upper airway. Not only is the upper airway under the impact of sudden lumen pressure swings in OSA, but also under the impact of vibration from chronic loud snoring, which makes pharyngeal tissues a likely site for microcirculation injury²².

Our objective is to determine whether OSA severity promotes upper airway microcirculation injury in non-obese OSA patients, as indicated by microcirculation remodeling and an increase in endothelial activation markers in the microcirculation walls and surrounding muscle tissue.

Methods

This study was approved by our Institutional Review Board (protocol number 965.007/2015 and 3.121.150/2019) and registered in the Brazilian National Unified Database for Human Research, No. CAAE 36571514.4.0000.0068. The original collection of our specimens was approved by our Institutional Review Board (protocol number 559/2005), for which all patients provided written informed consent. All methods were performed in accordance with the relevant guidelines and regulations.

We analyzed formalin-fixed paraffin-embedded blocks of deep pharyngeal muscle tissues removed with palatine tonsils during surgeries aiming to treat OSA, performed in our institution between the years 2005 and 2006. These tissue samples were used in a previous study by our group²³. Briefly, after the surgeries, the cold dissected tissues had been fixed in 10% buffered formalin for 24 h. Multiple 2 mm-thick cross sections of each tonsil-plus-muscle were made, and after fixation, sections were processed and paraffin-embedded.

Study population

Pharyngeal muscle samples came from 39 patients (age, 19–55 years) who underwent pharyngeal surgeries for OSA. The exclusion criteria were obesity (body mass index, BMI > 30 kg/m²), neuromuscular diseases, retroglossal obstructions, non-controlled or controlled for less than 1 year hypothyroidism, previous oropharyngeal surgeries, Down syndrome, craniofacial deformities and history of phlegmon or peritonsillar abscesses²³.

All patients had been tested with in-lab type I polysomnography within the 3 months prior to surgery. All polysomnograms were performed in the absence of acute upper airway inflammation and hypopneas were scored according to the criteria of that period (> 50% decrease in airflow or a clear decrease in airflow inferior to 50% associated with either an at least 4% oxygen desaturation or an arousal, lasting 10 s or longer)²⁴. All patients presented chronic anti-social loud snoring, excessive daytime sleepiness (Epworth sleepiness scale in all > 10)²⁵ and an apnea–hypopnea index (AHI) > 5 events/h. There were 17 patients with mild OSA (5 ≤ AHI < 15 events/h), 12 with moderate OSA (15 ≤ AHI ≤ 30 events/h) and 10 with severe OSA (AHI > 30 events/h).

Tissue preparation

The paraffin-embedded blocks were cut (2 or 3 blocks per patient) into 5 µm-thick sections and stained with hematoxylin and eosin to evaluate the adequacy of the specimen for the study. We studied sections of muscle fibers adhered to the tonsils, which are muscles from the lateral pharyngeal wall (palatopharyngeus and superior pharyngeal constrictor). Adequate sections had sufficient muscle tissue for analysis and an absence of artifacts such as bleeding and cauterization effects. The selected sections were then processed to identify the immunohistochemical expression of endothelial activation markers and evaluate microcirculation remodeling.

For structural analysis to evaluate microcirculation remodeling, 5 µm-thick sections were stained with Verhoeff-Masson trichrome.

We quantified the following endothelial activation markers: 8-isoprostane, VCAM-1, E-selectin, VEGF, ET-1, and ESM-1. For this, the sections were deparaffinized, and endogenous peroxidase was blocked with a 0.5% hydrogen peroxide in methanol solution for 10 min at room temperature. Pre-treatment of slides for antigen retrieval was performed with citrate (for 8-isoprostane, VEGF, and ET-1 staining), with TRIS–EDTA (for VCAM-1 and E-selectin staining) and with proteinase K (for ESM-1 staining). Sections were incubated overnight in a humid chamber with the following primary antibodies: 8-Isoprostane, polyclonal goat antibody (IgG) at a 1:500 dilution (catalogue IS20, Oxford Biomedical Research, Headington, UK); VCAM-1, monoclonal mouse antibody (IgG1) at a 1:100 dilution (catalogue sc-13160, Santa Cruz, Dallas, TX, USA); E-selectin, monoclonal mouse antibody (IgG2a) at a 1:50 dilution (catalogue sc-137054, Santa Cruz, Dallas, TX, USA); VEGF, monoclonal mouse antibody (IgG1) at a 1:8000 dilution (catalogue sc-7269, Santa Cruz, Dallas, TX, USA); ET-1, polyclonal goat antibody (IgG1) at a 1:300 dilution (catalogue sc-21625, Santa Cruz, Dallas, TX, USA); ESM-1, polyclonal rabbit antibody (IgG) at a 1:6000 dilution (catalogue ab224591, Abcam, Cambridge, UK). The secondary antibodies used were the Vectastain Elite® ABC-HRP kit, goat IgG (catalogue PK-6105, Vector Laboratories Inc, Newark, CA, USA) for 8-isoprostane and the Mouse and Rabbit Specific HRP/DAB IHC detection kit—micro-polymer (catalogue ab236466, Abcam, Cambridge, UK) for all the other markers. The antibodies were visualized with the chromogen 3,3′-diaminobenzidine (Sigma Chemical Co, St Louis, MO, USA). All sections were stained during the same staining session using antibodies coming from the same batch. Negative control sessions were performed by replacing primary antibody with phosphate-buffered saline. The sections were counterstained with Harris hematoxylin.

Image analysis and quantifications

All slides were scanned using a 3DHistec device (3DHistec, Budapest, Hungary), and image analysis was performed with Image Pro-Plus 4.1 software for Windows (Media Cybernetics Silver Spring, MD, USA).

We analyzed 6–9 vessels (arterioles) located within or in close proximity to the pharyngeal muscle bundles, but excluded those near the tonsil tissue. We did not necessarily analyze the same arterioles for all markers. We analyzed the expression of 8-isoprostane, VCAM-1, E-selectin, VEGF, ET-1, ESM-1, and the structural parameters in the arteriole walls at a 20× magnification (Fig. 1). In addition, we analyzed the expression of VEGF, ET-1, and ESM-1 within the surrounding muscle tissue, in an at least 2 mm² total area of transversally cut muscle fibers at a 10× magnification.

Fig. 1.

Illustrative case of immunohistochemical expression (brown staining) of endothelial activation markers in the microcirculation. All images from a male patient, 23 years-old, with an apnea–hypopnea index of 33.4 events/h. Markers: 8-isoprostane (A), vascular cell adhesion molecule 1, VCAM-1 (B), E-selectin (C), vascular endothelial growth factor, VEGF (D), endothelin 1, ET-1 (E), endothelial cell specific molecule 1, ESM-1 (F). Scale bar: 100 µm, 20× magnification. V, vessel.

Two experienced pathologists determined the positive color thresholding for each staining antibody by comparing 6–8 cases per marker with matched non-stained slices. The range of positive labeling generated a marker-specific file that was applied for each marker. Then, measurements were independently done by two examiners blinded to patient’s group (one examiner per marker). The expression of the markers (in the arteriole walls and the skeletal muscle) was calculated as integrated optical density (IOD)²⁶, which is a software generated value representing the product of mean color density (MCD) of the marker (range 0–255, 0 = black and 255 = white) and the sum of positive stained areas (µm²). We divided IOD per positive area to obtain the isolated MCD (from 0 to 255), then obtained the inverse results of MCD using 255 minus MCD (so that the more intensely stained areas would have a greater value) and multiplied this result by the positive area again. This value reflects both the presence and the intensity of the stain, and it was normalized by the external perimeter (for the vessel wall, µm²/µm, Fig. 2) or by the total measured area (for the skeletal muscle, µm²/µm², Fig. 3). For the vessels, we used the average value of all measured arterioles per case.

Fig. 2.

Illustrative case of immunohistochemical expression (brown staining) with Endothelial cell specific molecule-1 (ESM-1) in severe OSA. Microcirculation marked with ESM-1 before (A) and after (B) imaging processing. For each vessel, we calculated the integrated optical density (marked in red in panel B, black arrow) normalized by the outer perimeter (scale bar: 100 µm, 20× magnification).

Fig. 3.

Illustrative case of immunohistochemical expression (brown staining) with Endothelial cell specific molecule-1 (ESM-1) in severe OSA. Pharyngeal muscle marked with ESM-1 before (A) and after (B) imaging processing. Area of interest of the muscle is outlined in green and integrated optical density is marked in red, normalized by the measured area (scale bar: 100 µm, 10× magnification).

We measured the following structural characteristics in the arterioles: diameter (the shortest diameter between two points of the external elastic lamina), external perimeter, vessel area, lumen area, wall area (the difference between vessel and lumen area) and wall thickness (wall area divided by external perimeter, corresponding to mean wall thickness). We used the average value of all measured arterioles per case.

Statistical analysis

The distribution of the study variables was non-parametric, as indicated by the Kolmogorov–Smirnov test. Data were presented as median and interquartile ranges. We used the average values for each marker, per case. For comparisons among OSA severity groups, the Kruskal–Wallis test was used followed by Dunn’s post hoc test (for pairwise comparisons) with Bonferroni correction. For correlations with clinical variables, a Spearman’s correlation test was used. Statistical analysis was performed with the SPSS software version 21 (SPSS Inc.©, Chicago, IL). The level of significance was set at 5% (p < 0.05).

Results

The demographic and clinical characteristics of the 39 patients are shown in Table 1. Our patients were relatively young (mean age, 37.9 years, median age, 37 [28–48] years). The mean AHI for the group was 22.3 events/h (median, 16.4 [9.5–30] events/h) and the mean BMI was 26.0 kg/m² (median, 26.9 [23.9–28.4] kg/m²). There was no significant difference in age and BMI among the subgroups. Eight patients (20.5%) had comorbidities: 5 were smokers, 2 had diabetes and 1 had arterial hypertension, similarly distributed across all OSA severity subgroups.

Table 1.

Demographic, anthropometric and polysomnographic data in OSA subgroups. Data are presented as median and interquartile range.

Characteristic	Mild OSA	Moderate OSA	Severe OSA	P†
Cases, n	17	12	10
Sex (male/female), n	11 / 6	8 / 4	7 / 3	0.96
Age, years	31 (27–43)	40 (33—52)	42 (23—48)	0.23
AHI, events/hour	9.5 (7–10)	18.5 (17.2–24.9)	44.1 (33–59)	> 0.001*
Minimum SpO2, %	87.5 (84.5–92.0)	86.5 (82–90)	75.5 (69–87)	0.01*
BMI, kg/m2	25.8 (21.7–28.0)	27.5 (25.8–28.3)	27.7 (22.8–29.2)	0.25
Smoking, n	2	2	1
Diabetes, n	0	1	1
Hypertension, n	1	0	0

Significant values are in [bold].

AHI, apnea–hypopnea index; SpO₂, oxyhemoglobin saturation; BMI, body mass index.

^†Kruskal–Wallis test.

*p < 0.05.

Microcirculation remodeling

The structural characteristics of the microcirculation of the pharyngeal muscles are shown in Table 2. We measured the morphology of 319 arterioles (mean of 8 vessels per patient). There was a significant difference in wall thickness among the OSA severity groups. Wall thickness was significantly higher in patients with severe OSA compared to mild OSA (p = 0.034, Figs. 4 and 5). There was a positive correlation between AHI and arteriole wall thickness among OSA patients (r = 0.363, p = 0.025, 95% Confidence Interval, CI [0.064, 0.626], Fig. 6). This increase in wall thickness did not correlate with reducing in lumen measurements, characterizing an outward hypertrophy as AHI rises.

Table 2.

Structural characteristics of the microcirculation of the pharyngeal muscles in OSA subgroups.

Measure	Mild OSA	Moderate OSA	Severe OSA	P†
Diameter, µm	35.2 (28.4–51.1)	38.7 (30.8–46.0)	53.3 (36.9–69.7)	0.08
External perimeter, µm	163.8 (136.6–240.8)	192.6 (144.4–235.8)	240.2 (176.4–275.2)	0.14
Wall thickness, µm	4.0 (3.4–6.2)	5.6 (4.0–7.2)	6.8 (5.6–9.7)	0.04*
Total area, µm2	1696.1 (1203.2–5025.2)	2388.9 (1262.0–3759.3)	4174.8 (2113.4–5855.3)	0.17
Lumen area, µm2	830.8 (589.7–2023.1)	879.3 (538.2–1421.0)	1637.3 (612.3–2188.2)	0.44
Wall area, µm2	1030.8 (519.9–2046.2)	1509.6 (699.5–2239.0)	2303.4 (1510.4–4451.4)	0.13

Significant values are in [bold].

Data are presented as median and interquartile range.

Wall thickness in severe OSA > mild OSA (p = 0.034, Dunn’s post hoc test).

^†Kruskal–Wallis test.

*p < 0.05.

Fig. 4.

Pharyngeal muscle microcirculation wall thickness among the OSA severity groups. Dunn’s post hoc test.

Fig. 5.

Illustrative cases of microcirculation wall thickness (white arrows) in pharyngeal muscles in mild (A), moderate (B) and severe (C) OSA. Verhoeef Masson staining. Elastic tissue in black, muscle tissue in red and connective tissue in blue (scale bar: 100 µm, 20× magnification).

Fig. 6.

Correlation between apnea–hypopnea index (AHI) and pharyngeal muscle microcirculation wall thickness in patients with OSA. Spearman’s correlation test.

Endothelial activation markers

We quantified endothelial activation markers in a total of 1872 arterioles, with a mean of 312 vessels per marker. The expression of these markers in the vessel wall and in the surrounding skeletal muscle tissue is presented in Table 3. There was no difference in the expression of any of these markers among OSA severity groups, either in the arterioles or in the muscle of the upper airway. Overall, ESM-1 was the most expressed marker in the vessel walls and in the muscle, whereas 8-isoprostane was almost absent in the vessel walls of all groups.

Table 3.

Expression of endothelial activation markers in the vessel wall (arterioles) and in the skeletal muscle of the pharynx in OSA subgroups.

Endothelial activation marker		Mild OSA		Moderate OSA		Severe OSA		P†
VESSEL WALL (µm2/µm)	8-Isoprostane	0.4	(0.1–2.2)	0.3	(0–1.1)	1.0	(0.1–1.7)	0.47
	VCAM-1	2.5	(1.3–4.6)	0.9	(0.3–1.8)	2.8	(0.7–3.3)	0.09
	E-Selectin	177.2	(112.6–215.7)	219.9	(112.4–308.6)	112.2	(91.1–200.5)	0.36
	VEGF	15.7	(5.0–29.3)	12.9	(3.7–32.2)	19.9	(9.0–26.8)	0.55
	ET-1	117.6	(67.5–146.1)	107.7	(67.0–135.3)	84.0	(45.6–170.6)	0.66
	ESM-1	256.2	(196.7–360.4)	238.3	(135.3–364.4)	217.1	(167.1–358.6)	0.86
MUSCLE (µm2/µm2)	VEGF	6.6	(4.1–7.9)	6.5	(3.2–9.3)	5.5	(4.8–11.5)	0.88
	ET-1	6.6	(3.2–19.9)	8.0	(2.4–14.7)	4.1	(1.4–11.0)	0.44
	ESM-1	31.6	(21.6–39.6)	22.9	(18.0–37.5)	23.7	(18.1–41.9)	0.68

Data are presented as median and interquartile range.

VCAM-1: vascular cell adhesion molecule 1; VEGF: vascular endothelial growth factor; ET-1: endothelin 1; ESM-1: endothelial cell specific molecule 1, also known as Endocan.

Expression in vessel wall normalized by external perimeter (µm²/µm).

Expression in muscle normalized by total measured area (µm²/µm²).

^†Kruskal–Wallis test.

To verify the influence of smoking, hypertension, and diabetes on our findings, we compared OSA patients with (n = 8) and without (n = 31) any of these comorbidities. There was no difference between those groups regarding the expression of endothelial activation markers in the microcirculation wall and in the skeletal muscle, neither was there a difference in the microcirculation morphometry in this analysis (data not shown).

Discussion

To our knowledge, this is the first histological study evaluating the effect of OSA on pharyngeal muscle microcirculation remodeling. Arteriole wall was thicker in severe OSA compared to mild OSA, without lumen reduction. In addition, there was positive correlation between AHI and wall thickness among OSA patients. However, there was no increase in the endothelial activation markers according to OSA severity. This pointed to OSA promoting outward remodeling in upper airway microcirculation without detectable endothelial activation. We speculate that these findings may suggest the role of alternative pathophysiological traits, such as OSA-related perturbed shear stress inducing proliferative outward remodeling in upper airway microcirculation.

Endothelium is the pivotal component of the microcirculation as it regulates local vasodilation^12,14. The classical mechanism relating OSA with microcirculation changes is IH, which triggers local inflammatory pathways through increased oxidative stress⁴. Besides this chemical pathway, OSA may also impact microcirculation through a mechanical pathway, particularly in the upper and lower airway, due to repetitive bursts of negative airway pressures inducing swings in shear stress on local vessels^20,21.

Being exposed to both chemical and mechanical injuries triggered by OSA, pharyngeal surgical specimens seem an ideal site for studying early microcirculation consequences of OSA. Previous studies in humans examined the microcirculation in the forearm subcutaneous¹⁷ and gluteal¹⁸ biopsies in very limited number of very obese OSA patients (mean BMI from 35 to 39 kg/m²). One study demonstrated increased oxidant production (peroxynitrite, the product of nitric oxide and superoxide, a reactive oxygen species) in the microcirculatory endothelium, which decreased after 12 weeks of treatment with continuous positive airway pressure (CPAP)¹⁷. The other study verified upregulation in several inflammatory genes in the microcirculatory endothelium that decreased after 12 weeks of CPAP¹⁸. Our study includes a much larger sample and we did not find any increase in several endothelial activation markers in the microcirculation of patients with OSA. Some factors may explain this difference. First, we only studied non-obese patients, which is a paramount difference from other studies. In obesity, the adipose tissue itself releases inflammatory mediators that already induce endothelial activation and dysfunction¹⁶. For instance, there is evidence that the number of inflammatory markers in the uvula is linked to obesity, rather than to OSA²⁷. Second, different body tissues may present different oxygen swings following IH, thus triggering tissue-specific responses¹⁵. Finally, obstructive events disappear after patients wake up, discontinuing the trigger for endothelial activation. Many of those markers are cleared in a relatively short time. The half-life is around 16 min for 8-isoprostane²⁸, 33 min for VEGF²⁹, 1h for ESM-1³⁰, 2-4h for E-selectin³¹, 4h for VCAM-1³², and 7.5h for ET-1³³. Our specimens were collected during surgeries, hours after the individuals had awoken. This may have allowed transient elevations in endothelial activation markers to dissipate, potentially obscuring acute inflammatory changes. We also acknowledge that, despite our efforts to standardize tissue processing and staining, including the use of positive and negative controls, inherent variability in immunohistochemical staining should be considered when interpreting our results.

We have selected a range of immunohistochemistry markers to investigate various aspects of the cascade of reactions induced by oxidative stress in the endothelium, thereby increasing the likelihood of identifying a marker that reflects OSA severity in the upper airway. Specifically, we evaluated proteins associated with oxidation products (8-isoprostane), endothelial activation (E-selectin and VCAM-1), and responses to hypoxemia (ESM-1, ET-1, and VEGF). Our findings of non-increase in endothelial activation markers in the arterioles following OSA severity remained the same after searching markers through the whole surrounding skeletal muscle. This immunohistochemistry analysis matches the histological observations our group made when we first examined these samples, which had shown absence of inflammatory cell infiltrations within the muscle tissue in OSA subjects²³. Although previous studies have found increased inflammation in the upper airway by analyzing specimens from the mucosa or superficial pharyngeal layers^34–37, our samples were taken from deeper pharyngeal layers, which may be less affected by snoring vibrations but still impacted by pharyngeal pressure swings due to repetitive upper airway obstructions.

We could not find any other study linking microcirculation morphometry and OSA in humans. A very large study relating carotid intima-media thickness (CIMT) and OSA (in a group with a mean age of 68 years) concluded that the only sleep measure associated with an increase in CIMT was not the AHI, but a higher percentage of sleep time with SpO₂ < 90%³⁸. On the other hand, another large study in younger individuals (mean age, 48 years), found a significant positive correlation between AHI and CIMT (r = 0.33; p < 0.001)³⁹. Similarly, our non-obese OSA group (mean age, 37.9 years) showed a correlation between microcirculation wall thickness and AHI (r = 0.36, p = 0.025).

Outward hypertrophy linked to increased level of shear stress is associated with increased blood flow, but not increased blood pressure^40,41. Flow-induced outward remodeling is considered an adaptation, found in exercise training and pregnancy⁴⁰, and also essential for collateral growth following ischemia⁴². This remodeling seems related to proliferation of smooth muscle cells in the vessel wall^42,43. This seems to fit the condition we found in our study, where OSA yields mechanical forces over the microcirculation, without elevating endothelial activation markers. This may represent early microcirculation changes in OSA. Once OSA is left untreated, disease persistence over the years may prompt inward remodeling, particularly when associated with hypertension⁴⁰. Therefore, controlling OSA before inward microcirculation remodeling occurs would seem reasonable to improve cardiovascular outcomes in these patients.

We acknowledge our study has some limitations. Our sample size is somewhat limited, particularly in the severe OSA subgroup, which may have reduced our ability to detect subtle differences in endothelial activation markers. The lack of a control group consisting of non-snoring, non-OSA patients without inflammatory diseases of the pharynx limits the generalizability of our findings. Such a group is particularly difficult to obtain in this type of study, which requires in vivo invasive surgery. The alternative of collecting tissues from cadavers would introduce significant variability in the identification of markers due to the variability in post-mortem intervals. However, future studies could include cadavers to investigate whether the microcirculation remodeling observed in our study is specific to OSA. Additionally, the absence of patients with obesity or advanced age prevents us from extending our conclusions to these populations. We also did not evaluate patients with significant cardiovascular comorbidities or far advanced OSA. Future studies of the microcirculation in these latter groups could provide important information regarding the progression of the disease in the micro vessels.

Conclusion

We found outward arteriole thickening in the pharyngeal muscle of relatively young, non-obese adult patients with severe obstructive sleep apnea, which did not correlate to markers of endothelial dysfunction/activation. This possibly indicates an adaptive microcirculation response caused by perturbed shear stress linked to upper airway pressure shifts during sleep, suggesting the need for treating this condition to prevent the progression of these microcirculation changes.

Author contributions

Kristine Fahl: design of the work; acquisition, analysis, interpretation of data, drafted the work. Thais Mauad, MD: conception and design of the work, analysis, drafted the work. Jôse M. de Brito: acquisition, analysis, interpretation of data. Natalia S. X. Costa: acquisition, analysis of data. Danielle A. S. Dantas: acquisition of data. Heraldo Possolo de Souza: analysis of data. Roney O. Sampaio: analysis, interpretation of data. Luiz U. Sennes: conception and design of the work, analysis, revision of the work. Michel B. Cahali: conception and design of the work, analysis, interpretation of data, drafted the work.

Funding

This study was supported by grant# 2020/09089-8, Sao Paulo Research Foundation (FAPESP). FAPESP had no involvement in study design or manuscript preparation.

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

Michel B. Cahali and Luiz U. Sennes: investors in the Brazilian company Biologix, that markets a home sleep apnea test (not used in this study). Thais Mauad: supported by the Brazilian Research Council (CNPq 304277/2019-3). All the remaining authors declare no conflict of interest.