Airflow Characteristics of Empty Nose Syndrome: Insights From Patient‐Specific Computational Modeling of Airflow Distribution Patterns and Nasal Air‐Conditioning

ABSTRACT

Background

Computational fluid dynamics (CFD) is increasingly used to investigate the complex pathophysiology of empty nose syndrome (ENS). Although previous CFD studies reported middle‐meatus–dominant airflow and reduced air–mucosal interaction in ENS, whether airflow distribution alone explains symptoms or impaired nasal physiology remains unclear.

Methods

Patient‐specific CFD simulations were performed in 14 ENS patients using nasal cavity models reconstructed from computed tomography. Airflow distribution, nasal resistance, air temperature, relative humidity, and surface area‐to‐volume ratio (SAVR) were analyzed. Ten non‐ENS subjects with nasal septal deviation (NSD) without prior nasal surgery and five normal subjects were analyzed for comparison.

Results

Airflow distributions in ENS were heterogeneous, consisting of middle‐meatus‐dominant (9/14), inferior‐meatus–dominant (4/14), and mixed (1/14) patterns. ENS6Q scores tended to be higher in the inferior‐meatus–dominant group, whereas other variables did not differ between patterns. NSD subjects showed similarly diverse distributions (3/10, 3/10, and 4/10, respectively), indicating that skewed airflow distribution is not unique to ENS, although middle‐meatus–dominant airflow was more frequent in ENS. All normal subjects exhibited mixed distributions. ENS subjects showed lower and more variable choanal relative humidity than normal subjects, and relative humidity correlated positively with SAVR in ENS and NSD, suggesting that nasal air‐conditioning may be related to residual mucosal surface geometry. No parameter differed significantly between ENS and NSD or correlated with ENS6Q scores.

Conclusion

ENS is not characterized by a single airflow pattern but represents a heterogeneous spectrum of airflow characteristics, with similar phenotypes also occurring in non‐ENS subjects. Airflow distribution pattern alone was insufficient to explain ENS symptoms, and none of the evaluated global CFD parameters correlated significantly with ENS6Q scores, although the small sample precludes firm exclusion of clinically relevant associations. Preservation of mucosal surface area may nevertheless play an important role in nasal air‐conditioning. These findings support a physiology‐based and patient‐specific understanding of ENS.

Using patient‐specific computational fluid dynamics in 14 patients with empty nose syndrome (ENS), 10 subjects with nasal septal deviation, and five normal subjects, we found that ENS was not characterized by a single airflow pattern but by a heterogeneous spectrum of phenotypes that also occurred in non‐ENS subjects. ENS subjects showed reduced choanal humidity and temperature compared with normal subjects, and humidity correlated positively with the mucosal surface area‐to‐volume ratio, whereas none of the CFD‐derived parameters correlated with ENS6Q symptom scores. These findings indicate that airflow distribution alone is insufficient to explain ENS symptoms and support a physiology‐based, patient‐specific understanding of the condition.

1. Introduction

Empty nose syndrome (ENS) is an unusual and debilitating complication of nasal surgery, most often inferior turbinate volume reduction procedures and, less frequently, other turbinate or septal procedures. ENS presents a clinical paradox: despite a widely patent nasal cavity, patients experience severe nasal obstruction, dryness, crusting, and impaired perception of airflow [1, 2, 3]. The pathophysiology of ENS remains unresolved and is likely multifactorial; proposed contributors include altered aerodynamics, impaired air‐conditioning, mucosal and trigeminal sensory dysfunction, and psychological, functional, and central perceptual mechanisms, whose relative contributions remain uncertain [4].

The aerodynamic and air‐conditioning components of ENS pathophysiology explain why computational fluid dynamics (CFD)—which enables quantitative, patient‐specific evaluation of nasal airflow and air‐conditioning—has been increasingly adopted to investigate ENS [5, 6, 7]. Early CFD studies demonstrated altered airflow characteristics in ENS compared with healthy controls [8, 9]. Subsequent studies reported preferential airflow distribution toward the middle meatus and reduced wall shear stress along the inferior turbinate region, suggesting potential links between altered aerodynamics and sensory dysfunction [10, 11]. CFD analyses of turbinate surgery further showed that surgical modification can substantially alter airflow distribution and air‐conditioning performance [12, 13, 14], whereas recent virtual turbinectomy simulations suggested that turbinate removal does not inevitably reproduce ENS‐like aerodynamics [15].

Despite these advances, several gaps remain. Most CFD investigations have focused primarily on airflow distribution, whereas fewer studies have simultaneously evaluated airflow, air‐conditioning performance, and anatomical factors related to residual mucosal surface area. Furthermore, skewed airflow distribution patterns similar to those described in ENS may also occur in non‐ENS nasal conditions, raising the question of whether airflow distribution pattern alone can distinguish ENS pathophysiology. Because turbinate geometry governs airflow–surface interaction, it is plausible to consider that some other anatomical parameter such as the surface area‐to‐volume ratio (SAVR) may play a role in nasal heating and humidification. This study therefore aimed to conduct a more comprehensive evaluation of airflow characteristics in ENS by investigating the airflow distribution pattern along the nasal airway, nasal resistance, air‐conditioning performance, and SAVR using patient‐specific CFD analysis, additionally examining airflow characteristics in a non‐ENS cohort with nasal septal deviation (NSD) and normal subjects for comparison. By integrating aerodynamic, physiological, and anatomical perspectives, this study sought to gain better insight into the pathophysiological underpinnings of ENS.

2. Materials and Methods

2.1. Patient Selection

Fourteen patients (ENS 1–ENS 14) with a clinical diagnosis of ENS were enrolled in this study. Diagnosis was established by two experienced otolaryngologists based on postoperative history, characteristic symptoms, and clinical examination, together with completion of the ENS6Q questionnaire [16, 17] in accordance with current clinical practice. Patients with chronic rhinosinusitis, nasal polyposis, neoplastic disease, or allergic rhinitis were excluded.

The cohort consisted of 12 male and two female patients, with a mean age of 42.1 years (range, 28–59 years). ENS6Q scores ranged from 14 to 27 (mean ± SD, 19.4 ± 4.1), indicating moderate to severe symptom burden. Symptoms were bilateral in 8 patients and unilateral in six patients (left, n = 2; right, n = 4).

All patients had a history of prior inferior turbinate reduction, including bilateral inferior turbinate reduction in 10 patients and unilateral inferior turbinate reduction in four patients (left, n = 2; right, n = 2). Several patients had undergone additional procedures, including septoplasty, rhinoplasty, or endoscopic sinus surgery. The side of symptoms corresponded to the side of prior turbinate surgery.

For comparison, ten patients (NSD 1–NSD 10) with NSD who reported nasal obstruction but did not exhibit ENS symptoms were included. These subjects were scheduled to undergo septoplasty and had no prior history of nasal surgery. The NSD cohort consisted of five male and five female subjects with a mean age of 32.1 years.

Additionally, five normal subjects (Normal 1–Normal 5) were included as a reference cohort. These subjects had no history of nasal surgery and no clinical or radiologic evidence of sinonasal inflammatory or neoplastic disease. The normal cohort consisted of two male and three female subjects with a mean age of 45.8 years.

This study was approved by the Institutional Review Board of Asan Medical Center (2020‐0118). All data were anonymized, and informed consent was waived.

2.2. Construction of Nasal Cavity Models

Nasal cavity models were reconstructed from computed tomography (CT) scans of the osteo‐meatal unit (slice thickness, 0.625–0.75 mm) obtained from the study subjects. Segmentation was performed using Mimics v23.0 (Materialize, Leuven, Belgium), with external facial contours incorporated to improve the realism of airflow simulation through the nostrils [18, 19].

2.3. CFD Methodology

Simulations were performed using ANSYS Fluent R23.1. Turbulent airflow was modeled using the shear stress transport (SST) k–ω model with low‐Reynolds‐number correction. Second‐order upwind discretization schemes were applied for spatial terms. In addition to airflow, transport equations for heat and water vapor were solved to evaluate nasal air‐conditioning performance.

To obtain more realistic epithelial surface temperature distributions, a previously validated wall model assuming 100% relative humidity at the epithelial surface was employed [20, 21, 22]. The wall temperature distribution was determined using a conjugate heat transfer framework, thereby eliminating the need to prescribe an arbitrary constant surface temperature. Accordingly, the water vapor mass fraction at the epithelial surface was calculated to maintain 100% relative humidity using an in‐house user‐defined function implemented in ANSYS Fluent R23.1. Seven prism layers were generated adjacent to the wall to adequately resolve near‐wall gradients, while polyhedral elements were used in the remaining computational domain.

Mesh independence was assessed using a representative ENS subject (ENS 3). Simulations were performed with approximately 5.3, 7.6, 9.8, and 11.9 million mesh elements. Variations in velocity, pressure, temperature, and water vapor mass fraction were evaluated at representative locations in the nasal chamber, nasal valve region, middle meatus, and nasopharynx. Increasing the mesh size from 9.8 to 11.9 million elements resulted in changes of no more than approximately 1.25% in the evaluated variables. Accordingly, a mesh containing approximately 9.8 million elements was considered to provide sufficient spatial resolution for the objectives of the present study. The same size function was subsequently applied to the remaining nasal cavity models, resulting in mesh counts ranging from 7.7 to 10.5 million elements. Approximately 6000–9000 iterations were required for each simulation to achieve convergence, with residuals decreasing below 1 × 10⁻⁵.

At the inlet, ambient conditions were set to 25°C and 35% relative humidity. Outlet pressure was adjusted to achieve a constant inspiratory flow rate of 250 mL/s, representative of quiet breathing. All variables, except static pressure at the outlet, were extrapolated from the interior according to the pressure boundary condition implemented in ANSYS Fluent R23.1.

From each converged simulation, a consistent set of CFD‐derived parameters was extracted to characterize nasal airflow and air‐conditioning: (1) the airflow velocity distribution on coronal cross‐sections and the meatus through which inspiratory flow predominantly passed; (2) bilateral nasal resistance, defined as the transnasal pressure drop between the nostrils and the nasopharynx divided by the volumetric flow rate; (3) air temperature; and (4) relative humidity at the choanae as indices of heating and humidification performance, evaluated as patient‐level values over both choanae; (5) wall shear stress along the mucosal surface, representing the air–mucosa frictional interaction; and (6) the SAVR, defined as the total mucosal surface area of both nasal airways divided by their combined air volume, evaluated over the airway extending from the nostrils to the choanae as in our previous studies, as an anatomical index of the residual mucosal surface available for air–mucosa interaction [22, 23]. Streamlines were additionally visualized to qualitatively assess overall flow organization.

Each airway was classified into one of three airflow distribution patterns according to the meatus carrying the predominant inspiratory flow, assessed from the velocity distribution across the nasal meatuses on coronal cross‐sections: inferior‐meatus–dominant (Group IT), middle‐meatus–dominant (Group MT), or a balanced distribution without a single dominant meatus (Group Mixed). Classification was performed by visual inspection of coronal velocity distributions, by consensus of the CFD investigator and the senior rhinologist (Y.N. and Y.J.J.). In ENS patients with unilateral symptoms, the symptomatic side was assessed. In ENS patients with bilateral symptoms, and in the NSD and normal cohorts in which a symptomatic side could not be defined, both sides were inspected and the phenotype was assigned from the overall pattern: Group IT when a conspicuous inferior‐meatus–dominant flow was present, Group MT when flow on both sides was skewed toward the middle meatus, and Group Mixed otherwise. This phenotypic classification was applied to the ENS and NSD cohorts and used for the subsequent comparison of physiological parameters and symptom severity between airflow phenotypes.

2.4. Statistical Analysis

Continuous variables were compared between the ENS and NSD cohorts and between the ENS and normal cohorts, as well as between airflow phenotype groups (Group MT vs. Group IT) within the ENS and NSD cohorts, using two‐sided Welch's t tests. Mean differences with 95% confidence intervals (CIs) are reported as effect estimates. Correlations among CFD‐derived parameters, as well as their associations with ENS6Q scores, were assessed using Pearson correlation analysis, with 95% CIs for correlation coefficients. The distribution of airflow patterns (Group IT, Group MT, and Group Mixed) across the three cohorts was compared using the Fisher–Freeman–Halton exact test, and pairwise differences in the proportion of Group MT were assessed using the Fisher exact test. A p < 0.05 was considered statistically significant. Given the small sample size and multiple comparisons, phenotype‐level comparisons were considered exploratory, and nominal p values are reported without multiplicity adjustment.

3. Results

3.1. Airflow Distributions

Among the 14 ENS subjects, airflow predominantly favored the inferior meatus on the symptomatic side in four cases (ENS 2, ENS 4, ENS 7, and ENS 8), classified as Group IT. In contrast, nine subjects (ENS 1, ENS 5, ENS 6, and ENS 9–ENS 14) demonstrated airflow skewed toward the middle meatus (Group MT). One subject (ENS 3) exhibited airflow broadly distributed across both the inferior and middle meatuses and was therefore classified as a mixed phenotype (Group Mixed).

Velocity contours in selected anterior coronal planes for representative ENS subjects from each airflow phenotype (ENS 4 for Group IT, ENS 10 for Group MT, and ENS 3 for Group Mixed) are shown in Figure 1. For each subject, the coronal plane was selected to best represent the airflow distribution within the anterior nasal cavity. Considerable intersubject variability in airflow distribution was observed, indicating that nasal airflow distributions vary substantially among ENS patients. Although most subjects had undergone inferior turbinate reduction, individual anatomical differences resulted in markedly different airflow phenotypes.

FIGURE 1.

Velocity contours in selected anterior coronal planes for representative ENS subjects, grouped according to airflow phenotype: (A) inferior‐meatus–dominant airflow (Group IT), (B) middle‐meatus–dominant airflow (Group MT), and (C) mixed inferior–middle meatus distribution (Group Mixed). The symptomatic side is indicated by red dotted circles.

For comparison, airflow distributions were also examined in an NSD cohort without prior surgery and a normal cohort. Among 10 NSD subjects (NSD 1–NSD 10) without ENS symptoms, airflow distributions were similarly diverse. Three subjects (NSD 2, NSD 3, and NSD 5) demonstrated inferior‐meatus–dominant airflow (Group IT), three subjects (NSD 6, NSD 8, and NSD 9) exhibited middle‐meatus–dominant airflow (Group MT), and the remaining four subjects (NSD 1, NSD 4, NSD 7, and NSD 10) showed mixed airflow distributions (Group Mixed). Representative NSD subjects for each phenotype (NSD 5 for Group IT, NSD 8 for Group MT, and NSD 4 for Group Mixed) are illustrated in Figure 2A. In the normal cohort, all subjects exhibited mixed airflow distributions (Group Mixed), indicating that airflow was distributed across the middle, common and inferior meatuses as shown in Figure 2B with three representative subjects.

FIGURE 2.

Velocity contours in selected anterior coronal planes for comparison cohorts. (A) Representative NSD subjects grouped according to airflow phenotype. (B) Representative normal subjects demonstrating mixed airflow distributions (Group Mixed).

Overall, airflow distributions differed across the ENS, NSD, and normal cohorts (Figure 3). In the ENS group, middle‐meatus–dominant airflow accounted for approximately 64% of subjects (9/14), whereas inferior‐meatus–dominant and mixed phenotypes accounted for 29% (4/14) and 7% (1/14), respectively. In contrast, the NSD cohort showed a more evenly distributed pattern, with 30% middle‐meatus–dominant (3/10), 30% inferior‐meatus–dominant (3/10), and 40% mixed airflow distributions (4/10). In the normal cohort, mixed airflow distributions were observed in all subjects (5/5). This difference was statistically significant (Fisher–Freeman–Halton exact test, p = 0.004); the proportion of Group MT was significantly higher in ENS than in normal subjects (p = 0.033) but not significantly different from the NSD cohort (p = 0.21). These findings indicate that airflow skewed toward either the middle or inferior meatus can occur in both ENS and non‐ENS conditions.

FIGURE 3.

Distribution of airflow patterns (Group IT, Group MT, and Group Mixed) across the ENS (n = 14), NSD (n = 10), and normal (n = 5) cohorts. The overall distribution differed across cohorts (Fisher–Freeman–Halton exact test, p = 0.004). Brackets indicate pairwise comparisons of the Group MT proportion (Fisher exact test). *p < 0.05.

To illustrate airflow–surface interaction, representative subjects with the highest ENS6Q scores in Group IT and Group MT were examined (Figure 4). Subject ENS 4 from Group IT (ENS6Q = 27) demonstrated substantial airflow through an enlarged inferior meatus on the symptomatic side. Despite this airflow distribution, wall shear stress along the nasal floor of the inferior meatus remained low. Similarly, the contralateral asymptomatic side also exhibited low wall shear stress in the same region. In the representative Group MT subject (ENS 12, ENS6Q = 25), airflow was strongly directed toward the middle meatus on the symptomatic side; however, wall shear stress along the inferior nasal floor also remained low, similar to that observed in the Group IT subject.

FIGURE 4.

Unilateral streamline and wall shear stress distributions in representative ENS subjects with high ENS6Q scores. (A) Group IT subject ENS 4 and (B) Group MT subject ENS 12. For each subject, streamlines through the right airway and wall shear stress contours in sagittal and axial views are shown. Despite different airflow skewness patterns toward the inferior or middle meatus, wall shear stress along the inferior nasal floor remained low in both subjects.

Several physiological variables were compared between Group MT and Group IT in the ENS and NSD cohorts (Table 1). In an exploratory comparison of ENS subjects, the mean ENS6Q score was higher in Group IT (23.0 ± 3.3) than in Group MT (18.0 ± 3.8) (mean difference, 5.0; 95% CI, 0.08–9.92; nominal p = 0.047). Notably, three of the four Group IT subjects (ENS 4, ENS 2, and ENS 7) exhibited three of the four highest ENS6Q scores. Other physiological parameters, including nasal resistance, relative humidity, air temperature, and SAVR, tended to be higher in Group IT, although these differences did not reach statistical significance. In the NSD cohort, Group IT also showed higher mean values for physiological parameters, but no statistically significant differences between airflow phenotypes were observed.

TABLE 1.

Comparison of physiological parameters between middle‐meatus–dominant (Group MT) and inferior‐meatus–dominant (Group IT) airflow phenotypes in ENS and NSD cohorts.

I. ENS subjects (n = 14)
Parameters	Group MT (n = 9)	Group IT (n = 4)	Group mixed (n = 1)	Mean difference (95% CI)	p (MT vs. IT)
Nasal resistance (Pa·s/mL)	0.0281 ± 0.0208	0.0365 ± 0.0308	0.0174	0.0084 (−0.0373 to 0.0541)	0.644
Relative humidity at choanae (%)	84.1 ± 8.4	86.2 ± 8.0	83.4	2.1 (−9.9 to 14.0)	0.688
Air temperature at choanae (°C)	31.3 ± 1.0	31.7 ± 0.9	31.2	0.3 (−1.0 to 1.7)	0.573
SAVR (mm−1)	0.624 ± 0.212	0.874 ± 0.248	0.696	0.250 (−0.115 to 0.615)	0.139
ENS6Q score	18.0 ± 3.8	23.0 ± 3.3	18	5.0 (0.08–9.92)	0.047

II. NSD subjects (n = 10)
Parameters	Group MT (n = 3)	Group IT (n = 3)	Group mixed (n = 4)	Mean difference (95% CI)	p (MT vs. IT)
Nasal resistance (Pa·s/mL)	0.0266 ± 0.0076	0.0389 ± 0.0125	0.0339 ± 0.0184	0.0123 (−0.0131 to 0.0377)	0.232
Relative humidity at choanae (%)	85.8 ± 2.2	89.8 ± 4.6	90.2 ± 4.4	4.0 (−5.7 to 13.7)	0.274
Air temperature at choanae (°C)	30.9 ± 0.4	31.6 ± 0.6	31.5 ± 0.4	0.7 (−0.5 to 1.9)	0.175
SAVR (mm−1)	0.639 ± 0.071	0.957 ± 0.174	0.856 ± 0.175	0.318 (−0.055 to 0.691)	0.071

Note: ENS6Q scores are additionally shown for ENS subjects. p values represent comparisons between Group MT and Group IT. Group Mixed was included for descriptive purposes only due to limited sample size. Mean differences (Group IT—Group MT) with 95% confidence intervals (CIs) were derived from Welch's t tests; phenotype‐level comparisons are exploratory and nominal p values are shown. Bold values indicate statistical significance (p < 0.05).

Abbreviations: ENS, empty nose syndrome; NSD, nasal septal deviation; SAVR, surface area‐to‐volume ratio.

When comparing the overall ENS and NSD cohorts, no significant differences were observed in nasal resistance, relative humidity at choanae, air temperature, or SAVR (Table 2I). In contrast, compared with normal subjects, ENS subjects showed significantly lower relative humidity (p < 0.001) and air temperature (p < 0.01) at the choanae, whereas nasal resistance and SAVR did not differ significantly (Table 2II). ENS subjects also showed lower mean SAVR than NSD and normal subjects.

TABLE 2.

Comparison of physiological parameters between ENS and NSD cohorts and between ENS and normal cohorts.

I. ENS vs. NSD subjects
Parameter	ENS (n = 14)	NSD (n = 10)	Mean difference (95% CI)	p
Nasal resistance (Pa·s/mL)	0.0297 ± 0.0227	0.0332 ± 0.0136	−0.0035 (−0.0190 to 0.0119)	0.639
Relative humidity at choanae (%)	84.6 ± 7.7	88.8 ± 4.1	−4.1 (−9.2 to 0.9)	0.103
Air temperature at choanae (°C)	31.4 ± 0.9	31.3 ± 0.5	0.1 (−0.5 to 0.7)	0.718
SAVR (mm−1)	0.700 ± 0.235	0.821 ± 0.189	−0.121 (−0.301 to 0.059)	0.176

II. ENS vs. normal subjects
Parameter	ENS (n = 14)	Normal (n = 5)	Mean difference (95% CI)	p
Nasal resistance (Pa·s/mL)	0.0297 ± 0.0227	0.0315 ± 0.0097	−0.0018 (−0.0175 to 0.0140)	0.815
Relative humidity at choanae (%)	84.6 ± 7.7	93.5 ± 1.5	−8.9 (−13.5 to −4.3)	< 0.001
Air temperature at choanae (°C)	31.4 ± 0.9	32.2 ± 0.2	−0.8 (−1.4 to −0.2)	< 0.01
SAVR (mm−1)	0.700 ± 0.235	0.861 ± 0.118	−0.161 (−0.336 to 0.014)	0.069

Note: Mean differences (ENS − comparison cohort) with 95% confidence intervals (CIs) were derived from Welch's t tests. Bold values indicate statistical significance (p < 0.05).

Abbreviations: ENS, empty nose syndrome; NSD, nasal septal deviation; SAVR, surface area‐to‐volume ratio.

To further examine the relationship between nasal cavity geometry and air‐conditioning performance, the association between relative humidity at the choanae and SAVR was analyzed (Figure 5). Pearson correlation analysis demonstrated a moderate positive correlation between relative humidity and SAVR in ENS subjects (r = 0.663; 95% CI, 0.20–0.88; p < 0.01) and a strong positive correlation in NSD subjects (r = 0.920; 95% CI, 0.69–0.98; p < 0.001). Normal subjects formed a high‐humidity reference cluster, with all subjects showing choanal relative humidity values above 90%. These findings suggest that preserved mucosal surface geometry, reflected by higher SAVR, may be associated with more effective humidification, whereas ENS subjects showed lower and more variable humidification performance.

FIGURE 5.

Relationship between relative humidity at choanae and surface area‐to‐volume ratio (SAVR) in ENS, NSD, and normal subjects. Pearson correlation coefficients (r) and corresponding p values are indicated in the figure.

3.2. Correlation of ENS6Q Score With CFD Variables

Pearson correlation analysis was performed to evaluate relationships between ENS6Q scores and CFD‐derived variables, including nasal resistance, relative humidity, air temperature, and SAVR (Figure 6). None of these variables showed statistically significant correlations with ENS6Q score. Specifically, bilateral nasal resistance showed a weak positive correlation (r = 0.089; 95% CI, −0.46 to 0.59; p = 0.762), whereas relative humidity and air temperature demonstrated negligible correlations (r = −0.043; 95% CI, −0.56 to 0.50; p = 0.884 and r = 0.007; 95% CI, −0.53 to 0.54; p = 0.982, respectively). SAVR also showed no significant correlation with ENS6Q score (r = 0.232; 95% CI, −0.34 to 0.68; p = 0.425).

FIGURE 6.

Pearson correlation coefficients between ENS6Q score and CFD‐derived physiological parameters in ENS subjects. Scatter plots showing correlations between ENS6Q score and (A) bilateral nasal resistance, (B) relative humidity at choanae, (C) air temperature at choanae, and (D) SAVR. Solid lines represent linear regression fits. Pearson correlation coefficients (r) and corresponding p values are indicated in each panel.

4. Discussion

The present study demonstrates that ENS is associated with heterogeneous airflow phenotypes rather than a single canonical aerodynamic pattern. While middle‐meatus–dominant airflow was most frequently observed in ENS subjects (9/14), inferior‐meatus–dominant (4/14) and mixed (1/14) patterns were also identified. In contrast, all normal reference subjects demonstrated mixed airflow distributions (Group Mixed), suggesting that balanced airflow across the middle, common, and inferior meatuses may represent an unaltered physiological airflow distribution.

Importantly, similar airflow phenotypes were observed in the non‐ENS NSD cohort, which included middle‐meatus–dominant, inferior‐meatus–dominant, and mixed distributions. This indicates that airflow skewed toward either the middle or inferior meatus is not unique to ENS and that airflow pattern alone cannot distinguish ENS from other nasal conditions. Nevertheless, the proportion of Group MT was higher in ENS subjects (~64%) than in normal subjects (0%, p = 0.033) and numerically higher than in the NSD cohort (~30%), although the latter difference was not statistically significant (p = 0.21). This suggests that anatomical modification, particularly following inferior turbinate reduction, may increase the tendency for airflow distribution toward the middle meatus, consistent with prior CFD studies reporting a paradoxical middle‐meatus airflow jet in ENS after inferior turbinate reduction [10, 11].

These findings align with recent work [15], which demonstrated through virtual turbinectomy simulations that aggressive turbinate removal does not inevitably generate ENS‐like aerodynamics. This is further supported by Malik et al., who compared symptomatic ENS patients with patients who had undergone equally aggressive inferior turbinate reduction but remained asymptomatic. Despite comparable surgical extent, inferior airway cross‐sectional area, and nasal resistance, only the symptomatic ENS group exhibited a paradoxical reduction of inferior‐meatus airflow—in contrast to the increase in inferior‐meatus airflow typically produced by turbinate reduction [24]—leading the authors to conclude that the degree of turbinate resection alone does not determine ENS [13]. Neither surgical history nor airflow pattern alone is therefore sufficient to define typical CFD results of ENS.

ENS subjects classified as Group IT showed higher ENS6Q scores than Group MT (mean difference, 5.0; 95% CI, 0.08–9.92; nominal p = 0.047). Although this was the only phenotype‐level association to reach nominal significance, it should be interpreted cautiously given the small subgroup sizes and does not by itself establish that airflow pattern determines symptom severity. Under the previously proposed hypothesis that middle‐meatus–dominant airflow lowers nasal‐floor shear stress [11], inferior‐meatus–dominant airflow would instead be expected to preserve, or even increase, it. However, in the most symptomatic Group IT subject (Figure 4A), the increased inferior‐meatus airflow was not accompanied by higher nasal‐floor wall shear stress, which remained low—similarly low to that in the most symptomatic Group MT subject (Figure 4B). This indicates that reduced air–mucosal surface interaction can arise in both inferior‐ and middle‐meatus–dominant configurations, so the dominant airflow pathway alone does not guarantee effective mucosal stimulation of the inferior nasal floor. These wall shear stress observations in two selected subjects are illustrative and do not establish a group‐level relationship with symptoms. Notably, Group IT also had numerically higher relative humidity, air temperature, and SAVR values than Group MT despite the greater symptom burden (Table 1), so the symptom difference is not explained by these global CFD‐derived measures.

Although SAVR tended to be lower in ENS than in NSD subjects, this difference was not statistically significant (p = 0.176), suggesting that global CFD‐derived parameters alone may be insufficient to distinguish ENS from NSD.

Another important observation was the relationship between nasal air‐conditioning capacity and morphological characteristics. Relative humidity, a parameter closely related to the sensation of nasal dryness, demonstrated a positive correlation with SAVR. ENS subjects also tended to show lower relative humidity than NSD subjects, and relative humidity was significantly lower in ENS subjects than in normal subjects. In addition, air temperature at the choanae was likewise significantly reduced in ENS relative to normal subjects (p < 0.01), consistent with reduced warming capacity accompanying the impaired humidification. These findings suggest that impaired nasal air‐conditioning in ENS may be related to reduced residual mucosal surface geometry after turbinate surgery. Clinically, this supports the rationale for preserving turbinate mucosal surface area during nasal surgery to maintain humidification capacity.

This dependence of air‐conditioning on nasal cavity geometry is consistent with our group's previous CFD analyses in other nasal pathologies. In nasal septal perforation, SAVR was the principal determinant of intranasal relative humidity and air temperature, with correlations that exceeded the influence of the perforation itself [22]. A comparable dependence of humidity and temperature on SAVR was observed across caudal and posterior septal deviation phenotypes [23]. The present observation that relative humidity was positively correlated with SAVR in ENS and NSD subjects, with normal subjects forming a high‐humidity reference cluster, reinforces a consistent association between nasal geometry and air‐conditioning capacity across distinct nasal conditions.

None of the global CFD‐derived parameters, including nasal resistance, relative humidity, air temperature, or SAVR, showed significant correlations with ENS6Q scores. This reinforces the multifactorial nature of ENS and suggests that symptom severity may arise from complex interactions among aerodynamics, mucosal physiology, neural sensory function, individual anatomical variability, and psychological, functional, or central perceptual factors rather than from a single measurable CFD metric. Instead, these parameters appear to reflect group‐level physiological differences (ENS vs. normal) more than individual symptom severity. Psychological, functional, and central perceptual factors were not measured in the present study, and our findings neither establish nor exclude their causal roles.

From a clinical perspective, CFD analysis should be viewed not as a diagnostic classifier but as a tool for understanding the patient‐specific physiological conditions that may contribute to symptom generation. Such patient‐specific characterization may also inform future research. For example, it could be used to evaluate candidate sites for submucosal implant placement to restore the warming and humidification of inhaled air and to dampen airflow turbulence in the abnormally widened cavity—an approach reported to yield long‐term symptom improvement in ENS [25]. Whether CFD can guide implant placement in practice remains untested.

Several limitations should be acknowledged. First, the sample size was modest, and the findings should be interpreted cautiously given the heterogeneity of the cohort. In addition, the comparison cohorts did not include postoperative patients without ENS. Differences from the reference cohorts therefore cannot be attributed with certainty to ENS itself rather than to postoperative anatomical change. Second, preoperative imaging in the ENS cohort was unavailable, so we could not determine whether a predisposing airflow configuration existed before surgery. Third, the simulations assumed rigid airway walls, standardized mucosal wall‐model conditions, and a single steady‐state inspiratory condition; consequently, dynamic mucosal vascular and thermoregulatory responses, the nasal cycle, expiratory airflow, and the effects of varying breathing efforts were not captured. The simulated temperature and humidity therefore primarily reflect the influence of airway geometry under these assumptions and do not capture patient‐specific mucosal function. Fourth, mucosal sensory function and trigeminal sensitivity were not directly measured or linked to the aerodynamic findings. Fifth, airflow phenotype classification relied on qualitative visual inspection, and interrater reliability was not formally assessed. Finally, this was a single‐center, retrospective study, which may limit the generalizability of the results. Larger, prospective, multicenter studies incorporating dynamic physiological modeling and direct sensory assessment are warranted to validate these findings.

5. Conclusion

Patient‐specific CFD modeling demonstrated heterogeneous airflow distribution patterns in ENS, in contrast to the balanced airflow of normal subjects. However, similar phenotypes also occurred in the non‐ENS NSD cohort, indicating that airflow pattern alone cannot distinguish ENS or explain symptom severity. Nasal air‐conditioning was impaired in ENS relative to normal subjects, with reduced temperature and a humidification deficit that may be associated with residual mucosal surface area, as reflected by SAVR. None of the evaluated CFD‐derived parameters differed significantly between ENS and NSD or correlated with symptom severity, although the small sample size precludes firm exclusion of clinically relevant associations. These findings support the concept that ENS is a complex multifactorial condition that warrants an individualized understanding of its pathophysiology beyond simple aerodynamic classification.

Funding

The authors have nothing to report.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data available on request from the authors.