Computational fluid dynamics evaluation of posterior septectomy as a viable treatment option for large septal perforations

Abstract

Introduction

Numerous surgical techniques exist to treat nasal septal perforation (NSP). The surgical closure of large NSPs (>2cm) is still challenging. Posterior septectomy has been reported as a simple alternative to treat large NSP, yet its mechanisms for symptom relief are not clear, and if failed, its consequence cannot be easily reversed.

Methods

Ten NSP patients were recruited: five underwent posterior septectomy and five conventional flap or button repair. Computational Fluid dynamics (CFD) simulated the nasal aerodynamics based on CT scans. All patients had preoperative CT, however, only four had postoperative CT – two underwent posterior septectomy and the other two flap repair. We examined surgical outcomes and the nasal airflow features among the two treatment options.

Results

Both groups of patients had good outcomes based on chart review. Patients undergoing septectomy had significantly larger perforation size (2.32±0.87 vs 1.21±0.60 cm), higher flow rate across the perforation (47.8±28.6 vs. 18.3±12.2 ml/s) and higher wall shear stress (WSS) along the posterior perforation margin (1.39±0.52 vs. 1.15±0.58 Pa). The posterior WSS significantly correlate with cross-over flow velocity (r=0.77, p=0.009) and was reduced by almost 67% post-septectomy, and by 29% post-repair.

Conclusions

This is the first CFD analysis on NSP patient cohort. NSP resulted in flow disturbance and increased WSS that potentially lead to symptomatology. The removal of high stress points along the posterior margin may explain why posterior septectomy can be an effective treatment option. Aerodynamic abnormalities, in addition to perforation size and location, could serve as basis for future treatment decisions.

Introduction

Nasal septal perforation (NSP) affects up to 0.9% of the general population and may occur due to previous surgery, trauma, inflammatory disease, and nasal spray utilization ^{1, 2}. Symptoms accompanying septal perforation include whistling, crusting, epistaxis, nasal obstruction and postnasal drip. Conservative treatment options available to patients with NSPs include topical therapy, such as nasal irrigation and placement of emollients, and septal button placement (either in the office setting or in the operating suite). The surgical management of septal perforations is generally aimed at repairing the defect in order to mitigate the associated symptoms. The relevant strategies are non-standardized, but generally rely on various flaps and grafts. Repair of NSPs can be technically challenging, and the size of perforation is an important factor to consider. Large perforations (>2 cm) can have significantly higher failure rates (up to 30–70% in some studies) compared to small-to-moderate perforations (<2 cm) ^{2, 3}.

Similar to Beckmann et. al.³, we anecdotally noted a relative paucity of symptoms in patients who underwent posterior septectomy for extirpative procedures related to sinonasal and anterior cranial fossa neoplasms, especially when compared to patients with symptomatic septal perforations³. The symptomatology related to NSPs is thought to be at least partially due to alteration of the normal airflow ⁴ as well as dessication of the posterior margin of the perforation. Accordingly, studies that have attempted to remove the posterior margin of large NSP, via posterior septectomy, or that have pushed the posterior margin more posteriorly, have demonstrated symptomatic improvement ^{3, 5, 6}. To include this un-conventional and seemingly aggressive surgical approach in the treatment algorithm of NSP, more objective evidence is needed, especially to clarify the mechanism of how and why posterior septectomy may relieve NSP patients’ symptoms.

In addition, although alterations in normal airflow may be present in all or most NSPs, the factors that lead to symptomatology still remain unclear. CFD has been used to analyze the characteristics of various nasal sinus conditions or procedures in efforts to better understand the associated effects on nasal airflow ^7–9. However, there are relatively limited CFD data demonstrating the abnormal airflow patterns and the associated affects related to septal perforation in real patients – most prior published studies either use simulated perforations based on a healthy nose or have a sample size of one patient ^10–15.

The aim of this study is to describe the airflow patterns in a cohort of patients with symptomatic NSPs utilizing CFD analysis and to compare this with those of patients who have undergone conventional button or flap repairs and posterior septectomy.

Materials and Methods

Subjects

10 NSP patients were enrolled in the current study. The group consisted of 4 males and 6 females. Their ages ranged from 31–70 with a mean of 47.9, median of 50, and standard deviation of 12.7 years. After the diagnosis, one patient received button repair, four received flap repair surgeries, and five received posterior septectomy. The posterior septectomy is performed similarly as previously described by Beckman et. al. ³ All patients had relatively good outcomes based on chart review. One of the patient underwent posterior septectomy had failed a prior button repair. The study was approved by the institutional review boards of The Ohio State University.

CT scan and CFD model

All NSP patients had a preoperative CT scans, but only four had postoperative CT scans (2 posterior septectomy patients and 2 flap repair patients). These CT scans enabled the construction of CFD nasal airway models for each subject using methods described previously ^{9, 16, 17}, and only brief descriptions are provided here. The CT scans were imported into the commercial software AMIRA (Visualization Sciences Group, Hillsboro, OR, USA) to extract nasal cavity geometry. After necessary segmentation, smoothing, and correction for artifacts, a three-dimensional surface geometry of the nasal airway was generated. Then the commercial grid generator ICEM CFD (Ansys, Inc., Canonsburg, PA, USA) was applied to generate a computational mesh. A four-layer prism mesh was adopted at the boundary with a total height of ~0.2 mm near the mucosal surface to more accurately model the rapidly changing near-wall air velocity. A typical initial nasal cavity mesh with boundary layers contained between 1 million and 3 million finite elements. Then the initial meshes were refined by gradient adaptation and boundary adaptation until grid independence of the solutions was achieved. After the grid adaptation, the final nasal cavity mesh ranged from 1.5 million to 3.3 million hybrid finite elements.

Next, the solutions of the three-dimensional steady Navier-Stokes equations were obtained using the commercial software package FLUENT 16.2 (Ansys, Inc., Canonsburg, PA, USA), by applying a physiologically realistic pressure drop of 15 Pa between the nostrils and the nasal pharynx ¹⁸. This pressure drop of 15 Pa was chosen to simulate restfully breathing, a state that is most relevant to patients’ symptoms during routine daily life ¹⁹. Room air temperature of 20°C was set at the nostrils. Along the nasal mucosal walls, the usual no-slip velocity condition was applied, and the wall is assumed to be rigid and at constant temperature of 35°C. The numerical solutions of the continuity, momentum and energy equations were determined using the finite-volume method. A second-order upwind scheme was used for discretization. The SIMPLEC algorithm was used to link pressure and velocity. The discretized equations were then solved sequentially using a segregated solver. Convergence was obtained when the scaled residuals of continuity and momentum quantities were less than 10⁻⁵. The convergence residual of the energy equation was set as 10⁻⁸.

Results

Figure 1 summarizes the normalized location and size of NSPs in sagittal view for (a) 5 patients who underwent flap/button repair and (b) 5 patients who underwent posterior septectomy. Since most NSPs are not perfect circular shape, the size of the NSPs was defined by their largest diameter. Patients who underwent posterior septectomy had significantly larger size NSP diameter (2.32±0.87 cm) and more posterior located NSP than those who underwent repair (size 1.21±0.60 cm). Similarly, patients who underwent posterior septectomy had larger NSP cross sectional area than those undergoing repair (2.6±1.8 cm² vs. 0.8±0.6 cm²).

Figure 1.

Normalized location of the septal perforation in sagittal view for (a) 5 patients underwent flap/button repair and (b) 5 patients underwent posterior septectomy. The size of NSP were defined as the largest the diameter across the hole, and is color coded.

For all patients, the NSP resulted in a shunting of airflow through the perforation. Figure 2(a), demonstrates the shunting of inspired air from the left nostril through the NSP and into the right nasal cavity. The cross over flow can have a very high velocity (contour plot of the NSP in Figure 2(b)). For this particular case, the majority of the airflow crosses from left to right, as illustrated by the velocity component perpendicular to the NSP plane (X-velocity or crossing velocity) shown in Figure 2(c). To better analyze the impacts of this crossing airflow on nasal mucosa, the perforation site is isolated, and both wall shear stress (WSS) and heat flux at the perforation margins are quantified. Figure 2(d) and (e) showed highly elevated WSS and heat flux along the NSP margins, especially the posterior margins. To provide further detailed quantifications, we divided the NSP margins into four zones: anterior, posterior, cranial and caudal, as labeled in Figure 2(f).

Figure 2.

Representative views of the impacts of NSP on nasal airflow pattern (Patient 9). (a) Airflow streamlines of the left side cross the NSP to the right side of the nasal cavity. (b) Airflow velocity magnitude across the NSP. (c) Airflow velocity component perpendicular to the plane of the NSP (x-velocity or crossing velocity). The dark red color indicates the majority of crossing flow is from the left side cavity to the right-side along the -x direction. (d) Wall shear stress (WSS) distribution along the NSP margins that were highest posteriorly. (e) Heat flux distribution along the NSP margins. (f) NSP margins are divided into four regions: anterior, posterior, cranial and caudal.

Figure 3 demonstrates a quantitative comparison of the two groups of patients prior to surgery. Patients who underwent posterior septectomy had lower nasal resistance (233.6±61.1 ml/s@15 Pa) than those who received repair (268.0±99.1 ml/s@15 Pa). On the other hand, patients who underwent posterior septectomy had almost 2.6 fold higher volume flow rates crossing the NSP (47.8±28.6 ml/s) than those who underwent perforation repair (18.3±12.2 ml/s), Figure 3(b). For both groups, the posterior NSP margin consistently had the highest WSS and heat flux than the other regions (Figure 3(c)). Despite the 2.6 times increase in cross-over flow rate, however, patients who underwent posterior septectomy had only slightly higher WSS along the posterior NSP margin (1.39±0.52 Pa), compared to that of the repair group (1.15±0.58 Pa). Both WSS and heat flux along the posterior perforation margin significantly correlate with crossing velocity or x-velocity (for WSS r=0.77, p=0.009; for heat flux r=0.70, p=0.025), as shown in Figure 3(e–f)). No correlation was found between WSS or heat flux to NSP size, cross over volume flow rate, or any other variables in this study.

Figure 3.

Comparison of (a) the total nasal flow rate, (b) the volume flow rate through NSP, (c) regional wall shear stress (WSS) and (d) regional heat flux along NSP margins between patients who underwent flap/button repair (n=5) and patients who underwent posterior septectomy (n=5). Patients who underwent posterior septectomy had higher total nasal flow rate and significantly (almost 2.6 fold) higher volume flow rates crossing the NSP than those who underwent repair. For both groups, the posterior NSP margin consistently had the highest WSS and heat flux. Regression analysis indicated that both WSS (e) and heat flux (f) at the posterior region are significantly correlated with the crossing velocity (X-Velocity): for WSS r=0.77, p=0.009; for Heat flux r=0.70, p=0.025.

Velocity contours and streamline patterns within the NSPs (Figure 4) revealed that the velocity peak did not always appear in the center-to-posterior regions as suggested by previous studies ^{14, 15}. For instance, the overall velocity magnitude was highest in the dorsal region for patient 5. Streamline patterns of the airflow in the NSP planes showed a stationary vortex in six patients but not in other four patients. Neither peak velocity locations nor the presence of vortices within the NSP showed a correlation with the size of the NSP nor with flow rate/WSS/heat flux.

Figure 4.

The velocity contours (1^st and 3^rd columns) and streamline pattern (2^nd and 4^th columns) in the NSP for all patients involved in the current study. Patients 1–5 underwent flap/button repair and Patients 6–10 underwent posterior septectomy. Significant variations in distribution of airflow peaks and vortex were observed, with no relationship to the size or location of the NSP found.

Figure 5 shows the preoperative versus postoperative nasal airflow patterns for patient 6, who underwent posterior septectomy. The coronal velocity contour plots (2^nd row) revealed that the airflow distribution in the preoperative airways demonstrated several flow peaks around middle-to-inferior turbinate regions. However, postoperatively, the nasal airflow patterns were more evenly distributed; the WSS (3^rd row) was also significantly reduced. As demonstrated in the 4^th row of the figure, posterior septectomy removed the high WSS peak at the posterior region.

Figure 5.

Patient 6 underwent posterior septectomy in 2016. The preoperative CT scan showed a large perforation (NSP size =2.0cm). The postoperative CT demonstrated removal of nasal septum. CFD analysis showed the removal of nasal septum reduced the peak airflow velocity (dark red) around the middle and inferior turbinates (2^nd row) and reduced the WSS (3^rd row), especially the peak WSS at the posterior margins of the NSP (4^th row). To better visualize the model, the perforation site has been isolated and color coded, with the remainder of the nasal cavity rendered transparent.

Figure 6 summarizes the surgical outcomes of both the flap repair (n=2) and posterior septectomy (n=2) groups. Both surgical procedures reduced the peak WSS to a similar range (from 0.78 Pa to 0.55 Pa for the repair group vs. from 1.66 Pa to 0.54 Pa for the septectomy group). Similarly, the heat flux peak also decreased after the procedures for both groups (Repair: from 3587 W/m² to 2749 W/m²; Septectomy: from 2947 W/m² to 2260 W/m²). Finally, posterior septectomy significantly reduced the nasal resistance compared with flap/button repair (Repair: from 0.084 Pa·s/ml to 0.062 Pa·s/ml; Septectomy: from 0.084 Pa·s/ml to 0.037 Pa·s/ml).

Figure 6.

Comparison of the surgical outcomes between patients underwent flap repair (n=2) and patients underwent posterior septectomy (n=2). Both surgical procedures significantly reduced (a) wall shear stress (WSS), (b) heat flux near in NPS margins*, as well as (c) bilateral nasal resistance (CFD simulated). *After repair or septectomy, the well-defined perforation margin is gone, thus postoperative versus preoperative comparisons were based on WSS and heat flux peaks roughly within the previous NSP margins.

Discussion

In the current study, we provide the first quantitative CFD analysis in a cohort of NSP patients and examine the aerodynamic factors that may contribute to their symptoms and treatment outcomes. First, our results demonstrate that all symptomatic NSPs result in shunting of nasal airflow across the perforation with resultant higher WSS and heat flux along the margins of the perforation. WSS, defined as the shear force of airflow exerted onto a unit area of nasal mucosa, is an indicator for airflow-mucosa interactions. This WSS was greater along the posterior margin of all perforations, regardless of size. This is consistent not only with previous published CFD data, but also with clinical observations whereby symptomatic patients tend to experience more crusting and bleeding along the posterior margin of the NSP.

In previous CFD studies, Cannon et. al. ¹¹ evaluated the characteristics of 1 and 2 cm anterior, superior, and posterior circular septal perforations virtually created on a healthy nose. Their analysis demonstrated shunting of air across the perforation, which increases with perforation size in more anterior locations, but decreases as a function size at more posterior locations. The maximum WSSs were consistently located at the posterior margin of the perforation. Lindemann et. al. ¹⁴ also evaluated the characteristics of three simulated septal perforations (anterior caudal, anterior cranial, posterior caudal) virtual created on a healthy 50 year old male. Their simulation showed that a perforation in the anterior part of the septum with high flow velocities leads to increased turbulence, resulting in alteration of air distribution in the posterior nasal cavity. Similar to Cannon et. al., the highest cross-perforation velocity was found in the anterior-caudal perforation. The limitation of these and other prior CFD studies ^{11, 14} is that they were based on virtual perforations of healthy controls who may or may not have any symptoms.

Our results, based on a cohort of symptomatic patients with NSPs, confirmed that larger NSPs among the septectomy group resulted in a relatively high cross over flow rate (~2.6 times that of the repair group). WSS and heat flux was greater along the posterior margin all perforations, regardless of size. However, the peak values of WSS and heat flux did not correlate with cross-over flow rate nor NSP size. Rather, they correlated with cross over airflow velocity. This may explain why larger NSPs may sometimes be less symptomatic than smaller NSPs. Due to the larger cross sectional area of large NSPs, the volume of airflow rate across the perforation (cross-over flow rate) may be high, but spreading over area, the crossing velocity may be lower than that of smaller NSPs. However, not all large NSPs are symptom free, as evident in our septectomy patient group. This shows a potential limitation of only using size in treatment decisions.

The heat flux peak along the posterior margin also correlates with cross over velocity, but the septectomy group consistently had lower heat flux than repair group (Figure 3(d)). We believe that this is mainly due to the fact that patients who underwent posterior septectomy generally had NSPs located more posteriorly than that of the repair group. Air contacting the mucosa more posteriorly in the nose has already been warmed by the anterior mucosa along the flow path. This is a potential reason why posterior NSPs anecdotally tend to be less symptomatic.

Finally, our results provide the first evidence that both NSP repair and septectomy can significantly reduce the posterior margin WSS and heat flux, as illustrated in Figure 5 and Figure 6. The peak WSS reduced by almost 67% in the two post-septectomy patients and by 29% in the two post-repair patients. This may provide a mechanism by which posterior septectomy reduces symptoms associated with NSP. Essentially, posterior septectomy eliminates the focal point of airflow stress point along the posterior margin of the NSP.

Although technically simple, posterior septectomy is still recommended only when other options have failed, are impractical or impossible. Topical therapy involving nasal irrigation and emollients is the first step for symptomatic patients with NSPs. Subsequent therapy is currently chosen on an individual basis, accounting for the size and location of the perforation, the availability of tissue for reconstruction, previous surgery (including attempts at reconstruction of the NSP), mucosal integrity, previous radiation, and comorbid disease (sarcoidosis, granulomatosis with polyangiitis). Generally, we attempt repair of NSPs that are 2 cm or smaller when sufficient tissue is available. Alternatively, placement of a septal button may be considered. Only when the perforation is larger than 2 cm or when previous therapy (surgical or radiation) or comorbid disease (vasculitis, granulomatous disease, autoimmune disease) precludes a reasonable chance at successful repair, posterior septectomy is considered. Based on the results of this study and hopefully future large scale studies, we envision that objective findings from CFD simulations of NSP and virtual repair/septectomy may serve as additional basis for future personalized treatment decision.

Limitations

Some limitations to this study should be acknowledged. First of all, this is a retrospective study with limited sample size and with only four of ten patients receiving postoperative CT scans. Postoperative CT scans are often not clinically necessary for septal perf cases and rarely performed. To mitigate this limitation, future studies will be designed to incorporate postoperative CT scans for research purposes. The 2 groups (repair vs septectomy) are also rather different in their pre-existing conditions. But with sample size n=10 and 4 with post-op models, this is already the largest CFD cohort study on NSP that significantly advanced our knowledge on this topic. All previously CFD studies on NSP are case reports with n =1, or are based on virtual septal perforations created in healthy controls. In this study, we demonstrated the feasibility of the use of CFD to investigate the aerodynamic contribution to NSP symptomatology and the outcomes of a non-conventional treatment option for NSP. We hope that the preliminary results from this study will stimulate future scientific discussion and breed future large scale prospective studies. Another limitation of the study is that all enrolled NSP patients are symptomatic, thus we have no power to determine whether the variables we explored, eg. airflow rate shunting through the perforation, posterior margin WSS, heat flux, etc., indeed causes the symptoms. Therefore, for future investigation, we are actively recruiting an asymptomatic NSP cohort as well as patients who have undergone posterior septecotmy as a component of skull base tumor removal.

Conclusions

In this study, we used CFD technology to investigate ten patients with NSP (five underwent flap/button repair and five underwent posterior septectomy). The preoperative data demonstrate high WSS and heat flux at the posterior perforation margin that have a strong correlation with the x-velocity component perpendicular to the NSP plane, which may result in symptomology. Our results show that both flap repair and posterior septectomy can effectively reduce the WSS and heat flux along posterior margin. This provides the first CFD evidence of posterior septectomy as a viable treatment option for large NSPs where surgical closure had a low likelihood of success.