Modeling alterations in sinonasal physiology after skull base surgery

Sinonasal Outcomes of the Combined Transseptal/Transnasal Approach with Unilateral Nasoseptal Rescue Flap in Endoscopic Endonasal Transsphenoidal Surgery: A Propensity Score Matching Analysis

Objectives  The authors applied surgical techniques acquired during the use of endoscopic combined transseptal/transnasal approach to reduce approach-related morbidity and improve sinonasal outcomes. Study Design  This is a retrospective cohort study of a prospectively collected database. Setting  The study setting involves a tertiary referral center. Participants  A total of 86 patients who underwent endoscopic endonasal transsphenoidal surgery for newly diagnosed pituitary adenomas from April 2018 to March 2021 were included. Patients treated via the combined transseptal/transnasal approach served as the study group ( n  = 18); those treated via the bilateral transnasal approach comprised the control group ( n  = 68). From the control group, propensity score matching (PSM) analysis was further performed to account for potential confounders and selection bias. Main Outcome Measures  Paired analysis was performed for pre- and 6-month-postoperative time points in study group, control group, and PSM control group. Olfactory function was evaluated by Connecticut Chemosensory Clinical Research Center (CCCRC) test, Cross-Cultural Smell Identification Test (CCSIT), and sinonasal outcomes were assessed by Sino-Nasal Outcome Test-22 (SNOT-22). Results  In the study group, CCCRC ( p  = 0.517) and CCSIT ( p  = 0.497) did not show any significant difference before and after surgery. There was some improvement in the symptom score of SNOT-22, but it was not statistically significant ( p  = 0.115). In the control group adjusted with PSM, a significant decrease in olfaction ( p  = 0.047) was observed using CCCRC. The CCSIT score was also decreased but not significant ( p  = 0.163). Also, there was no difference in the improvement of SNOT-22 ( p  = 0.781). Conclusion  Our new surgical method preserves olfactory function without compromising surgical outcomes.

Is nasal airflow disrupted after endoscopic skull base surgery? A short review

Even the most delicate endonasal surgery for skull base lesion causes changes in the nasal cavity, some of them permanent. Morphological changes in the nasal cavity and their consequences (changes in nasal airflow) are often studied by advanced numerical analysis called computational fluid dynamics. This review summarizes current knowledge of endoscopic transsphenoidal skull base surgery effects on nasal airflow. Several studies have shown that endoscopic skull base surgery changes nasal anatomy to the extent that nasal airflow changes significantly postoperatively. Removing any intranasal structure increases the cross-sectional area of the respective nasal meatus, leading to increased nasal airflow in this area while airflow in the narrower periphery decreases. Middle turbinate resection increases airflow in the middle meatus and reduces airflow in the superior and inferior meatus. Small posterior septectomy does not cause a significant change in nasal airflow. Nasal septum deviation is an important factor in airflow changes. Current studies describe nasal changes after rather extensive procedures (e.g., middle turbinectomy, ethmoidectomy) that are unnecessary in routine pituitary adenoma surgery. No studies have compared changes using pre- and postoperative scans of the same patients after actual surgery.

Evaluation of nasal function after endoscopic endonasal surgery for pituitary adenoma: a computational fluid dynamics study

OBJECTIVE

To analyze the effect of different endoscopic endonasal approaches (EEAs) on nasal airflow and heating and humidification in patients with pituitary adenoma (PA) by computational fluid dynamics (CFD).

METHODS

A three-dimensional pre-surgical model (Pre) of the nasal cavity and 6 that were post-EEA surgery were created from computed tomography scans as follows: small posterior septectomy (0.5 cm, sPS), middle posterior septectomy (1.5 cm, mPS), large posterior septectomy (2.5 cm, lPS), and sPS with middle turbinate resection (sPS-MTR), mPS-MTR, and lPS-MTR. Simulations were performed by CFD to compare the changes in different models.

RESULTS

The temperature in the nasal vestibule rose more rapidly than in other parts of the nasal cavities in all models. There were no apparent differences in temperature and humidity among the models in sections anterior to the middle turbinate head (C6 section). MTR significantly influenced airflow distribution between the bilateral nasal cavities and the different parts of the nasal cavity, while changes in temperature and humidity in each section were mainly affected by MTR. The temperature and humidity of the choana and nasopharynx of each postoperative model were significantly different from those of the preoperative model and the change in values significantly correlated with the surface-to-volume ratio (SVR) of the airway.

CONCLUSIONS

Changes due to the different nasal structures caused different effects on nasal function following the use of EEA surgery for the treatment of PA. CFD provided a new approach to assess nasal function, promising to provide patients with individualized preoperative functional assessment and surgical planning.

Analysis of nasal air conditioning in subjects with unilateral cleft lip nasal deformity

This study evaluated the impact of unilateral cleft lip nasal deformity (uCLND) on the ability of the nasal passages to warm and humidify inspired environmental air using computational fluid dynamics (CFD) modeling. Nasal air conditioning was simulated at resting inspiration in ten individuals with uCLND and seven individuals with normal anatomy. The overall heat and water transfer through nasal mucosa was significantly greater (p = 0.02 for both heat and moisture fluxes) on the non-cleft side than on the cleft side. Unilateral median and interquartile range (IQR) for heat flux (W/m²) was 190.3 (IQR 59.9) on the non-cleft side, 160.9 (IQR 105.0) on the cleft side, and 170.7 (IQR 87.8) for normal subjects. For moisture flux (mg/(s·m²), they were 357.4 (IQR 112.9), 298.7 (IQR 200.3) and 320.8 (IQR 173.0), respectively. Significant differences of SAHF50 between cleft side of uCLND and normal existed except for anterior region. Nevertheless, air conditioning ability in subjects with uCLND was generally comparable to that of normal subjects.

Orally Inhaled Drug Particle Transport in Computerized Models of Laryngotracheal Stenosis

OBJECTIVE

Adjuvant management for laryngotracheal stenosis (LTS) may involve inhaled corticosteroids, but metered dose inhalers are designed for pulmonary drug delivery. Comprehensive analyses of drug particle deposition efficiency for orally inhaled corticosteroids in the stenosis of LTS subjects are lacking.

STUDY DESIGN

Descriptive research.

SETTING

Academic medical center.

METHODS

Anatomically realistic 3-dimensional reconstructions of the upper airway were created from computed tomography images of 4 LTS subjects-2 subglottic stenosis and 2 tracheal stenosis subjects. Computational fluid dynamics modeling was used to simulate airflow and drug particle transport in each airway. Three inhalation pressures were simulated, 10 Pa, 25 Pa, and 40 Pa. Drug particle transport was simulated for 100 to 950 nanoparticles and 1 to 50 micron-particles. Particles were released into the airway to mimic varying inhaler conditions with and without a spacer chamber.

RESULTS

Based on smallest to largest cross-sectional area ratio, the laryngotracheal stenotic segment shrunk by 57% and 47%, respectively, for subglottic stenosis models and by 53% for both tracheal stenosis models. Airflow resistance at the stenotic segment was lower in subglottic stenosis models than in tracheal stenosis models: 0.001 to 0.011 Pa.s/mL vs 0.024 to 0.082 Pa.s/mL. Drug depositions for micron-particles and nanoparticles at stenosis were 0.06% to 2.48% and 0.10% to 2.60% for subglottic stenosis and tracheal stenosis models, respectively. Particle sizes with highest stenotic deposition were 6 to 20 µm for subglottic stenosis models and 1 to 10 µm for tracheal stenosis models.

CONCLUSION

This study suggests that at most, 2.60% of inhaled drug particles deposit at the stenosis. Particle size ranges with highest stenotic deposition may not represent typical sizes emitted by inhalers.

Superior turbinate management and olfactory outcome after endoscopic endonasal transsphenoidal surgery for pituitary adenoma: a propensity score-matched cohort study

Background

Surgical management of the superior turbinate (ST) is required to access the sella in endoscopic endonasal transsphenoidal surgery (EETS) for pituitary adenoma. Two common ST management techniques include partial resection of the ST (PRST) and intentional lateralization of the ST (ILST). Given the concentrated distribution of the olfactory nerve fibers on the medial surface of the ST, in this study we aimed to ascertain whether PRST worsens the objective olfactory outcome when compared with ILST.

Methods

A retrospective, propensity score‒matched cohort study was performed at a tertiary referral center. A total of 232 adult patients undergoing EETS for pituitary adenoma were analyzed. The threshold test (STT) and the 12‐item identification test (SIT‐12) from “Sniffin’ Sticks” were administered for separate nostrils preoperatively and 6 months postoperatively.

Results

Of 232 patients, 109 had right‐sided PRST and 123 received right‐sided ILST. Propensity score matching—controlling for olfactory‐related confounding factors, including gender, age, medical comorbidities, surgical technique, and preoperative olfaction—resulted in 74 matched pairs. When comparing the 6‐month postoperative olfactory performance of the right nostril, the STT score was significantly lower in the PRST group than the ILST group (p = 0.036, η² for effect size estimate = 0.030), but the SIT‐12 scores were similar in the 2 groups (p = 0.325). Overall, the olfactory outcomes for the right nostril did not qualitatively differ between the PRST and ILST groups (p = 0.401).

Conclusion

Despite its association with threshold impairment, PRST in EETS does not seem to carry an additional risk of postoperative olfactory dysfunction.

Computational technology for nasal cartilage-related clinical research and application

Surgeons need to understand the effects of the nasal cartilage on facial morphology, the function of both soft tissues and hard tissues and nasal function when performing nasal surgery. In nasal cartilage-related surgery, the main goals for clinical research should include clarification of surgical goals, rationalization of surgical methods, precision and personalization of surgical design and preparation and improved convenience of doctor-patient communication. Computational technology has become an effective way to achieve these goals. Advances in three-dimensional (3D) imaging technology will promote nasal cartilage-related applications, including research on computational modelling technology, computational simulation technology, virtual surgery planning and 3D printing technology. These technologies are destined to revolutionize nasal surgery further. In this review, we summarize the advantages, latest findings and application progress of various computational technologies used in clinical nasal cartilage-related work and research. The application prospects of each technique are also discussed.

Impact of posterior septum resection on nasal airflow pattern and warming function

Background: The functional significance of the posterior segment of the septum is not fully understood.Aims/Objectives: The purpose of this study was to evaluate the impact of posterior septum resection with varying extents on nasal airflow pattern and warming function, and to determine the appropriate resection extent on posterior septum without disturbing airflow pattern.Material and Methods: Three dimensional models were established from the nasal CT scan of a healthy female volunteer without nasal disease. The models included the normal, as well as 1 cm, 2 cm, and 3 cm extent of posterior septum resection by virtual surgery. Using computational fluid dynamics, the airflow characteristics such as average and peak airflow velocity, average wall shear stress, average temperature and turbulence kinetic energy were recorded and compared among the four models.Results: By comparison with the normal model, the virtual surgeries with varying extents of posterior septum resection (1 cm, 2 cm, and 3 cm), demonstrated little alteration on indices of the average and peak airflow velocity, average wall shear stress, airflow temperature and turbulence kinetic energy. However, a statistical difference of the abovementioned indices existed when compared between varying planes of the nasal cavity (p < .001), although this did not appear to be significantly influenced by the degree of posterior septectomy.Conclusions and significance: From an aerodynamics perspective, little disturbance on airflow pattern and warming function of nasal cavity could be detected with increasing resection of posterior septum (up to 3 cm). These results may imply that resection of the posterior septum (up to 3 cm) has little impact on nasal function.

Computational Analysis of the Mature Unilateral Cleft Lip Nasal Deformity on Nasal Patency

Background

Nasal airway obstruction (NAO) due to nasal anatomic deformities is known to be more common among cleft patients than the general population, yet information is lacking regarding severity and variability of cleft-associated nasal obstruction relative to other conditions causing NAO. This preliminary study compares differences in NAO experienced by unilateral cleft lip nasal deformity (uCLND) subjects with noncleft subjects experiencing NAO.

Methods

Computational modeling techniques based on patient-specific computed tomography images were used to quantify the nasal airway anatomy and airflow dynamics in 21 subjects: 5 healthy normal subjects; 8 noncleft NAO subjects; and 8 uCLND subjects. Outcomes reported include Nasal Obstruction Symptom Evaluation (NOSE) scores, cross-sectional area, and nasal resistance.

Results

uCLND subjects had significantly larger cross-sectional area differences between the left and right nasal cavities at multiple cross sections compared with normal and NAO subjects. Median and interquartile range (IQR) NOSE scores between NAO and uCLND were 75 (IQR = 22.5) and 67.5 (IQR = 30), respectively. Airflow partition difference between both cavities were: median = 9.4%, IQR = 10.9% (normal); median = 31.9%, IQR = 25.0% (NAO); and median = 29.9%, IQR = 44.1% (uCLND). Median nasal resistance difference between left and right nasal cavities were 0.01 pa.s/ml (IQR = 0.03 pa.s/ml) for normal, 0.09 pa.s/ml (IQR = 0.16 pa.s/ml) for NAO and 0.08 pa.s/ml (IQR = 0.25 pa.s/ml) for uCLND subjects.

Conclusions

uCLND subjects demonstrated significant asymmetry between both sides of the nasal cavity. Furthermore, there exists substantial disproportionality in flow partition difference and resistance difference between cleft and noncleft sides among uCLND subjects, suggesting that both sides may be dysfunctional.

A hierarchical stepwise approach to evaluate nasal patency after virtual surgery for nasal airway obstruction

BACKGROUND

Despite advances in medicine and expenditures associated in treatment of nasal airway obstruction, 25-50% of patients undergoing nasal surgeries complain of persistent obstructive symptoms. Our objective is to develop a "stepwise virtual surgery" method that optimizes surgical outcomes for treatment of nasal airway obstruction.

METHODS

Pre-surgery radiographic images of two subjects with nasal airway obstruction were imported into Mimics imaging software package for three-dimension reconstruction of the airway. A hierarchical stepwise approach was used to create seven virtual surgery nasal models comprising individual (inferior turbinectomy or septoplasty) procedures and combined inferior turbinectomy and septoplasty procedures via digital modifications of each subject's pre-surgery nasal model. To evaluate the effects of these procedures on nasal patency, computational fluid dynamics modeling was used to perform steady-state laminar inspiratory airflow and heat transfer simulations in every model, at resting breathing. Airflow-related variables were calculated for virtual surgery models and compared with dataset containing results of healthy subjects with no symptoms of nasal obstruction.

FINDINGS

For Subject 1, nasal models with virtual septoplasty only and virtual septoplasty plus inferior turbinectomy on less obstructed side were within the healthy reference thresholds on both sides of the nasal cavity and across all three computed variables. For Subject 2, virtual septoplasty plus inferior turbinectomy on less obstructed side model produced the best result.

INTERPRETATION

The hierarchical stepwise approach implemented in this preliminary report demonstrates computational fluid dynamics modeling ability to evaluate the efficiency of different surgical procedures for nasal obstruction in restoring nasal patency to normative level.

Computational fluid dynamics after endoscopic endonasal skull base surgery-possible empty nose syndrome in the context of middle turbinate resection

BACKGROUND

Empty nose syndrome (ENS) is a rare and debilitating disease with a controversial definition, etiology, and treatment. One puzzling fact is that patients who undergo an endoscopic endonasal approach (EEA) often have resection of multiple anatomic structures, yet seldom develop ENS. In this pilot study, we analyzed and compared the computational fluid dynamics (CFD) and symptoms among post-EEA patients, ENS patients, and healthy subjects.

METHODS

Computed tomography scans of 4 post-EEA patients were collected and analyzed using CFD techniques. Two patients had significant ENS symptoms based on results of the Empty Nose Syndrome 6-item Questionnaire (score >11), whereas the other 2 were asymptomatic. As a reference, their results were compared with previously published CFD results of 27 non-EEA ENS patients and 42 healthy controls.

RESULTS

Post-EEA patients with ENS symptoms had a similar nasal airflow pattern as non-EEA ENS patients. This pattern differed significantly from that of EEA patients without ENS symptoms and healthy controls. Overall, groups with ENS symptoms exhibited airflow dominant in the middle meatus region and a significantly lower percentage of airflow in the inferior turbinate region (EEA with ENS, 17.74 ± 4.00% vs EEA without ENS, 51.25 ± 3.33% [t test, p < 0.02]; non-EEA ENS, 25.8 ± 17.6%; healthy subjects, 36.5 ± 15.9%) as well as lower peak wall shear stress (EEA with ENS, 0.30 ± 0.13 Pa vs EEA without ENS, 0.61 ± 0.03 Pa [p = 0.003]; non-EEA ENS, 0.58 ± 0.24 Pa; healthy subjects, 1.18 ± 0.81 Pa).

CONCLUSION

These results suggest that turbinectomy and/or posterior septectomy may have a varying functional impact and that ENS symptoms go beyond anatomy and correlate with aerodynamic changes. The findings open the door for CFD as a potential objective diagnosis tool for ENS.

Olfactory function in patients after transsphenoidal surgery for pituitary adenomas-a short review

Olfaction is an important sensory input that obviously affects many daily activities. However, olfactory dysfunction (hyposmia and anosmia) leads to a pronounced decrease in quality of life. Surprisingly, little attention has been paid to olfactory changes after transsphenoidal surgery for pituitary tumors. In this review, we summarize current knowledge of the effects of transsphenoidal pituitary surgery on olfaction and compare different surgical techniques. Based on selected studies, the endoscopic approach, in comparison with the microscopic approach, seems to be superior in terms of preservation of olfactory function, although the quality of data from these studies is generally poor. The best results were observed when the endoscopic approach was used without harvesting of the nasoseptal flap.

Investigating the effects of laryngotracheal stenosis on upper airway aerodynamics

OBJECTIVE

Very little is known about the impact of laryngotracheal stenosis (LTS) on inspiratory airflow and resistance, especially in air hunger states. This study investigates the effect of LTS on airway resistance and volumetric flow across three different inspiratory pressures.

METHODS

Head-and-neck computed tomography scans of 11 subjects from 2010 to 2016 were collected. Three-dimensional reconstructions of the upper airway from the nostrils to carina, including the oral cavity, were created for one subject with a normal airway and for 10 patients with LTS. Airflow simulations were conducted using computational fluid dynamics modeling at three different inspiratory pressures (10, 25, 40 pascals [Pa]) for all subjects under two scenarios: 1) inspiration through nostrils only (MC), and 2) through both nostrils and mouth (MO).

RESULTS

Volumetric flows in the normal subject at the three inspiratory pressures were considerably higher (MC: 11.8-26.1 L/min; MO: 17.2-36.9 L/min) compared to those in LTS (MC: 2.86-6.75 L/min; MO: 4.11-9.00 L/min). Airway resistances in the normal subject were 0.051 to 0.092 pascal seconds per milliliter (Pa.s)/mL (MC) and 0.035-0.065 Pa.s/mL (MO), which were approximately tenfold lower than those of subjects with LTS: 0.39 to 0.89 Pa.s/mL (MC) and 0.45 to 0.84 Pa.s/mL (MO). Furthermore, subjects with glottic stenosis had the greatest resistance, whereas subjects with subglottic stenosis had the greatest variability in resistance. Subjects with tracheal stenosis had the lowest resistance.

CONCLUSION

This pilot study demonstrates that LTS increases resistance and decreases airflow. Mouth breathing significantly improved airflow and resistance but cannot completely compensate for the effects of stenosis. Furthermore, location of stenosis appears to modulate the effect of the stenosis on resistance differentially.

LEVEL OF EVIDENCE

NA. Laryngoscope, 128:E141-E149, 2018.

Effects of changes in nasal volume on voice in patients after endoscopic endonasal transsphenoidal surgery

OBJECTIVE

To investigate a potential correlation between changes in voice and changes in nasal volume.

METHODS

The endoscopic endonasal transsphenoidal approach (EETSA) was performed on 120 patients between February 2009 and August 2016 by using the bilateral modified nasoseptal rescue flaps method. All the patients were subjected to pre- and postoperative paranasal computed tomography (CT) and voice evaluations by using acoustic analyses, a nasometer to measure the nasalance, and determination of the voice handicap index (VHI). Paranasal CT and the medical image processing software were used to calculate changes in nasal cavity volume in three nasal sections.

RESULTS

Enlargement of the nasal cavity after surgery was evident in all three areas (anteronasal, p < 0.001; midnasal, p = 0.005; postnasal, p = 0.025). In addition, EETSA resulted in significantly higher mean nasalance scores for the oronasal passage (p < 0.001) and nasal passage (p < 0.001); more frequency perturbation (jitter) (p < 0.001) and amplitude perturbation (shimmer) (p < 0.001); and higher grade, roughness, breathiness, asthenia, and strain scores (p < 0.001), and VHI (p = 0.01). However, only changes in the nasal volume after EETSA correlated with postnasal hypernasality (r = 0.2; p = 0.029).

CONCLUSION

Although changes in nasal volume, voice, and speech may develop after EETSA, we found that changes in nasal volume were not correlated with changes in any voice-quality measure. However, the postnasal cavity was the region most dramatically affected by EETSA, and postnasal volume changes after surgery may be associated with hypernasal speech.

Endoscopic Superior Ethmoidal Approach for Anterior Cranial Base Resection: Tailoring the Approach for Maximum Exposure with Preservation of Nasal Structures

BACKGROUND

Traditional endoscopic anterior cranial base resection involves the total removal of the ethmoidal cells, including the middle and superior turbinates. This is associated with increased volume of the nasal cavity postoperatively, with increased crusting and permanent change of the nasal airflow. Here we provide a step-by-step description of the technique and evaluate the feasibility of the superior ethmoidal approach for anterior cranial base resection with maximum exposure of the anterior cranial base while keeping the middle turbinates, uncinate processes, and ostiomeatal complexes intact.

METHODS

Three fresh cadaveric heads were used for dissections. This technique was successfully performed in 2 consecutive cases of large olfactory groove meningiomas.

RESULTS

In all anatomic dissections, satisfactory exposure of the cranial base was achieved while keeping the middle turbinate, uncinate process, ethmoid bulla, and middle meatus intact bilaterally. Successful resection of 2 consecutive cases of olfactory groove meningioma was performed using this approach.

CONCLUSIONS

The endoscopic superior ethmoidal approach for anterior cranial base resection is a feasible and safe approach that maximizes preservation of the nasal structures while providing optimal access to the anterior skull base. It can be used in pathologies that involve the anterior cranial base and do not involve the nasal structures.

Characterizing airflow profile in the postoperative maxillary sinus by using computational fluid dynamics modeling: A pilot study

BACKGROUND

Maxillary antrostomy is commonly performed during endoscopic sinus surgery. Little is known about the association surrounding recalcitrant maxillary sinusitis, antrostomy size, and intranasal airflow changes. Furthermore, the interaction between sinus mucosa and airflow is poorly understood. This study used computational fluid dynamics (CFD) modeling to investigate postoperative airflow characteristics between diseased and nondiseased maxillary sinuses in subjects with recurrent disease.

METHODS

A retrospective review of patients from a tertiary-level academic rhinology practice was performed. Seven subjects with endoscopic evidence of postoperative maxillary sinus disease that presented as chronic unilateral crusting at least 1 year after bilateral maxillary antrostomies were selected. A three-dimensional model of each subject's sinonasal cavity was created from postoperative computed tomographies and used for CFD analysis.

RESULTS

Although the variables investigated between diseased and nondiseased sides were not statistically significant, the diseased side in six subjects had a smaller antrostomy, and five of these subjects had both reduced nasal unilateral airflow and increased unilateral nasal resistance on the diseased side. The ratio of posterior wall shear stress (WSS) of the maxillary sinus to the total WSS was higher on the diseased side in six subjects. Results also showed strong correlations between antrostomy and CFD variables on the diseased side than on the nondiseased side.

CONCLUSION

This pilot study showed that the majority of the simulated sinonasal models exhibited common characteristics on the side with persistent disease, such as smaller antrostomy, reduced nasal airflow, increased nasal resistance, and increased posterior WSS. Although statistical significance was not established, this study provided preliminary insight into variables to consider in a larger cohort study.

Characterizing human nasal airflow physiologic variables by nasal index

Although variations in nasal index (NI) have been reported to represent adaptation to climatic conditions, assessments of NI with airflow variables have not been rigorously investigated. This study uses computational fluid dynamics modeling to investigate the relationship between NI and airflow variables in 16 subjects with normal nasal anatomy. Airflow simulations were conducted under constant inspiratory pressure. Nasal resistance (NR) against NI showed weak association from nostrils to anterior inferior turbinate (R(2)=0.26) and nostril to choanae (R(2)=0.12). NI accounted for 38% and 41% of the respective variation in wall shear stress (WSS) and heat flux (HF) at the nasal vestibule, and 52% and 49% of variability in WSS and HF across the entire nose. HF and WSS had strong correlation with NI<80, and weakly correlated with NI>80; these differences in HF and WSS for NI<80 and NI>80 were not statistically significant. Results suggest strong relationship between NI and both WSS and HF but not NR, particularly in subjects with NI<80.